Search Results

Search found 511 results on 21 pages for 'benchmark'.

Page 19/21 | < Previous Page | 15 16 17 18 19 20 21  | Next Page >

• Win7 x64 unresponsive for a minute or so. HD failing?

  - by Gaia

  On a fully updated Win7 x64, every so often the system stalls for a minute or so. This has been going on for a couple months now. By stalling I mean the mouse responds and I can move windows around, but any window, any program, that is open becomes whiteish when I select it AND any new programs will not open. It doesn't matter what kind of program it is. When the stall stops all clicks I made (open new programs for example) take effect. Nothing shows up consistently (as in every time this happens) in the event log. Today though I was able to find something, but it doesn't reveal much other than the "system was unresponsive". It's a 7009 for "A timeout was reached (30000 milliseconds) while waiting for the Windows Error Reporting Service service to connect." It doesn't matter if I have any USB devices plug-in or not. I've ran Microsoft Security Essentials and Malwarebytes. While the machine is unresponsive, I've noticed that Drive D (the other partition on the single internal HD in this laptop) is displayed like this in explorer. This never occurs with Drive C or any other drive on the machine. . SMART report for the physical drive: Read benchmark by HD Tune 5 Pro, probably the most telling piece of the puzzle. Isn't this alone enough to see there is a problem with the drive, regardless of whether the unresponsiveness is caused by such purported problem? Here is a short hardware report: Computer: LENOVO ThinkPad T520 CPU: Intel Core i5-2520M (Sandy Bridge-MB SV, J1) 2500 MHz (25.00x100.0) @ 797 MHz (8.00x99.7) Motherboard: LENOVO 423946U Chipset: Intel QM67 (Cougar Point) [B3] Memory: 8192 MBytes @ 664 MHz, 9.0-9-9-24 - 4096 MB PC10600 DDR3 SDRAM - Samsung M471B5273CH0-CH9 - 4096 MB PC10600 DDR3 SDRAM - Patriot Memory (PDP Systems) PSD34G13332S Graphics: Intel Sandy Bridge-MB GT2+ - Integrated Graphics Controller [D2/J1/Q0] [Lenovo] Intel HD Graphics 3000 (Sandy Bridge GT2+), 3937912 KB Drive: ST320LT007, 312.6 GB, Serial ATA 3Gb/s Sound: Intel Cougar Point PCH - High Definition Audio Controller [B2] Network: Intel 82579LM (Lewisville) Gigabit Ethernet Controller Network: Intel Centrino Advanced-N 6205 AGN 2x2 HMC OS: Microsoft Windows 7 Professional (x64) Build 7601 The drive less than 1 year old. Do I have a defective drive? Seagate Tools diag says there is nothing wrong with the drive... UPDATE: I noticed that the windows error reporting service entered the running state then the stopped state and the space between the two events was exactly 2 minutes. Which error it was trying to report I don't know. I check the "Reliability Monitor" and it shows no errors to be reported. I've disabled the windows error reporting service to see if the problem stops.

  Read the article

• APC File Cache not working but user cache is fine

  - by danishgoel

  I have just got a VPS (with cPanel/WHM) to test what gains i could get in my application with using apc file cache AND user cache. So firstly I got the PHP 5.3 compiled in as a DSO (apache module). Then installed APC via PECL through SSH. (First I tried with WHM Module installer, it also had the same problem, so I tried it via ssh) All seemed fine and phpinfo showed apc loaded and enabled. Then I checked with apc.php. All seemed OK But as I started testing my php application, the stats in apc for File Cache Information state: Cached Files 0 ( 0.0 Bytes) Hits 1 Misses 0 Request Rate (hits, misses) 0.00 cache requests/second Hit Rate 0.00 cache requests/second Miss Rate 0.00 cache requests/second Insert Rate 0.00 cache requests/second Cache full count 0 Which meant no PHP files were being cached, even though I had browsed through over 10 PHP files having multiple includes. So there must have been some Cached Files. But the user cache is functioning fine. User Cache Information Cached Variables 0 ( 0.0 Bytes) Hits 1000 Misses 1000 Request Rate (hits, misses) 0.84 cache requests/second Hit Rate 0.42 cache requests/second Miss Rate 0.42 cache requests/second Insert Rate 0.84 cache requests/second Cache full count 0 Its actually from an APC caching test script which tries to retrieve and store 1000 entries and gives me the times. A sort of simple benchmark. Can anyone help me here. Even though apc.cache_by_default = 1, no php files are being cached. This is my apc config Runtime Settings apc.cache_by_default 1 apc.canonicalize 1 apc.coredump_unmap 0 apc.enable_cli 0 apc.enabled 1 apc.file_md5 0 apc.file_update_protection 2 apc.filters apc.gc_ttl 3600 apc.include_once_override 0 apc.lazy_classes 0 apc.lazy_functions 0 apc.max_file_size 1M apc.mmap_file_mask apc.num_files_hint 1000 apc.preload_path apc.report_autofilter 0 apc.rfc1867 0 apc.rfc1867_freq 0 apc.rfc1867_name APC_UPLOAD_PROGRESS apc.rfc1867_prefix upload_ apc.rfc1867_ttl 3600 apc.serializer default apc.shm_segments 1 apc.shm_size 32M apc.slam_defense 1 apc.stat 1 apc.stat_ctime 0 apc.ttl 0 apc.use_request_time 1 apc.user_entries_hint 4096 apc.user_ttl 0 apc.write_lock 1 Also most php files are under 20KB, thus, apc.max_file_size = 1M is not the cause. I have also tried using 'apc_compile_file ' to force some files into opcode cache with no luck. I have also re-installed APC with Debugging enabled, but nothing shows in the error_log I have also tried setting mmap_file_mask to /dev/zero and /tmp/apc.xxxxxx, i have also set /tmp permissions to 777 to no avail Any clue anyone. Update: I have tried following things and none cause APC file cache to populate 1. set apc.enable_cli = 1 AND run a script from cli 2. Set apc.max_file_size = 5M (just in case) 3. switched php handler from dso to FastCGI in WHM (then switched it back to dso as it did not solve the problem) 4. Even tried restarting the container

  Read the article

• Lag spikes at full CPU usage, maybe video card

  - by Roberts

  I am posting this thread in hurry so few things may be missed (I will update tomorrow). My PC specs: Motherboard Name - Gigabyte GA-945PL-S3 CPU Type - DualCore Intel Core 2 Duo E4300, 1800 MHz (9 x 200) OS - Microsoft Windows 7 Ultimate OS Kernel Type - 32-bit OS Version - 6.1.7601 I bougth a new video card one month ago. GeForce 210. I didn't have any problems. I wanted to overclock it, in other words: "Play with it". So I installed Gigabyte EasyBoost from CD and overclocked the GPU 590 + 110 mhz, memory to max to 960mhz from 800mhz. Benchmarks showed a little bit bigger score. Then I overclocked shader clock from 1405 to [..] (don't remeber really). So I was playing Modern Warfare 2 when off sudden computer froze when I wanted to select team, I was afk before that. I had to reset CMOS. After that I had problems with Skype: unread messages and no sound. Then I figured it out that when ever I open EasyBoost - Skype starts to glitch again. Now I use EVGA Precission X. Now after a month, I cleaned computer and closed the case, it was open all the time. I started to overclock GPU clock only (just a bit) because there was no problems that would stop me. So sometimes on heavy CPU load graphics starts to lag. Dragging a window is painful to watch too. Sometimes the screen freezes for 5 to 10 seconds (I can see that hard disk activity is maximal). You may say that CPU fault it is, isn't it? But sometimes lag spikes starts randomly when CPU load is at maximum. All 3 benchmark softwares (PerformanceTest, NovaBench and MSI Kombustor) shows that performance of my video card has dropped about 25%. BUT! CPU score is lower too. I ignored these problems but when I refreshed Windows Experience Index I was shocked. Month before (in latvian language but not so hard to understand): Now (upgraded RAM): This happened when I tried to capture Minecraft with Fraps on underclocked GPU to 580mhz (def: 590mhz): All drivers are up to date. Average CPU temperature from 55°C to 75°C (at 70°C sometimes starts these lag spikes). Video card's tempratures are from 45°C to 60°C (very hard to reach 60°C). So my hope is that the video card is fine, cause this card is very new and I want to upgrade CPU anyways. Aplogies for my mistakes in vocabulary (I am trying to type this as fast I can).

  Read the article

• Tips for maximizing Nginx requests/sec?

  - by linkedlinked

  I'm building an analytics package, and project requirements state that I need to support 1 billion hits per day. Yep, "billion". In other words, no less than 12,000 hits per second sustained, and preferably some room to burst. I know I'll need multiple servers for this, but I'm trying to get maximum performance out of each node before "throwing more hardware at it". Right now, I have the hits-tracking portion completed, and well optimized. I pretty much just save the requests straight into Redis (for later processing with Hadoop). The application is Python/Django with a gunicorn for the gateway. My 2GB Ubuntu 10.04 Rackspace server (not a production machine) can serve about 1200 static files per second (benchmarked using Apache AB against a single static asset). To compare, if I swap out the static file link with my tracking link, I still get about 600 requests per second -- I think this means my tracker is well optimized, because it's only a factor of 2 slower than serving static assets. However, when I benchmark with millions of hits, I notice a few things -- No disk usage -- this is expected, because I've turned off all Nginx logs, and my custom code doesn't do anything but save the request details into Redis. Non-constant memory usage -- Presumably due to Redis' memory managing, my memory usage will gradually climb up and then drop back down, but it's never once been my bottleneck. System load hovers around 2-4, the system is still responsive during even my heaviest benchmarks, and I can still manually view http://mysite.com/tracking/pixel with little visible delay while my (other) server performs 600 requests per second. If I run a short test, say 50,000 hits (takes about 2m), I get a steady, reliable 600 requests per second. If I run a longer test (tried up to 3.5m so far), my r/s degrades to about 250. My questions -- a. Does it look like I'm maxing out this server yet? Is 1,200/s static files nginx performance comparable to what others have experienced? b. Are there common nginx tunings for such high-volume applications? I have worker threads set to 64, and gunicorn worker threads set to 8, but tweaking these values doesn't seem to help or harm me much. c. Are there any linux-level settings that could be limiting my incoming connections? d. What could cause my performance to degrade to 250r/s on long-running tests? Again, the memory is not maxing out during these tests, and HDD use is nil. Thanks in advance, all :)

  Read the article

• Looking For iPhone 4S Alternatives? Here Are 3 Smartphones You Should Consider

  - by Gopinath

  If you going to buy iPhone 4S on a two year contract in USA, Europe or Australia you may not find it expensive. But if you are planning to buy it in any other parts of the world, you will definitely feel the heat of ridiculous iPhone 4S price. In India iPhone 4S costs approximately costs $1000 which is 30% more than the price tag of an unlocked iPhone sold in USA. Personally I love iPhones as there is no match for the user experience provided by Apple as well as the wide range of really meaning applications available for iPhone. But it breaks heart to spend $1000 for a phone and I’m forced to look at alternates available in the market. Here are the four iPhone 4S alternates available in almost all the countries where we can buy iPhone 4S Google Galaxy Nexus The Galaxy Nexus is Google’s own Android smartphone manufactured by Samsung and sold under the brand name of Google Nexus. Galaxy Nexus is the pure Android phone available in the market without any bloat software or custom user interfaces like other Androids available in the market. Galaxy Nexus is also the first Android phone to be shipped with the latest version of Android OS, Ice Cream Sandwich. This phone is the benchmark for the rest of Android phones that are going to enter the market soon. In the words of Google this smartphone is called as “Galaxy Nexus: Simple. Beautiful. Beyond Smart.”.  BGR review summarizes the phone as This is almost comical at this point, but the Samsung Galaxy Nexus is my favourite Android device in the world. Easily replacing the HTC Rezound, the Motorola DROID RAZR, and Samsung Galaxy S II, the Galaxy Nexus champions in a brand new version of Android that pushes itself further than almost any other mobile OS in the industry. Samsung Galaxy S II The one single company that is able to sell more smartphones than Apple is Samsung. Samsung recently displaced Apple from the top smartphone seller spot and occupied it with loads of pride. Samsung’s Galaxy S II fits as one the best alternatives to Apple’s iPhone 4S with it’s beautiful design and remarkable performance. Engadget summarizes Samsung Galaxy S2 review as It’s the best Android smartphone yet, but more importantly, it might well be the best smartphone, period. Of course, a 4.3-inch screen size won’t suit everyone, no matter how stupendously thin the device that carries it may be, and we also can’t say for sure that the Galaxy S II would justify a long-term iOS user foresaking his investment into one ecosystem and making the leap to another. Nonetheless, if you’re asking us what smartphone to buy today, unconstrained by such externalities, the Galaxy S II would be the clear choice. Sometimes it’s just as simple as that. Nokia Lumia 800 Here comes unexpected Windows Phone in to the boxing ring. May be they are not as great as Androids available in the market today, but they are picking up very quickly. Especially the Nokia Lumia 800 seems to be first ever Windows Phone 7 aimed at competing serious with Androids and iPhones available in the market. There are reports that Nokia Lumia 800 is outselling all Androids in UK and few high profile tech blogs are calling it as the king of Windows Phone. Considering this phone while evaluating the alternative of iPhone 4S will not disappoint you. We assure. Droid RAZR Remember the Motorola Driod that swept entire Android market share couple of years ago? The first two version of Motorola Droids were the best in the market and they out performed almost every other Android phone those days. The invasion of Samsung Androids, Motorola lost it charm. With the recent release of Droid RAZR, Motorola seems to be in the right direction to reclaiming the prestige. Droid RAZR is the thinnest smartphone available in the market and it’s beauty is not just skin deep. Here is a review of the phone from Engadget blog the RAZR’s beauty is not only skin deep. The LTE radio, 1.2GHz dual-core processor and 1GB of RAM make sure this sleek number is ready to run with the big boys. It kept pace with, and in some cases clearly outclassed its high-end competition. Despite its deficiencies in the display department and underwhelming battery life, the RAZR looks to be a perfectly viable alternative when considering the similarly-pricey Rezound and Galaxy Nexus Further Reading So we have seen the four alternates of iPhone 4S available in the market and I personally love to buy a Samsung smartphone if I’m don’t have money to afford an iPhone 4S. If you are interested in deep diving into the alternates, here few links that help you do more research Apple iPhone 4S vs. Samsung Galaxy Nexus vs. Motorola Droid RAZR: How Their Specs Compare by Huffington Post Nokia Lumia 800 vs. iPhone 4S vs. Nexus Galaxy: Spec Smackdown by PC World Browser Speed Test: Nokia Lumia 800 vs. iPhone 4S vs. Samsung Galaxy S II – by Gizmodo iPhone 4S vs Samsung Galaxy S II by pocket lint Apple iPhone 4S vs. Samsung Galaxy S II by techie buzz This article titled,Looking For iPhone 4S Alternatives? Here Are 3 Smartphones You Should Consider, was originally published at Tech Dreams. Grab our rss feed or fan us on Facebook to get updates from us.

  Read the article

• Troubleshooting High-CPU Utilization for SQL Server

  - by Susantha Bathige

  The objective of this FAQ is to outline the basic steps in troubleshooting high CPU utilization on  a server hosting a SQL Server instance. The first and the most common step if you suspect high CPU utilization (or are alerted for it) is to login to the physical server and check the Windows Task Manager. The Performance tab will show the high utilization as shown below: Next, we need to determine which process is responsible for the high CPU consumption. The Processes tab of the Task Manager will show this information: Note that to see all processes you should select Show processes from all user. In this case, SQL Server (sqlserver.exe) is consuming 99% of the CPU (a normal benchmark for max CPU utilization is about 50-60%). Next we examine the scheduler data. Scheduler is a component of SQLOS which evenly distributes load amongst CPUs. The query below returns the important columns for CPU troubleshooting. Note – if your server is under severe stress and you are unable to login to SSMS, you can use another machine’s SSMS to login to the server through DAC – Dedicated Administrator Connection (see http://msdn.microsoft.com/en-us/library/ms189595.aspx for details on using DAC) SELECT scheduler_id ,cpu_id ,status ,runnable_tasks_count ,active_workers_count ,load_factor ,yield_count FROM sys.dm_os_schedulers WHERE scheduler_id See below for the BOL definitions for the above columns. scheduler_id – ID of the scheduler. All schedulers that are used to run regular queries have ID numbers less than 1048576. Those schedulers that have IDs greater than or equal to 1048576 are used internally by SQL Server, such as the dedicated administrator connection scheduler. cpu_id – ID of the CPU with which this scheduler is associated. status – Indicates the status of the scheduler. runnable_tasks_count – Number of workers, with tasks assigned to them that are waiting to be scheduled on the runnable queue. active_workers_count – Number of workers that are active. An active worker is never preemptive, must have an associated task, and is either running, runnable, or suspended. current_tasks_count - Number of current tasks that are associated with this scheduler. load_factor – Internal value that indicates the perceived load on this scheduler. yield_count – Internal value that is used to indicate progress on this scheduler.                                                                 Now to interpret the above data. There are four schedulers and each assigned to a different CPU. All the CPUs are ready to accept user queries as they all are ONLINE. There are 294 active tasks in the output as per the current_tasks_count column. This count indicates how many activities currently associated with the schedulers. When a  task is complete, this number is decremented. The 294 is quite a high figure and indicates all four schedulers are extremely busy. When a task is enqueued, the load_factor  value is incremented. This value is used to determine whether a new task should be put on this scheduler or another scheduler. The new task will be allocated to less loaded scheduler by SQLOS. The very high value of this column indicates all the schedulers have a high load. There are 268 runnable tasks which mean all these tasks are assigned a worker and waiting to be scheduled on the runnable queue.   The next step is  to identify which queries are demanding a lot of CPU time. The below query is useful for this purpose (note, in its current form,  it only shows the top 10 records). SELECT TOP 10 st.text  ,st.dbid  ,st.objectid  ,qs.total_worker_time  ,qs.last_worker_time  ,qp.query_plan FROM sys.dm_exec_query_stats qs CROSS APPLY sys.dm_exec_sql_text(qs.sql_handle) st CROSS APPLY sys.dm_exec_query_plan(qs.plan_handle) qp ORDER BY qs.total_worker_time DESC This query as total_worker_time as the measure of CPU load and is in descending order of the  total_worker_time to show the most expensive queries and their plans at the top:      Note the BOL definitions for the important columns: total_worker_time - Total amount of CPU time, in microseconds, that was consumed by executions of this plan since it was compiled. last_worker_time - CPU time, in microseconds, that was consumed the last time the plan was executed.   I re-ran the same query again after few seconds and was returned the below output. After few seconds the SP dbo.TestProc1 is shown in fourth place and once again the last_worker_time is the highest. This means the procedure TestProc1 consumes a CPU time continuously each time it executes.      In this case, the primary cause for high CPU utilization was a stored procedure. You can view the execution plan by clicking on query_plan column to investigate why this is causing a high CPU load. I have used SQL Server 2008 (SP1) to test all the queries used in this article.

  Read the article

• Measuring Usability with Common Industry Format (CIF) Usability Tests

  - by Applications User Experience

  Sean Rice, Manager, Applications User Experience A User-centered Research and Design Process The Oracle Fusion Applications user experience was five years in the making. The development of this suite included an extensive and comprehensive user experience design process: ethnographic research, low-fidelity workflow prototyping, high fidelity user interface (UI) prototyping, iterative formative usability testing, development feedback and iteration, and sales and customer evaluation throughout the design cycle. However, this process does not stop when our products are released. We conduct summative usability testing using the ISO 25062 Common Industry Format (CIF) for usability test reports as an organizational framework. CIF tests allow us to measure the overall usability of our released products.  These studies provide benchmarks that allow for comparisons of a specific product release against previous versions of our product and against other products in the marketplace. What Is a CIF Usability Test? CIF refers to the internationally standardized method for reporting usability test findings used by the software industry. The CIF is based on a formal, lab-based test that is used to benchmark the usability of a product in terms of human performance and subjective data. The CIF was developed and is endorsed by more than 375 software customer and vendor organizations led by the National Institute for Standards and Technology (NIST), a US government entity. NIST sponsored the CIF through the American National Standards Institute (ANSI) and International Organization for Standardization (ISO) standards-making processes. Oracle played a key role in developing the CIF. The CIF report format and metrics are consistent with the ISO 9241-11 definition of usability: “The extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use.” Our goal in conducting CIF tests is to measure performance and satisfaction of a representative sample of users on a set of core tasks and to help predict how usable a product will be with the larger population of customers. Why Do We Perform CIF Testing? The overarching purpose of the CIF for usability test reports is to promote incorporation of usability as part of the procurement decision-making process for interactive products. CIF provides a common format for vendors to report the methods and results of usability tests to customer organizations, and enables customers to compare the usability of our software to that of other suppliers. CIF also enables us to compare our current software with previous versions of our software. CIF Testing for Fusion Applications Oracle Fusion Applications comprises more than 100 modules in seven different product families. These modules encompass more than 400 task flows and 400 user roles. Due to resource constraints, we cannot perform comprehensive CIF testing across the entire product suite. Therefore, we had to develop meaningful inclusion criteria and work with other stakeholders across the applications development organization to prioritize product areas for testing. Ultimately, we want to test the product areas for which customers might be most interested in seeing CIF data. We also want to build credibility with customers; we need to be able to make the case to current and prospective customers that the product areas tested are representative of the product suite as a whole. Our goal is to test the top use cases for each product. The primary activity in the scoping process was to work with the individual product teams to identify the key products and business process task flows in each product to test. We prioritized these products and flows through a series of negotiations among the user experience managers, product strategy, and product management directors for each of the primary product families within the Oracle Fusion Applications suite (Human Capital Management, Supply Chain Management, Customer Relationship Management, Financials, Projects, and Procurement). The end result of the scoping exercise was a list of 47 proposed CIF tests for the Fusion Applications product suite.  Figure 1. A participant completes tasks during a usability test in Oracle’s Usability Labs Fusion Supplier Portal CIF Test The first Fusion CIF test was completed on the Supplier Portal application in July of 2011.  Fusion Supplier Portal is part of an integrated suite of Procurement applications that helps supplier companies manage orders, schedules, shipments, invoices, negotiations and payments. The user roles targeted for the usability study were Supplier Account Receivables Specialists and Supplier Sales Representatives, including both experienced and inexperienced users across a wide demographic range.  The test specifically focused on the following functionality and features: Manage payments – view payments Manage invoices – view invoice status and create invoices Manage account information – create new contact, review bank account information Manage agreements – find and view agreement, upload agreement lines, confirm status of agreement lines upload Manage purchase orders (PO) – view history of PO, request change to PO, find orders Manage negotiations – respond to request for a quote, check the status of a negotiation response These product areas were selected to represent the most important subset of features and functionality of the flow, in terms of frequency and criticality of use by customers. A total of 20 users participated in the usability study. The results of the Supplier Portal evaluation were favorable and exceeded our expectations. Figure 2. Fusion Supplier Portal Next Studies We plan to conduct two Fusion CIF usability studies per product family over the next nine months. The next product to be tested will be Self-service Procurement. End users are currently being recruited to participate in this usability study, and the test sessions are scheduled to begin during the last week of November.

  Read the article

• HDFC Bank's Journey to Oracle Private Database Cloud

  - by Nilesh Agrawal

  One of the key takeaways from a recent post by Sushil Kumar is the importance of business initiative that drives the transformational journey from legacy IT to enterprise private cloud. The journey that leads to a agile, self-service and efficient infrastructure with reduced complexity and enables IT to deliver services more closely aligned with business requirements. Nilanjay Bhattacharjee, AVP, IT of HDFC Bank presented a real-world case study based on one such initiative in his Oracle OpenWorld session titled "HDFC BANK Journey into Oracle Database Cloud with EM 12c DBaaS". The case study highlighted in this session is from HDFC Bank’s Lending Business Segment, which comprises roughly 50% of Bank’s top line. Bank’s Lending Business is always under pressure to launch “New Schemes” to compete and stay ahead in this segment and IT has to keep up with this challenging business requirement. Lending related applications are highly dynamic and go through constant changes and every single and minor change in each related application is required to be thoroughly UAT tested certified before they are certified for production rollout. This leads to a constant pressure in IT for rapid provisioning of UAT databases on an ongoing basis to enable faster time to market. Nilanjay joined Sushil Kumar, VP, Product Strategy, Oracle, during the Enterprise Manager general session at Oracle OpenWorld 2012. Let's watch what Nilanjay had to say about their recent Database cloud deployment. “Agility” in launching new business schemes became the key business driver for private database cloud adoption in the Bank. Nilanjay spent an hour discussing it during his session. Let's look at why Database-as-a-Service(DBaaS) model was need of the hour in this case  - Average 3 days to provision UAT Database for Loan Management Application Silo’ed UAT environment with Average 30% utilization Compliance requirement consume UAT testing resources DBA activities leads to $$ paid to SI for provisioning databases manually Overhead in managing configuration drift between production and test environments Rollout impact/delay on new business initiatives The private database cloud implementation progressed through 4 fundamental phases - Standardization, Consolidation, Automation, Optimization of UAT infrastructure. Project scoping was carried out and end users and stakeholders were engaged early on right from planning phase and including all phases of implementation. Standardization and Consolidation phase involved multiple iterations of planning to first standardize on infrastructure, db versions, patch levels, configuration, IT processes etc and with database level consolidation project onto Exadata platform. It was also decided to have existing AIX UAT DB landscape covered and EM 12c DBaaS solution being platform agnostic supported this model well. Automation and Optimization phase provided the necessary Agility, Self-Service and efficiency and this was made possible via EM 12c DBaaS. EM 12c DBaaS Self-Service/SSA Portal was setup with required zones, quotas, service templates, charge plan defined. There were 2 zones implemented - Exadata zone  primarily for UAT and benchmark testing for databases running on Exadata platform and second zone was for AIX setup to cover other databases those running on AIX. Metering and Chargeback/Showback capabilities provided business and IT the framework for cloud optimization and also visibility into cloud usage. More details on UAT cloud implementation, related building blocks and EM 12c DBaaS solution are covered in Nilanjay's OpenWorld session here. Some of the key Benefits achieved from UAT cloud initiative are - New business initiatives can be easily launched due to rapid provisioning of UAT Databases [ ~3 hours ] Drastically cut down $$ on SI for DBA Activities due to Self-Service Effective usage of infrastructure leading to  better ROI Empowering  consumers to provision database using Self-Service Control on project schedule with DB end date aligned to project plan submitted during provisioning Databases provisioned through Self-Service are monitored in EM and auto configured for Alerts and KPI Regulatory requirement of database does not impact existing project in queue This table below shows typical list of activities and tasks involved when a end user requests for a UAT database. EM 12c DBaaS solution helped reduce UAT database provisioning time from roughly 3 days down to 3 hours and this timing also includes provisioning time for database with production scale data (ranging from 250 G to 2 TB of data) - And it's not just about time to provision,  this initiative has enabled an agile, efficient and transparent UAT environment where end users are empowered with real control of cloud resources and IT's role is shifted as enabler of strategic services instead of being administrator of all user requests. The strong collaboration between IT and business community right from planning to implementation to go-live has played the key role in achieving this common goal of enterprise private cloud. Finally, real cloud is here and this cloud is accompanied with rain (business benefits) as well ! For more information, please go to Oracle Enterprise Manager  web page or  follow us at :  Twitter | Facebook | YouTube | Linkedin | Newsletter

  Read the article

• Why lock-free data structures just aren't lock-free enough

  - by Alex.Davies

  Today's post will explore why the current ways to communicate between threads don't scale, and show you a possible way to build scalable parallel programming on top of shared memory. The problem with shared memory Soon, we will have dozens, hundreds and then millions of cores in our computers. It's inevitable, because individual cores just can't get much faster. At some point, that's going to mean that we have to rethink our architecture entirely, as millions of cores can't all access a shared memory space efficiently. But millions of cores are still a long way off, and in the meantime we'll see machines with dozens of cores, struggling with shared memory. Alex's tip: The best way for an application to make use of that increasing parallel power is to use a concurrency model like actors, that deals with synchronisation issues for you. Then, the maintainer of the actors framework can find the most efficient way to coordinate access to shared memory to allow your actors to pass messages to each other efficiently. At the moment, NAct uses the .NET thread pool and a few locks to marshal messages. It works well on dual and quad core machines, but it won't scale to more cores. Every time we use a lock, our core performs an atomic memory operation (eg. CAS) on a cell of memory representing the lock, so it's sure that no other core can possibly have that lock. This is very fast when the lock isn't contended, but we need to notify all the other cores, in case they held the cell of memory in a cache. As the number of cores increases, the total cost of a lock increases linearly. A lot of work has been done on "lock-free" data structures, which avoid locks by using atomic memory operations directly. These give fairly dramatic performance improvements, particularly on systems with a few (2 to 4) cores. The .NET 4 concurrent collections in System.Collections.Concurrent are mostly lock-free. However, lock-free data structures still don't scale indefinitely, because any use of an atomic memory operation still involves every core in the system. A sync-free data structure Some concurrent data structures are possible to write in a completely synchronization-free way, without using any atomic memory operations. One useful example is a single producer, single consumer (SPSC) queue. It's easy to write a sync-free fixed size SPSC queue using a circular buffer*. Slightly trickier is a queue that grows as needed. You can use a linked list to represent the queue, but if you leave the nodes to be garbage collected once you're done with them, the GC will need to involve all the cores in collecting the finished nodes. Instead, I've implemented a proof of concept inspired by this intel article which reuses the nodes by putting them in a second queue to send back to the producer. * In all these cases, you need to use memory barriers correctly, but these are local to a core, so don't have the same scalability problems as atomic memory operations. Performance tests I tried benchmarking my SPSC queue against the .NET ConcurrentQueue, and against a standard Queue protected by locks. In some ways, this isn't a fair comparison, because both of these support multiple producers and multiple consumers, but I'll come to that later. I started on my dual-core laptop, running a simple test that had one thread producing 64 bit integers, and another consuming them, to measure the pure overhead of the queue. So, nothing very interesting here. Both concurrent collections perform better than the lock-based one as expected, but there's not a lot to choose between the ConcurrentQueue and my SPSC queue. I was a little disappointed, but then, the .NET Framework team spent a lot longer optimising it than I did. So I dug out a more powerful machine that Red Gate's DBA tools team had been using for testing. It is a 6 core Intel i7 machine with hyperthreading, adding up to 12 logical cores. Now the results get more interesting. As I increased the number of producer-consumer pairs to 6 (to saturate all 12 logical cores), the locking approach was slow, and got even slower, as you'd expect. What I didn't expect to be so clear was the drop-off in performance of the lock-free ConcurrentQueue. I could see the machine only using about 20% of available CPU cycles when it should have been saturated. My interpretation is that as all the cores used atomic memory operations to safely access the queue, they ended up spending most of the time notifying each other about cache lines that need invalidating. The sync-free approach scaled perfectly, despite still working via shared memory, which after all, should still be a bottleneck. I can't quite believe that the results are so clear, so if you can think of any other effects that might cause them, please comment! Obviously, this benchmark isn't realistic because we're only measuring the overhead of the queue. Any real workload, even on a machine with 12 cores, would dwarf the overhead, and there'd be no point worrying about this effect. But would that be true on a machine with 100 cores? Still to be solved. The trouble is, you can't build many concurrent algorithms using only an SPSC queue to communicate. In particular, I can't see a way to build something as general purpose as actors on top of just SPSC queues. Fundamentally, an actor needs to be able to receive messages from multiple other actors, which seems to need an MPSC queue. I've been thinking about ways to build a sync-free MPSC queue out of multiple SPSC queues and some kind of sign-up mechanism. Hopefully I'll have something to tell you about soon, but leave a comment if you have any ideas.

  Read the article

• Talend Enterprise Data Integration overperforms on Oracle SPARC T4

  - by Amir Javanshir

  The SPARC T microprocessor, released in 2005 by Sun Microsystems, and now continued at Oracle, has a good track record in parallel execution and multi-threaded performance. However it was less suited for pure single-threaded workloads. The new SPARC T4 processor is now filling that gap by offering a 5x better single-thread performance over previous generations. Following our long-term relationship with Talend, a fast growing ISV positioned by Gartner in the “Visionaries” quadrant of the “Magic Quadrant for Data Integration Tools”, we decided to test some of their integration components with the T4 chip, more precisely on a T4-1 system, in order to verify first hand if this new processor stands up to its promises. Several tests were performed, mainly focused on: Single-thread performance of the new SPARC T4 processor compared to an older SPARC T2+ processor Overall throughput of the SPARC T4-1 server using multiple threads The tests consisted in reading large amounts of data --ten's of gigabytes--, processing and writing them back to a file or an Oracle 11gR2 database table. They are CPU, memory and IO bound tests. Given the main focus of this project --CPU performance--, bottlenecks were removed as much as possible on the memory and IO sub-systems. When possible, the data to process was put into the ZFS filesystem cache, for instance. Also, two external storage devices were directly attached to the servers under test, each one divided in two ZFS pools for read and write operations. Multi-thread: Testing throughput on the Oracle T4-1 The tests were performed with different number of simultaneous threads (1, 2, 4, 8, 12, 16, 32, 48 and 64) and using different storage devices: Flash, Fibre Channel storage, two stripped internal disks and one single internal disk. All storage devices used ZFS as filesystem and volume management. Each thread read a dedicated 1GB-large file containing 12.5M lines with the following structure: customerID;FirstName;LastName;StreetAddress;City;State;Zip;Cust_Status;Since_DT;Status_DT 1;Ronald;Reagan;South Highway;Santa Fe;Montana;98756;A;04-06-2006;09-08-2008 2;Theodore;Roosevelt;Timberlane Drive;Columbus;Louisiana;75677;A;10-05-2009;27-05-2008 3;Andrew;Madison;S Rustle St;Santa Fe;Arkansas;75677;A;29-04-2005;09-02-2008 4;Dwight;Adams;South Roosevelt Drive;Baton Rouge;Vermont;75677;A;15-02-2004;26-01-2007 […] The following graphs present the results of our tests: Unsurprisingly up to 16 threads, all files fit in the ZFS cache a.k.a L2ARC : once the cache is hot there is no performance difference depending on the underlying storage. From 16 threads upwards however, it is clear that IO becomes a bottleneck, having a good IO subsystem is thus key. Single-disk performance collapses whereas the Sun F5100 and ST6180 arrays allow the T4-1 to scale quite seamlessly. From 32 to 64 threads, the performance is almost constant with just a slow decline. For the database load tests, only the best IO configuration --using external storage devices-- were used, hosting the Oracle table spaces and redo log files. Using the Sun Storage F5100 array allows the T4-1 server to scale up to 48 parallel JVM processes before saturating the CPU. The final result is a staggering 646K lines per second insertion in an Oracle table using 48 parallel threads. Single-thread: Testing the single thread performance Seven different tests were performed on both servers. Given the fact that only one thread, thus one file was read, no IO bottleneck was involved, all data being served from the ZFS cache. Read File ? Filter ? Write File: Read file, filter data, write the filtered data in a new file. The filter is set on the “Status” column: only lines with status set to “A” are selected. This limits each output file to about 500 MB. Read File ? Load Database Table: Read file, insert into a single Oracle table. Average: Read file, compute the average of a numeric column, write the result in a new file. Division & Square Root: Read file, perform a division and square root on a numeric column, write the result data in a new file. Oracle DB Dump: Dump the content of an Oracle table (12.5M rows) into a CSV file. Transform: Read file, transform, write the result data in a new file. The transformations applied are: set the address column to upper case and add an extra column at the end, which is the concatenation of two columns. Sort: Read file, sort a numeric and alpha numeric column, write the result data in a new file. The following table and graph present the final results of the tests: Throughput unit is thousand lines per second processed (K lines/second). Improvement is the % of improvement between the T5140 and T4-1. Test T4-1 (Time s.) T5140 (Time s.) Improvement T4-1 (Throughput) T5140 (Throughput) Read/Filter/Write 125 806 645% 100 16 Read/Load Database 195 1111 570% 64 11 Average 96 557 580% 130 22 Division & Square Root 161 1054 655% 78 12 Oracle DB Dump 164 945 576% 76 13 Transform 159 1124 707% 79 11 Sort 251 1336 532% 50 9 The improvement of single-thread performance is quite dramatic: depending on the tests, the T4 is between 5.4 to 7 times faster than the T2+. It seems clear that the SPARC T4 processor has gone a long way filling the gap in single-thread performance, without sacrifying the multi-threaded capability as it still shows a very impressive scaling on heavy-duty multi-threaded jobs. Finally, as always at Oracle ISV Engineering, we are happy to help our ISV partners test their own applications on our platforms, so don't hesitate to contact us and let's see what the SPARC T4-based systems can do for your application! "As describe in this benchmark, Talend Enterprise Data Integration has overperformed on T4. I was generally happy to see that the T4 gave scaling opportunities for many scenarios like complex aggregations. Row by row insertion in Oracle DB is faster with more than 650,000 rows per seconds without using any bulk Oracle capabilities !" Cedric Carbone, Talend CTO.

  Read the article

• MySQL Connect 9 Days Away – Optimizer Sessions

  - by Bertrand Matthelié

  72 1024x768 Normal 0 false false false EN-US X-NONE X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-qformat:yes; mso-style-parent:""; mso-padding-alt:0cm 5.4pt 0cm 5.4pt; mso-para-margin:0cm; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:11.0pt; font-family:"Calibri","sans-serif"; mso-ascii-font-family:Calibri; mso-ascii-theme-font:minor-latin; mso-fareast-font-family:"Times New Roman"; mso-fareast-theme-font:minor-fareast; mso-hansi-font-family:Calibri; mso-hansi-theme-font:minor-latin; mso-bidi-font-family:"Times New Roman"; mso-bidi-theme-font:minor-bidi;} Following my previous blog post focusing on InnoDB talks at MySQL Connect, let us review today the sessions focusing on the MySQL Optimizer: Saturday, 11.30 am, Room Golden Gate 6: MySQL Optimizer Overview—Olav Sanstå, Oracle The goal of MySQL optimizer is to take a SQL query as input and produce an optimal execution plan for the query. This session presents an overview of the main phases of the MySQL optimizer and the primary optimizations done to the query. These optimizations are based on a combination of logical transformations and cost-based decisions. Examples of optimization strategies the presentation covers are the main query transformations, the join optimizer, the data access selection strategies, and the range optimizer. For the cost-based optimizations, an overview of the cost model and the data used for doing the cost estimations is included. Saturday, 1.00 pm, Room Golden Gate 6: Overview of New Optimizer Features in MySQL 5.6—Manyi Lu, Oracle Many optimizer features have been added into MySQL 5.6. This session provides an introduction to these great features. Multirange read, index condition pushdown, and batched key access will yield huge performance improvements on large data volumes. Structured explain, explain for update/delete/insert, and optimizer tracing will help users analyze and speed up queries. And last but not least, the session covers subquery optimizations in Release 5.6. Saturday, 7.00 pm, Room Golden Gate 4: BoF: Query Optimizations: What Is New and What Is Coming? This BoF presents common techniques for query optimization, covers what is new in MySQL 5.6, and provides a discussion forum in which attendees can tell the MySQL optimizer team which optimizations they would like to see in the future. Sunday, 1.15 pm, Room Golden Gate 8: Query Performance Comparison of MySQL 5.5 and MySQL 5.6—Øystein Grøvlen, Oracle MySQL Release 5.6 contains several improvements in the query optimizer that create improved performance for complex queries. This presentation looks at how MySQL 5.6 improves the performance of many of the queries in the DBT-3 benchmark. Based on the observed improvements, the presentation discusses what makes the specific queries perform better in Release 5.6. It describes the relevant new optimization techniques and gives examples of the types of queries that will benefit from these techniques. Sunday, 4.15 pm, Room Golden Gate 4: Powerful EXPLAIN in MySQL 5.6—Evgeny Potemkin, Oracle The EXPLAIN command of MySQL has long been a very useful tool for understanding how MySQL will execute a query. Release 5.6 of the MySQL database offers several new additions that give more-detailed information about the query plan and make it easier to understand at the same time. This presentation gives an overview of new EXPLAIN features: structured EXPLAIN in JSON format, EXPLAIN for INSERT/UPDATE/DELETE, and optimizer tracing. Examples in the session give insights into how you can take advantage of the new features. They show how these features supplement and relate to each other and to classical EXPLAIN and how and why the MySQL server chooses a particular query plan. You can check out the full program here as well as in the September edition of the MySQL newsletter. Not registered yet? You can still save US$ 300 over the on-site fee – Register Now!

  Read the article

• .NET 4: &ldquo;Slim&rdquo;-style performance boost!

  - by Vitus

  RTM version of .NET 4 and Visual Studio 2010 is available, and now we can do some test with it. Parallel Extensions is one of the most valuable part of .NET 4.0. It’s a set of good tools for easily consuming multicore hardware power. And it also contains some “upgraded” sync primitives – Slim-version. For example, it include updated variant of widely known ManualResetEvent. For people, who don’t know about it: you can sync concurrency execution of some pieces of code with this sync primitive. Instance of ManualResetEvent can be in 2 states: signaled and non-signaled. Transition between it possible by Set() and Reset() methods call. Some shortly explanation: Thread 1 Thread 2 Time mre.Reset(); mre.WaitOne(); //code execution 0 //wating //code execution 1 //wating //code execution 2 //wating //code execution 3 //wating mre.Set(); 4 //code execution //… 5 Upgraded version of this primitive is ManualResetEventSlim. The idea in decreasing performance cost in case, when only 1 thread use it. Main concept in the “hybrid sync schema”, which can be done as following:   internal sealed class SimpleHybridLock : IDisposable { private Int32 m_waiters = 0; private AutoResetEvent m_waiterLock = new AutoResetEvent(false);   public void Enter() { if (Interlocked.Increment(ref m_waiters) == 1) return; m_waiterLock.WaitOne(); }   public void Leave() { if (Interlocked.Decrement(ref m_waiters) == 0) return; m_waiterLock.Set(); }   public void Dispose() { m_waiterLock.Dispose(); } } It’s a sample from Jeffry Richter’s book “CLR via C#”, 3rd edition. Primitive SimpleHybridLock have two public methods: Enter() and Leave(). You can put your concurrency-critical code between calls of these methods, and it would executed in only one thread at the moment. Code is really simple: first thread, called Enter(), increase counter. Second thread also increase counter, and suspend while m_waiterLock is not signaled. So, if we don’t have concurrent access to our lock, “heavy” methods WaitOne() and Set() will not called. It’s can give some performance bonus. ManualResetEvent use the similar idea. Of course, it have more “smart” technics inside, like a checking of recursive calls, and so on. I want to know a real difference between classic ManualResetEvent realization, and new –Slim. I wrote a simple “benchmark”: class Program { static void Main(string[] args) { ManualResetEventSlim mres = new ManualResetEventSlim(false); ManualResetEventSlim mres2 = new ManualResetEventSlim(false);   ManualResetEvent mre = new ManualResetEvent(false);   long total = 0; int COUNT = 50;   for (int i = 0; i < COUNT; i++) { mres2.Reset(); Stopwatch sw = Stopwatch.StartNew();   ThreadPool.QueueUserWorkItem((obj) => { //Method(mres, true); Method2(mre, true); mres2.Set(); }); //Method(mres, false); Method2(mre, false);   mres2.Wait(); sw.Stop();   Console.WriteLine("Pass {0}: {1} ms", i, sw.ElapsedMilliseconds); total += sw.ElapsedMilliseconds; }   Console.WriteLine(); Console.WriteLine("==============================="); Console.WriteLine("Done in average=" + total / (double)COUNT); Console.ReadLine(); }   private static void Method(ManualResetEventSlim mre, bool value) { for (int i = 0; i < 9000000; i++) { if (value) { mre.Set(); } else { mre.Reset(); } } }   private static void Method2(ManualResetEvent mre, bool value) { for (int i = 0; i < 9000000; i++) { if (value) { mre.Set(); } else { mre.Reset(); } } } } I use 2 concurrent thread (the main thread and one from thread pool) for setting and resetting ManualResetEvents, and try to run test COUNT times, and calculate average execution time. Here is the results (I get it on my dual core notebook with T7250 CPU and Windows 7 x64): ManualResetEvent ManualResetEventSlim Difference is obvious and serious – in 10 times! So, I think preferable way is using ManualResetEventSlim, because not always on calling Set() and Reset() will be called “heavy” methods for working with Windows kernel-mode objects. It’s a small and nice improvement! ;)

  Read the article

• How to Tell If Your Computer is Overheating and What to Do About It

  - by Chris Hoffman

  Heat is a computer’s enemy. Computers are designed with heat dispersion and ventilation in mind so they don’t overheat. If too much heat builds up, your computer may become unstable or suddenly shut down. The CPU and graphics card produce much more heat when running demanding applications. If there’s a problem with your computer’s cooling system, an excess of heat could even physically damage its components. Is Your Computer Overheating? When using a typical computer in a typical way, you shouldn’t have to worry about overheating at all. However, if you’re encountering system instability issues like abrupt shut downs, blue screens, and freezes — especially while doing something demanding like playing PC games or encoding video — your computer may be overheating. This can happen for several reasons. Your computer’s case may be full of dust, a fan may have failed, something may be blocking your computer’s vents, or you may have a compact laptop that was never designed to run at maximum performance for hours on end. Monitoring Your Computer’s Temperature First, bear in mind that different CPUs and GPUs (graphics cards) have different optimal temperature ranges. Before getting too worried about a temperature, be sure to check your computer’s documentation — or its CPU or graphics card specifications — and ensure you know the temperature ranges your hardware can handle. You can monitor your computer’s temperatures in a variety of different ways. First, you may have a way to monitor temperature that is already built into your system. You can often view temperature values in your computer’s BIOS or UEFI settings screen. This allows you to quickly see your computer’s temperature if Windows freezes or blue screens on you — just boot the computer, enter the BIOS or UEFI screen, and check the temperatures displayed there. Note that not all BIOSes or UEFI screens will display this information, but it is very common. There are also programs that will display your computer’s temperature. Such programs just read the sensors inside your computer and show you the temperature value they report, so there are a wide variety of tools you can use for this, from the simple Speccy system information utility to an advanced tool like SpeedFan. HWMonitor also offer this feature, displaying a wide variety of sensor information. Be sure to look at your CPU and graphics card temperatures. You can also find other temperatures, such as the temperature of your hard drive, but these components will generally only overheat if it becomes extremely hot in the computer’s case. They shouldn’t generate too much heat on their own. If you think your computer may be overheating, don’t just glance as these sensors once and ignore them. Do something demanding with your computer, such as running a CPU burn-in test with Prime 95, playing a PC game, or running a graphical benchmark. Monitor the computer’s temperature while you do this, even checking a few hours later — does any component overheat after you push it hard for a while? Preventing Your Computer From Overheating If your computer is overheating, here are some things you can do about it: Dust Out Your Computer’s Case: Dust accumulates in desktop PC cases and even laptops over time, clogging fans and blocking air flow. This dust can cause ventilation problems, trapping heat and preventing your PC from cooling itself properly. Be sure to clean your computer’s case occasionally to prevent dust build-up. Unfortunately, it’s often more difficult to dust out overheating laptops. Ensure Proper Ventilation: Put the computer in a location where it can properly ventilate itself. If it’s a desktop, don’t push the case up against a wall so that the computer’s vents become blocked or leave it near a radiator or heating vent. If it’s a laptop, be careful to not block its air vents, particularly when doing something demanding. For example, putting a laptop down on a mattress, allowing it to sink in, and leaving it there can lead to overheating — especially if the laptop is doing something demanding and generating heat it can’t get rid of. Check if Fans Are Running: If you’re not sure why your computer started overheating, open its case and check that all the fans are running. It’s possible that a CPU, graphics card, or case fan failed or became unplugged, reducing air flow. Tune Up Heat Sinks: If your CPU is overheating, its heat sink may not be seated correctly or its thermal paste may be old. You may need to remove the heat sink and re-apply new thermal paste before reseating the heat sink properly. This tip applies more to tweakers, overclockers, and people who build their own PCs, especially if they may have made a mistake when originally applying the thermal paste. This is often much more difficult when it comes to laptops, which generally aren’t designed to be user-serviceable. That can lead to trouble if the laptop becomes filled with dust and needs to be cleaned out, especially if the laptop was never designed to be opened by users at all. Consult our guide to diagnosing and fixing an overheating laptop for help with cooling down a hot laptop. Overheating is a definite danger when overclocking your CPU or graphics card. Overclocking will cause your components to run hotter, and the additional heat will cause problems unless you can properly cool your components. If you’ve overclocked your hardware and it has started to overheat — well, throttle back the overclock! Image Credit: Vinni Malek on Flickr     

  Read the article

• Polite busy-waiting with WRPAUSE on SPARC

  - by Dave

  Unbounded busy-waiting is an poor idea for user-space code, so we typically use spin-then-block strategies when, say, waiting for a lock to be released or some other event. If we're going to spin, even briefly, then we'd prefer to do so in a manner that minimizes performance degradation for other sibling logical processors ("strands") that share compute resources. We want to spin politely and refrain from impeding the progress and performance of other threads — ostensibly doing useful work and making progress — that run on the same core. On a SPARC T4, for instance, 8 strands will share a core, and that core has its own L1 cache and 2 pipelines. On x86 we have the PAUSE instruction, which, naively, can be thought of as a hardware "yield" operator which temporarily surrenders compute resources to threads on sibling strands. Of course this helps avoid intra-core performance interference. On the SPARC T2 our preferred busy-waiting idiom was "RD %CCR,%G0" which is a high-latency no-nop. The T4 provides a dedicated and extremely useful WRPAUSE instruction. The processor architecture manuals are the authoritative source, but briefly, WRPAUSE writes a cycle count into the the PAUSE register, which is ASR27. Barring interrupts, the processor then delays for the requested period. There's no need for the operating system to save the PAUSE register over context switches as it always resets to 0 on traps. Digressing briefly, if you use unbounded spinning then ultimately the kernel will preempt and deschedule your thread if there are other ready threads than are starving. But by using a spin-then-block strategy we can allow other ready threads to run without resorting to involuntary time-slicing, which operates on a long-ish time scale. Generally, that makes your application more responsive. In addition, by blocking voluntarily we give the operating system far more latitude regarding power management. Finally, I should note that while we have OS-level facilities like sched_yield() at our disposal, yielding almost never does what you'd want or naively expect. Returning to WRPAUSE, it's natural to ask how well it works. To help answer that question I wrote a very simple C/pthreads benchmark that launches 8 concurrent threads and binds those threads to processors 0..7. The processors are numbered geographically on the T4, so those threads will all be running on just one core. Unlike the SPARC T2, where logical CPUs 0,1,2 and 3 were assigned to the first pipeline, and CPUs 4,5,6 and 7 were assigned to the 2nd, there's no fixed mapping between CPUs and pipelines in the T4. And in some circumstances when the other 7 logical processors are idling quietly, it's possible for the remaining logical processor to leverage both pipelines. Some number T of the threads will iterate in a tight loop advancing a simple Marsaglia xor-shift pseudo-random number generator. T is a command-line argument. The main thread loops, reporting the aggregate number of PRNG steps performed collectively by those T threads in the last 10 second measurement interval. The other threads (there are 8-T of these) run in a loop busy-waiting concurrently with the T threads. We vary T between 1 and 8 threads, and report on various busy-waiting idioms. The values in the table are the aggregate number of PRNG steps completed by the set of T threads. The unit is millions of iterations per 10 seconds. For the "PRNG step" busy-waiting mode, the busy-waiting threads execute exactly the same code as the T worker threads. We can easily compute the average rate of progress for individual worker threads by dividing the aggregate score by the number of worker threads T. I should note that the PRNG steps are extremely cycle-heavy and access almost no memory, so arguably this microbenchmark is not as representative of "normal" code as it could be. And for the purposes of comparison I included a row in the table that reflects a waiting policy where the waiting threads call poll(NULL,0,1000) and block in the kernel. Obviously this isn't busy-waiting, but the data is interesting for reference. _table { border:2px black dotted; margin: auto; width: auto; } _tr { border: 2px red dashed; } _td { border: 1px green solid; } _table { border:2px black dotted; margin: auto; width: auto; } _tr { border: 2px red dashed; } td { background-color : #E0E0E0 ; text-align : right ; } th { text-align : left ; } td { background-color : #E0E0E0 ; text-align : right ; } th { text-align : left ; } Aggregate progress T = #worker threads Wait Mechanism for 8-T threadsT=1T=2T=3T=4T=5T=6T=7T=8 Park thread in poll() 32653347334833483348334833483348 no-op 415 831 124316482060249729303349 RD %ccr,%g0 "pause" 14262429269228623013316232553349 PRNG step 412 829 124616702092251029303348 WRPause(8000) 32443361333133483349334833483348 WRPause(4000) 32153308331533223347334833473348 WRPause(1000) 30853199322432513310334833483348 WRPause(500) 29173070315032223270330933483348 WRPause(250) 26942864294930773205338833483348 WRPause(100) 21552469262227902911321433303348

  Read the article

• Oracle NoSQL Database Exceeds 1 Million Mixed YCSB Ops/Sec

  - by Charles Lamb

  We ran a set of YCSB performance tests on Oracle NoSQL Database using SSD cards and Intel Xeon E5-2690 CPUs with the goal of achieving 1M mixed ops/sec on a 95% read / 5% update workload. We used the standard YCSB parameters: 13 byte keys and 1KB data size (1,102 bytes after serialization). The maximum database size was 2 billion records, or approximately 2 TB of data. We sized the shards to ensure that this was not an "in-memory" test (i.e. the data portion of the B-Trees did not fit into memory). All updates were durable and used the "simple majority" replica ack policy, effectively 'committing to the network'. All read operations used the Consistency.NONE_REQUIRED parameter allowing reads to be performed on any replica. In the past we have achieved 100K ops/sec using SSD cards on a single shard cluster (replication factor 3) so for this test we used 10 shards on 15 Storage Nodes with each SN carrying 2 Rep Nodes and each RN assigned to its own SSD card. After correcting a scaling problem in YCSB, we blew past the 1M ops/sec mark with 8 shards and proceeded to hit 1.2M ops/sec with 10 shards.  Hardware Configuration We used 15 servers, each configured with two 335 GB SSD cards. We did not have homogeneous CPUs across all 15 servers available to us so 12 of the 15 were Xeon E5-2690, 2.9 GHz, 2 sockets, 32 threads, 193 GB RAM, and the other 3 were Xeon E5-2680, 2.7 GHz, 2 sockets, 32 threads, 193 GB RAM.  There might have been some upside in having all 15 machines configured with the faster CPU, but since CPU was not the limiting factor we don't believe the improvement would be significant. The client machines were Xeon X5670, 2.93 GHz, 2 sockets, 24 threads, 96 GB RAM. Although the clients had 96 GB of RAM, neither the NoSQL Database or YCSB clients require anywhere near that amount of memory and the test could have just easily been run with much less. Networking was all 10GigE. YCSB Scaling Problem We made three modifications to the YCSB benchmark. The first was to allow the test to accommodate more than 2 billion records (effectively int's vs long's). To keep the key size constant, we changed the code to use base 32 for the user ids. The second change involved to the way we run the YCSB client in order to make the test itself horizontally scalable.The basic problem has to do with the way the YCSB test creates its Zipfian distribution of keys which is intended to model "real" loads by generating clusters of key collisions. Unfortunately, the percentage of collisions on the most contentious keys remains the same even as the number of keys in the database increases. As we scale up the load, the number of collisions on those keys increases as well, eventually exceeding the capacity of the single server used for a given key.This is not a workload that is realistic or amenable to horizontal scaling. YCSB does provide alternate key distribution algorithms so this is not a shortcoming of YCSB in general. We decided that a better model would be for the key collisions to be limited to a given YCSB client process. That way, as additional YCSB client processes (i.e. additional load) are added, they each maintain the same number of collisions they encounter themselves, but do not increase the number of collisions on a single key in the entire store. We added client processes proportionally to the number of records in the database (and therefore the number of shards). This change to the use of YCSB better models a use case where new groups of users are likely to access either just their own entries, or entries within their own subgroups, rather than all users showing the same interest in a single global collection of keys. If an application finds every user having the same likelihood of wanting to modify a single global key, that application has no real hope of getting horizontal scaling. Finally, we used read/modify/write (also known as "Compare And Set") style updates during the mixed phase. This uses versioned operations to make sure that no updates are lost. This mode of operation provides better application behavior than the way we have typically run YCSB in the past, and is only practical at scale because we eliminated the shared key collision hotspots.It is also a more realistic testing scenario. To reiterate, all updates used a simple majority replica ack policy making them durable. Scalability Results In the table below, the "KVS Size" column is the number of records with the number of shards and the replication factor. Hence, the first row indicates 400m total records in the NoSQL Database (KV Store), 2 shards, and a replication factor of 3. The "Clients" column indicates the number of YCSB client processes. "Threads" is the number of threads per process with the total number of threads. Hence, 90 threads per YCSB process for a total of 360 threads. The client processes were distributed across 10 client machines. Shards KVS Size Clients Mixed (records) Threads OverallThroughput(ops/sec) Read Latencyav/95%/99%(ms) Write Latencyav/95%/99%(ms) 2 400m(2x3) 4 90(360) 302,152 0.76/1/3 3.08/8/35 4 800m(4x3) 8 90(720) 558,569 0.79/1/4 3.82/16/45 8 1600m(8x3) 16 90(1440) 1,028,868 0.85/2/5 4.29/21/51 10 2000m(10x3) 20 90(1800) 1,244,550 0.88/2/6 4.47/23/53

  Read the article

• Oracle 11g R2 1???????~????????(Exadata??)?????

  - by Yusuke.Yamamoto

  ??2010?11?17???Oracle Database 11g Release2(R2) ???????1???? ????Oracle Database 11g R2 ?????????????????????????? ???? 2010/11/17:????? 2011/01/07:???????(Exadata/??) 2011/01/18:???????(Exadata/?????????????) 2011/02/22:???????(Exadata/?????:IT Leaders ????????) 2011/04/21:?????? 2011/04/21:???????(????????????) 2011/04/21:???????(Exadata/???????????????????????????????????) 2011/06/27:Oracle Exadata Database Machine ????1,000??? ?? Oracle Database 11g R2 ??????? Oracle Database 11g ?????????(????) ??????? Oracle Database 11g R2(???/????) Oracle Database 11g R2 ??????? ?? ??? 2009?11?11? Oracle Exadata Database Machine Version 2 ???? 2009?11?17? Oracle Database 11g R2 ???? 2010?02?01? ?????????????????????????????? 2010?03?31? SAP ? Oracle Database 11g R2 ??????????ISV????????·??????????? 2010?05?18? Windows Server 2008 R2 / Windows 7 ?????????Oracle Database 10g R2 ??? 2010?06?23? Oracle Application Express 4.0 ???? 2010?07?09? ?? Windows RDBMS ?????(2009?)????????? 2010?08?17? TPC-C Benchmark Price/Performance ???????? 2010?09?13? Patch Set 11.2.0.2 for Linux ????(??) 2010?10?20? Oracle Exadata Database Machine X2 ???? 2010?11?17? Oracle Database 11g R2 ????1?? 2010?11?19? ?? Windows RDBMS ?????(2010????)????????????? 2011?03?29? Oracle SQL Developer 3.0 ???? 2011?06?27? Oracle Exadata Database Machine ????1,000????????????????·?????????????? Oracle Database 11g ?????????(????) ????????????????????????????????(????)? ????(??????????) ??????????(???) ????? ????(???) ?????·???????·??? ????? ????·??????·?? ???? ???????(??????????????)|???99.999%???????500???????????? - ITpro ??????????? ????(????) ???(???) ????????(???) ??????(???????????) Oracle Exadata Database Machine ????? Oracle Database 11g ??(????)? ??????????????????????????????????? ????(??????) ????????????? ?????·???????·??? ??(??????????????) ?????(??????????) ?????????(????????) ?????????? ????(???????) ?????? ????/????·???????? ???????????(???????/NTT??????????) ????????????? ???? ???????????? ?????? ??? ?????|DWH?????????????? - IT Leaders(????????)|DWH?????????????? - IT Leaders ????(???????????) Customer Voice ????:????IT?????24??365????????????????????? ?Oracle9i Database ?????????????????????Oracle Database 11g ???????????????????????? Oracle9i Database ???????????????? Customer Voice ??????:Oracle Database 11g????????????????????? ?Oracle ASM ???????????????????I/O????????????????????????????????????? ??????? Oracle Database 11g R2(???/????) ???????????????? Oracle 11g R2 ????????? - IT Leaders ??????????11g R2?5???? - ??SE????Oracle??? - Think IT ????????????????????????~Oracle Database 11g Release2 ????????? - oracletech.jp ??????????? Oracle Database 11g Release 2(11gR2)|??????????? ???????|???????????

  Read the article

• Oracle 11gR2:????(RAC/Exadata), ??, ????

  - by Yusuke.Yamamoto

  Oracle Database 11g ?????????????(RAC/Exadata)????????????? ???? 2010/11/17:????? 2011/01/07:???????:Exadata/?? 2011/01/18:???????:Exadata/????????????? 2011/04/21:???????:???????????? 2011/04/21:???????:Exadata/??????????????????????????????????? 2011/06/27:??????:Oracle Exadata Database Machine ????1,000??? 2011/07/06:???????:Exadata/????(?????????????????) 2011/07/08:??????:SAP?Oracle Exadata Database Machine??? 2011/07/15:???????:Exadata/????NTT????KDDI????????????? 2011/07/19:???????:Exadata/????(?????????????????) 2011/08/29:???????:Exadata/?????????(?????????) 2011/08/29:???????:Exadata/????? 2011/08/29:???????:Exadata/????(??????????????) 2011/10/27:???????:Exadata/??? 2011/11/08:???????:Exadata/NTT??? 2012/02/16:???????:Exadata/??????????????? 2012/03/21:???????:Exadata/?????????????(?????????????????) 2012/03/22:???????:Exadata/????? ?? Oracle Database 11g R2 ??????? Oracle Database 11g ?????????(????) Oracle Database 11g ?????????(????:Exadata?) ??????? Oracle Database 11g R2(???/????) Oracle Database 11g R2 ??????? ?? ??? 2009?11?11? Oracle Exadata Database Machine Version 2 ???? 2009?11?17? Oracle Database 11g R2 ???? 2010?02?01? ?????????????????????????????? 2010?03?31? SAP ? Oracle Database 11g R2 ??????????ISV????????·??????????? 2010?05?18? Windows Server 2008 R2 / Windows 7 ?????????Oracle Database 10g R2 ??? 2010?06?23? Oracle Application Express 4.0 ???? 2010?07?09? ?? Windows RDBMS ?????(2009?)????????? 2010?08?17? TPC-C Benchmark Price/Performance ???????? 2010?09?13? Patch Set 11.2.0.2 for Linux ???? 2010?10?20? Oracle Exadata Database Machine X2 ???? 2010?11?17? Oracle Database 11g R2 ????1?? 2010?11?19? ?? Windows RDBMS ?????(2010????)????????????? 2011?03?29? Oracle SQL Developer 3.0 ???? 2011?06?27? Oracle Exadata Database Machine ????1,000????????????????·?????????????? 2011?07?08? SAP ? Oracle Exadata Database Machine ??? 2011?09?01? Oracle Database Express Edition 11g Release 2 ???? 2011?09?23? Patch Set 11.2.0.3 for Linux ???? 2011?11?14? Oracle Database Appliance ???? Oracle Database 11g ?????????(????) ????????????????????????????????(????)? ?[RAC]:Oracle Real Application Clusters(RAC) ???????? ????(??????????) [RAC] ??????????(???) ????? [RAC] ????(???) ?????·???????·??? [RAC] ????? ????·??????·?? ???? ???????(??????????????) [RAC]|???99.999%???????500???????????? - ITpro ??????????? [RAC] ????(????) [RAC] ???(???) ????????(???) [RAC] ??????(???????????) [RAC] Oracle Database 11g ?????????(????:Exadata?) ????????????????????????????????(????)? ?Exadata ??????Oracle Database 11g / Oracle Real Application Clusters(RAC) ?? ?()??????????????? KDDI(??????????????) NTT??? NTT???(???????????) ??????????????? ??? ????(??????) ????????????? ?????·???????·??? ??(??????????????) ?????(??????????) ?????????(SCSK) ?????(????????????) ??????????(???????????) ????(???????) ??????? ?????? ????/????·???????? ???????????(???????/NTT??????????) ????? ?????????????(????/????????????) ???? ????????????|?????? ????|?????????????2013?2????3??????????? - ITpro ??? ?????? ??? ?????(SCSK)|DWH?????????????? - IT Leaders ????(???????????)|???????????·??????????????????????? - oracledatabase.jp Customer Voice ????:????IT?????24??365????????????????????? ?Oracle9i Database ?????????????????????Oracle Database 11g ???????????????????????? Oracle9i Database ???????????????? Customer Voice ??????:Oracle Database 11g????????????????????? ?Oracle ASM ???????????????????I/O????????????????????????????????????? ??????? Oracle Database 11g R2(???/????) ???????????????? Oracle 11g R2 ????????? - IT Leaders ??????????11g R2?5???? - ??SE????Oracle??? - Think IT ????????????????????????~Oracle Database 11g Release2 ????????? - oracletech.jp ??????????? Oracle Database 11g Release 2(11gR2)|??????????? Oracle Exadata|??????????? ???????|???????????

  Read the article

• PostgreSQL to Data-Warehouse: Best approach for near-real-time ETL / extraction of data

  - by belvoir

  Background: I have a PostgreSQL (v8.3) database that is heavily optimized for OLTP. I need to extract data from it on a semi real-time basis (some-one is bound to ask what semi real-time means and the answer is as frequently as I reasonably can but I will be pragmatic, as a benchmark lets say we are hoping for every 15min) and feed it into a data-warehouse. How much data? At peak times we are talking approx 80-100k rows per min hitting the OLTP side, off-peak this will drop significantly to 15-20k. The most frequently updated rows are ~64 bytes each but there are various tables etc so the data is quite diverse and can range up to 4000 bytes per row. The OLTP is active 24x5.5. Best Solution? From what I can piece together the most practical solution is as follows: Create a TRIGGER to write all DML activity to a rotating CSV log file Perform whatever transformations are required Use the native DW data pump tool to efficiently pump the transformed CSV into the DW Why this approach? TRIGGERS allow selective tables to be targeted rather than being system wide + output is configurable (i.e. into a CSV) and are relatively easy to write and deploy. SLONY uses similar approach and overhead is acceptable CSV easy and fast to transform Easy to pump CSV into the DW Alternatives considered .... Using native logging (http://www.postgresql.org/docs/8.3/static/runtime-config-logging.html). Problem with this is it looked very verbose relative to what I needed and was a little trickier to parse and transform. However it could be faster as I presume there is less overhead compared to a TRIGGER. Certainly it would make the admin easier as it is system wide but again, I don't need some of the tables (some are used for persistent storage of JMS messages which I do not want to log) Querying the data directly via an ETL tool such as Talend and pumping it into the DW ... problem is the OLTP schema would need tweaked to support this and that has many negative side-effects Using a tweaked/hacked SLONY - SLONY does a good job of logging and migrating changes to a slave so the conceptual framework is there but the proposed solution just seems easier and cleaner Using the WAL Has anyone done this before? Want to share your thoughts?

  Read the article

• Fulltext search for django : Mysql not so bad ? (vs sphinx, xapian)

  - by Eric

  I am studying fulltext search engines for django. It must be simple to install, fast indexing, fast index update, not blocking while indexing, fast search. After reading many web pages, I put in short list : Mysql MYISAM fulltext, djapian/python-xapian, and django-sphinx I did not choose lucene because it seems complex, nor haystack as it has less features than djapian/django-sphinx (like fields weighting). Then I made some benchmarks, to do so, I collected many free books on the net to generate a database table with 1 485 000 records (id,title,body), each record is about 600 bytes long. From the database, I also generated a list of 100 000 existing words and shuffled them to create a search list. For the tests, I made 2 runs on my laptop (4Go RAM, Dual core 2.0Ghz): the first one, just after a server reboot to clear all caches, the second is done juste after in order to test how good are cached results. Here are the "home made" benchmark results : 1485000 records with Title (150 bytes) and body (450 bytes) Mysql 5.0.75/Ubuntu 9.04 Fulltext : ========================================================================== Full indexing : 7m14.146s 1 thread, 1000 searchs with single word randomly taken from database : First run : 0:01:11.553524 next run : 0:00:00.168508 Mysql 5.5.4 m3/Ubuntu 9.04 Fulltext : ========================================================================== Full indexing : 6m08.154s 1 thread, 1000 searchs with single word randomly taken from database : First run : 0:01:11.553524 next run : 0:00:00.168508 1 thread, 100000 searchs with single word randomly taken from database : First run : 9m09s next run : 5m38s 1 thread, 10000 random strings (random strings should not be found in database) : just after the 100000 search test : 0:00:15.007353 1 thread, boolean search : 1000 x (+word1 +word2) First run : 0:00:21.205404 next run : 0:00:00.145098 Djapian Fulltext : ========================================================================== Full indexing : 84m7.601s 1 thread, 1000 searchs with single word randomly taken from database with prefetch : First run : 0:02:28.085680 next run : 0:00:14.300236 python-xapian Fulltext : ========================================================================== 1 thread, 1000 searchs with single word randomly taken from database : First run : 0:01:26.402084 next run : 0:00:00.695092 django-sphinx Fulltext : ========================================================================== Full indexing : 1m25.957s 1 thread, 1000 searchs with single word randomly taken from database : First run : 0:01:30.073001 next run : 0:00:05.203294 1 thread, 100000 searchs with single word randomly taken from database : First run : 12m48s next run : 9m45s 1 thread, 10000 random strings (random strings should not be found in database) : just after the 100000 search test : 0:00:23.535319 1 thread, boolean search : 1000 x (word1 word2) First run : 0:00:20.856486 next run : 0:00:03.005416 As you can see, Mysql is not so bad at all for fulltext search. In addition, its query cache is very efficient. Mysql seems to me a good choice as there is nothing to install (I need just to write a small script to synchronize an Innodb production table to a MyISAM search table) and as I do not really need advanced search feature like stemming etc... Here is the question : What do you think about Mysql fulltext search engine vs sphinx and xapian ?

  Read the article

• Floating point vs integer calculations on modern hardware

  - by maxpenguin

  I am doing some performance critical work in C++, and we are currently using integer calculations for problems that are inherently floating point because "its faster". This causes a whole lot of annoying problems and adds a lot of annoying code. Now, I remember reading about how floating point calculations were so slow approximately circa the 386 days, where I believe (IIRC) that there was an optional co-proccessor. But surely nowadays with exponentially more complex and powerful CPUs it makes no difference in "speed" if doing floating point or integer calculation? Especially since the actual calculation time is tiny compared to something like causing a pipeline stall or fetching something from main memory? I know the correct answer is to benchmark on the target hardware, what would be a good way to test this? I wrote two tiny C++ programs and compared their run time with "time" on Linux, but the actual run time is too variable (doesn't help I am running on a virtual server). Short of spending my entire day running hundreds of benchmarks, making graphs etc. is there something I can do to get a reasonable test of the relative speed? Any ideas or thoughts? Am I completely wrong? The programs I used as follows, they are not identical by any means: #include <iostream> #include <cmath> #include <cstdlib> #include <time.h> int main( int argc, char** argv ) { int accum = 0; srand( time( NULL ) ); for( unsigned int i = 0; i < 100000000; ++i ) { accum += rand( ) % 365; } std::cout << accum << std::endl; return 0; } Program 2: #include <iostream> #include <cmath> #include <cstdlib> #include <time.h> int main( int argc, char** argv ) { float accum = 0; srand( time( NULL ) ); for( unsigned int i = 0; i < 100000000; ++i ) { accum += (float)( rand( ) % 365 ); } std::cout << accum << std::endl; return 0; } Thanks in advance!

  Read the article

• Profile memory-performance for part of an rails project

  - by Florian Pilz

  I want to test the profile usage of an important library-class of my rails-project. It uses ActiveRecord so I need all rails dependencies to profile it. As far as I know, I need a patched ruby (rubygc) so script/profile and script/benchmark can track memory usage. I tried to follow this official guide to patch the source code of ruby 1.8.6 (p399) and 1.8.7 (p248), but both fail with the following message: patching file gc.c Hunk #2 succeeded at 50 with fuzz 2 (offset 2 lines). Hunk #3 succeeded at 87 with fuzz 2 (offset 6 lines). Hunk #4 succeeded at 153 with fuzz 1 (offset 45 lines). Hunk #5 succeeded at 409 with fuzz 2 (offset 274 lines). Hunk #6 FAILED at 462. Hunk #7 FAILED at 506. Hunk #8 FAILED at 520. Hunk #9 FAILED at 745. Hunk #10 FAILED at 754. Hunk #11 FAILED at 923. Hunk #12 succeeded at 711 (offset 46 lines). Hunk #13 succeeded at 730 (offset 46 lines). Hunk #14 succeeded at 766 (offset 55 lines). Hunk #15 succeeded at 1428 (offset 87 lines). Hunk #16 succeeded at 1492 (offset 89 lines). Hunk #17 FAILED at 1541. Hunk #18 FAILED at 1551. Hunk #19 succeeded at 1571 (offset 91 lines). Hunk #20 succeeded at 1592 (offset 91 lines). Hunk #21 succeeded at 1601 (offset 91 lines). Hunk #22 succeeded at 1826 (offset 108 lines). Hunk #23 succeeded at 1843 (offset 108 lines). Hunk #24 succeeded at 1926 (offset 108 lines). Hunk #25 succeeded at 2118 (offset 108 lines). Hunk #26 succeeded at 2563 (offset 100 lines). Hunk #27 succeeded at 2611 with fuzz 1 (offset 102 lines). Hunk #28 succeeded at 2628 (offset 102 lines). 8 out of 28 hunks FAILED -- saving rejects to file gc.c.rej patching file intern.h Hunk #1 succeeded at 268 (offset 15 lines). I also tried to use ruby-prof, but I always get the error "uninitialized constant RubyProf::Test". I don't know how to use the gem "memory" and neither "memprof" nor "bleak_house" could be installed successfully. If I get a patched ruby running, I should be fine. But any other possibility to profile the memory of library classes are welcome. Thanks for helping!

  Read the article

• Can GPU capabilities impact virtual machine performance?

  - by Dave White

  While this many not seem like a programming question directly, it impacts my development activities and so it seems like it belongs here. It seems that more and more developers are turning to virtual environments for development activities on their computers, SharePoint development being a prime example. Also, as a trainer, I have virtual training environments for all of the classes that I teach. I recently purchased a new Dell E6510 to travel around with. It has the i7 620M (Dual core, HyperThreaded cpu running at 2.66GHz) and 8 GB of memory. Reading the spec sheet, it sounded like it would be a great laptop to carry around and run virtual machines on. Getting the laptop though, I've been pretty disappointed with the user experience of developing in a virtual machine. Giving the Virtual Machine 4 GB of memory, it was slow and I could type complete sentences and watch the VM "catchup". My company has training laptops that we provide for our classes. They are Dell Precision M6400 Intel Core 2 Duo P8700 running at 2.54Ghz with 8 GB of memory and the experience on this laptops is night and day compared to the E6510. They are crisp and you barely aware that you are running in a virtual environment. Since the E6510 should be faster in all categories than the M6400, I couldn't understand why the new laptop was slower, so I did a component by component comparison and the only place where the E6510 is less performant than the M6400 is the graphics department. The M6400 is running a nVidia FX 2700m GPU and the E6510 is running a nVidia 3100M GPU. Looking at benchmarks of the two GPUs suggest that the FX 2700M is twice as fast as the 3100M. http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html 3100M = 111th (E6510) FX 2700m = 47th (Precision M6400) Radeon HD 5870 = 8th (Alienware) The host OS is Windows 7 64bit as is the guest OS, running in Virtual Box 3.1.8 with Guest Additions installed on the guest. The IDE being used in the virtual environment is VS 2010 Premium. So after that long setup, my question is: Is the GPU significantly impacting the virtual machine's performance or are there other factors that I'm not looking at that I can use to boost the vm's performance? Do we now have to consider GPU performance when purchasing laptops where we expect to use virtualized development environments? Thanks in advance. Cheers, Dave

  Read the article

• AJAX response not valid in C++ but Apache

  - by fehergeri

  I want to make a server written in C++ to power my game. I learned the basics of sockets and wrote a basic chat program that worked well. Now I want to create an HTTP server like Apache, but only for the AJAX request-response part. I think just for the beginning i copied one Apache response text, and i sent the exact response with the C++ server program. The problem that is that the browser (Firefox) connnects to the apache and everything works fine, except all of the requests get a correct response. But if i send this with the C++ client, then FireBug tells me that the response status is OK (200) but there is no actual response text. (How is this possible?) This response-text is exactly the same what apache sends. I made a bit-bit comparison and they were the same. The php file wich is the original response <?php echo "AS";echo rand(0,9); ?> And the origional source code: Socket.h http://pastebin.com/bW9qxtrR Socket.cpp http://pastebin.com/S3c8RFM7 main.cpp http://pastebin.com/ckExuXsR index.html http://pastebin.com/mcfEEqPP < this is the requester file. ajax.js http://pastebin.com/uXJe9hVC benchmark.js http://pastebin.com/djSYtKg9 jQuery is not needed. The main.cpp there is lot of trash code like main3 and main4 functions, these do not affect the result. I know that the response stuff in the C++ code is not really good because the connection closing is not the best; I will fix that later now I want to send a success response first. UPDATE: now i tested today a lot again and i find out there is no problem with the socket. I used the fiddler program to capture the the good answer and to capture the bad. They were the same. After this i turned off my socket application, and forced fiddler to auto respond, and the answer from the 'bad' answer still bat. So after that i replaced the bad with the good and nothing happedned. The bad answer with the good text still bad on the :8888 port but the other on the original :80 port was good, but they were absolutly the same and the same program sended it (fiddler) i think there is something missing if the response is not on the same server address (even not the same port). UPDATE: oh my god! i cant send ajax request to a remote server. now i know this.

  Read the article

• Slow MySQL query....only sometimes

  - by Shane N

  I have a query that's used in a reporting system of ours that sometimes runs quicker than a second, and other times takes 1 to 10 minutes to run. Here's the entry from the slow query log: # Query_time: 543 Lock_time: 0 Rows_sent: 0 Rows_examined: 124948974 use statsdb; SELECT count(distinct Visits.visitorid) as 'uniques' FROM Visits,Visitors WHERE Visits.visitorid=Visitors.visitorid and candidateid in (32) and visittime>=1275721200 and visittime<=1275807599 and (omit=0 or omit>=1275807599) AND Visitors.segmentid=9 AND Visits.visitorid NOT IN (SELECT Visits.visitorid FROM Visits,Visitors WHERE Visits.visitorid=Visitors.visitorid and candidateid in (32) and visittime<1275721200 and (omit=0 or omit>=1275807599) AND Visitors.segmentid=9); It's basically counting unique visitors, and it's doing that by counting the visitors for today and then substracting those that have been here before. If you know of a better way to do this, let me know. I just don't understand why sometimes it can be so quick, and other times takes so long - even with the same exact query under the same server load. Here's the EXPLAIN on this query. As you can see it's using the indexes I've set up: id select_type table type possible_keys key key_len ref rows Extra 1 PRIMARY Visits range visittime_visitorid,visitorid visittime_visitorid 4 NULL 82500 Using where; Using index 1 PRIMARY Visitors eq_ref PRIMARY,cand_visitor_omit PRIMARY 8 statsdb.Visits.visitorid 1 Using where 2 DEPENDENT SUBQUERY Visits ref visittime_visitorid,visitorid visitorid 8 func 1 Using where 2 DEPENDENT SUBQUERY Visitors eq_ref PRIMARY,cand_visitor_omit PRIMARY 8 statsdb.Visits.visitorid 1 Using where I tried to optimize the query a few weeks ago and came up with a variation that consistently took about 2 seconds, but in practice it ended up taking more time since 90% of the time the old query returned much quicker. Two seconds per query is too long because we are calling the query up to 50 times per page load, with different time periods. Could the quick behavior be due to the query being saved in the query cache? I tried running 'RESET QUERY CACHE' and 'FLUSH TABLES' between my benchmark tests and I was still getting quick results most of the time. Note: last night while running the query I got an error: Unable to save result set. My initial research shows that may be due to a corrupt table that needs repair. Could this be the reason for the behavior I'm seeing? In case you want server info: Accessing via PHP 4.4.4 MySQL 4.1.22 All tables are InnoDB We run optimize table on all tables weekly The sum of both the tables used in the query is 500 MB MySQL config: key_buffer = 350M max_allowed_packet = 16M thread_stack = 128K sort_buffer = 14M read_buffer = 1M bulk_insert_buffer_size = 400M set-variable = max_connections=150 query_cache_limit = 1048576 query_cache_size = 50777216 query_cache_type = 1 tmp_table_size = 203554432 table_cache = 120 thread_cache_size = 4 wait_timeout = 28800 skip-external-locking innodb_file_per_table innodb_buffer_pool_size = 3512M innodb_log_file_size=100M innodb_log_buffer_size=4M

  Read the article

• Local Variables take 7x longer to access than global variables?

  - by ItzWarty

  I was trying to benchmark the gain/loss of "caching" math.floor, in hopes that I could make calls faster. Here was the test: <html> <head> <script> window.onload = function() { var startTime = new Date().getTime(); var k = 0; for(var i = 0; i < 1000000; i++) k += Math.floor(9.99); var mathFloorTime = new Date().getTime() - startTime; startTime = new Date().getTime(); window.mfloor = Math.floor; k = 0; for(var i = 0; i < 1000000; i++) k += window.mfloor(9.99); var globalFloorTime = new Date().getTime() - startTime; startTime = new Date().getTime(); var mfloor = Math.floor; k = 0; for(var i = 0; i < 1000000; i++) k += mfloor(9.99); var localFloorTime = new Date().getTime() - startTime; document.getElementById("MathResult").innerHTML = mathFloorTime; document.getElementById("globalResult").innerHTML = globalFloorTime; document.getElementById("localResult").innerHTML = localFloorTime; }; </script> </head> <body> Math.floor: <span id="MathResult"></span>ms <br /> var mathfloor: <span id="globalResult"></span>ms <br /> window.mathfloor: <span id="localResult"></span>ms <br /> </body> </html> My results from the test: [Chromium 5.0.308.0]: Math.floor: 49ms var mathfloor: 271ms window.mathfloor: 40ms [IE 8.0.6001.18702] Math.floor: 703ms var mathfloor: 9890ms [LOL!] window.mathfloor: 375ms [Firefox [Minefield] 3.7a4pre] Math.floor: 42ms var mathfloor: 2257ms window.mathfloor: 60ms [Safari 4.0.4[531.21.10] ] Math.floor: 92ms var mathfloor: 289ms window.mathfloor: 90ms [Opera 10.10 build 1893] Math.floor: 500ms var mathfloor: 843ms window.mathfloor: 360ms [Konqueror 4.3.90 [KDE 4.3.90 [KDE 4.4 RC1]]] Math.floor: 453ms var mathfloor: 563ms window.mathfloor: 312ms The variance is random, of course, but for the most part In all cases [this shows time taken]: [takes longer] mathfloor Math.floor window.mathfloor [is faster] Why is this? In my projects i've been using var mfloor = Math.floor, and according to my not-so-amazing benchmarks, my efforts to "optimize" actually slowed down the script by ALOT... Is there any other way to make my code more "efficient"...? I'm at the stage where i basically need to optimize, so no, this isn't "premature optimization"...

  Read the article

< Previous Page | 15 16 17 18 19 20 21  | Next Page >