Search Results

Search found 48586 results on 1944 pages for 'page performance'.

facebook twitter digg google delicious technorati stumbleupon myspace wordpress linkedin gmail igoogle windows live tumblr viadeo yahoo buzz yahoo mail yahoo bookmarks favorites email print

Page 46/1944 | < Previous Page | 42 43 44 45 46 47 48 49 50 51 52 53  | Next Page >

  • Improve disk read performance (multiple files) with threading

    - by pablo
    I need to find a method to read a big number of small files (about 300k files) as fast as possible. Reading them sequentially using FileStream and reading the entire file in a single call takes between 170 and 208 seconds (you know, you re-run, disk cache plays its role and time varies). Then I tried using PInvoke with CreateFile/ReadFile and using FILE_FLAG_SEQUENTIAL_SCAN, but I didn't appreciate any changes. I tried with several threads (divide the big set in chunks and have every thread reading its part) and this way I was able to improve speed just a little bit (not even a 5% with every new thread up to 4). Any ideas on how to find the most effective way to do this?

    Read the article

  • Performance of Serialized Objects in C++

    - by jm1234567890
    Hi Everyone, I'm wondering if there is a fast way to dump an STL set to disk and then read it back later. The internal structure of a set is a binary tree, so if I serialize it naively, when I read it back the program will have to go though the process of inserting each element again. I think this is slow even if it is read back in correct order, correct me if I am wrong. Is there a way to "dump" the memory containing the set into disk and then read it back later? That is, keep everything in binary format, thus avoiding the re-insertion. Do the boost serialization tools do this? Thanks! EDIT: oh I should probably read, http://www.parashift.com/c++-faq-lite/serialization.html I will read it now... no it doesn't really help

    Read the article

  • postgres min function performance

    - by wutzebaer
    hi i need the lowest value for runnerId this query: SELECT "runnerId" FROM betlog WHERE "marketId" = '107416794' ; takes 80ms (1968 result rows) this SELECT min("runnerId") FROM betlog WHERE "marketId" = '107416794' ; takes 1600ms is there a faster way to find the minimum, or should i calc the min in my java programm? "Result (cost=100.88..100.89 rows=1 width=0)" " InitPlan 1 (returns $0)" " -> Limit (cost=0.00..100.88 rows=1 width=9)" " -> Index Scan using runneridindex on betlog (cost=0.00..410066.33 rows=4065 width=9)" " Index Cond: ("runnerId" IS NOT NULL)" " Filter: ("marketId" = 107416794::bigint)" CREATE INDEX marketidindex ON betlog USING btree ("marketId" COLLATE pg_catalog."default"); another idea SELECT "runnerId" FROM betlog WHERE "marketId" = '107416794' ORDER BY "runnerId" LIMIT 1 >1600ms SELECT "runnerId" FROM betlog WHERE "marketId" = '107416794' ORDER BY "runnerId" >>100ms how can a limit slow the query down?

    Read the article

  • C# Performance on Errors

    - by pm_2
    It would appear that catching an error is slower that performing a check prior to the error (for example a TryParse). The related questions that prompt this observation are here and here. Can anyone tell me why this is so - why is it more costly to catch an error that to perform one or many checks of the data to prevent the error?

    Read the article

  • Objective-C vs JavaScript loop performance

    - by micadelli
    I have a PhoneGap mobile application that I need to generate an array of match combinations. In JavaScript side, the code hanged pretty soon when the array of which the combinations are generated from got a bit bigger. So, I thought I'll make a plugin to generate the combinations, passing the array of javascript objects to native side and loop it there. To my surprise the following codes executes in 150 ms (JavaScript) whereas in native side (Objective-C) it takes ~1000 ms. Does anyone know any tips for speeding up those executing times? When players exceeds 10, i.e. the length of the array of teams equals 252 it really gets slow. Those execution times mentioned above are for 10 players / 252 teams. Here's the JavaScript code: for (i = 0; i < GAME.teams.length; i += 1) { for (j = i + 1; j < GAME.teams.length; j += 1) { t1 = GAME.teams[i]; t2 = GAME.teams[j]; if ((t1.mask & t2.mask) === 0) { GAME.matches.push({ Team1: t1, Team2: t2 }); } } } ... and here's the native code: NSArray *teams = [[NSArray alloc] initWithArray: [options objectForKey:@"teams"]]; NSMutableArray *t = [[NSMutableArray alloc] init]; int mask_t1; int mask_t2; for (NSInteger i = 0; i < [teams count]; i++) { for (NSInteger j = i + 1; j < [teams count]; j++) { mask_t1 = [[[teams objectAtIndex:i] objectForKey:@"mask"] intValue]; mask_t2 = [[[teams objectAtIndex:j] objectForKey:@"mask"] intValue]; if ((mask_t1 & mask_t2) == 0) { [t insertObject:[teams objectAtIndex:i] atIndex:0]; [t insertObject:[teams objectAtIndex:j] atIndex:1]; /* NSArray *newCombination = [[NSArray alloc] initWithObjects: [teams objectAtIndex:i], [teams objectAtIndex:j], nil]; */ [combinations addObject:t]; } } } ... the array in question (GAME.teams) looks like this: { count = 2; full = 1; list = ( { index = 0; mask = 1; name = A; score = 0; }, { index = 1; mask = 2; name = B; score = 0; } ); mask = 3; name = A; }, { count = 2; full = 1; list = ( { index = 0; mask = 1; name = A; score = 0; }, { index = 2; mask = 4; name = C; score = 0; } ); mask = 5; name = A; },

    Read the article

  • Python performance profiling (file close)

    - by user1853986
    First of all thanks for your attention. My question is how to reduce the execution time of my code. Here is the relevant code. The below code is called in iteration from the main. def call_prism(prism_input_file,random_length): prism_output_file = "path.txt" cmd = "prism %s -simpath %d %s" % (prism_input_file,random_length,prism_output_file) p = os.popen(cmd) p.close() return prism_output_file def main(prism_input_file, number_of_strings): ... for n in range(number_of_strings): prism_output_file = call_prism(prism_input_file,z[n]) ... return I used statistics from the "profile statistics browser" when I profiled my code. The "file close" system command took the maximum time (14.546 seconds). The call_prism routine is called 10 times. But the number_of_strings is usually in thousands, so, my program takes lot of time to complete. Let me know if you need more information. By the way I tried with subprocess, too. Thanks.

    Read the article

  • How can I improve the performance of this algorithm

    - by Justin
    // Checks whether the array contains two elements whose sum is s. // Input: A list of numbers and an integer s // Output: return True if the answer is yes, else return False public static boolean calvalue (int[] numbers, int s){ for (int i=0; i< numbers.length; i++){ for (int j=i+1; j<numbers.length;j++){ if (numbers[i] < s){ if (numbers[i]+numbers[j] == s){ return true; } } } } return false; }

    Read the article

  • PHP array performance

    - by dfo
    Hi, this is my first question on Stackoverflow, please bear with me. I'm testing an algorithm for 2d bin packing and I've chosen PHP to mock it up as it's my bread-and-butter language nowadays. As you can see on http://themworks.com/pack_v0.2/oopack.php?ol=1 it works pretty well, but you need to wait around 10-20 seconds for 100 rectangles to pack. For some hard to handle sets it would hit the php's 30s runtime limit. I did some profiling and it shows that most of the time my script goes through different parts of a small 2d array with 0's and 1's in it. It either checks if certain cell equals to 0/1 or sets it to 0/1. It can do such operations million times and each times it takes few microseconds. I guess I could use an array of booleans in a statically typed language and things would be faster. Or even make an array of 1 bit values. I'm thinking of converting the whole thing to some compiled language. Is PHP just not good for it? If I do need to convert it to let's say C++, how good are the automatic converters? My script is just a lot of for loops with basic arrays and objects manipulations. Thank you! Edit. This function gets called more than any other. It reads few properties of a very simple object, and goes through a very small part of a smallish array to check if there's any element not equal to 0. function fits($bin, $file, $x, $y) { $flag = true; $xw = $x + $file->get_width();; $yh = $y + $file->get_height(); for ($i = $x; $i < $xw; $i++) { for ($j = $y; $j < $yh; $j++) { if ($bin[$i][$j] !== 0) { $flag = false; break; } } if (!$flag) break; } return $flag; }

    Read the article

  • C/C++ variable length automatic array performance

    - by aaa
    hello. Is there significant cpu/memory overhead associated with using automatic arrays with g++/Intel on 64-bit x86 linux platform? int function(int N) { double array[N]; overhead compared to allocating array before hand (assuming function is called multiple times) overhead compared to using new overhead compared to using malloc range of N maybe from 1kb to 16kb roughly, stack overrun is not a problem Thank you

    Read the article

  • Which should I use? (performance)

    - by Yim
    I want to know a simple thing: when setting up a style that is inherited by all its children, is it recommended most specific? (even if you don't care others having this style) Structure: html body parent_content wrapper p I don't care having parent_content or wrapper having the style I do care changing the html or body style (or all p) So what should I use? #parent_content{ color:#555; } #parent_content p{ color:#555; } #wrapper{ color:#555; } ... Also, some links to tutorials about this would be great

    Read the article

  • Will logging debugging incur a performance hit if I don't turn debugging on?

    - by romandas
    On a Cisco device, I know that enabling debugging can incur a performance hit since debugging has such a high priority on the CPU. I know that to log debugging, you have to set logging up to the debugging level (logging buffered 4096 debugging, for example) and also enable debugging on some feature. Does configuring the logging debugging incur the performance hit even if you don't enable debugging on some feature, or would it be safe (assuming you want and can handle all the logging events via syslog) to configure 'logging buffered 4096 debugging' to have maximum logging available if/when someone uses debug?

    Read the article

  • Sysadmin 101: How can I figure out why my server crashes and monitor performance?

    - by bflora
    I have a Drupal-powered site that seems to have neverending performance problems. It was butt-slow about 5 months ago. I brought in some guys who installed nginx for anonymous visitors, ajaxified a few queries so they wouldn't fire during page load, and helped me find a few bottlenecks in the code. For about a month, the site was significantly faster, though not "fast" by any stretch of the word. Meanwhile, I'm now shelling out $400/month to Slicehost to host a site that gets less than 5,000/uniques a day. Yes, you read that right. Go Drupal. Recently the site started crashing again and is slow again. I can't afford to hire people to come in, study my code from top to bottom, and make changes that may or may not help anymore. And I can't afford to throw more hardware at the problem. So I need to figure out what the problem is myself. Questions: When apache crashes, is it possible to find out what caused it to crash? There has to be a way, right? If so, how can I do this? Is there software I can use that will tell me which process caused my server to die? (e.g. "Apache crashed because someone visited page X." or "Apache crashed because you were importing too many RSS items from feed X.") There's got to be a way to learn this, right? What's a good, noob-friendly way to monitor my current apache performance? My developer friends tell me to "just use Top, dude," but Top shows me a bunch of numbers without any context. I have no clue what qualifies as a bad number or a good number in Top, or which processes are relevant and which aren't. Are there any noob-friendly server monitoring tools out there? Ideally, I could have a page that would give me a color-coded indicator about how apache is performing and then show me a list of processes or pages that are sucking right now. This way, I could know when performance is bad and then what's causing it to be so bad. Why does PHP memory matter? My apparently has a 30MB memory foot print. Will it run faster if I bring that number down? Thanks for any advice. I spent a year or so trying to boost my advertising income so I could hire a contractor to solve my performance woes. I didn't want to have to learn all this sysadmin voodoo. I'm now resigned to the fact that might not have a choice.

    Read the article

  • How to increase performance of Acer Aspire One 751h netbook?

    - by Wolfarian
    Hello! I have bought my new netbook Acer Aspire One 751h some days ago and was very unpleased with it performance - videotalking in skype is almost unuseable, watching videos on YouTube(even in standart definition) is like watching slideshow and all netbook have increadible lags if I'm running more then 4-5 programms in one time. So, can somebody tell me how to impruve the performance of the netbook(OS - WinXP SP3)? And can you say me where to control power managment, please? Thank you!

    Read the article

  • Weird nfs performance: 1 thread better than 8, 8 better than 2!

    - by Joe
    I'm trying to determine the cause of poor nfs performance between two Xen Virtual Machines (client & server) running on the same host. Specifically, the speed at which I can sequentially read a 1GB file on the client is much lower than what would be expected based on the measured network connection speed between the two VMs and the measured speed of reading the file directly on the server. The VMs are running Ubuntu 9.04 and the server is using the nfs-kernel-server package. According to various NFS tuning resources, changing the number of nfsd threads (in my case kernel threads) can affect performance. Usually this advice is framed in terms of increasing the number from the default of 8 on heavily-used servers. What I find in my current configuration: RPCNFSDCOUNT=8: (default): 13.5-30 seconds to cat a 1GB file on the client so 35-80MB/sec RPCNFSDCOUNT=16: 18s to cat the file 60MB/s RPCNFSDCOUNT=1: 8-9 seconds to cat the file (!!?!) 125MB/s RPCNFSDCOUNT=2: 87s to cat the file 12MB/s I should mention that the file I'm exporting is on a RevoDrive SSD mounted on the server using Xen's PCI-passthrough; on the server I can cat the file in under seconds ( 250MB/s). I am dropping caches on the client before each test. I don't really want to leave the server configured with just one thread as I'm guessing that won't work so well when there are multiple clients, but I might be misunderstanding how that works. I have repeated the tests a few times (changing the server config in between) and the results are fairly consistent. So my question is: why is the best performance with 1 thread? A few other things I have tried changing, to little or no effect: increasing the values of /proc/sys/net/ipv4/ipfrag_low_thresh and /proc/sys/net/ipv4/ipfrag_high_thresh to 512K, 1M from the default 192K,256K increasing the value of /proc/sys/net/core/rmem_default and /proc/sys/net/core/rmem_max to 1M from the default of 128K mounting with client options rsize=32768, wsize=32768 From the output of sar -d I understand that the actual read sizes going to the underlying device are rather small (<100 bytes) but this doesn't cause a problem when reading the file locally on the client. The RevoDrive actually exposes two "SATA" devices /dev/sda and /dev/sdb, then dmraid picks up a fakeRAID-0 striped across them which I have mounted to /mnt/ssd and then bind-mounted to /export/ssd. I've done local tests on my file using both locations and see the good performance mentioned above. If answers/comments ask for more details I will add them.

    Read the article

  • Performance boost for MacBook: Hybrid hard drive or 4GB RAM?

    - by user13572
    I have an aluminium 13" MacBook with 2GB or RAM and 5400RPM 500GB hard drive. The main tasks I perform are developing iPhone and Mac apps in Xcode and websites in Coda. I want to improve the performance so I am considering buying 4GB of RAM or a 500GB Seagate solid-state hybrid drive. What is likely to provide the biggest performance boost?

    Read the article

  • Does chunk size affect the read performance of a Linux md software RAID1 array?

    - by OldWolf
    This came up in relation to this question on determining chunk size of an existing RAID array. The general consensus seems to be that chunk size does not apply to RAID1 as it is not striped. On the other hand, the Linux RAID Wiki claims that it will have an affect on read performance. However, I cannot find any benchmarks testing/proving that. Can anyone point to conclusive documentation that it either does or does not affect read performance?

    Read the article

  • Performance boast for MacBook: Hybrid hard drive or 4GB RAM?

    - by user13572
    I have an aluminium 13" MacBook with 2GB or RAM and 5400RPM 500GB hard drive. The main tasks I perform are developing iPhone and Mac apps in Xcode and websites in Coda. I want to improve the performance so I am considering buying 4GB of RAM or a 500GB Seagate solid-state hybrid drive. What is likely to provide the biggest performance boast?

    Read the article

  • JavaScript functions broke my page jumps?

    - by Mark
    Hi everybody, I'm trying to develop a personal website and running into some issues. At the top of my page I had a horizontal table that contained 6 different images (buttons) that linked to different sections of the page. I wanted to spice up the buttons with JavaScript and make the image of the button change onMouseOver and onMouseOut so I found a way to do so online with some JavaScript but the problem now is that the buttons no longer jump to the appropriate section of the page when using Google Chrome or Safari, but in IE 9 they still do work correctly... Here is my old code that worked properly: <html> ... <body> <center> <table border="1" style="color:#D80000;text-align:center"> <tr> <td> <a href="#about"><img src="images/aboutMe.png" alt="About Me"></a> </td> <td> <a href="#courses"><img src="images/courseList.png" alt="Courses"></a> </td> <td> <a href="#work"><img src="images/workHistory.png" alt="Work History"></a> </td> <td> <a href="#places"><img src="images/placesBeen.png" alt="Places I've Been"></a> </td> <td> <a href="#games"><img src="images/gamesPlay.png" alt="Games I Play"></a> </td> <td> <a href="#contact"><img src="images/contactMe.png" alt="Contact"></a> </td> </tr> </table> </center> ... </body> </html> Here is my new code that is now broken: <html> <head> ... <script type="text/javascript"> <!-- if (document.images) { <!-- Preload original Images --> var about_init= new Image(); about_init.src="images/aboutMe.png"; var courses_init= new Image(); courses_init.src="images/courseList.png"; var work_init= new Image(); work_init.src="images/workHistory.png"; var places_init= new Image(); places_init.src="images/placesBeen.png"; var games_init= new Image(); games_init.src="images/gamesPlay.png"; var contact_init= new Image(); contact_init.src="images/contactMe.png"; <!--Preload images for mouseover --> var about_new= new Image(); about_new.src="images/aboutAlt.png"; var courses_new= new Image(); courses_new.src="images/courseAlt.png"; var work_new= new Image(); work_new.src="images/workAlt.png"; var places_new= new Image(); places_new.src="images/placesAlt.png"; var games_new= new Image(); games_new.src="images/gamesAlt.png"; var contact_new= new Image(); contact_new.src="images/contactAlt.png"; } function change_it(the_name) { if (document.images) { document.images[the_name].src= eval(the_name+"_new.src"); } } function change_back(the_name) { if (document.images) { document.images[the_name].src= eval(the_name+"_init.src"); } } //--> </script> </head> <body> ... <center> <table border="1" style="color:#D80000;text-align:center"> <tr> <td> <a href="#about" onMouseOver="change_it('about')" onMouseOut="change_back('about')"><img src="images/aboutMe.png" name="about" id="about" border="0" alt="About Me"></a> </td> <td> <a href="#courses" onMouseOver="change_it('courses')" onMouseOut="change_back('courses')"><img src="images/courseList.png" name="courses" id="courses" border="0" alt="Courses"></a> </td> <td> <a href="#work" onMouseOver="change_it('work')" onMouseOut="change_back('work')"><img src="images/workHistory.png" name="work" id="work" border="0" alt="Work History"></a> </td> <td> <a href="#places" onMouseOver="change_it('places')" onMouseOut="change_back('places')"><img src="images/placesBeen.png" name="places" id="places" border="0" alt="Places I've Been"></a> </td> <td> <a href="#games" onMouseOver="change_it('games')" onMouseOut="change_back('games')"><img src="images/gamesPlay.png" name="games" id="games" border="0" alt="Games I Play"></a> </td> <td> <a href="#contact" onMouseOver="change_it('contact')" onMouseOut="change_back('contact')"><img src="images/contactMe.png" name="contact" id="contact" border="0" alt="Contact"></a> </td> </tr> </table> </center> ... </body> </html> The images properly switch when the mouse is moved over/off them, but the jump to the corresponding page sections seems to be broken, it seems like they're all jumping to the #about section now. Once again, the weirder thing is that this is only the case when I'm using Google Chrome on my laptop running Windows 7 or Safari using my iPhone 4 but when I use the new Internet Explorer 9 on my laptop running Windows 7 all the buttons still link correctly. Any idea what's breaking the links?

    Read the article

  • Improving Partitioned Table Join Performance

    - by Paul White
    The query optimizer does not always choose an optimal strategy when joining partitioned tables. This post looks at an example, showing how a manual rewrite of the query can almost double performance, while reducing the memory grant to almost nothing. Test Data The two tables in this example use a common partitioning partition scheme. The partition function uses 41 equal-size partitions: CREATE PARTITION FUNCTION PFT (integer) AS RANGE RIGHT FOR VALUES ( 125000, 250000, 375000, 500000, 625000, 750000, 875000, 1000000, 1125000, 1250000, 1375000, 1500000, 1625000, 1750000, 1875000, 2000000, 2125000, 2250000, 2375000, 2500000, 2625000, 2750000, 2875000, 3000000, 3125000, 3250000, 3375000, 3500000, 3625000, 3750000, 3875000, 4000000, 4125000, 4250000, 4375000, 4500000, 4625000, 4750000, 4875000, 5000000 ); GO CREATE PARTITION SCHEME PST AS PARTITION PFT ALL TO ([PRIMARY]); There two tables are: CREATE TABLE dbo.T1 ( TID integer NOT NULL IDENTITY(0,1), Column1 integer NOT NULL, Padding binary(100) NOT NULL DEFAULT 0x,   CONSTRAINT PK_T1 PRIMARY KEY CLUSTERED (TID) ON PST (TID) );   CREATE TABLE dbo.T2 ( TID integer NOT NULL, Column1 integer NOT NULL, Padding binary(100) NOT NULL DEFAULT 0x,   CONSTRAINT PK_T2 PRIMARY KEY CLUSTERED (TID, Column1) ON PST (TID) ); The next script loads 5 million rows into T1 with a pseudo-random value between 1 and 5 for Column1. The table is partitioned on the IDENTITY column TID: INSERT dbo.T1 WITH (TABLOCKX) (Column1) SELECT (ABS(CHECKSUM(NEWID())) % 5) + 1 FROM dbo.Numbers AS N WHERE n BETWEEN 1 AND 5000000; In case you don’t already have an auxiliary table of numbers lying around, here’s a script to create one with 10 million rows: CREATE TABLE dbo.Numbers (n bigint PRIMARY KEY);   WITH L0 AS(SELECT 1 AS c UNION ALL SELECT 1), L1 AS(SELECT 1 AS c FROM L0 AS A CROSS JOIN L0 AS B), L2 AS(SELECT 1 AS c FROM L1 AS A CROSS JOIN L1 AS B), L3 AS(SELECT 1 AS c FROM L2 AS A CROSS JOIN L2 AS B), L4 AS(SELECT 1 AS c FROM L3 AS A CROSS JOIN L3 AS B), L5 AS(SELECT 1 AS c FROM L4 AS A CROSS JOIN L4 AS B), Nums AS(SELECT ROW_NUMBER() OVER (ORDER BY (SELECT NULL)) AS n FROM L5) INSERT dbo.Numbers WITH (TABLOCKX) SELECT TOP (10000000) n FROM Nums ORDER BY n OPTION (MAXDOP 1); Table T1 contains data like this: Next we load data into table T2. The relationship between the two tables is that table 2 contains ‘n’ rows for each row in table 1, where ‘n’ is determined by the value in Column1 of table T1. There is nothing particularly special about the data or distribution, by the way. INSERT dbo.T2 WITH (TABLOCKX) (TID, Column1) SELECT T.TID, N.n FROM dbo.T1 AS T JOIN dbo.Numbers AS N ON N.n >= 1 AND N.n <= T.Column1; Table T2 ends up containing about 15 million rows: The primary key for table T2 is a combination of TID and Column1. The data is partitioned according to the value in column TID alone. Partition Distribution The following query shows the number of rows in each partition of table T1: SELECT PartitionID = CA1.P, NumRows = COUNT_BIG(*) FROM dbo.T1 AS T CROSS APPLY (VALUES ($PARTITION.PFT(TID))) AS CA1 (P) GROUP BY CA1.P ORDER BY CA1.P; There are 40 partitions containing 125,000 rows (40 * 125k = 5m rows). The rightmost partition remains empty. The next query shows the distribution for table 2: SELECT PartitionID = CA1.P, NumRows = COUNT_BIG(*) FROM dbo.T2 AS T CROSS APPLY (VALUES ($PARTITION.PFT(TID))) AS CA1 (P) GROUP BY CA1.P ORDER BY CA1.P; There are roughly 375,000 rows in each partition (the rightmost partition is also empty): Ok, that’s the test data done. Test Query and Execution Plan The task is to count the rows resulting from joining tables 1 and 2 on the TID column: SET STATISTICS IO ON; DECLARE @s datetime2 = SYSUTCDATETIME();   SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID;   SELECT DATEDIFF(Millisecond, @s, SYSUTCDATETIME()); SET STATISTICS IO OFF; The optimizer chooses a plan using parallel hash join, and partial aggregation: The Plan Explorer plan tree view shows accurate cardinality estimates and an even distribution of rows across threads (click to enlarge the image): With a warm data cache, the STATISTICS IO output shows that no physical I/O was needed, and all 41 partitions were touched: Running the query without actual execution plan or STATISTICS IO information for maximum performance, the query returns in around 2600ms. Execution Plan Analysis The first step toward improving on the execution plan produced by the query optimizer is to understand how it works, at least in outline. The two parallel Clustered Index Scans use multiple threads to read rows from tables T1 and T2. Parallel scan uses a demand-based scheme where threads are given page(s) to scan from the table as needed. This arrangement has certain important advantages, but does result in an unpredictable distribution of rows amongst threads. The point is that multiple threads cooperate to scan the whole table, but it is impossible to predict which rows end up on which threads. For correct results from the parallel hash join, the execution plan has to ensure that rows from T1 and T2 that might join are processed on the same thread. For example, if a row from T1 with join key value ‘1234’ is placed in thread 5’s hash table, the execution plan must guarantee that any rows from T2 that also have join key value ‘1234’ probe thread 5’s hash table for matches. The way this guarantee is enforced in this parallel hash join plan is by repartitioning rows to threads after each parallel scan. The two repartitioning exchanges route rows to threads using a hash function over the hash join keys. The two repartitioning exchanges use the same hash function so rows from T1 and T2 with the same join key must end up on the same hash join thread. Expensive Exchanges This business of repartitioning rows between threads can be very expensive, especially if a large number of rows is involved. The execution plan selected by the optimizer moves 5 million rows through one repartitioning exchange and around 15 million across the other. As a first step toward removing these exchanges, consider the execution plan selected by the optimizer if we join just one partition from each table, disallowing parallelism: SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = 1 AND $PARTITION.PFT(T2.TID) = 1 OPTION (MAXDOP 1); The optimizer has chosen a (one-to-many) merge join instead of a hash join. The single-partition query completes in around 100ms. If everything scaled linearly, we would expect that extending this strategy to all 40 populated partitions would result in an execution time around 4000ms. Using parallelism could reduce that further, perhaps to be competitive with the parallel hash join chosen by the optimizer. This raises a question. If the most efficient way to join one partition from each of the tables is to use a merge join, why does the optimizer not choose a merge join for the full query? Forcing a Merge Join Let’s force the optimizer to use a merge join on the test query using a hint: SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID OPTION (MERGE JOIN); This is the execution plan selected by the optimizer: This plan results in the same number of logical reads reported previously, but instead of 2600ms the query takes 5000ms. The natural explanation for this drop in performance is that the merge join plan is only using a single thread, whereas the parallel hash join plan could use multiple threads. Parallel Merge Join We can get a parallel merge join plan using the same query hint as before, and adding trace flag 8649: SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID OPTION (MERGE JOIN, QUERYTRACEON 8649); The execution plan is: This looks promising. It uses a similar strategy to distribute work across threads as seen for the parallel hash join. In practice though, performance is disappointing. On a typical run, the parallel merge plan runs for around 8400ms; slower than the single-threaded merge join plan (5000ms) and much worse than the 2600ms for the parallel hash join. We seem to be going backwards! The logical reads for the parallel merge are still exactly the same as before, with no physical IOs. The cardinality estimates and thread distribution are also still very good (click to enlarge): A big clue to the reason for the poor performance is shown in the wait statistics (captured by Plan Explorer Pro): CXPACKET waits require careful interpretation, and are most often benign, but in this case excessive waiting occurs at the repartitioning exchanges. Unlike the parallel hash join, the repartitioning exchanges in this plan are order-preserving ‘merging’ exchanges (because merge join requires ordered inputs): Parallelism works best when threads can just grab any available unit of work and get on with processing it. Preserving order introduces inter-thread dependencies that can easily lead to significant waits occurring. In extreme cases, these dependencies can result in an intra-query deadlock, though the details of that will have to wait for another time to explore in detail. The potential for waits and deadlocks leads the query optimizer to cost parallel merge join relatively highly, especially as the degree of parallelism (DOP) increases. This high costing resulted in the optimizer choosing a serial merge join rather than parallel in this case. The test results certainly confirm its reasoning. Collocated Joins In SQL Server 2008 and later, the optimizer has another available strategy when joining tables that share a common partition scheme. This strategy is a collocated join, also known as as a per-partition join. It can be applied in both serial and parallel execution plans, though it is limited to 2-way joins in the current optimizer. Whether the optimizer chooses a collocated join or not depends on cost estimation. The primary benefits of a collocated join are that it eliminates an exchange and requires less memory, as we will see next. Costing and Plan Selection The query optimizer did consider a collocated join for our original query, but it was rejected on cost grounds. The parallel hash join with repartitioning exchanges appeared to be a cheaper option. There is no query hint to force a collocated join, so we have to mess with the costing framework to produce one for our test query. Pretending that IOs cost 50 times more than usual is enough to convince the optimizer to use collocated join with our test query: -- Pretend IOs are 50x cost temporarily DBCC SETIOWEIGHT(50);   -- Co-located hash join SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID OPTION (RECOMPILE);   -- Reset IO costing DBCC SETIOWEIGHT(1); Collocated Join Plan The estimated execution plan for the collocated join is: The Constant Scan contains one row for each partition of the shared partitioning scheme, from 1 to 41. The hash repartitioning exchanges seen previously are replaced by a single Distribute Streams exchange using Demand partitioning. Demand partitioning means that the next partition id is given to the next parallel thread that asks for one. My test machine has eight logical processors, and all are available for SQL Server to use. As a result, there are eight threads in the single parallel branch in this plan, each processing one partition from each table at a time. Once a thread finishes processing a partition, it grabs a new partition number from the Distribute Streams exchange…and so on until all partitions have been processed. It is important to understand that the parallel scans in this plan are different from the parallel hash join plan. Although the scans have the same parallelism icon, tables T1 and T2 are not being co-operatively scanned by multiple threads in the same way. Each thread reads a single partition of T1 and performs a hash match join with the same partition from table T2. The properties of the two Clustered Index Scans show a Seek Predicate (unusual for a scan!) limiting the rows to a single partition: The crucial point is that the join between T1 and T2 is on TID, and TID is the partitioning column for both tables. A thread that processes partition ‘n’ is guaranteed to see all rows that can possibly join on TID for that partition. In addition, no other thread will see rows from that partition, so this removes the need for repartitioning exchanges. CPU and Memory Efficiency Improvements The collocated join has removed two expensive repartitioning exchanges and added a single exchange processing 41 rows (one for each partition id). Remember, the parallel hash join plan exchanges had to process 5 million and 15 million rows. The amount of processor time spent on exchanges will be much lower in the collocated join plan. In addition, the collocated join plan has a maximum of 8 threads processing single partitions at any one time. The 41 partitions will all be processed eventually, but a new partition is not started until a thread asks for it. Threads can reuse hash table memory for the new partition. The parallel hash join plan also had 8 hash tables, but with all 5,000,000 build rows loaded at the same time. The collocated plan needs memory for only 8 * 125,000 = 1,000,000 rows at any one time. Collocated Hash Join Performance The collated join plan has disappointing performance in this case. The query runs for around 25,300ms despite the same IO statistics as usual. This is much the worst result so far, so what went wrong? It turns out that cardinality estimation for the single partition scans of table T1 is slightly low. The properties of the Clustered Index Scan of T1 (graphic immediately above) show the estimation was for 121,951 rows. This is a small shortfall compared with the 125,000 rows actually encountered, but it was enough to cause the hash join to spill to physical tempdb: A level 1 spill doesn’t sound too bad, until you realize that the spill to tempdb probably occurs for each of the 41 partitions. As a side note, the cardinality estimation error is a little surprising because the system tables accurately show there are 125,000 rows in every partition of T1. Unfortunately, the optimizer uses regular column and index statistics to derive cardinality estimates here rather than system table information (e.g. sys.partitions). Collocated Merge Join We will never know how well the collocated parallel hash join plan might have worked without the cardinality estimation error (and the resulting 41 spills to tempdb) but we do know: Merge join does not require a memory grant; and Merge join was the optimizer’s preferred join option for a single partition join Putting this all together, what we would really like to see is the same collocated join strategy, but using merge join instead of hash join. Unfortunately, the current query optimizer cannot produce a collocated merge join; it only knows how to do collocated hash join. So where does this leave us? CROSS APPLY sys.partitions We can try to write our own collocated join query. We can use sys.partitions to find the partition numbers, and CROSS APPLY to get a count per partition, with a final step to sum the partial counts. The following query implements this idea: SELECT row_count = SUM(Subtotals.cnt) FROM ( -- Partition numbers SELECT p.partition_number FROM sys.partitions AS p WHERE p.[object_id] = OBJECT_ID(N'T1', N'U') AND p.index_id = 1 ) AS P CROSS APPLY ( -- Count per collocated join SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals; The estimated plan is: The cardinality estimates aren’t all that good here, especially the estimate for the scan of the system table underlying the sys.partitions view. Nevertheless, the plan shape is heading toward where we would like to be. Each partition number from the system table results in a per-partition scan of T1 and T2, a one-to-many Merge Join, and a Stream Aggregate to compute the partial counts. The final Stream Aggregate just sums the partial counts. Execution time for this query is around 3,500ms, with the same IO statistics as always. This compares favourably with 5,000ms for the serial plan produced by the optimizer with the OPTION (MERGE JOIN) hint. This is another case of the sum of the parts being less than the whole – summing 41 partial counts from 41 single-partition merge joins is faster than a single merge join and count over all partitions. Even so, this single-threaded collocated merge join is not as quick as the original parallel hash join plan, which executed in 2,600ms. On the positive side, our collocated merge join uses only one logical processor and requires no memory grant. The parallel hash join plan used 16 threads and reserved 569 MB of memory:   Using a Temporary Table Our collocated merge join plan should benefit from parallelism. The reason parallelism is not being used is that the query references a system table. We can work around that by writing the partition numbers to a temporary table (or table variable): SET STATISTICS IO ON; DECLARE @s datetime2 = SYSUTCDATETIME();   CREATE TABLE #P ( partition_number integer PRIMARY KEY);   INSERT #P (partition_number) SELECT p.partition_number FROM sys.partitions AS p WHERE p.[object_id] = OBJECT_ID(N'T1', N'U') AND p.index_id = 1;   SELECT row_count = SUM(Subtotals.cnt) FROM #P AS p CROSS APPLY ( SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals;   DROP TABLE #P;   SELECT DATEDIFF(Millisecond, @s, SYSUTCDATETIME()); SET STATISTICS IO OFF; Using the temporary table adds a few logical reads, but the overall execution time is still around 3500ms, indistinguishable from the same query without the temporary table. The problem is that the query optimizer still doesn’t choose a parallel plan for this query, though the removal of the system table reference means that it could if it chose to: In fact the optimizer did enter the parallel plan phase of query optimization (running search 1 for a second time): Unfortunately, the parallel plan found seemed to be more expensive than the serial plan. This is a crazy result, caused by the optimizer’s cost model not reducing operator CPU costs on the inner side of a nested loops join. Don’t get me started on that, we’ll be here all night. In this plan, everything expensive happens on the inner side of a nested loops join. Without a CPU cost reduction to compensate for the added cost of exchange operators, candidate parallel plans always look more expensive to the optimizer than the equivalent serial plan. Parallel Collocated Merge Join We can produce the desired parallel plan using trace flag 8649 again: SELECT row_count = SUM(Subtotals.cnt) FROM #P AS p CROSS APPLY ( SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals OPTION (QUERYTRACEON 8649); The actual execution plan is: One difference between this plan and the collocated hash join plan is that a Repartition Streams exchange operator is used instead of Distribute Streams. The effect is similar, though not quite identical. The Repartition uses round-robin partitioning, meaning the next partition id is pushed to the next thread in sequence. The Distribute Streams exchange seen earlier used Demand partitioning, meaning the next partition id is pulled across the exchange by the next thread that is ready for more work. There are subtle performance implications for each partitioning option, but going into that would again take us too far off the main point of this post. Performance The important thing is the performance of this parallel collocated merge join – just 1350ms on a typical run. The list below shows all the alternatives from this post (all timings include creation, population, and deletion of the temporary table where appropriate) from quickest to slowest: Collocated parallel merge join: 1350ms Parallel hash join: 2600ms Collocated serial merge join: 3500ms Serial merge join: 5000ms Parallel merge join: 8400ms Collated parallel hash join: 25,300ms (hash spill per partition) The parallel collocated merge join requires no memory grant (aside from a paltry 1.2MB used for exchange buffers). This plan uses 16 threads at DOP 8; but 8 of those are (rather pointlessly) allocated to the parallel scan of the temporary table. These are minor concerns, but it turns out there is a way to address them if it bothers you. Parallel Collocated Merge Join with Demand Partitioning This final tweak replaces the temporary table with a hard-coded list of partition ids (dynamic SQL could be used to generate this query from sys.partitions): SELECT row_count = SUM(Subtotals.cnt) FROM ( VALUES (1),(2),(3),(4),(5),(6),(7),(8),(9),(10), (11),(12),(13),(14),(15),(16),(17),(18),(19),(20), (21),(22),(23),(24),(25),(26),(27),(28),(29),(30), (31),(32),(33),(34),(35),(36),(37),(38),(39),(40),(41) ) AS P (partition_number) CROSS APPLY ( SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals OPTION (QUERYTRACEON 8649); The actual execution plan is: The parallel collocated hash join plan is reproduced below for comparison: The manual rewrite has another advantage that has not been mentioned so far: the partial counts (per partition) can be computed earlier than the partial counts (per thread) in the optimizer’s collocated join plan. The earlier aggregation is performed by the extra Stream Aggregate under the nested loops join. The performance of the parallel collocated merge join is unchanged at around 1350ms. Final Words It is a shame that the current query optimizer does not consider a collocated merge join (Connect item closed as Won’t Fix). The example used in this post showed an improvement in execution time from 2600ms to 1350ms using a modestly-sized data set and limited parallelism. In addition, the memory requirement for the query was almost completely eliminated  – down from 569MB to 1.2MB. The problem with the parallel hash join selected by the optimizer is that it attempts to process the full data set all at once (albeit using eight threads). It requires a large memory grant to hold all 5 million rows from table T1 across the eight hash tables, and does not take advantage of the divide-and-conquer opportunity offered by the common partitioning. The great thing about the collocated join strategies is that each parallel thread works on a single partition from both tables, reading rows, performing the join, and computing a per-partition subtotal, before moving on to a new partition. From a thread’s point of view… If you have trouble visualizing what is happening from just looking at the parallel collocated merge join execution plan, let’s look at it again, but from the point of view of just one thread operating between the two Parallelism (exchange) operators. Our thread picks up a single partition id from the Distribute Streams exchange, and starts a merge join using ordered rows from partition 1 of table T1 and partition 1 of table T2. By definition, this is all happening on a single thread. As rows join, they are added to a (per-partition) count in the Stream Aggregate immediately above the Merge Join. Eventually, either T1 (partition 1) or T2 (partition 1) runs out of rows and the merge join stops. The per-partition count from the aggregate passes on through the Nested Loops join to another Stream Aggregate, which is maintaining a per-thread subtotal. Our same thread now picks up a new partition id from the exchange (say it gets id 9 this time). The count in the per-partition aggregate is reset to zero, and the processing of partition 9 of both tables proceeds just as it did for partition 1, and on the same thread. Each thread picks up a single partition id and processes all the data for that partition, completely independently from other threads working on other partitions. One thread might eventually process partitions (1, 9, 17, 25, 33, 41) while another is concurrently processing partitions (2, 10, 18, 26, 34) and so on for the other six threads at DOP 8. The point is that all 8 threads can execute independently and concurrently, continuing to process new partitions until the wider job (of which the thread has no knowledge!) is done. This divide-and-conquer technique can be much more efficient than simply splitting the entire workload across eight threads all at once. Related Reading Understanding and Using Parallelism in SQL Server Parallel Execution Plans Suck © 2013 Paul White – All Rights Reserved Twitter: @SQL_Kiwi

    Read the article

facebook twitter digg google delicious technorati stumbleupon myspace wordpress linkedin gmail igoogle windows live tumblr viadeo yahoo buzz yahoo mail yahoo bookmarks favorites email print

< Previous Page | 42 43 44 45 46 47 48 49 50 51 52 53  | Next Page >