Search Results

Search found 21107 results on 845 pages for 'size optimization'.

Page 9/845 | < Previous Page | 5 6 7 8 9 10 11 12 13 14 15 16  | Next Page >

  • How to set the size of multiple buttons on content

    - by rob
    I am using three buttons along with the list view.These three buttons are added to the layout using the TableLayout and TableRow.The first button takes more space on content, so the other two becomes smaller in size than first one. When the application runs, it doesn't even show the two other buttons. How can I make the size of three buttons equal and display all of them when application runs?Give me some example please. Thanks

    Read the article

  • SQL query performance optimization (TimesTen)

    - by Sergey Mikhanov
    Hi community, I need some help with TimesTen DB query optimization. I made some measures with Java profiler and found the code section that takes most of the time (this code section executes the SQL query). What is strange that this query becomes expensive only for some specific input data. Here’s the example. We have two tables that we are querying, one represents the objects we want to fetch (T_PROFILEGROUP), another represents the many-to-many link from some other table (T_PROFILECONTEXT_PROFILEGROUPS). We are not querying linked table. These are the queries that I executed with DB profiler running (they are the same except for the ID): Command> select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; < 1169655247309537280 > < 1169655249792565248 > < 1464837997699399681 > 3 rows found. Command> select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1466585677823868928; < 1169655247309537280 > 1 row found. This is what I have in the profiler: 12:14:31.147 1 SQL 2L 6C 10825P Preparing: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272 12:14:31.147 2 SQL 4L 6C 10825P sbSqlCmdCompile ()(E): (Found already compiled version: refCount:01, bucket:47) cmdType:100, cmdNum:1146695. 12:14:31.147 3 SQL 4L 6C 10825P Opening: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; 12:14:31.147 4 SQL 4L 6C 10825P Fetching: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; 12:14:31.148 5 SQL 4L 6C 10825P Fetching: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; 12:14:31.148 6 SQL 4L 6C 10825P Fetching: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; 12:14:31.228 7 SQL 4L 6C 10825P Fetching: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; 12:14:31.228 8 SQL 4L 6C 10825P Closing: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1464837998949302272; 12:14:35.243 9 SQL 2L 6C 10825P Preparing: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1466585677823868928 12:14:35.243 10 SQL 4L 6C 10825P sbSqlCmdCompile ()(E): (Found already compiled version: refCount:01, bucket:44) cmdType:100, cmdNum:1146697. 12:14:35.243 11 SQL 4L 6C 10825P Opening: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1466585677823868928; 12:14:35.243 12 SQL 4L 6C 10825P Fetching: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1466585677823868928; 12:14:35.243 13 SQL 4L 6C 10825P Fetching: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1466585677823868928; 12:14:35.243 14 SQL 4L 6C 10825P Closing: select G.M_ID from T_PROFILECONTEXT_PROFILEGROUPS CG, T_PROFILEGROUP G where CG.M_ID_EID = G.M_ID and CG.M_ID_OID = 1466585677823868928; It’s clear that the first query took almost 100ms, while the second was executed instantly. It’s not about queries precompilation (the first one is precompiled too, as same queries happened earlier). We have DB indices for all columns used here: T_PROFILEGROUP.M_ID, T_PROFILECONTEXT_PROFILEGROUPS.M_ID_OID and T_PROFILECONTEXT_PROFILEGROUPS.M_ID_EID. My questions are: Why querying the same set of tables yields such a different performance for different parameters? Which indices are involved here? Is there any way to improve this simple query and/or the DB to make it faster? UPDATE: to give the feeling of size: Command> select count(*) from T_PROFILEGROUP; < 183840 > 1 row found. Command> select count(*) from T_PROFILECONTEXT_PROFILEGROUPS; < 2279104 > 1 row found.

    Read the article

  • How to determine the size of a project (lines of code, function points, other)

    - by sixtyfootersdude
    How would you evaluate project size? Part A: Before you start a project. Part B: For a complete project. I am interested in comparing unrelated projects. Here are some options: 1) Lines of code. I know that this is not a good metric of productivity but is this a reasonable measure of project size? If I wanted to estimate how long it would take to recreate a project would this be a reasonable way to do it? How many lines of code should I estimate a day? 2) Function Points. Functions points are defined as the number of: inputs outputs inquires internal files external interfaces Anyone have a veiw point on whether this is a good measure? Is there a way to **actually do this? Does anyone have another solution? Hours taken seems like it could be a useful metric but not solely. If I ask you what is a "bigger program" and give you two programs how would you approach the question? I have seen several discussions of this on stackover flow but most discuss how to measure programmer productivity. I am more interested in project size.

    Read the article

  • Which libraries use the "We Know Where You Live" optimization for std::make_shared?

    - by KnowItAllWannabe
    Over two years ago, Stephan T. Lavavej described a space-saving optimization he implemented in Microsoft's implementation of std::make_shared, and I know from speaking with him that Microsoft has nothing against other library implementations adopting this optimization. If you know for sure whether other libraries (e.g., for Gnu C++, Clang, Intel C++, plus Boost (for boost::make_shared)) have adopted this implementation, please contribute an answer. I don't have ready access to that many make_shared implementations, nor am I wild about digging into the bowels of the ones I have to see if they've implemented the WKWYL optimization, but I'm hoping that SO readers know the answers for some libraries off-hand. I know from looking at the code that as of Boost 1.52, the WKWYL optimization had not been implemented, but Boost is now up to version 1.55. Note that this optimization is different from std::make_shared's ability to avoid a dedicated heap allocation for the reference count used by std::shared_ptr. For a discussion of the difference between WKWYL and that optimication, consult this question.

    Read the article

  • ntfsresize volume and size information

    - by antonio
    I am going to resize my sda2 NTFS partition. When gathering info with ntfsresize, I get: ntfsresize --info /dev/sda2 ntfsresize v2013.1.13 (libntfs-3g) Device name : /dev/sda2 NTFS volume version: 3.1 Cluster size : 4096 bytes Current volume size: 21999993344 bytes (22000 MB) Current device size: 23622320128 bytes (23623 MB) Checking filesystem consistency ... Accounting clusters ... Space in use : 10673 MB (48.5%) Collecting resizing constraints ... You might resize at 10672590848 bytes or 10673 MB (freeing 11327 MB). Please make a test run using both the -n and -s options before real resizing! Can you tell me what is the difference between volume and device size? As for device size, 23622320128 bytes / 1000^2 = 23622.3 MB. Why is 23623 MB reported instead of 23622? Note that parted confirms this value: parted /dev/sda2 unit MB p Model: Unknown (unknown) Disk /dev/sda2: 23622MB Sector size (logical/physical): 512B/512B Partition Table: loop Disk Flags: Number Start End Size File system Flags 1 0.00MB 23622MB 23622MB ntfs

    Read the article

  • Help with code optimization

    - by Ockonal
    Hello, I've written a little particle system for my 2d-application. Here is raining code: // HPP ----------------------------------- struct Data { float x, y, x_speed, y_speed; int timeout; Data(); }; std::vector<Data> mData; bool mFirstTime; void processDrops(float windPower, int i); // CPP ----------------------------------- Data::Data() : x(rand()%ScreenResolutionX), y(0) , x_speed(0), y_speed(0), timeout(rand()%130) { } void Rain::processDrops(float windPower, int i) { int posX = rand() % mWindowWidth; mData[i].x = posX; mData[i].x_speed = WindPower*0.1; // WindPower is float mData[i].y_speed = Gravity*0.1; // Gravity is 9.8 * 19.2 // If that is first time, process drops randomly with window height if (mFirstTime) { mData[i].timeout = 0; mData[i].y = rand() % mWindowHeight; } else { mData[i].timeout = rand() % 130; mData[i].y = 0; } } void update(float windPower, float elapsed) { // If this is first time - create array with new Data structure objects if (mFirstTime) { for (int i=0; i < mMaxObjects; ++i) { mData.push_back(Data()); processDrops(windPower, i); } mFirstTime = false; } for (int i=0; i < mMaxObjects; i++) { // Sleep until uptime > 0 (To make drops fall with randomly timeout) if (mData[i].timeout > 0) { mData[i].timeout--; } else { // Find new x/y positions mData[i].x += mData[i].x_speed * elapsed; mData[i].y += mData[i].y_speed * elapsed; // Find new speeds mData[i].x_speed += windPower * elapsed; mData[i].y_speed += Gravity * elapsed; // Drawing here ... // If drop has been falled out of the screen if (mData[i].y > mWindowHeight) processDrops(windPower, i); } } } So the main idea is: I have some structure which consist of drop position, speed. I have a function for processing drops at some index in the vector-array. Now if that's first time of running I'm making array with max size and process it in cycle. But this code works slower that all another I have. Please, help me to optimize it. I tried to replace all int with uint16_t but I think it doesn't matter.

    Read the article

  • Matlab: Optimization by perturbing variable

    - by S_H
    My main script contains following code: %# Grid and model parameters nModel=50; nModel_want=1; nI_grid1=5; Nth=1; nRow.Scale1=5; nCol.Scale1=5; nRow.Scale2=5^2; nCol.Scale2=5^2; theta = 90; % degrees a_minor = 2; % range along minor direction a_major = 5; % range along major direction sill = var(reshape(Deff_matrix_NthModel,nCell.Scale1,1)); % variance of the coarse data matrix of size nRow.Scale1 X nCol.Scale1 %# Covariance computation % Scale 1 for ihRow = 1:nRow.Scale1 for ihCol = 1:nCol.Scale1 [cov.Scale1(ihRow,ihCol),heff.Scale1(ihRow,ihCol)] = general_CovModel(theta, ihCol, ihRow, a_minor, a_major, sill, 'Exp'); end end % Scale 2 for ihRow = 1:nRow.Scale2 for ihCol = 1:nCol.Scale2 [cov.Scale2(ihRow,ihCol),heff.Scale2(ihRow,ihCol)] = general_CovModel(theta, ihCol/(nCol.Scale2/nCol.Scale1), ihRow/(nRow.Scale2/nRow.Scale1), a_minor, a_major, sill/(nRow.Scale2*nCol.Scale2), 'Exp'); end end %# Scale-up of fine scale values by averaging [covAvg.Scale2,var_covAvg.Scale2,varNorm_covAvg.Scale2] = general_AverageProperty(nRow.Scale2/nRow.Scale1,nCol.Scale2/nCol.Scale1,1,nRow.Scale1,nCol.Scale1,1,cov.Scale2,1); I am using two functions, general_CovModel() and general_AverageProperty(), in my main script which are given as following: function [cov,h_eff] = general_CovModel(theta, hx, hy, a_minor, a_major, sill, mod_type) % mod_type should be in strings angle_rad = theta*(pi/180); % theta in degrees, angle_rad in radians R_theta = [sin(angle_rad) cos(angle_rad); -cos(angle_rad) sin(angle_rad)]; h = [hx; hy]; lambda = a_minor/a_major; D_lambda = [lambda 0; 0 1]; h_2prime = D_lambda*R_theta*h; h_eff = sqrt((h_2prime(1)^2)+(h_2prime(2)^2)); if strcmp(mod_type,'Sph')==1 || strcmp(mod_type,'sph') ==1 if h_eff<=a cov = sill - sill.*(1.5*(h_eff/a_minor)-0.5*((h_eff/a_minor)^3)); else cov = sill; end elseif strcmp(mod_type,'Exp')==1 || strcmp(mod_type,'exp') ==1 cov = sill-(sill.*(1-exp(-(3*h_eff)/a_minor))); elseif strcmp(mod_type,'Gauss')==1 || strcmp(mod_type,'gauss') ==1 cov = sill-(sill.*(1-exp(-((3*h_eff)^2/(a_minor^2))))); end and function [PropertyAvg,variance_PropertyAvg,NormVariance_PropertyAvg]=... general_AverageProperty(blocksize_row,blocksize_col,blocksize_t,... nUpscaledRow,nUpscaledCol,nUpscaledT,PropertyArray,omega) % This function computes average of a property and variance of that averaged % property using power averaging PropertyAvg=zeros(nUpscaledRow,nUpscaledCol,nUpscaledT); %# Average of property for k=1:nUpscaledT, for j=1:nUpscaledCol, for i=1:nUpscaledRow, sum=0; for a=1:blocksize_row, for b=1:blocksize_col, for c=1:blocksize_t, sum=sum+(PropertyArray((i-1)*blocksize_row+a,(j-1)*blocksize_col+b,(k-1)*blocksize_t+c).^omega); % add all the property values in 'blocksize_x','blocksize_y','blocksize_t' to one variable end end end PropertyAvg(i,j,k)=(sum/(blocksize_row*blocksize_col*blocksize_t)).^(1/omega); % take average of the summed property end end end %# Variance of averageed property variance_PropertyAvg=var(reshape(PropertyAvg,... nUpscaledRow*nUpscaledCol*nUpscaledT,1),1,1); %# Normalized variance of averageed property NormVariance_PropertyAvg=variance_PropertyAvg./(var(reshape(... PropertyArray,numel(PropertyArray),1),1,1)); Question: Using Matlab, I would like to optimize covAvg.Scale2 such that it matches closely with cov.Scale1 by perturbing/varying any (or all) of the following variables 1) a_minor 2) a_major 3) theta Thanks.

    Read the article

  • Set page size using WIA (with scanner)

    - by Salvador Sarpi
    Hello, I'm using WIA to acquire images from a scanner with C#. I can scan the papers, but I can't set up the page size correctly, it always defaults to A4 and I need to use Letter or Legal sometimes. I tried with the WIA_DPS_PAGE_SIZE property, but when I try to set a value, I always get an error, that the value is out of the interval (tried a lot of possible values). I wan't to be able to use WIA_DPS_PAGE_SIZE = WIA_PAGE_AUTO (for automatic page size), but I can't find anything on the web related to this. Does anyone know a solution? thanks!

    Read the article

  • UIImageView size becomes equal to the size of image it holds after autoresizing

    - by abura
    Hello, I faced a strange problem. My task is to provide scaling an image to fill the UIImageView frame. The UIImageView is a subview of another view. The code is following: CGRect frame=CGRectMake(0,0,viewSize.width,viewSize.height); UIView* backView=[[UIView alloc] initWithFrame:frame]; backView.autoresizesSubviews=YES; backView.autoresizingMask=UIViewAutoresizingFlexibleWidth; UIImageView* imgView=[[UIImageView alloc]initWithFrame:viewFrame]; imgView.autoresizingMask = UIViewAutoresizingFlexibleWidth | UIViewAutoresizingFlexibleHeight; imgView.contentMode=UIViewContentModeScaleToFill; [subcellBackView addSubview:imgView]; [imgView release]; When the image arrives, it is assigned the the imgView: imgView.image=img; The image is scaled according to the imgView's size. Everything is all right until the iPhone (simulator) rotation. After first autoresizing the imgView size becomes equal to the image size (which is much greater) and overlaps its superview frame. The superview frame reacts properly to the orientation change. Could you please help?

    Read the article

  • How John Got 15x Improvement Without Really Trying

    - by rchrd
    The following article was published on a Sun Microsystems website a number of years ago by John Feo. It is still useful and worth preserving. So I'm republishing it here.  How I Got 15x Improvement Without Really Trying John Feo, Sun Microsystems Taking ten "personal" program codes used in scientific and engineering research, the author was able to get from 2 to 15 times performance improvement easily by applying some simple general optimization techniques. Introduction Scientific research based on computer simulation depends on the simulation for advancement. The research can advance only as fast as the computational codes can execute. The codes' efficiency determines both the rate and quality of results. In the same amount of time, a faster program can generate more results and can carry out a more detailed simulation of physical phenomena than a slower program. Highly optimized programs help science advance quickly and insure that monies supporting scientific research are used as effectively as possible. Scientific computer codes divide into three broad categories: ISV, community, and personal. ISV codes are large, mature production codes developed and sold commercially. The codes improve slowly over time both in methods and capabilities, and they are well tuned for most vendor platforms. Since the codes are mature and complex, there are few opportunities to improve their performance solely through code optimization. Improvements of 10% to 15% are typical. Examples of ISV codes are DYNA3D, Gaussian, and Nastran. Community codes are non-commercial production codes used by a particular research field. Generally, they are developed and distributed by a single academic or research institution with assistance from the community. Most users just run the codes, but some develop new methods and extensions that feed back into the general release. The codes are available on most vendor platforms. Since these codes are younger than ISV codes, there are more opportunities to optimize the source code. Improvements of 50% are not unusual. Examples of community codes are AMBER, CHARM, BLAST, and FASTA. Personal codes are those written by single users or small research groups for their own use. These codes are not distributed, but may be passed from professor-to-student or student-to-student over several years. They form the primordial ocean of applications from which community and ISV codes emerge. Government research grants pay for the development of most personal codes. This paper reports on the nature and performance of this class of codes. Over the last year, I have looked at over two dozen personal codes from more than a dozen research institutions. The codes cover a variety of scientific fields, including astronomy, atmospheric sciences, bioinformatics, biology, chemistry, geology, and physics. The sources range from a few hundred lines to more than ten thousand lines, and are written in Fortran, Fortran 90, C, and C++. For the most part, the codes are modular, documented, and written in a clear, straightforward manner. They do not use complex language features, advanced data structures, programming tricks, or libraries. I had little trouble understanding what the codes did or how data structures were used. Most came with a makefile. Surprisingly, only one of the applications is parallel. All developers have access to parallel machines, so availability is not an issue. Several tried to parallelize their applications, but stopped after encountering difficulties. Lack of education and a perception that parallelism is difficult prevented most from trying. I parallelized several of the codes using OpenMP, and did not judge any of the codes as difficult to parallelize. Even more surprising than the lack of parallelism is the inefficiency of the codes. I was able to get large improvements in performance in a matter of a few days applying simple optimization techniques. Table 1 lists ten representative codes [names and affiliation are omitted to preserve anonymity]. Improvements on one processor range from 2x to 15.5x with a simple average of 4.75x. I did not use sophisticated performance tools or drill deep into the program's execution character as one would do when tuning ISV or community codes. Using only a profiler and source line timers, I identified inefficient sections of code and improved their performance by inspection. The changes were at a high level. I am sure there is another factor of 2 or 3 in each code, and more if the codes are parallelized. The study’s results show that personal scientific codes are running many times slower than they should and that the problem is pervasive. Computational scientists are not sloppy programmers; however, few are trained in the art of computer programming or code optimization. I found that most have a working knowledge of some programming language and standard software engineering practices; but they do not know, or think about, how to make their programs run faster. They simply do not know the standard techniques used to make codes run faster. In fact, they do not even perceive that such techniques exist. The case studies described in this paper show that applying simple, well known techniques can significantly increase the performance of personal codes. It is important that the scientific community and the Government agencies that support scientific research find ways to better educate academic scientific programmers. The inefficiency of their codes is so bad that it is retarding both the quality and progress of scientific research. # cacheperformance redundantoperations loopstructures performanceimprovement 1 x x 15.5 2 x 2.8 3 x x 2.5 4 x 2.1 5 x x 2.0 6 x 5.0 7 x 5.8 8 x 6.3 9 2.2 10 x x 3.3 Table 1 — Area of improvement and performance gains of 10 codes The remainder of the paper is organized as follows: sections 2, 3, and 4 discuss the three most common sources of inefficiencies in the codes studied. These are cache performance, redundant operations, and loop structures. Each section includes several examples. The last section summaries the work and suggests a possible solution to the issues raised. Optimizing cache performance Commodity microprocessor systems use caches to increase memory bandwidth and reduce memory latencies. Typical latencies from processor to L1, L2, local, and remote memory are 3, 10, 50, and 200 cycles, respectively. Moreover, bandwidth falls off dramatically as memory distances increase. Programs that do not use cache effectively run many times slower than programs that do. When optimizing for cache, the biggest performance gains are achieved by accessing data in cache order and reusing data to amortize the overhead of cache misses. Secondary considerations are prefetching, associativity, and replacement; however, the understanding and analysis required to optimize for the latter are probably beyond the capabilities of the non-expert. Much can be gained simply by accessing data in the correct order and maximizing data reuse. 6 out of the 10 codes studied here benefited from such high level optimizations. Array Accesses The most important cache optimization is the most basic: accessing Fortran array elements in column order and C array elements in row order. Four of the ten codes—1, 2, 4, and 10—got it wrong. Compilers will restructure nested loops to optimize cache performance, but may not do so if the loop structure is too complex, or the loop body includes conditionals, complex addressing, or function calls. In code 1, the compiler failed to invert a key loop because of complex addressing do I = 0, 1010, delta_x IM = I - delta_x IP = I + delta_x do J = 5, 995, delta_x JM = J - delta_x JP = J + delta_x T1 = CA1(IP, J) + CA1(I, JP) T2 = CA1(IM, J) + CA1(I, JM) S1 = T1 + T2 - 4 * CA1(I, J) CA(I, J) = CA1(I, J) + D * S1 end do end do In code 2, the culprit is conditionals do I = 1, N do J = 1, N If (IFLAG(I,J) .EQ. 0) then T1 = Value(I, J-1) T2 = Value(I-1, J) T3 = Value(I, J) T4 = Value(I+1, J) T5 = Value(I, J+1) Value(I,J) = 0.25 * (T1 + T2 + T5 + T4) Delta = ABS(T3 - Value(I,J)) If (Delta .GT. MaxDelta) MaxDelta = Delta endif enddo enddo I fixed both programs by inverting the loops by hand. Code 10 has three-dimensional arrays and triply nested loops. The structure of the most computationally intensive loops is too complex to invert automatically or by hand. The only practical solution is to transpose the arrays so that the dimension accessed by the innermost loop is in cache order. The arrays can be transposed at construction or prior to entering a computationally intensive section of code. The former requires all array references to be modified, while the latter is cost effective only if the cost of the transpose is amortized over many accesses. I used the second approach to optimize code 10. Code 5 has four-dimensional arrays and loops are nested four deep. For all of the reasons cited above the compiler is not able to restructure three key loops. Assume C arrays and let the four dimensions of the arrays be i, j, k, and l. In the original code, the index structure of the three loops is L1: for i L2: for i L3: for i for l for l for j for k for j for k for j for k for l So only L3 accesses array elements in cache order. L1 is a very complex loop—much too complex to invert. I brought the loop into cache alignment by transposing the second and fourth dimensions of the arrays. Since the code uses a macro to compute all array indexes, I effected the transpose at construction and changed the macro appropriately. The dimensions of the new arrays are now: i, l, k, and j. L3 is a simple loop and easily inverted. L2 has a loop-carried scalar dependence in k. By promoting the scalar name that carries the dependence to an array, I was able to invert the third and fourth subloops aligning the loop with cache. Code 5 is by far the most difficult of the four codes to optimize for array accesses; but the knowledge required to fix the problems is no more than that required for the other codes. I would judge this code at the limits of, but not beyond, the capabilities of appropriately trained computational scientists. Array Strides When a cache miss occurs, a line (64 bytes) rather than just one word is loaded into the cache. If data is accessed stride 1, than the cost of the miss is amortized over 8 words. Any stride other than one reduces the cost savings. Two of the ten codes studied suffered from non-unit strides. The codes represent two important classes of "strided" codes. Code 1 employs a multi-grid algorithm to reduce time to convergence. The grids are every tenth, fifth, second, and unit element. Since time to convergence is inversely proportional to the distance between elements, coarse grids converge quickly providing good starting values for finer grids. The better starting values further reduce the time to convergence. The downside is that grids of every nth element, n > 1, introduce non-unit strides into the computation. In the original code, much of the savings of the multi-grid algorithm were lost due to this problem. I eliminated the problem by compressing (copying) coarse grids into continuous memory, and rewriting the computation as a function of the compressed grid. On convergence, I copied the final values of the compressed grid back to the original grid. The savings gained from unit stride access of the compressed grid more than paid for the cost of copying. Using compressed grids, the loop from code 1 included in the previous section becomes do j = 1, GZ do i = 1, GZ T1 = CA(i+0, j-1) + CA(i-1, j+0) T4 = CA1(i+1, j+0) + CA1(i+0, j+1) S1 = T1 + T4 - 4 * CA1(i+0, j+0) CA(i+0, j+0) = CA1(i+0, j+0) + DD * S1 enddo enddo where CA and CA1 are compressed arrays of size GZ. Code 7 traverses a list of objects selecting objects for later processing. The labels of the selected objects are stored in an array. The selection step has unit stride, but the processing steps have irregular stride. A fix is to save the parameters of the selected objects in temporary arrays as they are selected, and pass the temporary arrays to the processing functions. The fix is practical if the same parameters are used in selection as in processing, or if processing comprises a series of distinct steps which use overlapping subsets of the parameters. Both conditions are true for code 7, so I achieved significant improvement by copying parameters to temporary arrays during selection. Data reuse In the previous sections, we optimized for spatial locality. It is also important to optimize for temporal locality. Once read, a datum should be used as much as possible before it is forced from cache. Loop fusion and loop unrolling are two techniques that increase temporal locality. Unfortunately, both techniques increase register pressure—as loop bodies become larger, the number of registers required to hold temporary values grows. Once register spilling occurs, any gains evaporate quickly. For multiprocessors with small register sets or small caches, the sweet spot can be very small. In the ten codes presented here, I found no opportunities for loop fusion and only two opportunities for loop unrolling (codes 1 and 3). In code 1, unrolling the outer and inner loop one iteration increases the number of result values computed by the loop body from 1 to 4, do J = 1, GZ-2, 2 do I = 1, GZ-2, 2 T1 = CA1(i+0, j-1) + CA1(i-1, j+0) T2 = CA1(i+1, j-1) + CA1(i+0, j+0) T3 = CA1(i+0, j+0) + CA1(i-1, j+1) T4 = CA1(i+1, j+0) + CA1(i+0, j+1) T5 = CA1(i+2, j+0) + CA1(i+1, j+1) T6 = CA1(i+1, j+1) + CA1(i+0, j+2) T7 = CA1(i+2, j+1) + CA1(i+1, j+2) S1 = T1 + T4 - 4 * CA1(i+0, j+0) S2 = T2 + T5 - 4 * CA1(i+1, j+0) S3 = T3 + T6 - 4 * CA1(i+0, j+1) S4 = T4 + T7 - 4 * CA1(i+1, j+1) CA(i+0, j+0) = CA1(i+0, j+0) + DD * S1 CA(i+1, j+0) = CA1(i+1, j+0) + DD * S2 CA(i+0, j+1) = CA1(i+0, j+1) + DD * S3 CA(i+1, j+1) = CA1(i+1, j+1) + DD * S4 enddo enddo The loop body executes 12 reads, whereas as the rolled loop shown in the previous section executes 20 reads to compute the same four values. In code 3, two loops are unrolled 8 times and one loop is unrolled 4 times. Here is the before for (k = 0; k < NK[u]; k++) { sum = 0.0; for (y = 0; y < NY; y++) { sum += W[y][u][k] * delta[y]; } backprop[i++]=sum; } and after code for (k = 0; k < KK - 8; k+=8) { sum0 = 0.0; sum1 = 0.0; sum2 = 0.0; sum3 = 0.0; sum4 = 0.0; sum5 = 0.0; sum6 = 0.0; sum7 = 0.0; for (y = 0; y < NY; y++) { sum0 += W[y][0][k+0] * delta[y]; sum1 += W[y][0][k+1] * delta[y]; sum2 += W[y][0][k+2] * delta[y]; sum3 += W[y][0][k+3] * delta[y]; sum4 += W[y][0][k+4] * delta[y]; sum5 += W[y][0][k+5] * delta[y]; sum6 += W[y][0][k+6] * delta[y]; sum7 += W[y][0][k+7] * delta[y]; } backprop[k+0] = sum0; backprop[k+1] = sum1; backprop[k+2] = sum2; backprop[k+3] = sum3; backprop[k+4] = sum4; backprop[k+5] = sum5; backprop[k+6] = sum6; backprop[k+7] = sum7; } for one of the loops unrolled 8 times. Optimizing for temporal locality is the most difficult optimization considered in this paper. The concepts are not difficult, but the sweet spot is small. Identifying where the program can benefit from loop unrolling or loop fusion is not trivial. Moreover, it takes some effort to get it right. Still, educating scientific programmers about temporal locality and teaching them how to optimize for it will pay dividends. Reducing instruction count Execution time is a function of instruction count. Reduce the count and you usually reduce the time. The best solution is to use a more efficient algorithm; that is, an algorithm whose order of complexity is smaller, that converges quicker, or is more accurate. Optimizing source code without changing the algorithm yields smaller, but still significant, gains. This paper considers only the latter because the intent is to study how much better codes can run if written by programmers schooled in basic code optimization techniques. The ten codes studied benefited from three types of "instruction reducing" optimizations. The two most prevalent were hoisting invariant memory and data operations out of inner loops. The third was eliminating unnecessary data copying. The nature of these inefficiencies is language dependent. Memory operations The semantics of C make it difficult for the compiler to determine all the invariant memory operations in a loop. The problem is particularly acute for loops in functions since the compiler may not know the values of the function's parameters at every call site when compiling the function. Most compilers support pragmas to help resolve ambiguities; however, these pragmas are not comprehensive and there is no standard syntax. To guarantee that invariant memory operations are not executed repetitively, the user has little choice but to hoist the operations by hand. The problem is not as severe in Fortran programs because in the absence of equivalence statements, it is a violation of the language's semantics for two names to share memory. Codes 3 and 5 are C programs. In both cases, the compiler did not hoist all invariant memory operations from inner loops. Consider the following loop from code 3 for (y = 0; y < NY; y++) { i = 0; for (u = 0; u < NU; u++) { for (k = 0; k < NK[u]; k++) { dW[y][u][k] += delta[y] * I1[i++]; } } } Since dW[y][u] can point to the same memory space as delta for one or more values of y and u, assignment to dW[y][u][k] may change the value of delta[y]. In reality, dW and delta do not overlap in memory, so I rewrote the loop as for (y = 0; y < NY; y++) { i = 0; Dy = delta[y]; for (u = 0; u < NU; u++) { for (k = 0; k < NK[u]; k++) { dW[y][u][k] += Dy * I1[i++]; } } } Failure to hoist invariant memory operations may be due to complex address calculations. If the compiler can not determine that the address calculation is invariant, then it can hoist neither the calculation nor the associated memory operations. As noted above, code 5 uses a macro to address four-dimensional arrays #define MAT4D(a,q,i,j,k) (double *)((a)->data + (q)*(a)->strides[0] + (i)*(a)->strides[3] + (j)*(a)->strides[2] + (k)*(a)->strides[1]) The macro is too complex for the compiler to understand and so, it does not identify any subexpressions as loop invariant. The simplest way to eliminate the address calculation from the innermost loop (over i) is to define a0 = MAT4D(a,q,0,j,k) before the loop and then replace all instances of *MAT4D(a,q,i,j,k) in the loop with a0[i] A similar problem appears in code 6, a Fortran program. The key loop in this program is do n1 = 1, nh nx1 = (n1 - 1) / nz + 1 nz1 = n1 - nz * (nx1 - 1) do n2 = 1, nh nx2 = (n2 - 1) / nz + 1 nz2 = n2 - nz * (nx2 - 1) ndx = nx2 - nx1 ndy = nz2 - nz1 gxx = grn(1,ndx,ndy) gyy = grn(2,ndx,ndy) gxy = grn(3,ndx,ndy) balance(n1,1) = balance(n1,1) + (force(n2,1) * gxx + force(n2,2) * gxy) * h1 balance(n1,2) = balance(n1,2) + (force(n2,1) * gxy + force(n2,2) * gyy)*h1 end do end do The programmer has written this loop well—there are no loop invariant operations with respect to n1 and n2. However, the loop resides within an iterative loop over time and the index calculations are independent with respect to time. Trading space for time, I precomputed the index values prior to the entering the time loop and stored the values in two arrays. I then replaced the index calculations with reads of the arrays. Data operations Ways to reduce data operations can appear in many forms. Implementing a more efficient algorithm produces the biggest gains. The closest I came to an algorithm change was in code 4. This code computes the inner product of K-vectors A(i) and B(j), 0 = i < N, 0 = j < M, for most values of i and j. Since the program computes most of the NM possible inner products, it is more efficient to compute all the inner products in one triply-nested loop rather than one at a time when needed. The savings accrue from reading A(i) once for all B(j) vectors and from loop unrolling. for (i = 0; i < N; i+=8) { for (j = 0; j < M; j++) { sum0 = 0.0; sum1 = 0.0; sum2 = 0.0; sum3 = 0.0; sum4 = 0.0; sum5 = 0.0; sum6 = 0.0; sum7 = 0.0; for (k = 0; k < K; k++) { sum0 += A[i+0][k] * B[j][k]; sum1 += A[i+1][k] * B[j][k]; sum2 += A[i+2][k] * B[j][k]; sum3 += A[i+3][k] * B[j][k]; sum4 += A[i+4][k] * B[j][k]; sum5 += A[i+5][k] * B[j][k]; sum6 += A[i+6][k] * B[j][k]; sum7 += A[i+7][k] * B[j][k]; } C[i+0][j] = sum0; C[i+1][j] = sum1; C[i+2][j] = sum2; C[i+3][j] = sum3; C[i+4][j] = sum4; C[i+5][j] = sum5; C[i+6][j] = sum6; C[i+7][j] = sum7; }} This change requires knowledge of a typical run; i.e., that most inner products are computed. The reasons for the change, however, derive from basic optimization concepts. It is the type of change easily made at development time by a knowledgeable programmer. In code 5, we have the data version of the index optimization in code 6. Here a very expensive computation is a function of the loop indices and so cannot be hoisted out of the loop; however, the computation is invariant with respect to an outer iterative loop over time. We can compute its value for each iteration of the computation loop prior to entering the time loop and save the values in an array. The increase in memory required to store the values is small in comparison to the large savings in time. The main loop in Code 8 is doubly nested. The inner loop includes a series of guarded computations; some are a function of the inner loop index but not the outer loop index while others are a function of the outer loop index but not the inner loop index for (j = 0; j < N; j++) { for (i = 0; i < M; i++) { r = i * hrmax; R = A[j]; temp = (PRM[3] == 0.0) ? 1.0 : pow(r, PRM[3]); high = temp * kcoeff * B[j] * PRM[2] * PRM[4]; low = high * PRM[6] * PRM[6] / (1.0 + pow(PRM[4] * PRM[6], 2.0)); kap = (R > PRM[6]) ? high * R * R / (1.0 + pow(PRM[4]*r, 2.0) : low * pow(R/PRM[6], PRM[5]); < rest of loop omitted > }} Note that the value of temp is invariant to j. Thus, we can hoist the computation for temp out of the loop and save its values in an array. for (i = 0; i < M; i++) { r = i * hrmax; TEMP[i] = pow(r, PRM[3]); } [N.B. – the case for PRM[3] = 0 is omitted and will be reintroduced later.] We now hoist out of the inner loop the computations invariant to i. Since the conditional guarding the value of kap is invariant to i, it behooves us to hoist the computation out of the inner loop, thereby executing the guard once rather than M times. The final version of the code is for (j = 0; j < N; j++) { R = rig[j] / 1000.; tmp1 = kcoeff * par[2] * beta[j] * par[4]; tmp2 = 1.0 + (par[4] * par[4] * par[6] * par[6]); tmp3 = 1.0 + (par[4] * par[4] * R * R); tmp4 = par[6] * par[6] / tmp2; tmp5 = R * R / tmp3; tmp6 = pow(R / par[6], par[5]); if ((par[3] == 0.0) && (R > par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * tmp5; } else if ((par[3] == 0.0) && (R <= par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * tmp4 * tmp6; } else if ((par[3] != 0.0) && (R > par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * TEMP[i] * tmp5; } else if ((par[3] != 0.0) && (R <= par[6])) { for (i = 1; i <= imax1; i++) KAP[i] = tmp1 * TEMP[i] * tmp4 * tmp6; } for (i = 0; i < M; i++) { kap = KAP[i]; r = i * hrmax; < rest of loop omitted > } } Maybe not the prettiest piece of code, but certainly much more efficient than the original loop, Copy operations Several programs unnecessarily copy data from one data structure to another. This problem occurs in both Fortran and C programs, although it manifests itself differently in the two languages. Code 1 declares two arrays—one for old values and one for new values. At the end of each iteration, the array of new values is copied to the array of old values to reset the data structures for the next iteration. This problem occurs in Fortran programs not included in this study and in both Fortran 77 and Fortran 90 code. Introducing pointers to the arrays and swapping pointer values is an obvious way to eliminate the copying; but pointers is not a feature that many Fortran programmers know well or are comfortable using. An easy solution not involving pointers is to extend the dimension of the value array by 1 and use the last dimension to differentiate between arrays at different times. For example, if the data space is N x N, declare the array (N, N, 2). Then store the problem’s initial values in (_, _, 2) and define the scalar names new = 2 and old = 1. At the start of each iteration, swap old and new to reset the arrays. The old–new copy problem did not appear in any C program. In programs that had new and old values, the code swapped pointers to reset data structures. Where unnecessary coping did occur is in structure assignment and parameter passing. Structures in C are handled much like scalars. Assignment causes the data space of the right-hand name to be copied to the data space of the left-hand name. Similarly, when a structure is passed to a function, the data space of the actual parameter is copied to the data space of the formal parameter. If the structure is large and the assignment or function call is in an inner loop, then copying costs can grow quite large. While none of the ten programs considered here manifested this problem, it did occur in programs not included in the study. A simple fix is always to refer to structures via pointers. Optimizing loop structures Since scientific programs spend almost all their time in loops, efficient loops are the key to good performance. Conditionals, function calls, little instruction level parallelism, and large numbers of temporary values make it difficult for the compiler to generate tightly packed, highly efficient code. Conditionals and function calls introduce jumps that disrupt code flow. Users should eliminate or isolate conditionls to their own loops as much as possible. Often logical expressions can be substituted for if-then-else statements. For example, code 2 includes the following snippet MaxDelta = 0.0 do J = 1, N do I = 1, M < code omitted > Delta = abs(OldValue ? NewValue) if (Delta > MaxDelta) MaxDelta = Delta enddo enddo if (MaxDelta .gt. 0.001) goto 200 Since the only use of MaxDelta is to control the jump to 200 and all that matters is whether or not it is greater than 0.001, I made MaxDelta a boolean and rewrote the snippet as MaxDelta = .false. do J = 1, N do I = 1, M < code omitted > Delta = abs(OldValue ? NewValue) MaxDelta = MaxDelta .or. (Delta .gt. 0.001) enddo enddo if (MaxDelta) goto 200 thereby, eliminating the conditional expression from the inner loop. A microprocessor can execute many instructions per instruction cycle. Typically, it can execute one or more memory, floating point, integer, and jump operations. To be executed simultaneously, the operations must be independent. Thick loops tend to have more instruction level parallelism than thin loops. Moreover, they reduce memory traffice by maximizing data reuse. Loop unrolling and loop fusion are two techniques to increase the size of loop bodies. Several of the codes studied benefitted from loop unrolling, but none benefitted from loop fusion. This observation is not too surpising since it is the general tendency of programmers to write thick loops. As loops become thicker, the number of temporary values grows, increasing register pressure. If registers spill, then memory traffic increases and code flow is disrupted. A thick loop with many temporary values may execute slower than an equivalent series of thin loops. The biggest gain will be achieved if the thick loop can be split into a series of independent loops eliminating the need to write and read temporary arrays. I found such an occasion in code 10 where I split the loop do i = 1, n do j = 1, m A24(j,i)= S24(j,i) * T24(j,i) + S25(j,i) * U25(j,i) B24(j,i)= S24(j,i) * T25(j,i) + S25(j,i) * U24(j,i) A25(j,i)= S24(j,i) * C24(j,i) + S25(j,i) * V24(j,i) B25(j,i)= S24(j,i) * U25(j,i) + S25(j,i) * V25(j,i) C24(j,i)= S26(j,i) * T26(j,i) + S27(j,i) * U26(j,i) D24(j,i)= S26(j,i) * T27(j,i) + S27(j,i) * V26(j,i) C25(j,i)= S27(j,i) * S28(j,i) + S26(j,i) * U28(j,i) D25(j,i)= S27(j,i) * T28(j,i) + S26(j,i) * V28(j,i) end do end do into two disjoint loops do i = 1, n do j = 1, m A24(j,i)= S24(j,i) * T24(j,i) + S25(j,i) * U25(j,i) B24(j,i)= S24(j,i) * T25(j,i) + S25(j,i) * U24(j,i) A25(j,i)= S24(j,i) * C24(j,i) + S25(j,i) * V24(j,i) B25(j,i)= S24(j,i) * U25(j,i) + S25(j,i) * V25(j,i) end do end do do i = 1, n do j = 1, m C24(j,i)= S26(j,i) * T26(j,i) + S27(j,i) * U26(j,i) D24(j,i)= S26(j,i) * T27(j,i) + S27(j,i) * V26(j,i) C25(j,i)= S27(j,i) * S28(j,i) + S26(j,i) * U28(j,i) D25(j,i)= S27(j,i) * T28(j,i) + S26(j,i) * V28(j,i) end do end do Conclusions Over the course of the last year, I have had the opportunity to work with over two dozen academic scientific programmers at leading research universities. Their research interests span a broad range of scientific fields. Except for two programs that relied almost exclusively on library routines (matrix multiply and fast Fourier transform), I was able to improve significantly the single processor performance of all codes. Improvements range from 2x to 15.5x with a simple average of 4.75x. Changes to the source code were at a very high level. I did not use sophisticated techniques or programming tools to discover inefficiencies or effect the changes. Only one code was parallel despite the availability of parallel systems to all developers. Clearly, we have a problem—personal scientific research codes are highly inefficient and not running parallel. The developers are unaware of simple optimization techniques to make programs run faster. They lack education in the art of code optimization and parallel programming. I do not believe we can fix the problem by publishing additional books or training manuals. To date, the developers in questions have not studied the books or manual available, and are unlikely to do so in the future. Short courses are a possible solution, but I believe they are too concentrated to be much use. The general concepts can be taught in a three or four day course, but that is not enough time for students to practice what they learn and acquire the experience to apply and extend the concepts to their codes. Practice is the key to becoming proficient at optimization. I recommend that graduate students be required to take a semester length course in optimization and parallel programming. We would never give someone access to state-of-the-art scientific equipment costing hundreds of thousands of dollars without first requiring them to demonstrate that they know how to use the equipment. Yet the criterion for time on state-of-the-art supercomputers is at most an interesting project. Requestors are never asked to demonstrate that they know how to use the system, or can use the system effectively. A semester course would teach them the required skills. Government agencies that fund academic scientific research pay for most of the computer systems supporting scientific research as well as the development of most personal scientific codes. These agencies should require graduate schools to offer a course in optimization and parallel programming as a requirement for funding. About the Author John Feo received his Ph.D. in Computer Science from The University of Texas at Austin in 1986. After graduate school, Dr. Feo worked at Lawrence Livermore National Laboratory where he was the Group Leader of the Computer Research Group and principal investigator of the Sisal Language Project. In 1997, Dr. Feo joined Tera Computer Company where he was project manager for the MTA, and oversaw the programming and evaluation of the MTA at the San Diego Supercomputer Center. In 2000, Dr. Feo joined Sun Microsystems as an HPC application specialist. He works with university research groups to optimize and parallelize scientific codes. Dr. Feo has published over two dozen research articles in the areas of parallel parallel programming, parallel programming languages, and application performance.

    Read the article

  • CSS size transition to stay inside parent div

    - by IlyaD
    I want to have an image frame that when I hove over it the image inside will zoom in a little (I am using size transition), but the frame will stay the same size. What happens now that even if the frame has a fixed width and height it is stilled zoomed with the image HTML: <div class="img-wrapper"> <img class="thumbnail" src="http://placekitten.com/400/200"> </div> and CSS .img-wrapper { width: 400px; } .thumbnail { width: 400px; } .thumbnail { -webkit-transition: all 0.2s ease-out; -moz-transition: all 0.2s ease-out; transition: all 0.2s ease-out; } .thumbnail:hover { width: 500px; -webkit-transition: all 0.2s ease-out; -moz-transition: all 0.2s ease-out; transition: all 0.2s ease-out; } http://codepen.io/pen/KCJny

    Read the article

  • Cannot sort a row of size 8130, which is greater than the allowable maximum of 8094

    - by Sri Kumar
    Hello All, SELECT DISTINCT tblJobReq.JobReqId, tblJobReq.JobStatusId, tblJobClass.JobClassId, tblJobClass.Title, tblJobReq.JobClassSubTitle, tblJobAnnouncement.JobClassDesc, tblJobAnnouncement.EndDate, tblJobAnnouncement.AgencyMktgVerbage, tblJobAnnouncement.SpecInfo, tblJobAnnouncement.Benefits, tblSalary.MinRateSal, tblSalary.MaxRateSal, tblSalary.MinRateHour, tblSalary.MaxRateHour, tblJobClass.StatementEval, tblJobReq.ApprovalDate, tblJobReq.RecruiterId, tblJobReq.AgencyId FROM ((tblJobReq LEFT JOIN tblJobAnnouncement ON tblJobReq.JobReqId =tblJobAnnouncement.JobReqId) INNER JOIN tblJobClass ON tblJobReq.JobClassId = tblJobClass.JobClassId) LEFT JOIN tblSalary ON tblJobClass.SalaryCode = tblSalary.SalaryCode WHERE (tblJobReq.JobClassId in (SELECT JobClassId from tblJobClass WHERE tblJobClass.Title like '%Family Therapist%')) When i try to execute the query it results in the following error. Cannot sort a row of size 8130, which is greater than the allowable maximum of 8094 I checked and didn't find any solution. The only way is to truncate (substring())the "tblJobAnnouncement.JobClassDesc" in the query which has column size of around 8000. Do we have any work around so that i need not truncate the values. Or Can this query be optimised? Any setting in SQL Server 2000?

    Read the article

  • XHR FF POST size limit

    - by usurper
    Hi, My XHR POST REQUEST is cut off. When I try to reload my page information is missing. Firebugs sends following message: ... Firebug request size limit has been reached by Firebug. ... My question is: What are my options? Would it work if I declare the content.length in the header? I added a line to my apache config file and restarted it: LimitRequestBody 0 I increased the size of transfer files in mysql config file Or it is a browser issue? The only solution I could think of was to cut the data in pieces and transmit the array one by one but I don't like this idea. The content length is 91691 according to firebug. Any suggestions?

    Read the article

  • When optimizing database queries, what exactly is the relationship between number of queries and siz

    - by williamjones
    To optimize application speed, everyone always advises to minimize the number of queries an application makes to the database, consolidating them into fewer queries that retrieve more wherever possible. However, this also always comes with the caution that data transferred is still data transferred, and just because you are making fewer queries doesn't make the data transferred free. I'm in a situation where I can over-include on the query in order to cut down the number of queries, and simply remove the unwanted data in the application code. Is there any type of a rule of thumb on how much of a cost there is to each query, to know when to optimize number of queries versus size of queries? I've tried to Google for objective performance analysis data, but surprisingly haven't been able to find anything like that. Clearly this relationship will change for factors such as when the database grows in size, making this somewhat individualized, but surely this is not so individualized that a broad sense of the landscape can't be drawn out? I'm looking for general answers, but for what it's worth, I'm running an application on Heroku.com, which means Ruby on Rails with a Postgres database.

    Read the article

  • ajax response byte size

    - by Alex Pacurar
    Im using jQuery's getJSONP and I want to log the duration of the call and the size of the response to be able to have some statistics about the usage of my application. This is a cross domain ajax call, so I need to use JSONP, but as the JSONP call is not done with an XMLHttpRequest object, the complete callback from jquery's ajax doesnt pass the response content. So my question is how to get the response size (content lenght) from a JSONP call. $.ajaxSetup( { complete:function(x,e) { log(x.responseText.length, x.responseText); } } here x is a XMLHttpRequest object for a JSON call , but for JSONP call is undefined.

    Read the article

  • Size and position of the C# form.

    - by Vytas999
    I use Visual Studio .NET to create a Windows-based application. The application includes a form named CForm. CForm contains 15 controls that enable users to set basic configuration options for the application. I design these controls to dynamically adjust when users resize CForm. The controls automatically update their size and position on the form as the form is resized. The initial size of the form should be 650 x 700 pixels. If CForm is resized to be smaller then 500 x 600 pixels, the controls will not be displayed correctly. How I must ensure that users cannot resize CForm to be smaller than 500 x 600 pixels..?

    Read the article

  • How to change font size in a pickerview?

    - by user183804
    I'm fairly new to iPhone programming and am trying to implement a double-component PickerView. At this point the picker works fine in terms of taking user input, created with Interface Builder. However, I need to change the font size to accommodate the text length in each column. I very much would appreciate any link to a straightforward method to creating a multi-component picker with an adjustable font size. Seems like it is not possible if Interface Builder is used to create the picker. So far I have not found any code links that address this issue in detail. Thanks in advance.

    Read the article

  • Is there a way that I can hard code a const XmlNameTable to be reused by all of my XmlTextReader(s)?

    - by highone
    Before I continue I would just like to say I know that "Premature optimization is the root of all evil." However this program is only a hobby project and I enjoy trying to find ways to optimize it. That being said, I was reading an article on improving xml performance and it recommended sharing "the XmlNameTable class that is used to store element and attribute names across multiple XML documents of the same type to improve performance." I wasn't able to find any information about doing this in my googling, so it is likely that this is either not possible, a no-no, or a stupid question, but what's the harm in asking?

    Read the article

  • Change the size of the panel dynamically

    - by C. Karunarathne
    Hi! I am implementing an application that needs to drag and drop image boxes in a panel.Image boxes are added dynamically from the program and so I have set the autoscroll property to true in panel.But when I have drag out the boxes at bottom in the panel the panel size got reduced.I have put autosize property false in panel.The panel is docked in another panel.I want to set the size of panel at run time.How can I achieve this. public form1(int[,] dummy, int columnSize, int rowSize) { this.dummy= dummy; numOfColumns = columnSize; numOfRows = rowSize; getData(); addIds = addIdArray; data = mylist; InitializeComponent(); //panel1.MinimumSize = new Size(columnSize * 40, rowSize * 40); //panel1.Height = rowSize * 40; //panel1.Width = columnSize * 40; //panel4.Height = rowSize * 40; //panel4.Width = columnSize * 40; int x, y; Structures.EmptyRectSpace space = new Structures.EmptyRectSpace(); for (int i = 0; i < data.Count; i++)// set picture boxes { space = (Structures.EmptyRectSpace)data[i]; x = space.startingJ; y = space.startingI; int h, w; h = space.length; w = space.width; p = new PictureBox(); p.Width = w * 40; p.Height = h * 40; p.BackColor = Color.DarkGreen; p.Image = Properties.Resources.v; p.BorderStyle = BorderStyle.FixedSingle; p.Name = addIdArray[i].ToString(); p.Location = new Point((x + 1 - w) * 40, (y + 1 - h) * 40); this.panel1.Controls.Add(p); } foreach (Control c in this.panel1.Controls) { if (c is PictureBox) { c.MouseDown += new MouseEventHandler(pictureBox1_MouseDown); } } this.panel1.DragOver += new System.Windows.Forms.DragEventHandler(this.panel1_DragOver); panel1.DragOver += new DragEventHandler(panel1_DragOver); panel1.DragDrop += new DragEventHandler(panel1_DragDrop); panel1.AllowDrop = true; panel2.AllowDrop = true; foreach (Control c in this.panel2.Controls) { c.MouseDown += new MouseEventHandler(pictureBox1_MouseDown); } this.panel2.DragOver += new System.Windows.Forms.DragEventHandler(this.panel2_DragOver); panel2.DragOver += new DragEventHandler(panel2_DragOver); panel2.DragDrop += new DragEventHandler(panel2_DragDrop); } This is the constructor of the form that contained panel.When it loaded the picture boxes must be added to the panel and there drag drop events of panel are implemented. Please give me a helping hand..

    Read the article

  • web page db query optimisation

    - by morpheous
    I am putting together a web page which is quite 'expensive' in terms of Db hits. I dont want to start optimizing at this stage - though with me trying to hit a deadline, I may end up not optimising at all. Currently the page requires 18 (thats right eighteen) hits to the db. I am already using joins, and some of the queries are UNIONed to minimize the trips to the db. My local dev machine can handle this (page is not slow) however, I feel if I release this into the wild, the number of queries will quickly overwhelm my database (mySQL). I could always use memcache or something similar, but I would much rather continue with my other dev work that needs to be completed before the deadline - at least retrieving the page work, its simply a matter of optimization. My question therefore is - is 18 db queries for a single page retrieval completely outrageous - (i.e. I should put everything on hold and optimize the hell of the retrieval logic), or shall I continue as normal, meet the deadline and release on schedule and see what happens?

    Read the article

  • Best way to handle MySQL date for performance with thousands of users

    - by bitLost
    I am currently part of a team designing a site that will potentially have thousands of users who will be doing a number of date related searches. During the design phase we have been trying to determine which makes more sense for performance optimization. Should we store the datetime field as a mysql datetime. Or should be break it up into a number of fields (year, month, day, hour, minute, ...) The question is with a large data set and a potentially large set of users, would we gain performance wise breaking the datetime into multiple fields and saving on relying on mysql date functions? Or is mysql already optimized for this?

    Read the article

  • Get dimensions for dataset matlab, Size Function

    - by Tim
    Hello I have been trying to display the dimensions of a dataset, below is the following matlab code that i used...but it gives me the wrong output....i'e number of rows and columns values are wrong....any help regarding this issue would be highly appreciated [file_input,pathname] = uigetfile( ... {'*.txt', 'Text (*.txt)'; ... '*.xls', 'Excel (*.xls)'; ... '*.*', 'All Files (*.*)'}, ... 'Select files'); [pathstr, name, ext, versn] = fileparts(file_input) r = size(name,1); c = size(name,2); disp(r) disp(c)

    Read the article

< Previous Page | 5 6 7 8 9 10 11 12 13 14 15 16  | Next Page >