Search Results

Search found 5298 results on 212 pages for 'marching cubes algorithm'.

Page 116/212 | < Previous Page | 112 113 114 115 116 117 118 119 120 121 122 123  | Next Page >

  • 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

  • Are there advantages for using recursion over iteration - other than sometimes readability and elegance?

    - by Prog
    I am about to make two assumptions. Please correct me if they're wrong: There isn't a recursive algorithm without an iterative equivalent. Iteration is always cheaper performance-wise than recursion (at least in general purpose languages such as Java, C++, Python etc.). If it's true that recursion is always more costly than iteration, and that it can always be replaced with an iterative algorithm (in languages that allow it) - than I think that the two remaining reasons to use recursion are: elegance and readability. Some algorithms are expressed more elegantly with recursion. E.g. scanning a binary tree. However apart from that, are there any reasons to use recursion over iteration? Does recursion have advantages over iteration other than sometimes elegance and readability?

    Read the article

  • Validate java SAML signature from C#

    - by Adrya
    How can i validate in .Net C# a SAML signature created in Java? Here is the SAML Signature that i get from Java: <ds:Signature xmlns:ds="http://www.w3.org/2000/09/xmldsig#"> <ds:SignedInfo> <ds:CanonicalizationMethod Algorithm="http://www.w3.org/2001/10/xml-exc-c14n#"> </ds:CanonicalizationMethod> <ds:SignatureMethod Algorithm="http://www.w3.org/2000/09/xmldsig#rsa-sha1"> </ds:SignatureMethod> <ds:Reference URI="#_e8bcba9d1c76d128938bddd5ae8c68e1"> <ds:Transforms> <ds:Transform Algorithm="http://www.w3.org/2000/09/xmldsig#enveloped-signature"> </ds:Transform> <ds:Transform Algorithm="http://www.w3.org/2001/10/xml-exc-c14n#"> <ec:InclusiveNamespaces xmlns:ec="http://www.w3.org/2001/10/xml-exc-c14n#" PrefixList="code ds kind rw saml samlp typens #default xsd xsi"> </ec:InclusiveNamespaces> </ds:Transform> </ds:Transforms> <ds:DigestMethod Algorithm="http://www.w3.org/2000/09/xmldsig#sha1"> </ds:DigestMethod> <ds:DigestValue>zEL7mB0Wkl+LtjMViO1imbucXiE=</ds:DigestValue> </ds:Reference> </ds:SignedInfo> <ds:SignatureValue> jpIX3WbX9SCFnqrpDyLj4TeJN5DGIvlEH+o/mb9M01VGdgFRLtfHqIm16BloApUPg2dDafmc9DwL Pyvs3TJ/hi0Q8f0ucaKdIuw+gBGxWFMcj/U68ZuLiv7U+Qe7i4ZA33rWPorkE82yfMacGf6ropPt v73mC0bpBP1ubo5qbM4= </ds:SignatureValue> <ds:KeyInfo> <ds:X509Data> <ds:X509Certificate> MIIDBDCCAeygAwIBAgIIC/ktBs1lgYcwDQYJKoZIhvcNAQEFBQAwNzERMA8GA1UEAwwIQWRtaW5D QTExFTATBgNVBAoMDEVKQkNBIFNhbXBsZTELMAkGA1UEBhMCU0UwHhcNMDkwMjIzMTAwMzEzWhcN MTgxMDE1MDkyNTQyWjBaMRQwEgYDVQQDDAsxMC41NS40MC42MTEbMBkGA1UECwwST24gRGVtYW5k IFBsYXRmb3JtMRIwEAYDVQQLDAlPbiBEZW1hbmQxETAPBgNVBAsMCFNvZnR3YXJlMIGfMA0GCSqG SIb3DQEBAQUAA4GNADCBiQKBgQCk5EqiedxA6WEE9N2vegSCqleFpXMfGplkrcPOdXTRLLOuRgQJ LEsOaqspDFoqk7yJgr7kaQROjB9OicSH7Hhsu7HbdD6N3ntwQYoeNZ8nvLSSx4jz21zvswxAqw1p DoGl3J6hks5owL4eYs2yRHvqgqXyZoxCccYwc4fYzMi42wIDAQABo3UwczAdBgNVHQ4EFgQUkrpk yryZToKXOXuiU2hNsKXLbyIwDAYDVR0TAQH/BAIwADAfBgNVHSMEGDAWgBSiviFUK7DUsjvByMfK g+pm4b2s7DAOBgNVHQ8BAf8EBAMCBaAwEwYDVR0lBAwwCgYIKwYBBQUHAwEwDQYJKoZIhvcNAQEF BQADggEBAKb94tnK2obEyvw8ZJ87u7gvkMxIezpBi/SqXTEBK1by0NHs8VJmdDN9+aOvC5np4fOL fFcRH++n6fvemEGgIkK3pOmNL5WiPpbWxrx55Yqwnr6eLsbdATALE4cgyZWHl/E0uVO2Ixlqeygw XTfg450cCWj4yfPTVZ73raKaDTWZK/Tnt7+ulm8xN+YWUIIbtW3KBQbGomqOzpftALyIKLVtBq7L J0hgsKGHNUnssWj5dt3bYrHgzaWLlpW3ikdRd67Nf0c1zOEgKHNEozrtRKiLLy+3bIiFk0CHImac 1zeqLlhjrG3OmIsIjxc1Vbc0+E+z6Unco474oSGf+D1DO+Y= </ds:X509Certificate> </ds:X509Data> </ds:KeyInfo> </ds:Signature> I know to parse SAML, i need to validate the signature. I tried this: public bool VerifySignature() { X509Certificate2 certificate = null; XmlDocument doc = new XmlDocument(); XmlElement xmlAssertionElement = this.GetXml(doc); doc.AppendChild(xmlAssertionElement); // Create a new SignedXml object and pass it // the XML document class. SamlSignedXml signedXml = new SamlSignedXml(xmlAssertionElement); // Get signature XmlElement xmlSignature = this.Signature; if (xmlSignature == null) { return false; } // Load the signature node. signedXml.LoadXml(xmlSignature); // Get the certificate used to sign the assertion if information about this // certificate is available in the signature of the assertion. foreach (KeyInfoClause clause in signedXml.KeyInfo) { if (clause is KeyInfoX509Data) { if (((KeyInfoX509Data)clause).Certificates.Count > 0) { certificate = (X509Certificate2)((KeyInfoX509Data)clause).Certificates[0]; } } } if (certificate == null) { return false; } return signedXml.CheckSignature(certificate, true); } It validates the signature of a SAML signed in .Net but not of this Java one.

    Read the article

  • Modified Strategy Design Pattern

    - by Samuel Walker
    I've started looking into Design Patterns recently, and one thing I'm coding would suit the Strategy pattern perfectly, except for one small difference. Essentially, some (but not all) of my algorithms, need an extra parameter or two passed to them. So I'll either need to pass them an extra parameter when I invoke their calculate method or store them as variables inside the ConcreteAlgorithm class, and be able to update them before I call the algorithm. Is there a design pattern for this need / How could I implement this while sticking to the Strategy Pattern? I've considered passing the client object to all the algorithms, and storing the variables in there, then using that only when the particular algorithm needs it. However, I think this is both unwieldy, and defeats the point of the strategy pattern. Just to be clear I'm implementing in Java, and so don't have the luxury of optional parameters (which would solve this nicely).

    Read the article

  • Developing Schema Compare for Oracle (Part 3): Ghost Objects

    - by Simon Cooper
    In the previous blog post, I covered how we solved the problem of dependencies between objects and between schemas. However, that isn’t the end of the issue. The dependencies algorithm I described works when you’re querying live databases and you can get dependencies for a particular schema direct from the server, and that’s all well and good. To throw a (rather large) spanner in the works, Schema Compare also has the concept of a snapshot, which is a read-only compressed XML representation of a selection of schemas that can be compared in the same way as a live database. This can be useful for keeping historical records or a baseline of a database schema, or comparing a schema on a computer that doesn’t have direct access to the database. So, how do snapshots interact with dependencies? Inter-database dependencies don't pose an issue as we store the dependencies in the snapshot. However, comparing a snapshot to a live database with cross-schema dependencies does cause a problem; what if the live database has a dependency to an object that does not exist in the snapshot? Take a basic example schema, where you’re only populating SchemaA: SOURCE   TARGET (using snapshot) CREATE TABLE SchemaA.Table1 ( Col1 NUMBER REFERENCES SchemaB.Table1(col1));   CREATE TABLE SchemaA.Table1 ( Col1 VARCHAR2(100)); CREATE TABLE SchemaB.Table1 ( Col1 NUMBER PRIMARY KEY);   CREATE TABLE SchemaB.Table1 ( Col1 VARCHAR2(100)); In this case, we want to generate a sync script to synchronize SchemaA.Table1 on the database represented by the snapshot. When taking a snapshot, database dependencies are followed, but because you’re not comparing it to anything at the time, the comparison dependencies algorithm described in my last post cannot be used. So, as you only take a snapshot of SchemaA on the target database, SchemaB.Table1 will not be in the snapshot. If this snapshot is then used to compare against the above source schema, SchemaB.Table1 will be included in the source, but the object will not be found in the target snapshot. This is the same problem that was solved with comparison dependencies, but here we cannot use the comparison dependencies algorithm as the snapshot has not got any information on SchemaB! We've now hit quite a big problem - we’re trying to include SchemaB.Table1 in the target, but we simply do not know the status of this object on the database the snapshot was taken from; whether it exists in the database at all, whether it’s the same as the target, whether it’s different... What can we do about this sorry state of affairs? Well, not a lot, it would seem. We can’t query the original database, as it may not be accessible, and we cannot assume any default state as it could be wrong and break the script (and we currently do not have a roll-back mechanism for failed synchronizes). The only way to fix this properly is for the user to go right back to the start and re-create the snapshot, explicitly including the schemas of these 'ghost' objects. So, the only thing we can do is flag up dependent ghost objects in the UI, and ask the user what we should do with it – assume it doesn’t exist, assume it’s the same as the target, or specify a definition for it. Unfortunately, such functionality didn’t make the cut for v1 of Schema Compare (as this is very much an edge case for a non-critical piece of functionality), so we simply flag the ghost objects up in the sync wizard as unsyncable, and let the user sort out what’s going on and edit the sync script as appropriate. There are some things that we do do to alleviate somewhat this rather unhappy situation; if a user creates a snapshot from the source or target of a database comparison, we include all the objects registered from the database, not just the ones in the schemas originally selected for comparison. This includes any extra dependent objects registered through the comparison dependencies algorithm. If the user then compares the resulting snapshot against the same database they were comparing against when it was created, the extra dependencies will be included in the snapshot as required and everything will be good. Fortunately, this problem will come up quite rarely, and only when the user uses snapshots and tries to sync objects with unknown cross-schema dependencies. However, the solution is not an easy one, and lead to some difficult architecture and design decisions within the product. And all this pain follows from the simple decision to allow schema pre-filtering! Next: why adding a column to a table isn't as easy as you would think...

    Read the article

  • Is it true that the Google Spider gives the most relevance of a search result to the first 68 characters of the <title>?

    - by leeand00
    I am reading documentation about my CMS and it states that an HTML page <title> tag is really important in SEO. It states that the Google Spider gives the most relevance to the first 68 characters of a site title. (68 characters being the number of characters that Google will display in it's search engine result pages,) Can anyone verify this is still true? I read in The Information Diet that content farms were getting too good at gaming Google's algorithm for collecting and posting SERPs and so google had to change the search algorithm.

    Read the article

  • Obtaining In game Warcraft III data to program standalone AI

    - by Slav
    I am implementing common purpose behavioral algorithm and would like to test it under my lovely Warcraft III game and watch how it will fight against real players. The problem is how to obtain information about in game state (units, structures, environment, etc. ). Algorithm needs access to hard drive and possibly distributed computing, that's why JASS (WC3 Editor language) usage doesn't solve the issue. Direct 3D hooking is an approach, but it wasn't done for WC3 yet and significant drawback is inability to watch online at how AI performs since it uses the viewport to issue commands. How in game data can be obtained to a different process in a fastest and easiest way? Thank you.

    Read the article

  • Algorithms and Programmer's day-to-day job

    - by Lior Kogan
    As of July 10, 2012, Stack Overflow contains 3,345,864 questions, out of which 20,840 questions are tagged as "Algorithm" - this is less than 0.6% ! I find it disturbing. Many programmers have several years of academic education in computer science / software engineering. Most of them are smart... When asked, most would say that they love algorithms. Computer programming is generally about solving problems using algorithms... Yet, only 1 of 160 questions is tagged as algorithm related. What does it say about our profession?

    Read the article

  • Fast pixelshader 2D raytracing

    - by heishe
    I'd like to do a simple 2D shadow calculation algorithm by rendering my environment into a texture, and then use raytracing to determine what pixels of the texture are not visible to the point light (simply handed to the shader as a vec2 position) . A simple brute force algorithm per pixel would looks like this: line_segment = line segment between current pixel of texture and light source For each pixel in the texture: { if pixel is not just empty space && pixel is on line_segment output = black else output = normal color of the pixel } This is, of course, probably not the fastest way to do it. Question is: What are faster ways to do it or what are some optimizations that can be applied to this technique?

    Read the article

  • Managed code and the Shell – Do?

    Back in 2006 I wrote a blog post titled: Managed code and the Shell – Don't!. Please visit that post to see why that advice was given.The crux of the issue has been addressed in the latest CLR via In-Process Side-by-Side Execution. In addition to the MSDN documentation I just linked, there is also an MSDN article on the topic: In-Process Side-by-Side.Now, even though the major technical impediment seems to be removed, I don’t know if Microsoft is now officially supporting managed extensions to the shell. Either way, I noticed a CodePlex project that is marching ahead to enable exactly that: Managed Mini Shell Extension Framework. Not much activity there, but maybe it will grow once .NET 4 is released... Comments about this post welcome at the original blog.

    Read the article

  • Conways Game of Life C#

    - by Darren Young
    Hi, Not sure if this is the correct place for this question or SO - mods please move if necessary. I am going to have a go at creating GoL over the weekend as a little test project : http://en.wikipedia.org/wiki/Conway's_Game_of_Life I understand the algorithm, however I just wanted to check regarding the implementation, from maybe somebody that has tried it. Essentially, my first (basic) implementation, will be a static grid at a set speed. If I understand correctly, these are the steps I will need: Initial seed Create 2d array with initial set up Foreach iteration, create temporary array, calculating each cells new state based on the Game of Life algorithm Assign temp array to proper array. Redraw grid from proper array. My concerns are over speed. When I am populating the grid from the array, would it simply be a case of looping through the array, assigning on or off to each grid cell and then redraw the grid? Am I on the correct path?

    Read the article

  • Card deck and sparse matrix interview questions

    - by MrDatabase
    I just had a technical phone screen w/ a start-up. Here's the technical questions I was asked ... and my answers. What do think of these answers? Feel free to post better answers :-) Question 1: how would you represent a standard 52 card deck in (basically any language)? How would you shuffle the deck? Answer: use an array containing a "Card" struct or class. Each instance of card has some unique identifier... either it's position in the array or a unique integer member variable in the range [0, 51]. Shuffle the cards by traversing the array once from index zero to index 51. Randomly swap ith card with "another card" (I didn't remember how this shuffle algorithm works exactly). Watch out for using the same probability for each card... that's a gotcha in this algorithm. I mentioned the algorithm is from Programming Pearls. Question 2: how to represent a large sparse matrix? the matrix can be very large... like 1000x1000... but only a relatively small number (~20) of the entries are non-zero. Answer: condense the array into a list of the non-zero entries. for a given entry (i,j) in the array... "map" (i,j) to a single integer k... then use k as a key into a dictionary or hashtable. For the 1000x1000 sparse array map (i,j) to k using something like f(i, j) = i + j * 1001. 1001 is just one plus the maximum of all i and j. I didn't recall exactly how this mapping worked... but the interviewer got the idea (I think). Are these good answers? I'm wondering because after I finished the second question the interviewer said the dreaded "well that's all the questions I have for now." Cheers!

    Read the article

  • creating the nodes for path finding during run time - more like path making and more

    - by bigbadbabybear
    i'm making my 1st game. i'm using javascript as i currently want to learn to make games without needing to learn another language but this is more of a general game dev question its a 2d turn-based tile/grid game. you can check it here http://www.patinterotest.tk/ it creates a movable area when you hover a player and it implements the A* algo for moving the player. The Problem: i want to make the 'dynamic movable area creation' already implement a limited number of steps for a player. The Questions: what is a good way to do this? is there another algorithm to use for this? the A* algorithm needs a start and destination, with what i want to do i don't have a destination or should i just limit the iteration of the A* algo to the steps variable? hopefully you understand the problem & questions easily

    Read the article

  • Prefer algorithms to hand-written loops?

    - by FredOverflow
    Which of the following to you find more readable? The hand-written loop: for (std::vector<Foo>::const_iterator it = vec.begin(); it != vec.end(); ++it) { bar.process(*it); } Or the algorithm invocation: #include <algorithm> #include <functional> std::for_each(vec.begin(), vec.end(), std::bind1st(std::mem_fun_ref(&Bar::process), bar)); I wonder if std::for_each is really worth it, given such a simple example already requires so much code. What are your thoughts on this matter?

    Read the article

  • What if we run out of stack space in C# or Python?

    - by dotneteer
    Supposing we are running a recursive algorithm on a very large data set that requires, say, 1 million recursive calls. Normally, one would solve such a large problem by converting recursion to a loop and a stack, but what if we do not want to or cannot rewrite the algorithm? Python has the sys.setrecursionlimit(int) method to set the number of recursions. However, this is only part of the story; the program can still run our of stack space. C# does not have a equivalent method. Fortunately, both C# and Python have option to set the stack space when creating a thread. In C#, there is an overloaded constructor for the Thread class that accepts a parameter for the stack size: Thread t = new Thread(work, stackSize); In Python, we can set the stack size by calling: threading.stack_size(67108864) We can then run our work under a new thread with increased stack size.

    Read the article

  • Count unique visitors by group of visited places

    - by Mathieu
    I'm facing the problem of counting the unique visitors of groups of places. Here is the situation: I have visitors that can visit places. For example, that can be internet users visiting web pages, or customers going to restaurants. A visitor can visit as much places as he wishes, and a place can be visited by several visitors. A visitor can come to the same place several times. The places belong to groups. A group can obviously contain several places, and places can belong to several groups. Given that, for each visitor, we can have a list of visited places, how can I have the number of unique visitors per group of places? Example: I have visitors A, B, C and D; and I have places x, y and z. I have these visiting lists: [ A -> [x,x,y,x], B -> [], C -> [z,z], D -> [y,x,x,z] ] Having these number of unique visitors per place is quite easy: [ x -> 2, // A and D visited x y -> 2, // A and D visited y z -> 2 // C and D visited z ] But if I have these groups: [ G1 -> [x,y,z], G2 -> [x,z], G3 -> [x,y] ] How can I have this information? [ G1 -> 3, // A, C and D visited x or y or z G2 -> 3, // A, C and D visited x or z G3 -> 2 // A and D visited x or y ] Additional notes : There are so many places that it is not possible to store information about every possible group; It's not a problem if approximation are made. I don't need 100% precision. Having a fast algorithm that tells me that there were 12345 visits in a group instead of 12543 is better than a slow algorithm telling the exact number. Let's say there can be ~5% deviation. Is there an algorithm or class of algorithms that addresses this type of problem?

    Read the article

  • yield – Just yet another sexy c# keyword?

    - by George Mamaladze
    yield (see NSDN c# reference) operator came I guess with .NET 2.0 and I my feeling is that it’s not as wide used as it could (or should) be.   I am not going to talk here about necessarity and advantages of using iterator pattern when accessing custom sequences (just google it).   Let’s look at it from the clean code point of view. Let's see if it really helps us to keep our code understandable, reusable and testable.   Let’s say we want to iterate a tree and do something with it’s nodes, for instance calculate a sum of their values. So the most elegant way would be to build a recursive method performing a classic depth traversal returning the sum.           private int CalculateTreeSum(Node top)         {             int sumOfChildNodes = 0;             foreach (Node childNode in top.ChildNodes)             {                 sumOfChildNodes += CalculateTreeSum(childNode);             }             return top.Value + sumOfChildNodes;         }     “Do One Thing” Nevertheless it violates one of the most important rules “Do One Thing”. Our  method CalculateTreeSum does two things at the same time. It travels inside the tree and performs some computation – in this case calculates sum. Doing two things in one method is definitely a bad thing because of several reasons: ·          Understandability: Readability / refactoring ·          Reuseability: when overriding - no chance to override computation without copying iteration code and vice versa. ·          Testability: you are not able to test computation without constructing the tree and you are not able to test correctness of tree iteration.   I want to spend some more words on this last issue. How do you test the method CalculateTreeSum when it contains two in one: computation & iteration? The only chance is to construct a test tree and assert the result of the method call, in our case the sum against our expectation. And if the test fails you do not know wether was the computation algorithm wrong or was that the iteration? At the end to top it all off I tell you: according to Murphy’s Law the iteration will have a bug as well as the calculation. Both bugs in a combination will cause the sum to be accidentally exactly the same you expect and the test will PASS. J   Ok let’s use yield! That’s why it is generally a very good idea not to mix but isolate “things”. Ok let’s use yield!           private int CalculateTreeSumClean(Node top)         {             IEnumerable<Node> treeNodes = GetTreeNodes(top);             return CalculateSum(treeNodes);         }             private int CalculateSum(IEnumerable<Node> nodes)         {             int sumOfNodes = 0;             foreach (Node node in nodes)             {                 sumOfNodes += node.Value;             }             return sumOfNodes;         }           private IEnumerable<Node> GetTreeNodes(Node top)         {             yield return top;             foreach (Node childNode in top.ChildNodes)             {                 foreach (Node currentNode in GetTreeNodes(childNode))                 {                     yield return currentNode;                 }             }         }   Two methods does not know anything about each other. One contains calculation logic another jut the iteration logic. You can relpace the tree iteration algorithm from depth traversal to breath trevaersal or use stack or visitor pattern instead of recursion. This will not influence your calculation logic. And vice versa you can relace the sum with product or do whatever you want with node values, the calculateion algorithm is not aware of beeng working on some tree or graph.  How about not using yield? Now let’s ask the question – what if we do not have yield operator? The brief look at the generated code gives us an answer. The compiler generates a 150 lines long class to implement the iteration logic.       [CompilerGenerated]     private sealed class <GetTreeNodes>d__0 : IEnumerable<Node>, IEnumerable, IEnumerator<Node>, IEnumerator, IDisposable     {         ...        150 Lines of generated code        ...     }   Often we compromise code readability, cleanness, testability, etc. – to reduce number of classes, code lines, keystrokes and mouse clicks. This is the human nature - we are lazy. Knowing and using such a sexy construct like yield, allows us to be lazy, write very few lines of code and at the same time stay clean and do one thing in a method. That's why I generally welcome using staff like that.   Note: The above used recursive depth traversal algorithm is possibly the compact one but not the best one from the performance and memory utilization point of view. It was taken to emphasize on other primary aspects of this post.

    Read the article

  • Checking whether php script was resolved by optimal way

    - by user2135931
    Can anybody give some advise how to check the arbitrary php code on optimal solution. For example I create a simple algorithm and sent it on the special resource. After proccessing my algorithm this resource give me result whether my code is nice. If no it give me some advice and tell what is wrong (maybe I forgot check devision by zero etc). I looked for php code analyzer but could't find any variants. Maybe someone give me a resource where I can research this problem. Thanks in advance!

    Read the article

  • Can anyone help solve this complex algorithmic problem?

    - by Locaaaaa
    I got this question in an interview and I was not able to solve it. You have a circular road, with N number of gas stations. You know the ammount of gas that each station has. You know the ammount of gas you need to GO from one station to the next one. Your car starts with 0. The question is: Create an algorithm, to know from which gas station you must start driving. As an exercise to me, I would translate the algorithm to C#.

    Read the article

  • How can i get latency when using Game Center?

    - by Freddy
    I'm pretty new to network programming. Basically I'm using game center for making a relatively simple iPhone game using Game-center p2p. However i'm now working on a algorithm to improve the multiplayer performance. But, I need to know how long it took for a package to travel from one device to the another device (latency) for the algorithm to work good. As for now, I have solved the problem by sending a double with time interval since 1970 in the package and then I compare it with the time at the other device. However I have heard that the NSDate methods is connected to the internet, which also will cause latency so the time interval would not be perfectly correct. What is the ideal way to check for how long it take for a package to be sent?

    Read the article

  • Narrow-phase collision detection algorithms

    - by Marian Ivanov
    There are three phases of collision detection. Broadphase: It loops between all objecs that can interact, false positives are allowed, if it would speed up the loop. Narrowphase: Determines whether they collide, and sometimes, how, no false positives Resolution: Resolves the collision. The question I'm asking is about the narrowphase. There are multiple algorithms, differing in complexity and accuracy. Hitbox intersection: This is an a-posteriori algorithm, that has the lowest complexity, but also isn't too accurate, Color intersection: Hitbox intersection for each pixel, a-posteriori, pixel-perfect, not accuratee in regards to time, higher complexity Separating axis theorem: This is used more often, accurate for triangles, however, a-posteriori, as it can't find the edge, when taking last frame in account, it's more stable Linear raycasting: A-priori algorithm, useful for semi-realistic-looking physics, finds the intersection point, even more accurate than SAT, but with more complexity Spline interpolation: A-priori, even more accurate than linear rays, even more coplexity. There are probably many more that I've forgot about. The question is, in when is it better to use SAT, when rays, when splines, and whether there is anything better.

    Read the article

  • How can calculus and linear algebra be useful to a system programmer?

    - by Victor
    I found a website saying that calculus and linear algebra are necessary for System Programming. System Programming, as far as I know, is about osdev, drivers, utilities and so on. I just can't figure out how calculus and linear algebra can be helpful on that. I know that calculus has several applications in science, but in this particular field of programming I just can't imagine how calculus can be so important. The information was on this site: http://www.wikihow.com/Become-a-Programmer Edit: Some answers here are explaining about algorithm complexity and optimization. When I made this question I was trying to be more specific about the area of System's Programming. Algorithm complexity and optimization can be applied to any area of programming not just System's Programming. That may be why I wasn't able to came up with such thinking at the time of the question.

    Read the article

  • find second smallest element in Fibonacci Heap

    - by Longeyes
    I need to describe an algorithm that finds the second smallest element in a Fibonacci-Heap using the Operations: Insert, ExtractMin, DecreaseKey and GetMin. The last one is an algorithm previously implemented to find and return the smallest element of the heap. I thought I'd start by extracting the minimum, which results in its children becoming roots. I could then use GetMin to find the second smallest element. But it seems to me that I'm overlooking other cases because I don't know when to use Insert and DecreaseKey, and the way the question is phrased seems to suggest I should need them.

    Read the article

  • Application qos involving priority and bandwidth

    - by Steve Peng
    Our manager wants us to do applicaiton qos which is quite different from the well-known system qos. We have many services of three types, they have priorites, the manager wants to suspend low priority services requests when there are not enough bandwidth for high priority services. But if the high priority services requests decrease, the bandwidth for low priority services should increase and low priority service requests are allowed again. There should be an algorithm involving priority and bandwidth. I don't know how to design the algorithm, is there any example on the internet? Somebody can give suggestion? Thanks. UPDATE All these services are within a same process. We are setting the maximum bandwidth for the three types of services via ports of services via TC (TC is the linux qos tool whose name means traffic control).

    Read the article

< Previous Page | 112 113 114 115 116 117 118 119 120 121 122 123  | Next Page >