profiling

问题 Following the tutorial on TensorFlow, I am trying to understand run-time statistics using tensorboard. I find that the compute time of a High-level node representing a name scope is not equal to the sum of compute times of its sub-nodes. Why isn't it the same? For example, in the attached snapshot: Compute time of ConvLayer2 = 75.5 ms, while the Sub-nodes compute time = 55.2 (conv) + 1.73 (add) + 1 (other nodes) = 57.9 ms Snapshot of ConvLayer2 import numpy as np import tensorflow as tf from

问题 What tools are available to profile a .NET program by measuring function execution times and generating graphs to visualize the time spent at various points in the call graph? 回答1: AQTime and dotTrace are two very good commerical profilers. A free option would be ProfileSharp, though I have had little luck with it. Microsoft provides the CLR Profiler as well, which works well, but has fewer features. 回答2: It'll cost you but Ants Performance Profiler will do the job. 回答3: CLR Profiler 回答4:

问题 This question already has answers here : Closed 6 years ago . Possible Duplicate: Are doubles faster than floats in c#? I wrote simple benchmark to check how much performance i can get changing double datatype to float in my application. Here is my code: // my form: // one textbox: textbox1 (MultiLine property set to true) // one button: button1 with event button1_Click private void button1_Click(object sender, EventArgs e) { int num = 10000000; float[] floats1 = new float[num]; float[]

问题 I have several points stored in an array. I need to find bounds of that points ie. the rectangle which bounds all the points. I know how to solve this in plain Python. I would like to know is there a better way than the naive max, min over the array or built-in method to solve the problem. points = [[1, 3], [2, 4], [4, 1], [3, 3], [1, 6]] b = bounds(points) # the function I am looking for # now b = [[1, 1], [4, 6]] 回答1: My approach to getting performance is to push things down to C level

问题 I have several points stored in an array. I need to find bounds of that points ie. the rectangle which bounds all the points. I know how to solve this in plain Python. I would like to know is there a better way than the naive max, min over the array or built-in method to solve the problem. points = [[1, 3], [2, 4], [4, 1], [3, 3], [1, 6]] b = bounds(points) # the function I am looking for # now b = [[1, 1], [4, 6]] 回答1: My approach to getting performance is to push things down to C level

问题 Do you think that running set of automatic tests based on valgrind's tool suite makes sense? Did you hear about or see such setup in action? What automatic (free from human intuition) actions could such setup perform? 回答1: This would make sense if you were checking for memory problems / bad code as part of unit testing or final build testing. There may be two approaches: writing a test tool that will use valgrind's API through its library, pretty much creating a custom front-end replacing the

问题 Just wondering what the best tool is to really check JS scripts and look at ways of improving overall performance to the "utmost maximum" in terms of size and speed? 回答1: I like using Firebug's profiler for improving overall speed. It'll show you how many times each function is called, how long it took to execute (average and overall), and the percentage of the total JS execution time the function took. I'm not a big fan of micro-optimization, so I don't use any tools to get the "utmost

问题 I have producer and consumer threads in my application and I need to profile them to see the performance of the threads, time taken before each goes to sleep and waits, etc., and take corrective action to improve the over all efficiency of the application. Any suggestions on how to go about this? 回答1: Personally I use YourKit java profiler. It has an excellent thread profiler tool that graphically shows the state of each thread at any given time, relative to one another (among other things).

问题 This question is not specific to C++, AFAIK certain runtimes like Java RE can do profiled-guided optimization on the fly, I'm interested in that too. MSDN describes PGO like this: I instrument my program and run it under profiler, then the compiler uses data gathered by profiler to automatically reorganize branching and loops in such way that branch misprediction is reduced and most often run code is placed compactly to improve its locality Now obviously profiling result will depend on a

问题 We have a lot of reports that are generated via VBA & Excel. Only a small percentage of the reports are actual calculations - the majority of the work is sql calls and formatting/writing of cells. The longest of which takes several hours, the majority takes around 20-30 mins each. The VBA/Excel code plugs into a dll that the VB6 desktop apps use - it's here that all the sql calls are made. While I am sure that there is room for improvement here, it's not this that concerns me - the desktop