rogerallen / cvg Goto Github PK
View Code? Open in Web Editor NEWCPU vs GPU SGEMM perf test
cvg's Introduction
cvg - CPU vs. GPU SGEMM performance test by Roger Allen & Rick Wirch Feb 29, 2016 This is just a toy perf app for developers. I wanted a simple way to compare CPU vs. GPU SGEMM performance using Intel MKL, NVIDIA CUBLAS & CUBLAS-XT libraries. It started out simple, but like so many things, got a bit more complex over time... Since this is for developers, please just read the code to see what it is doing. NOTE: built assuming Intel MKL V11.3 Example usage & output: usage: cvg <options> options: -t g|s : test type: GEMM, SYRK+GEMM (default GEMM) -s c|g|x : API select: CPU (Intel MKL), GPU (CUBLAS), GPU (CUBLAS-XT) -p s|d : test precision: single (default) or double -l # : loops of the algorithm (default 1) -m # : matrix M dimension (default 1024) -n # : matrix N dimension (default 1024) -k # : matrix K dimension (default 2048) -b # : CUBLAS-XT block_dim (default 1024) -A # : alpha GEMM parameter (default 1.11) -B # : beta GEMM parameter (default 0.91) -g # : [CUBLAS-XT only] gpu id(s) to use. can specify multiple -g options. (default=0) ================================================================================ CPU SGEMM: ================================================================================ > x64/Debug/cvg.exe -s c -l 4 -m 6400 -n 6400 -k 26000 Intel MKL sgemm: loops=4 M=6400 N=6400 K=26000 Result: 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 ...[snip]... 0.00000 28719.44922 22169.48633 22673.35742 20154.22656 12596.54492 20154.33008 22673.64648 ...[snip]... 0.00000 22169.48633 24184.95898 20154.36719 20154.42969 10076.95996 20154.09180 20154.15234 ...[snip]... 0.00000 22673.35742 20154.36719 28719.94336 18138.64063 12596.54395 22169.84375 16627.07617 ...[snip]... 0.00000 20154.22656 20154.42969 18138.64063 24185.12891 10076.95898 16123.51660 22169.87695 ...[snip]... 0.00000 12596.54492 10076.95996 12596.54395 10076.95898 12596.49316 10076.95898 12596.54492 ...[snip]... 0.00000 20154.33008 20154.09180 22169.84375 16123.51660 10076.95898 24185.12891 18138.64258 ...[snip]... 0.00000 22673.64648 20154.15234 16627.07617 22169.87695 12596.54492 18138.64258 28719.93555 ...[snip]... ...[snip]... SGEMM: [6400x26000] * [26000x6400] + [6400x6400] seconds: 24.031000 Gigabytes: 1.5 Gigaflops: 8519.7 Gigaflops/s: 354.5 ================================================================================ GPU CUBLAS SGEMM: ================================================================================ > x64/Debug/cvg.exe -s g -l 4 -m 6400 -n 6400 -k 26000 NVIDIA CUBLAS sgemm: loops=4 M=6400 N=6400 K=26000 alpha=1.110000 beta=0.910000 Device Number: 0 Device name: GeForce GTX 960 Memory Clock Rate (KHz): 3505000 Memory Bus Width (bits): 128 Peak Memory Bandwidth (GB/s): 112.160000 Result: 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 ...[snip]... 0.00000 28720.24805 22170.63477 22672.59961 20156.65430 12596.54688 20155.56250 22672.87500 ...[snip]... 0.00000 22170.63477 24186.18359 20156.79102 20155.93164 10076.96289 20155.51758 20155.42578 ...[snip]... 0.00000 22672.59961 20156.79102 28720.79492 18140.14648 12596.54688 22171.04297 16626.54688 ...[snip]... 0.00000 20156.65430 20155.93164 18140.14648 24186.32227 10076.96289 16121.77734 22171.04492 ...[snip]... 0.00000 12596.54688 10076.96289 12596.54688 10076.96289 12596.54688 10076.96289 12596.54590 ...[snip]... 0.00000 20155.56250 20155.51758 22171.04297 16121.77734 10076.96289 24186.32422 18140.14844 ...[snip]... 0.00000 22672.87500 20155.42578 16626.54688 22171.04492 12596.54590 18140.14844 28720.79492 ...[snip]... ...[snip]... SGEMM: [6400x26000] * [26000x6400] * 1.11 + [6400x6400] * 0.91 seconds: 4.842000 Gigabytes: 1.5 Gigaflops: 8519.7 Gigaflops/s: 1759.5 ON DEVICE TIME:Result: 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 ...[snip]... 0.00000 28720.24805 22170.63477 22672.59961 20156.65430 12596.54688 20155.56250 22672.87500 ...[snip]... 0.00000 22170.63477 24186.18359 20156.79102 20155.93164 10076.96289 20155.51758 20155.42578 ...[snip]... 0.00000 22672.59961 20156.79102 28720.79492 18140.14648 12596.54688 22171.04297 16626.54688 ...[snip]... 0.00000 20156.65430 20155.93164 18140.14648 24186.32227 10076.96289 16121.77734 22171.04492 ...[snip]... 0.00000 12596.54688 10076.96289 12596.54688 10076.96289 12596.54688 10076.96289 12596.54590 ...[snip]... 0.00000 20155.56250 20155.51758 22171.04297 16121.77734 10076.96289 24186.32422 18140.14844 ...[snip]... 0.00000 22672.87500 20155.42578 16626.54688 22171.04492 12596.54590 18140.14844 28720.79492 ...[snip]... ...[snip]... SGEMM: [6400x26000] * [26000x6400] * 1.11 + [6400x6400] * 0.91 seconds: 3.328000 Gigabytes: 1.5 Gigaflops: 8519.7 Gigaflops/s: 2560.0 ================================================================================ GPU CUBLASXT SGEMM ================================================================================ > x64/Debug/cvg.exe -s x -l 4 -m 6400 -n 6400 -k 26000 -b 8192 NVIDIA CUBLASXT sgemm: loops=4 M=6400 N=6400 K=26000 block_dim=8192 Device Number: 0 Device name: GeForce GTX 960 Memory Clock Rate (KHz): 3505000 Memory Bus Width (bits): 128 Peak Memory Bandwidth (GB/s): 112.160000 Result: 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 ...[snip]... 0.00000 28719.88281 22169.25781 22673.02734 20154.70898 12596.54688 20154.59180 22673.30469 ...[snip]... 0.00000 22169.25781 24185.44727 20154.84766 20153.60938 10076.96289 20153.19922 20154.45313 ...[snip]... 0.00000 22673.02734 20154.84766 28720.43359 18138.39258 12596.54590 22169.67188 16626.87891 ...[snip]... 0.00000 20154.70898 20153.60938 18138.39258 24185.58398 10076.96289 16123.07813 22169.67383 ...[snip]... 0.00000 12596.54688 10076.96289 12596.54590 10076.96289 12596.54590 10076.96289 12596.54590 ...[snip]... 0.00000 20154.59180 20153.19922 22169.67188 16123.07813 10076.96289 24185.58008 18138.39063 ...[snip]... 0.00000 22673.30469 20154.45313 16626.87891 22169.67383 12596.54590 18138.39063 28720.43750 ...[snip]... ...[snip]... SGEMM: [6400x26000] * [26000x6400] + [6400x6400] seconds: 5.296000 Gigabytes: 1.5 Gigaflops: 8519.7 Gigaflops/s: 1608.7
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.