Differences
This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
gpu [2017/03/16 16:30] kocmanek [Servers with GPU units] |
gpu [2017/11/13 09:43] bojar [Rules] |
||
---|---|---|---|
Line 4: | Line 4: | ||
===== Servers with GPU units ===== | ===== Servers with GPU units ===== | ||
+ | GPU cluster '' | ||
- | | machine | GPU; [[https:// | + | | machine |
- | | titan-gpu | + | | iridium |
- | | twister1 | + | | titan-gpu |
- | | twister2 | + | | twister1; twister2; kronos |
- | | kronos-dev | Tesla K40c; cc3.5 | + | | dll1; dll2 | GeForce GTX 1080 |
- | | iridium | + | | titan |
- | | victoria | + | | dll3; dll4; dll5 | GeForce GTX 1080 Ti | |
- | | arc | GeForce GT 630; cc3.0 | 1 | 2 GB | Ales's desktop machine | | + | | dll6 | GeForce GTX 1080 Ti | |
- | | athena | + | |
- | | dll1 | GeForce GTX 1080; cc6.1 | 8 | 8 GB each core | | | + | |
- | | dll2 | GeForce GTX 1080; cc6.1 | 8 | 8 GB each core | | | + | |
- | not used at the moment: GeForce GTX 570 (from twister2) | + | Desktop machines: |
+ | | machine | ||
+ | | victoria; arc | GeForce GT 630 | cc3.0 | 1 | 2 GB | desktop machine | | ||
+ | | athena | ||
+ | |||
+ | Not used at the moment: GeForce GTX 570 (from twister2) | ||
All machines have CUDA8.0 and should support both Theano and TensorFlow. | All machines have CUDA8.0 and should support both Theano and TensorFlow. | ||
- | Summary of future plans: | + | ===== Rules ===== |
- | * Current Troja servers won't get any GPUs (the only option would be [[http://www.czc.cz/hp-quadro-k1200-4gb/171662/ | + | * First, read [[internal:Linux network]] and [[:Grid]]. |
- | * The old Quadro K2000 we have is a much more low end piece, so we can' | + | * All the rules from [[:Grid]] apply, even more strictly than for CPU because there are too many GPU users and not as many GPUs available. So as a reminder: always use GPUs via '' |
- | | + | * Always specify the number of GPU cards (e.g. '' |
- | * It is impossible (wasteful in terms of space and forbidden by a dean regulation) to put non-rack machines to our servers rooms. So we won' | + | * If you need more than one GPU card (on a single machine), always require as many CPU cores ('' |
- | * Yes, there are grant applications under review which include rack machines with GPUs, e.g. 5x2 or something like that; more will be known in 2017. | + | * For interactive jobs, you can use '' |
+ | ===== How to use cluster ===== | ||
- | === Individual acquisitions: | + | ==== Set-up CUDA and CUDNN ==== |
- | There is an easy way to get one high-end GPU: [[https:// | + | You can add following command into your ~/.bashrc |
- | Take care, however, to coordinate the grant applications a little, so that not too many arrive from UFAL within a short time: these grants are explicitly | + | CUDNN_version=6.0 |
+ | CUDA_version=8.0 | ||
+ | CUDA_DIR_OPT=/opt/cuda-$CUDA_version | ||
+ | if [ -d "$CUDA_DIR_OPT" | ||
+ | CUDA_DIR=$CUDA_DIR_OPT | ||
+ | export CUDA_HOME=$CUDA_DIR | ||
+ | export THEANO_FLAGS=" | ||
+ | export PATH=$PATH: | ||
+ | export LD_LIBRARY_PATH=$LD_LIBRARY_PATH: | ||
+ | export CPATH=$CUDA_DIR/ | ||
+ | fi | ||
- | Known NVIDIA Academic Hardware Grants: | + | ==== TensorFlow Environment ==== |
- | * Ondřej Plátek - granted (2015) | + | Majority people at UFAL use TensorFlow. To start using it you need to create python virtual environment |
- | * Jan Hajič jr. - granted (early 2016) | + | |
- | * Jindra Helcl - planning | + | |
+ | pip install tensorflow | ||
+ | pip install tensorflow-gpu | ||
+ | | ||
+ | You can use prepared environment by adding into your ~/.bashrc | ||
+ | export PATH=/ | ||
+ | |||
+ | And then you can activate your environment: | ||
+ | |||
+ | source activate tf1 | ||
+ | source activate tf1cpu | ||
+ | |||
+ | This environment have TensorFlow 1.0 and all necessary requirements for NeuralMonkey. | ||
+ | |||
+ | ==== Pytorch Environment ==== | ||
+ | |||
+ | If you want to use pytorch, there is a ready-made environment in | ||
+ | |||
+ | / | ||
| | ||
+ | It does rely on the CUDA and CuDNN setup above. | ||
+ | |||
+ | ==== Using cluster ==== | ||
+ | |||
+ | As an alternative to '' | ||
+ | |||
+ | qsubmit --gpumem=2G --queue=" | ||
+ | | ||
+ | It is recommended to use priority -100 if you are not rushing for the results and don't need to leap over your colleagues jobs. | ||
+ | ==== Basic commands ==== | ||
+ | |||
+ | lspci | ||
+ | # is any such hardware there? | ||
+ | nvidia-smi | ||
+ | # more details, incl. running processes on the GPU | ||
+ | # nvidia-* are typically located in /usr/bin | ||
+ | watch nvidia-smi | ||
+ | # For monitoring GPU activity in a separate terminal (thanks to Jindrich Libovicky for this!) | ||
+ | nvcc --version | ||
+ | # this should tell CUDA version | ||
+ | # nvcc is typically installed in / | ||
+ | theano-test | ||
+ | # dela to vubec neco uzitecneho? :-) | ||
+ | # theano-* are typically located in / | ||
+ | / | ||
+ | # shows CUDA capability etc. | ||
+ | ssh dll1; ~popel/ | ||
+ | # who occupies which card on a given machine | ||
+ | | ||
+ | === Select GPU device === | ||
+ | |||
+ | The variable CUDA_VISIBLE_DEVICES constrains tensorflow and other toolkits to compute only on the selected GPUs. **Do not set this variable yourself** (unless debugging SGE), it is set for you automatically by SGE if you ask for some GPUs (see above). | ||
+ | |||
+ | To list available devices, use: | ||
+ | / | ||
===== Performance tests ===== | ===== Performance tests ===== | ||
Line 47: | Line 112: | ||
* [[http:// | * [[http:// | ||
- | In the following table is the experiment conducted by Tom Kocmi. You can replicate experiment: / | + | In the following table is the experiment conducted by Tom Kocmi. You can replicate experiment: / |
- | + | ||
- | I am preparing department-wide | + | |
- | * Athena (GTX 1080) - 2 hodiny 32 minut | + | |
- | * Twister (Tesla K40c) - 6 hodin 46 minut | + | |
| machine | Setup; CPU/GPU; [[https:// | | machine | Setup; CPU/GPU; [[https:// | ||
| athena | | athena | ||
- | | dll2 | + | | dll2 | GeForce GTX 1080; cc6.1 | |
+ | | titan | GeForce GTX 1080 Ti | | ||
| dll1 | (2 GPU) GeForce GTX 1080; cc6.1 | | | dll1 | (2 GPU) GeForce GTX 1080; cc6.1 | | ||
- | | dll2 | + | | dll2 | GeForce GTX 1080; cc6.1 | |
| titan-gpu | | titan-gpu | ||
| kronos-dev | Tesla K40c; cc3.5 | | | kronos-dev | Tesla K40c; cc3.5 | | ||
Line 68: | Line 130: | ||
| arc | GeForce GT 630; cc3.0 | 103:42:30 | (approximated after 66 hours) | | | arc | GeForce GT 630; cc3.0 | 103:42:30 | (approximated after 66 hours) | | ||
| lucifer4 | | lucifer4 | ||
- | | victoria | ||
- | A comparison with Ondrej' | + | === Second Benchmark === |
- | * dll2 (2xGPU) takes 13m for one reporting period | + | |
- | * achilles2 (4xCPU with 8 CPUs reserved) takes 24m for one reporting period | + | |
+ | The previous benchmark only compares the speed of processing units within the GPUs and do not take into account the size of memory. Therefore I have conducted another benchmark, this time for each graphic card I have increased the batch size as much as possible so the model still could fit into the GPU (the previous benchmark model had batch size 20). This way the results should be more representative of the power for each GPU. | ||
- | ===== Installed toolkits ===== | + | | GPU; Cuda capability |
+ | | Tesla K40c; cc3.5 | 12 GB | | ||
+ | | GeForce GTX 1080 Ti; cc6.1 | 11 GB | 00:55:56 | 2300 | dll5 | | ||
+ | | GeForce GTX 1080; cc6.1 | 8 GB | 01:10:57 | 1700 | dll1 | | ||
+ | | GeForce GTX Titan Z; cc3.5 | 6 GB | 02:20:47 | 1100 | titan-gpu | | ||
+ | | Quadro K2000; cc3.0 | 2 GB | 28:15:26 | 50 | iridium | | ||
- | //This should mention where each interesting toolkit lives (on a particular machine).// | + | ===== Links ===== |
- | ==== TensorFlow ==== | + | * [[https:// |
- | [[https:// | ||
- | OP: I created | + | GPU specs for those GPUs we have: |
+ | * [[http://www.nvidia.com/content/PDF/ | ||
- | === Select GPU device | + | ==== Individual acquisitions: |
- | Use variable CUDA_VISIBLE_DEVICES | + | There is an easy way to get one high-end GPU: [[https:// |
- | < | + | |
- | To list available devices, use: | + | Take care, however, to coordinate the grant applications a little, so that not too many arrive from UFAL within a short time: these grants are explicitly |
- | < | + | |
- | ===== Basic commands ===== | + | Known NVIDIA Academic Hardware Grants: |
- | | + | * Ondřej Plátek |
- | # is any such hardware there? | + | |
- | nvidia-smi | + | |
- | # more details, incl. running processes on the GPU | + | |
- | # nvidia-* are typically located in /usr/bin | + | |
- | watch nvidia-smi | + | |
- | # For monitoring GPU activity in a separate terminal | + | |
- | | + | |
- | # this should tell CUDA version | + | |
- | # nvcc is typically installed in / | + | |
- | theano-test | + | |
- | # dela to vubec neco uzitecneho? :-) | + | |
- | # theano-* are typically located in / | + | |
- | / | + | |
- | # shows CUDA capability etc. | + | |
- | ===== Links ===== | ||
- | * [[https:// | ||
- | |||
- | GPU specs for those GPUs we have: | ||
- | * [[http:// |