| Both sides previous revision
Previous revision
Next revision
|
Previous revision
|
gpu [2018/10/02 16:41]
kocmi [Set-up CUDA and CUDNN] |
gpu [2024/10/02 15:21] (current)
popel |
| |
| This page summarizes which UFAL servers have some GPU card, and suggests basic diagnostic commands, paths to installed tools, etc., simply everything necessary at the very beginning of using GPUs for experiments. | This page summarizes which UFAL servers have some GPU card, and suggests basic diagnostic commands, paths to installed tools, etc., simply everything necessary at the very beginning of using GPUs for experiments. |
| | |
| | **TODO: IN 2022 MOVING FROM SGE TO SLURM** (see [[slurm|slurm guidelines]]) -- **commands like ''qsub'' and ''qstat'' will stop working!** |
| | |
| | **IN 2024: Newly, all the documentation is at a dedicated wiki https://ufal.mff.cuni.cz/lrc (you need to use ufal and [[internal:welcome-at-ufal#small-linguistic-password|small-linguistic password]] to access the wiki from outside of the UFAL network).*** |
| |
| ===== Servers with GPU units ===== | ===== Servers with GPU units ===== |
| GPU cluster ''gpu-ms.q'' at Malá Strana: | GPU cluster ''gpu-ms.q'' at Malá Strana: |
| | machine | GPU type | GPU driver version | [[https://en.wikipedia.org/wiki/CUDA#GPUs_supported|cc]] | GPU cnt | GPU RAM (GB) | machine RAM (GB)| AVX | | |
| | dll1 | GeForce GTX 1080 | 396.24 | 6.1 | 8 | 8 | 249 | yes | | | machine | GPU type | GPU driver version | [[https://en.wikipedia.org/wiki/CUDA#GPUs_supported|cc]] | GPU cnt | GPU RAM (GB) | machine RAM (GB)| |
| | dll2 | GeForce GTX 1080 | 396.24 | 6.1 | 8 | 8 | 249 | yes | | | dll1 | Quadro RTX 5000 | 455.23.05 | 7.5 | 8 | 16 | 366.2 | |
| | dll3 | GeForce GTX 1080 Ti | 396.24 | 6.1 | 9 | 11 | 249 | yes | | | dll3 | NVIDIA RTX A4000 | 510.73.08 | 8.6 | 10 | 16 | 248.8 | |
| | dll4 | GeForce GTX 1080 Ti | 396.24 | 6.1 | 10 | 11 | 249 | yes | | | dll4 | GeForce GTX 1080 Ti | 455.23.05 | 6.1 | 10 | 11 | 248.8 | |
| | dll5 | GeForce GTX 1080 Ti | 396.24 | 6.1 | 10 | 11 | 249 | yes | | | dll5 | GeForce GTX 1080 Ti | 455.23.05 | 6.1 | 10 | 11 | 248.8 | |
| | dll6 | GeForce GTX 1080 Ti | 396.24 | 6.1 | 9 | 11 | 123 | yes | | | dll6 | NVIDIA RTX A4000 | 510.73.08 | 8.6 | 8 | 16 | 248.8 | |
| | kronos | GeForce GTX 1080 Ti | 396.24 | 6.1 | 1 | 11 | 123 | yes | | | dll7 | NVIDIA RTX A4000 | 510.73.08 | 8.6 | 8 | 16 | 248.8 | |
| | titan1 | GeForce GTX 1080 | 396.24 | 6.1 | 1 | 8 | 30 | yes | | | dll8 | Quadro RTX 5000 | 455.23.05 | 7.5 | 8 | 16 | 366.2 | |
| | titan2 | Tesla K40c | 396.24 | 3.5 | 1 | 11 | 30 | yes | | | dll9 | GeForce RTX 3090 | 455.23.05 | 8.6 | 4 | 25 | 183.0 | |
| | twister1 | Tesla K40c | 396.24 | 3.5 | 1 | 11 | 45 | no | | | dll10 | GeForce RTX 3090 | 455.23.05 | 8.6 | 4 | 25 | 183.0 | |
| | twister2 | Tesla K40c | 396.24 | 3.5 | 1 | 11 | 45 | no | | |
| | GPU cluster ''gpu-troja.q'' at Troja: |
| | |
| | | machine | GPU type | GPU driver version | [[https://en.wikipedia.org/wiki/CUDA#GPUs_supported|cc]] | GPU cnt | GPU RAM (GB) | machine RAM (GB)| |
| | | tdll1 | Quadro P5000 | 455.23.05 | 6.1 | 8 | 16 | 245.0 | |
| | | tdll2 | Quadro P5000 | 455.23.05 | 6.1 | 8 | 16 | 245.0 | |
| | | tdll3 | Quadro P5000 | 455.23.05 | 6.1 | 8 | 16 | 245.0 | |
| | | tdll4 | Quadro P5000 | 455.23.05 | 6.1 | 8 | 16 | 245.0 | |
| | | tdll5 | Quadro P5000 | 455.23.05 | 6.1 | 8 | 16 | 245.0 | |
| |
| Desktop machines: | Desktop machines: |
| | athena | GeForce GTX 1080 | cc6.1 | 1 | 8 GB | Tom's desktop machine | | | athena | GeForce GTX 1080 | cc6.1 | 1 | 8 GB | Tom's desktop machine | |
| |
| Not used at the moment: GeForce GTX 570 (from twister2) | |
| Multiple versions of CUDA library are accessible on each machine together with cudnn. Theano and TensorFlow is supported. | Multiple versions of CUDA library are accessible on each machine together with cudnn. Theano and TensorFlow is supported. |
| |
| [[https://ufaladm2.ufal.hide.ms.mff.cuni.cz/munin/ufal.hide.ms.mff.cuni.cz/lrc-master.ufal.hide.ms.mff.cuni.cz/index.html#dll|GPU usage rolling graphs]] | [[http://ufaladm2.ufal.hide.ms.mff.cuni.cz/munin/ufal.hide.ms.mff.cuni.cz/lrc-master.ufal.hide.ms.mff.cuni.cz/index.html#dll|GPU usage rolling graphs]] |
| |
| |
| ===== Rules ===== | ===== Rules ===== |
| * First, read [[internal:Linux network]] and [[:Grid]]. | * First, read [[internal:Linux network]] and [[:Grid]]. |
| * 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 ''qsub'' (or ''qrsh''), never via ''ssh''. You can ssh to any machine e.g. to run ''nvidia-smi'' or ''htop'', but not to start computing on GPU. Don't forget to specify you RAM requirements with e.g. ''-l mem_free=8G,act_mem_free=8G,h_vmem=12G''. | * 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 ''qsub'' (or ''qrsh''), never via ''ssh''. You can ssh to any machine e.g. to run ''nvidia-smi'' or ''htop'', but not to start computing on GPU. Don't forget to specify you RAM requirements with e.g. ''-l mem_free=8G,act_mem_free=8G,h_data=12G''. |
| * Always specify the number of GPU cards (e.g. ''gpu=1''), the minimal Cuda capability you need (e.g. ''gpu_cc_min3.5=1'') and your GPU memory requirements (e.g. ''gpu_ram=2G''). Thus e.g. <code>qsub -q gpu-ms.q -l gpu=1,gpu_cc_min3.5=1,gpu_ram=2G</code> | * **Note that you need to use ''h_data'' instead of ''h_vmem'' for GPU jobs.** CUDA driver allocates a lot of "unused" virtual memory (tens of GB per card), which is counted in ''h_vmem'', but not in ''h_data''. All usual allocations (''malloc'', ''new'', Python allocations) seem to be included in ''h_data''. |
| * If you need more than one GPU card (on a single machine), always require as many CPU cores (''-pe smp X'') as many GPU cards you need. E.g. <code>qsub -q gpu-ms.q -l gpu=4,gpu_cc_min3.5=1,gpu_ram=7G -pe smp 4</code> **Warning**: currently, this does not work, so you can omit the ''-pe smp X'' part. Milan Fučík is working on a fix. | * Always specify the number of GPU cards (e.g. ''gpu=1'') and your GPU memory requirements (e.g. ''gpu_ram=2G''). Thus e.g. <code>qsub -q 'gpu*' -l gpu=1,gpu_ram=2G</code> |
| * For interactive jobs, you can use ''qrsh'', but make sure to end your job as soon as you don't need the GPU (so don't use qrsh for long training). **Warning: ''-pty yes bash'' is necessary**, otherwise the variable ''$CUDA_VISIBLE_DEVICES'' will not be set correctly. E.g. <code>qrsh -q gpu-ms.q -l gpu=1,gpu_ram=2G -pty yes bash</code>In general: don't reserve a GPU (as described above) without actually using it for longer time. (E.g. try separating steps which need GPU and steps which do not and execute those separately on our GPU resp. CPU cluster.) Ondřej Bojar has a script /home/bojar/tools/servers/watch_gpus for watching reserved but unused GPU on most machines which will e-mail you, but don't rely on in only. | * If you need more than one GPU card (on a single machine), always require at least as many CPU cores (''-pe smp X'') as many GPU cards you need. E.g. <code>qsub -q 'gpu*' -l gpu=4,gpu_ram=7G -pe smp 4</code> |
| * Note that the dll machines have typically 10 cards, but "just" 250 GB RAM (DLL6 has only 128 GB). So the expected (maximal) ''mem_free'' requirement for jobs with 1 GPU is 25GB. If your 1-GPU job takes e.g. 80 GB and you submit three such jobs on the same machine, you have effectively blocked the whole machine and seven GPUs remain unused. If you really need to submit more high-memory jobs, send each on a different machine. | * For interactive jobs, you can use ''qrsh'', but make sure to end your job as soon as you don't need the GPU (so don't use qrsh for long training). **Warning: ''-pty yes bash -l'' is necessary**, otherwise the variable ''$CUDA_VISIBLE_DEVICES'' will not be set correctly. E.g. <code>qrsh -q 'gpu*' -l gpu=1,gpu_ram=2G -pty yes bash -l</code>In general: don't reserve a GPU (as described above) without actually using it for longer time. (E.g. try separating steps which need GPU and steps which do not and execute those separately on our GPU resp. CPU cluster.) Ondřej Bojar has a script /home/bojar/tools/servers/watch_gpus for watching reserved but unused GPU on most machines which will e-mail you, but don't rely on it only. |
| | * Note that the dll machines have typically 10 cards, but "just" 250 GB RAM (DLL7 has only 128 GB). So the expected (maximal) ''mem_free'' requirement for jobs with 1 GPU is 25GB. If your 1-GPU job takes e.g. 80 GB and you submit three such jobs on the same machine, you have effectively blocked the whole machine and seven GPUs remain unused. If you really need to submit more high-memory jobs, send each on a different machine. |
| | * If you know an approximate runtime of your job, please specify it with ''-l s_rt=hh:mm:ss'' - this is a soft constraint so your job won't be killed if it runs longer than specified. It will help SGE to better schedule the jobs, especially multi-gpu reservations (see ''qconf -ssconf''). |
| |
| ===== How to use cluster ===== | ===== How to use cluster ===== |
| You need to set library path from your ''~/.bashrc'': | You need to set library path from your ''~/.bashrc'': |
| |
| CUDNN_version=7.0 | CUDA_version=11.1 |
| CUDA_version=9.0 | CUDNN_version=8.0 |
| CUDA_DIR_OPT=/opt/cuda/$CUDA_version | CUDA_DIR_OPT=/opt/cuda/$CUDA_version |
| if [ -d "$CUDA_DIR_OPT" ] ; then | if [ -d "$CUDA_DIR_OPT" ] ; then |
| fi | fi |
| |
| * When not using Theano, just Tensorflow this can be simplified to ''export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda/9.0/lib64:/opt/cuda/9.0/cudnn/7.0/lib64''. | * When not using Theano, just Tensorflow this can be simplified to ''export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda/10.1/lib64:/opt/cuda/11.1/cudnn/8.0/lib64''. |
| * Note that the ''cudnn'' library is compiled for specific version of ''cuda''. If you need specific version of ''cudnn'', you can look in ''/opt/cuda/$CUDA_version/cudnn/'' whether it is available for a given ''$CUDA_version''. | * Note that the ''cudnn'' library is compiled for specific version of ''cuda''. If you need specific version of ''cudnn'', you can look in ''/opt/cuda/$CUDA_version/cudnn/'' whether it is available for a given ''$CUDA_version''. |
| |
| This environment have TensorFlow 1.8.0 and all necessary requirements for NeuralMonkey. | This environment have TensorFlow 1.8.0 and all necessary requirements for NeuralMonkey. |
| |
| ==== Pytorch Environment ==== | ==== PyTorch Environment ==== |
| |
| If you want to use pytorch, there is a ready-made environment in | Install PyTorch following the instructions on https://pytorch.org. |
| |
| /home/hajicj/anaconda3/envs/pytorch/bin | At the time of writing, the recommended setup for CUDA 11.1 (supported by all GPU nodes) is: |
| | |
| It does rely on the CUDA and CuDNN setup above. | |
| |
| | pip3 install torch==1.10.0+cu111 torchvision==0.11.1+cu111 torchaudio==0.10.0+cu111 -f https://download.pytorch.org/whl/cu111/torch_stable.html |
| ==== Using cluster ==== | ==== Using cluster ==== |
| |
| qsubmit --gpumem=2G --queue="gpu-ms.q" WHAT_SHOULD_BE_RUN | qsubmit --gpumem=2G --queue="gpu-ms.q" WHAT_SHOULD_BE_RUN |
| | |
| It is recommended to use priority lower than the default -100 if you are not rushing for the results and don't need to leap over your colleagues jobs. | It is recommended to use priority lower than the default -100 if you are not rushing for the results and don't need to leap over your colleagues jobs. Please, do not use priority between -99 to 0 for jobs taking longer than a few hours, unless it is absolutely necessary for your work. |
| ==== Basic commands ==== | ==== Basic commands ==== |
| |
| watch nvidia-smi | watch nvidia-smi |
| # For monitoring GPU activity in a separate terminal (thanks to Jindrich Libovicky for this!) | # For monitoring GPU activity in a separate terminal (thanks to Jindrich Libovicky for this!) |
| | # You can also use nvidia-smi -l TIME |
| nvcc --version | nvcc --version |
| # this should tell CUDA version | # this should tell CUDA version |
