Differences

This shows you the differences between two versions of the page.

--- gpu [2017/10/11 13:07]
bojar people should not set CUDA_VISIBLE_DEVICES
+++ gpu [2018/06/12 12:42]
vodrazka [Servers with GPU units]
@@ Line 4: / Line 4: @@
 ===== Servers with GPU units =====
-GPU cluster ''gpu.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)|
+| dll1 |  GeForce GTX 1080 |  396.24 |  6.1 |  8 |  8 |  249 |
+| dll2 |  GeForce GTX 1080 |  396.24 |  6.1 |  8 |  8 |  249 |
+| dll3 |  GeForce GTX 1080 Ti |  396.24 |  6.1 |  9 |  11 |  249 |
+| dll4 |  GeForce GTX 1080 Ti |  396.24 |  6.1 |  10 |  11 |  249 |
+| dll5 |  GeForce GTX 1080 Ti |  396.24 |  6.1 |  10 |  11 |  249 |
+| dll6 |  GeForce GTX 1080 Ti |  396.24 |  6.1 |  9 |  11 |  123 |
+To be migrated to new cluster:
+| titan-gpu |  GeForce GTX TITAN Z |  381.22 |  3.5 |  2 |  6 |  31 |
+| kronos |  GeForce GTX 1080 Ti |  384.81 |  6.1 |  1 |  11 |  125 |
+| titan |  GeForce GTX 1080 |  381.22 |  6.1 |  1 |  8 |  31 |
+| twister1 |  Tesla K40c |  ? |  ? |  1 | 11 |  47 |
+| twister2 |  Tesla K40c |  384.81 |  3.5 |  1 |  11 |  47 |
-| machine                    | GPU type | [[https://en.wikipedia.org/wiki/CUDA#GPUs_supported|cc]] | GPUs | GPU RAM | Comment |
-| iridium                    | Quadro K2000        |  cc3.0 |   1|   2 GB |  |
-| titan-gpu                  | GeForce GTX Titan Z |  cc3.5 |   2|   6 GB |  |
-| twister1; twister2; kronos | Tesla K40c          |  cc3.5 |   1|  12 GB |  |
-| dll1; dll2                 | GeForce GTX 1080    |  cc6.1 |   8|   8 GB |  |
-| titan                      | GeForce GTX 1080    |  cc6.1 |   1|   8 GB |  |
-| dll3; dll4; dll5           | GeForce GTX 1080 Ti |  cc6.1 |  10|  11 GB | dll3 has only 9 GPUs since 2017/07 |
-| dll6                       | GeForce GTX 1080 Ti |  cc6.1 |   3|  11 GB |  |
 Desktop machines:
@@ Line 23: / Line 31: @@
 All machines have CUDA8.0 and should support both Theano and TensorFlow.
-=== Disk space ===
+[[https://ufaladm2.ufal.hide.ms.mff.cuni.cz/munin/ufal.hide.ms.mff.cuni.cz/lrc-headnode.ufal.hide.ms.mff.cuni.cz/index.html#dll|GPU usage rolling graphs]]
-All the GPU machines are at Malá Strana (not at Troja), so you should not use ''/lnet/tspec/work/'', but you should use:
-- ''/lnet/spec/work/'' (alias ''/net/work/'') - Lustre disk space at Malá Strana
-- ''/net/cluster/TMP'' - NFS hard disk for temporary files, so slower than Lustre for most tasks
-- ''/net/cluster/SSD'' - also NFS, but faster then TMP because of SSD
-- ''/COMP.TMP'' - local (for each machine) space for temporary files (use it instead of ''/tmp''; over-filling ''/COMP.TMP'' should not halt the system).
-=== Individual acquisitions: NVIDIA Academic Hardware Grants ==
-There is an easy way to get one high-end GPU: [[https://developer.nvidia.com/academic_gpu_seeding|ask NVIDIA for an Academic Hardware Grant]]. All it takes is writing a short grant application (at most ~2 hrs of work from scratch; if you have a GAUK, ~15 minutes of copy-pasting). Due to the GPU housing issues (mainly rack space and cooling), Milan F. said we should request the Tesla-line cards (2017 check with Milan about this issue). If you want to have a look at an application, feel free to ask at hajicj@ufal.mff.cuni.cz :)
+===== Rules =====
+  * 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''.
+  * 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.q -l gpu=1,gpu_cc_min3.5=1,gpu_ram=2G</code>
+  * 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.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.
+  * 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 ''$CUDA_VISIBLE_DEVICES'' will not be set correctly. E.g. <code>qrsh -q gpu.q -l gpu=1,gpu_ram=2G -pty yes bash</code>In general: don't reserve a GPU (as described above) without actually using it for longer time. (E.g. try separating steps which need GPU and steps which do not and execute those separately on our GPU resp. CPU cluster.) Ondřej Bojar has a script /home/bojar/tools/servers/watch_gpus for watching reserved but unused GPU on most machines which will e-mail you, but don't rely on in only.
+  * Note that the dll machines have typically 10 cards, but "just" 250 GB RAM (DLL6 has only 128 GB). So the expected (maximal) ''mem_free'' requirement for jobs with 1 GPU is 25GB. If your 1-GPU job takes e.g. 80 GB and you submit three such jobs on the same machine, you have effectively blocked the whole machine and seven GPUs remain unused. If you really need to submit more high-memory jobs, send each on different machine.
-Take care, however, to coordinate the grant applications a little, so that not too many arrive from UFAL within a short time: these grants are explicitly //not// intended to build GPU clusters, they are "seeding" grants meant for researchers to try out GPUs (and fall in love with them, and buy a cluster later). If you are planning to submit the hardware grant, have submitted one, or have already been awarded one, please add yourself here.
+===== How to use cluster =====
-Known NVIDIA Academic Hardware Grants:
+==== Set-up CUDA and CUDNN ====
-  * Ondřej Plátek - granted (2015)
+You should add the following commands into your ~/.bashrc
-  * Jan Hajič jr. - granted (early 2016)
+  CUDNN_version=6.0
+  CUDA_version=8.0
+  CUDA_DIR_OPT=/opt/cuda-$CUDA_version
+  if [ -d "$CUDA_DIR_OPT" ] ; then
+    CUDA_DIR=$CUDA_DIR_OPT
+    export CUDA_HOME=$CUDA_DIR
+    export THEANO_FLAGS="cuda.root=$CUDA_HOME,device=gpu,floatX=float32"
+    export PATH=$PATH:$CUDA_DIR/bin
+    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDA_DIR/cudnn/$CUDNN_version/lib64:$CUDA_DIR/lib64
+    export CPATH=$CUDA_DIR/cudnn/$CUDNN_version/include:$CPATH
+  fi
+When not using Theano, just Tensorflow this can be simplified to ''export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda-8.0/cudnn/6.0/lib64:/opt/cuda-8.0/lib64''. Note that on some machines (dll*, twister*), this is the current default even without setting LD_LIBRARY_PATH, but on other machines (kronos, titan, titan-gpu, iridium) you need to set LD_LIBRARY_PATH explicitly.
-===== How to use cluster =====
+TensorFlow 1.5 precompiled binaries need CUDA 9.0, for this you need to
-In this section will be explained how to use cluster properly.
+  export LD_LIBRARY_PATH=/opt/cuda-9.0/lib64/:/opt/cuda/cudnn/7.0/lib64/
-==== Set-up CUDA and CUDNN ====
+You also need to use ''qsub -q gpu.q@dll[256]'' because only those machines have drivers which support CUDA 9.
-You can add following command into your ~/.bashrc
+**THE NEW CLUSTER (SGE 8.1.9)**
-  CUDNN_version=6.0
+Multiple versions of ''cuda'' can be accessed in ''/opt/cuda''. **Compared to the old cluster there is a difference in setting the CUDA_DIR_OPT variable!!**
-  CUDA_version=8.0
-  CUDA_DIR_OPT=/opt/cuda-$CUDA_version
+You need to set library path from your ''~/.bashrc'':
+  CUDNN_version=7.0
+  CUDA_version=9.0
+  CUDA_DIR_OPT=/opt/cuda/$CUDA_version
   if [ -d "$CUDA_DIR_OPT" ] ; then
     CUDA_DIR=$CUDA_DIR_OPT
@@ Line 63: / Line 85: @@
     export CPATH=$CUDA_DIR/cudnn/$CUDNN_version/include:$CPATH
   fi
+  * When not using Theano, just Tensorflow this can be simplified to ''export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/cuda/9.0/lib64:/opt/cuda/9.0/cudnn/7.0/lib64''.
+  * There is no default and you always need to set ''LD_LIBRARY_PATH'' explicitly.
+  * Note that ''cudnn'' library is compiled for specific version of ''cuda''. If you need specific version of ''cudnn'', you can look in ''/opt/cuda/$CUDA_version/cudnn/'' whether it is available for given ''$CUDA_version''.
 ==== TensorFlow Environment ====
@@ Line 77: / Line 109: @@
 And then you can activate your environment:
-  source activate tf1
+  source activate tf18
-  source activate tf1cpu
+  source activate tf18cpu
-This environment have TensorFlow 1.0 and all necessary requirements for NeuralMonkey.
+This environment have TensorFlow 1.8.0 and all necessary requirements for NeuralMonkey.
-==== Using cluster ====
+==== Pytorch Environment ====
-Rule number one, always use the GPU queue (never log in machine by ssh). Always use qsub or qsubmit with proper arguments.
+If you want to use pytorch, there is a ready-made environment in
-For testing and using the cluster interactively you can use qrsh (this should not be used for long running experiments since the console is not closed on the end of the experiment). Following command will assign you a GPU and creates interactive console.
+  /home/hajicj/anaconda3/envs/pytorch/bin
-  qrsh -q gpu.q -l gpu=1,gpu_ram=2G -pty yes bash
-For running experiments you must use qsub command:
+It does rely on the CUDA and CuDNN setup above.
-  qsub -q gpu.q -l gpu=1,gpu_cc_min3.5=1,gpu_ram=2G WHAT_SHOULD_BE_RUN
+==== Using cluster ====
-Cleaner way to use cluster is with /home/bojar/tools/shell/qsubmit
+As an alternative to ''qsub'', you can use /home/bojar/tools/shell/qsubmit
   qsubmit --gpumem=2G --queue="gpu.q" WHAT_SHOULD_BE_RUN
-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.
+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.
 ==== Basic commands ====
@@ Line 137: / Line 167: @@
 | titan      | GeForce GTX 1080 Ti                |   10:45:11 | (new result with correct CUDA version) |
 | dll1       | (2 GPU) GeForce GTX 1080; cc6.1    |   12:34:34 | Probably only one GPU was used |
+| twister2   | Quadro P5000                       |   13:19:00 |  |
 | dll2       | GeForce GTX 1080; cc6.1            |   13:01:05 | Only one GPU was used |
 | titan-gpu  | (2 GPU) GeForce GTX Titan Z; cc3.5 |   16:05:24 | Probably only one GPU was used |
@@ Line 154: / Line 185: @@
 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.
-| GPU; Cuda capability       | GPU RAM |  Walltime | Batch size | Machine |
+| GPU; Cuda capability       | GPU RAM |  Walltime | Batch size | Machine   |
-| Tesla K40c; cc3.5          |   12 GB |           |        |  |
+| GeForce GTX 1080 Ti; cc6.1 |   11 GB |  00:55:56 |       2300 | dll5      |
-| 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 1080; cc6.1    |    8 GB |  01:10:57 |       1700 | dll1 |
+| Quadro P5000               |   16 GB |  01:17:00 |       3400 | twister2  |
 | 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 |
+| Quadro K2000; cc3.0        |    2 GB |  28:15:26 |         50 | iridium   |
 ===== Links =====
@@ Line 168: / Line 199: @@
 GPU specs for those GPUs we have:
   * [[http://www.nvidia.com/content/PDF/kepler/Tesla-K40-Active-Board-Spec-BD-06949-001_v03.pdf|Tesla K40c]]
+==== Individual acquisitions: NVIDIA Academic Hardware Grants ====
+There is an easy way to get one high-end GPU: [[https://developer.nvidia.com/academic_gpu_seeding|ask NVIDIA for an Academic Hardware Grant]]. All it takes is writing a short grant application (at most ~2 hrs of work from scratch; if you have a GAUK, ~15 minutes of copy-pasting). Due to the GPU housing issues (mainly rack space and cooling), Milan F. said we should request the Tesla-line cards (2017 check with Milan about this issue). If you want to have a look at an application, feel free to ask at hajicj@ufal.mff.cuni.cz :)
+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 //not// intended to build GPU clusters, they are "seeding" grants meant for researchers to try out GPUs (and fall in love with them, and buy a cluster later). If you are planning to submit the hardware grant, have submitted one, or have already been awarded one, please add yourself here.
+Known NVIDIA Academic Hardware Grants:
+  * Ondřej Plátek - granted (2015)
+  * Jan Hajič jr. - granted (early 2016)

[ Back to the navigation ] [ Back to the content ]

Institute of Formal and Applied Linguistics Wiki

Differences