Differences

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

--- gpu [2017/01/04 13:19]
kocmanek
+++ gpu [2017/10/12 13:42]
ufal [How to use cluster]
@@ Line 4: / Line 4: @@
 ===== Servers with GPU units =====
+GPU cluster ''gpu.q'' at Malá Strana:
-| machine | GPU; [[https://en.wikipedia.org/wiki/CUDA#Supported_GPUs|Capability]] [cc]  | cores | GPU RAM | Comment |
+| machine                    | GPU type | [[https://en.wikipedia.org/wiki/CUDA#GPUs_supported|cc]] | GPUs | GPU RAM | Comment |
-| titan-gpu  | GeForce GTX Titan Z; cc3.5 | 2 | 6 GB each core |        |
+| iridium                    | Quadro K2000        |  cc3.0 |   1|   2 GB |  |
-| twister1   | Tesla K40c; cc3.5          | 1 | 12 GB          |        |
+| titan-gpu                  | GeForce GTX Titan Z |  cc3.5 |   2|   6 GB |  |
-| twister2   | Tesla K40c; cc3.5          | 1 | 12 GB          |        |
+| twister1; twister2; kronos | Tesla K40c          |  cc3.5 |   1|  12 GB |  |
-| kronos-dev | Tesla K40c; cc3.5          | 1 | 12 GB          |        |
+| dll1; dll2                 | GeForce GTX 1080    |  cc6.1 |   8|   8 GB |  |
-| iridium    | Quadro K2000; cc3.0        | 1 | 2 GB           |        |
+| titan                      | GeForce GTX 1080    |  cc6.1 |   1|   8 GB |  |
-| victoria   | GeForce GT 630; cc3.0      | 1 | 2 GB           | Ondrej Bojar's desktop machine |
+| dll3; dll4; dll5           | GeForce GTX 1080 Ti |  cc6.1 |  10|  11 GB | dll3 has only 9 GPUs since 2017/07 |
-| arc        | GeForce GT 630; cc3.0      | 1 | 2 GB           | Ales's desktop machine |
+| dll6                       | GeForce GTX 1080 Ti |  cc6.1 |   3|  11 GB |  |
-| athena     | GeForce GTX 1080; cc6.1    | 1 | 8 GB           | Tom's desktop machine |
-| dll1     | GeForce GTX 1080; cc6.1    | 2 | 8 GB each core |  |
-| dll2     | GeForce GTX 1080; cc6.1    | 2 | 8 GB each core |  |
-not used at the moment: GeForce GTX 570 (from twister2)
+Desktop machines:
-All machines have CUDA8.0 and should support both Theano and TensorFlow.
+| machine                    | GPU type | [[https://en.wikipedia.org/wiki/CUDA#GPUs_supported|cc]] | GPUs | GPU RAM | Comment |
+| victoria; arc              | GeForce GT 630   | cc3.0 |  1 |  2 GB | desktop machine |
+| athena                     | GeForce GTX 1080 | cc6.1 |  1 |  8 GB | Tom's desktop machine |
-Summary of future plans:
+Not used at the moment: GeForce GTX 570 (from twister2)
-  * Current Troja servers won't get any GPUs (the only option would be [[http://www.czc.cz/hp-quadro-k1200-4gb/171662/produkt?ppcbee-adtext-variant=Produkt%3B+kategorie+%2B+cena%3B+Pobo%C4%8Dky&gclid=CKbKkbrWrswCFQUq0wodHDELCw|Quadro K1200 4GB]], horribly cost-inefficient)
+All machines have CUDA8.0 and should support both Theano and TensorFlow.
-  * The old Quadro K2000 we have is a much more low end piece, so we can't test is in Troja.
-  * There is MetaCentrum which also has GPUs, so testing can be done there.
-  * 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't be buying GeForce GTX 1080 (~20000CZK, out of stock now), for a non-rack machine since we most likely don't have any available.
-  * 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.
+=== Disk space ===
+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. If you want to have a look at an application, feel free to ask at hajicj@ufal.mff.cuni.cz :)
+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.
@@ Line 38: / Line 40: @@
   * Ondřej Plátek - granted (2015)
   * Jan Hajič jr. - granted (early 2016)
-  * Jindra Helcl - planning to apply (fall 2016)
-===== Performance tests =====
+===== How to use cluster =====
-* [[http://www.trustedreviews.com/nvidia-geforce-gtx-1080-review-performance-benchmarks-and-conclusion-page-2| 980 vs 1080 vs Titan X (not the Titan Z we have)]]
+In this section will be explained how to use cluster properly.
-In the following table is the experiment conducted by Tom Kocmi. You can replicate experiment: /a/merkur3/kocmanek/GPUBenchmark (you will need to prepare environment of TensorFlow11 or use my ANACONDA)
+==== Set-up CUDA and CUDNN ====
-I am preparing department-wide benchmark, but meanwhile the results for different experiment:
+You can add following command into your ~/.bashrc
- * Athena (GTX 1080) - 2 hodiny 32 minut
- * Twister (Tesla K40c) - 6 hodin 46 minut
-| machine | Setup; CPU/GPU; [[https://en.wikipedia.org/wiki/CUDA#Supported_GPUs|Capability]] [cc] | Walltime | Note |
+  CUDNN_version=6.0
-| athena     | GeForce GTX 1080; cc6.1            |  9:55:58 | Tom's desktop  |
+  CUDA_version=8.0
-| dll1       | (2 GPU) GeForce GTX 1080; cc6.1    |  12:34:34 | Probably only one GPU was used |
+  CUDA_DIR_OPT=/opt/cuda-$CUDA_version
-| dll2       | (2 GPU) GeForce GTX 1080; cc6.1    |  13:01:05 | Only one GPU was used |
+  if [ -d "$CUDA_DIR_OPT" ] ; then
-| titan-gpu  | (2 GPU) GeForce GTX Titan Z; cc3.5 |  16:05:24 | Probably only one GPU was used |
+    CUDA_DIR=$CUDA_DIR_OPT
-| kronos-dev | Tesla K40c; cc3.5                  |  22:41:01 |  |
+    export CUDA_HOME=$CUDA_DIR
-| twister2   | Tesla K40c; cc3.5                  |  22:43:10 |  |
+    export THEANO_FLAGS="cuda.root=$CUDA_HOME,device=gpu,floatX=float32"
-| twister1   | Tesla K40c; cc3.5                  |  24:19:45 |  |
+    export PATH=$PATH:$CUDA_DIR/bin
-| helena1    | 16x CPU                            |  46:33:14 |  |
+    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDA_DIR/cudnn/$CUDNN_version/lib64:$CUDA_DIR/lib64
-| belzebub   | 16x CPU                            |  52:36:56 |  |
+    export CPATH=$CUDA_DIR/cudnn/$CUDNN_version/include:$CPATH
-| iridium    | Quadro K2000; cc3.0                |  59:47:58 |  |
+  fi
-| helena7     | 8x CPU                            |  61:10:00 | approximated (still running) |
-| arc        | GeForce GT 630; cc3.0              |  103:42:30 | (approximated after 66 hours) |
-| lucifer4   | 8x CPU                             |  134:41:22 |  |
-| victoria   | GeForce GT 630; cc3.0              |  --- | never run, same GPU as Arc |
+==== TensorFlow Environment ====
-A comparison with Ondrej's small data set:
+Majority people at UFAL use TensorFlow. To start using it you need to create python virtual environment (virtualenv or use Anaconda for it). Into the environment you must place TensorFlow. The TF is either in CPU or GPU version.
-  * dll2 (2xGPU) takes 13m for one reporting period
-  * achilles2 (4xCPU with 8 CPUs reserved) takes 24m for one reporting period
+  pip install tensorflow
+  pip install tensorflow-gpu
+You can use prepared environment by adding into your ~/.bashrc
-===== Installed toolkits =====
+  export PATH=/a/merkur3/kocmanek/ANACONDA/bin:$PATH
-//This should mention where each interesting toolkit lives (on a particular machine).//
+And then you can activate your environment:
-==== TensorFlow ====
+  source activate tf1
+  source activate tf1cpu
-[[https://redmine.ms.mff.cuni.cz/projects/mmmt/repository/revisions/6a064187fc6959db9b77cf2d5350c5f4918a8067/entry/prepare_env.sh|This script]] installs TensorFlow 0.7.1 (and all other dependencies we need for Multimodal Translation) into `tf' and `tf-gpu' virtual environments. The GPU environment can be loaded by calling <code>source tf-gpu/bin/activate-gpu</code>
+This environment have TensorFlow 1.0 and all necessary requirements for NeuralMonkey.
-OP: I created [[https://gist.github.com/oplatek/323b63b8f116cd3d78c0f492f78cc289|script]] which install Tensorflow 0.8 and test it if it uses GPU. TF is installed into `user` or `global` installation either for `python3.4` or `python2.7`
+==== Pytorch Environment ====
-=== Select GPU device ===
+If you want to use pytorch, there is a ready-made environment in
-Use variable CUDA_VISIBLE_DEVICES to constrain tensorflow to compute only on the selected one. For the use of first GPU use:
+  /home/hajicj/anaconda3/envs/pytorch/bin
-<code>export CUDA_VISIBLE_DEVICES=0</code>
+It does rely on the CUDA and CuDNN setup above.
-To list available devices, use:
+==== Using cluster ====
-<code>/opt/cuda/samples/1_Utilities/deviceQuery/deviceQuery | grep ^Device</code>
-===== Basic commands =====
+Rule number one, always use the GPU queue (never log in machine by ssh). Always use qsub or qsubmit with proper arguments.
+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.
+  qrsh -q gpu.q -l gpu=1,gpu_ram=2G -pty yes bash
+For running experiments you must use qsub command:
+  qsub -q gpu.q -l gpu=1,gpu_cc_min3.5=1,gpu_ram=2G WHAT_SHOULD_BE_RUN
+Cleaner way to use cluster is with /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.
+==== Basic commands ====
   lspci
@@ Line 110: / Line 124: @@
   /usr/local/cuda/samples/1_Utilities/deviceQuery/deviceQuery
     # shows CUDA capability etc.
+  ssh dll1; ~popel/bin/gpu_allocations
+    # 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:
+  /opt/cuda/samples/1_Utilities/deviceQuery/deviceQuery | grep ^Device
+===== Performance tests =====
+* [[http://www.trustedreviews.com/nvidia-geforce-gtx-1080-review-performance-benchmarks-and-conclusion-page-2| 980 vs 1080 vs Titan X (not the Titan Z we have)]]
+In the following table is the experiment conducted by Tom Kocmi. You can replicate experiment: /a/merkur3/kocmanek/Projects/GPUBenchmark (you will need to prepare environment of TensorFlow11 or use my ANACONDA). The benchmark uses 2GB model of seq2seq machine translation in Neural Monkey (De > EN). If not specified, the benchmark had an access only to one GPU.
+| machine | Setup; CPU/GPU; [[https://en.wikipedia.org/wiki/CUDA#Supported_GPUs|Capability]] [cc] | Walltime | Note |
+| athena     | GeForce GTX 1080; cc6.1            |    9:55:58 | Tom's desktop  |
+| dll2       | GeForce GTX 1080; cc6.1            |   10:19:40 |  |
+| 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 |
+| 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 |
+| kronos-dev | Tesla K40c; cc3.5                  |   22:41:01 |  |
+| twister2   | Tesla K40c; cc3.5                  |   22:43:10 |  |
+| twister1   | Tesla K40c; cc3.5                  |   24:19:45 |  |
+| helena1    | 16x cores CPU                      |   46:33:14 |  |
+| belzebub   | 16x cores CPU                      |   52:36:56 |  |
+| iridium    | Quadro K2000; cc3.0                |   59:47:58 |  |
+| helena7    | 8x cores CPU                       |   60:39:17 |  |
+| arc        | GeForce GT 630; cc3.0              |  103:42:30 | (approximated after 66 hours) |
+| lucifer4   | 8x cores CPU                       |  134:41:22 |  |
+=== Second Benchmark ===
+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 |
+| 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 |
 ===== Links =====

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

Institute of Formal and Applied Linguistics Wiki

Differences