ZaidQureshi opened a new issue #11282: Surprisngly low Traninging performance on V100 URL: https://github.com/apache/incubator-mxnet/issues/11282 ## Description I am getting surprisingly low performance when running the examples/image-classification/train_imagenet.py example on my Tesla V100 gpu. I am getting roughly 540 img/sec with a batch size of 64 with synthetic data when training either Alexnet or Resnet50. I was expecting Alexnet training performance to be much more than this. ## Environment info (Required) ``` What to do: ----------Python Info---------- Version : 3.5.2 Compiler : GCC 5.4.0 20160609 Build : ('default', 'Nov 23 2017 16:37:01') Arch : ('64bit', 'ELF') ------------Pip Info----------- Version : 10.0.1 Directory : /usr/local/lib/python3.5/dist-packages/pip ----------MXNet Info----------- Version : 1.3.0 Directory : /usr/local/lib/python3.5/dist-packages/mxnet Commit Hash : b434b8ec18f774c99b0830bd3ca66859212b4911 ----------System Info---------- Platform : Linux-4.13.0-45-generic-x86_64-with-Ubuntu-16.04-xenial system : Linux node : css-host-8 release : 4.13.0-45-generic version : #50~16.04.1-Ubuntu SMP Wed May 30 11:18:27 UTC 2018 ----------Hardware Info---------- machine : x86_64 processor : x86_64 Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 40 On-line CPU(s) list: 0-39 Thread(s) per core: 2 Core(s) per socket: 10 Socket(s): 2 NUMA node(s): 2 Vendor ID: GenuineIntel CPU family: 6 Model: 79 Model name: Intel(R) Xeon(R) CPU E5-2640 v4 @ 2.40GHz Stepping: 1 CPU MHz: 1200.189 CPU max MHz: 3400.0000 CPU min MHz: 1200.0000 BogoMIPS: 4799.72 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 256K L3 cache: 25600K NUMA node0 CPU(s): 0-9,20-29 NUMA node1 CPU(s): 10-19,30-39 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti retpoline intel_ppin intel_pt spec_ctrl tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm rdt_a rdseed adx smap xsaveopt cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts ----------Network Test---------- Setting timeout: 10 Timing for Conda: https://repo.continuum.io/pkgs/free/, DNS: 0.0255 sec, LOAD: 0.1334 sec. Timing for MXNet: https://github.com/apache/incubator-mxnet, DNS: 0.0100 sec, LOAD: 0.5690 sec. Timing for Gluon Tutorial(en): http://gluon.mxnet.io, DNS: 0.3449 sec, LOAD: 1.6452 sec. Timing for PYPI: https://pypi.python.org/pypi/pip, DNS: 0.0099 sec, LOAD: 0.3464 sec. Timing for FashionMNIST: https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/dataset/fashion-mnist/train-labels-idx1-ubyte.gz, DNS: 0.2326 sec, LOAD: 0.6122 sec. Timing for Gluon Tutorial(cn): https://zh.gluon.ai, DNS: 0.3419 sec, LOAD: 2.3122 sec. ``` Package used (Python/R/Scala/Julia): I'm using Python. ## Build info (Required if built from source) Compiler (gcc/clang/mingw/visual studio): gcc MXNet commit hash: b434b8ec18f774c99b0830bd3ca66859212b4911 Build config: I installed the mxnet-cu91 python package ## Output Message: Alexnet: ``` INFO:root:start with arguments Namespace(batch_size=64, benchmark=1, data_nthreads=40, data_train=None, data_train_idx='', data_val=None, data_val_idx='', disp_batches=20, dtype='float16', gc_threshold=0.5, gc_type='none', gpus='2', image_shape='3,224,224', initializer='default', kv_store='device', load_epoch=None, loss='', lr=0.1, lr_factor=0.1, lr_step_epochs='30,60', macrobatch_size=0, max_random_aspect_ratio=0.25, max_random_h=36, max_random_l=50, max_random_rotate_angle=10, max_random_s=50, max_random_scale=1, max_random_shear_ratio=0.1, min_random_scale=1, model_prefix='alexnet', mom=0.9, monitor=0, network='resnet', num_classes=1000, num_epochs=80, num_examples=1281167, num_layers=50, optimizer='sgd', pad_size=0, random_crop=1, random_mirror=1, rgb_mean='123.68,116.779,103.939', save_period=1, test_io=0, top_k=0, warmup_epochs=5, warmup_strategy='linear', wd=0.0001) [11:08:27] src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:107: Running performance tests to find the best convolution algorithm, this can take a while... (setting env variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to disable) INFO:root:Epoch[0] Batch [20] Speed: 535.12 samples/sec accuracy=0.455357 INFO:root:Epoch[0] Batch [40] Speed: 551.94 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [60] Speed: 554.26 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [80] Speed: 549.69 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [100] Speed: 548.01 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [120] Speed: 541.15 samples/sec accuracy=1.000000 ``` Resnet 50: ``` INFO:root:start with arguments Namespace(batch_size=64, benchmark=1, data_nthreads=40, data_train=None, data_train_idx='', data_val=None, data_val_idx='', disp_batches=20, dtype='float16', gc_threshold=0.5, gc_type='none', gpus='2', image_shape='3,224,224', initializer='default', kv_store='device', load_epoch=None, loss='', lr=0.1, lr_factor=0.1, lr_step_epochs='30,60', macrobatch_size=0, max_random_aspect_ratio=0.25, max_random_h=36, max_random_l=50, max_random_rotate_angle=10, max_random_s=50, max_random_scale=1, max_random_shear_ratio=0.1, min_random_scale=1, model_prefix='resnet', mom=0.9, monitor=0, network='resnet', num_classes=1000, num_epochs=80, num_examples=1281167, num_layers=50, optimizer='sgd', pad_size=0, random_crop=1, random_mirror=1, rgb_mean='123.68,116.779,103.939', save_period=1, test_io=0, top_k=0, warmup_epochs=5, warmup_strategy='linear', wd=0.0001) [11:07:24] src/operator/nn/./cudnn/./cudnn_algoreg-inl.h:107: Running performance tests to find the best convolution algorithm, this can take a while... (setting env variable MXNET_CUDNN_AUTOTUNE_DEFAULT to 0 to disable) INFO:root:Epoch[0] Batch [20] Speed: 528.37 samples/sec accuracy=0.433036 INFO:root:Epoch[0] Batch [40] Speed: 540.25 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [60] Speed: 539.98 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [80] Speed: 546.55 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [100] Speed: 550.18 samples/sec accuracy=1.000000 INFO:root:Epoch[0] Batch [120] Speed: 543.24 samples/sec accuracy=1.000000 ``` ## Minimum reproducible example (If you are using your own code, please provide a short script that reproduces the error. Otherwise, please provide link to the existing example.) ## Steps to reproduce (Paste the commands you ran that produced the error.) 1. python3 train_imagenet.py --gpus 2 --model alexnet --test-io 0 --data-nthreads 40 --benchmark 1 --batch-size 64 --dtype float16 2. python3 train_imagenet.py --gpus 2 --model resnet --num-layers 50 --test-io 0 --data-nthreads 40 --benchmark 1 --batch-size 64 --dtype float16 ## What have you tried to solve it? 1. I have tried various batch sizes and this is the highest throughput I can get. 2. I have tried turning off data augmentation but still no effect 3. I have tried gluon and when I give it --dtype float16 the throughput seems to be much better but I think the support for flaot16 isn't complete as it causes an error. I'll make another issue request about this.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services