Getting the "tensorflow:Layer lstm will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU." warning while training my LSTM model on Apple Silicon M2. The training is just too slow. How can I get the best out of this chip for my task?
p.s. (1) I've already installed the tensorflow-macos and tensorflow-metal packages alongside the tensorflow-deps package provided in the Apple channel of conda.
(2) My model is not the deepest one too as it consists of 1 LSTM layer with 64 units, and 1 Dense layer with 64 units.
(3) My machine's main specs:
macOS Ventura 13.2.1 (the latest stable one)
Apple Silicon M2 (8-Core CPU, 10-Core GPU, and 16-Core Neural Engine)
16 GB Unified Memory
Many thanks in advance for your care & comments.
Since you're using Apple Silicon, cuDNN most likely isn't the culprit here.
Have you tried training on CPU and compare the time cost? Your model isn't large, so the overhead of dispatching work to GPU should be the leading cause here. As your model gets larger, the overhead tends to get amortized. See the Troubleshooting section on this page.
Related
I ran the following code on a Tesla K80, which as I understand consists of 2 GK210 graphics cards, each with 12GB of on chip ram, connected by something called a PLX switch. I am confused how at the pytorch level, the fact that there are two graphics cards is hidden from the user
import torch
torch.cuda.device_count() # 1
(my hunch is that tensorflow provides this same abstraction)
Follow up questions:
If I am training a model with pytorch, and I run nvidia-smi and see that the GPU is fully utilized, I would assume this means that both GK210's are at 100% utilization. How does pytorch distribute kernels across the two GK210's, and can I have faith that this is being done efficiently? (I.e. being done in a way that minimized data transfer between the two cards). Any resources that explain how this works would be much appreciated.
If I were writing a CUDA application, could I pin a CUDA stream to each card, and explicitly manage data transfers between the two cards?
Okay, so I've worked on a bunch of Deep Learning projects and internships now and I've never had to do heavy training. But lately I've been thinking of doing some Transfer Learning for which I'll need to run my code on a GPU. Now I have a system with Windows 10 and a dedicated NVIDIA GeForce 940M GPU. I've been doing a lot of research online, but I'm still a bit confused. I haven't installed the NVIDIA Cuda Toolkit or cuDNN or tensorflow-gpu on my system yet. I currently use tensorflow and pytorch to train my DL models. Here are my queries -
When I define a tensor in tf or pytorch, it is a cpu tensor by default. So, all the training I've been doing so far has been on the CPU. So, if I make sure to install the correct versions of Cuda and cuDNN and tensorflow-gpu (specifically for tensorflow), I can run my models on my GPU using tf-gpu and pytorch and that's it? (I'm aware of the torch.cuda.is_available() in pytorch to ensure pytorch can access my GPU and the device_lib module in tf to check if my gpu is visible to tensorflow)(I'm also aware of the fact that tf doesnt support all Nvidia GPUs)
Why does tf have a separate module for GPU support? PyTorch doesnt seem to have that and all you need to do is cast your tensor from cpu() to cuda() to switch between them.
Why install cuDNN? I know it is a high-level API CUDA built for support to train Deep Neural Nets on the GPU. But do tf-gpu and torch use these in the backend while training on the gpu?
After tf == 1.15, did they combine CPU and GPU support all into one package?
First of all unfortunately 940M is a kinda weak GPU for training. I suggest you use Google colab for faster training but of course, it would be faster than the CPU. So here my answers to your four questions.
1-) Yes if you install the requirements correctly, then you can run on GPU. You can manually place your data to your GPU as well. You can check implementations on TensorFlow. In PyTorch, you should specify the device that you want to use. As you said you should do device = torch.device("cuda" if args.cuda else "cpu") then for models and data you should always call .to(device) Then it will automatically use GPU if available.
2-) PyTorch also needs extra installation (module) for GPU support. However, with recent updates both TF and PyTorch are easy to use for GPU compatible code.
3-) Both Tensorflow and PyTorch is based on cuDNN. You can use them without cuDNN but as far as I know, it hurts the performance but I'm not sure about this topic.
4-) No they are still different packages. tensorflow-gpu==1.15 and tensorflow==1.15 what they did with tf2, was making the tensorflow more like Keras. So it is more simplified then 1.15 or before.
Rest was already answered by regarding 3) cudNN optimizes layer and such operations on hardware level and those implementations are pure black magic. It is incredibly hard to write CUDA code that properly utilizes your GPU (how load data into the GPU, how to actually perform them using matrices etc. )
I'm running a CNN with keras-gpu and tensorflow-gpu with a NVIDIA GeForce RTX 2080 Ti on Windows 10. My computer has a Intel Xeon e5-2683 v4 CPU (2.1 GHz). I'm running my code through Jupyter (most recent Anaconda distribution). The output in the command terminal shows that the GPU is being utilized, however the script I'm running takes longer than I expect to train/test on the data and when I open the task manager it looks like the GPU utilization is very low. Here's an image:
Note that the CPU isn't being utilized and nothing else on the task manager suggests anything is being fully utilized. I don't have an ethernet connection and am connected to Wifi (don't think this effects anything but I'm not sure with Jupyter since it runs through the web broswers). I'm training on a lot of data (~128GB) which is all loaded into the RAM (512GB). The model I'm running is a fully convolutional neural network (basically a U-Net architecture) with 566,290 trainable parameters. Things I tried so far:
1. Increasing batch size from 20 to 10,000 (increases GPU usage from ~3-4% to ~6-7%, greatly decreases training time as expected).
2. Setting use_multiprocessing to True and increasing number of workers in model.fit (no effect).
I followed the installation steps on this website: https://www.pugetsystems.com/labs/hpc/The-Best-Way-to-Install-TensorFlow-with-GPU-Support-on-Windows-10-Without-Installing-CUDA-1187/#look-at-the-job-run-with-tensorboard
Note that this installation specifically DOESN'T install CuDNN or CUDA. I've had trouble in the past with getting tensorflow-gpu running with CUDA (although I haven't tried in over 2 years so maybe it's easier with the latest versions) which is why I used this installation method.
Is this most likely the reason why the GPU isn't being fully utilized (no CuDNN/CUDA)? Does it have something to do with the dedicated GPU memory usage being a bottleneck? Or maybe something to do with the network architecture I'm using (number of parameters, etc.)?
Please let me know if you need any more information about my system or the code/data I'm running on to help diagnose. Thanks in advance!
EDIT: I noticed something interesting in the task manager. An epoch with batch size of 10,000 takes around 200s. For the last ~5s of each epoch, the GPU usage increases to ~15-17% (up from ~6-7% for the first 195s of each epoch). Not sure if this helps or indicates there's a bottleneck somewhere besides the GPU.
You for sure need to install CUDA/Cudnn to fully utilize GPU with tensorflow. You can double check that the packages are installed correctly and if the GPU is available to tensorflow/keras by using
import tensorflow as tf
tf.config.list_physical_devices("GPU")
and the output should look something like [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]
if the device is available.
If you've installed CUDA/Cudnn correctly then all you need to do is change copy --> cuda in the dropdown menu in the task manager which will show the number of active cuda cores. The other indicators for the GPU will not be active when running tf/keras because there is no video encoding/decoding etc to be done; it is simply using the cuda cores on the GPU so the only way to track GPU usage is to look at the cuda utilization (when considering monitoring from the task manager)
I would first start by running one of the short "tests" to ensure Tensorflow is utilizing the GPU. For example, I prefer #Salvador Dali's answer in that linked question
import tensorflow as tf
with tf.device('/gpu:0'):
a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
c = tf.matmul(a, b)
with tf.Session() as sess:
print (sess.run(c))
If Tensorflow is indeed using your GPU you should see the result of the matrix multplication printed. Otherwise a fairly long stack trace stating that "gpu:0" cannot be found.
If this all works well that I would recommend utilizing Nvidia's smi.exe utility. It is available on both Windows and Linux and AFAIK installs with the Nvidia driver. On a windows system it is located at
C:\Program Files\NVIDIA Corporation\NVSMI\nvidia-smi.exe
Open a windows command prompt and navigate to that directory. Then run
nvidia-smi.exe -l 3
This will show you a screen like so, that updates every three seconds.
Here we can see various information about the state of the GPUs and what they are doing. Of specific interest in this case is the "Pwr: Usage/Cap" and "Volatile GPU-Util" columns. If your model is indeed using the/a GPU these columns should increase "instantaneously" once you start training the model.
You most likely will see an increase in fan speed and temperature unless you have a very nice cooling solution. In the bottom of the printout you should also see a Process with a name akin to "python" or "Jupityr" running.
If this fails to provide an answers as to the slow training times than I would surmise the issue lies with the model and code itself. And I think its is actually the case here. Specifically viewing the Windows Task Managers listing for "Dedicated GPU Memory Usage" pinged at basically maximum.
If you have tried #KDecker's and #OverLordGoldDragon's solution, low GPU usage is still there, I would suggest first investigating your data pipeline. The following two figures are from tensorflow official guides data performance, they are well illustrated how data pipeline will affect the GPU efficiency.
As you can see, prepare data in parallel with the training will increase the GPU usage. In this situation, CPU processing is becoming the bottleneck. You need to find a mechanism to hide the latency of preprocessing, such as changing the number of processes, size of butter etc. The efficiency of CPU should match the efficiency of the GPU. In this way, the GPU will be maximally utilized.
Take a look at Tensorpack, and it has detailed tutorials of how to speed up your input data pipeline.
Everything works as expected; your dedicated memory usage is nearly maxed, and neither TensorFlow nor CUDA can use shared memory -- see this answer.
If your GPU runs OOM, the only remedy is to get a GPU with more dedicated memory, or decrease model size, or use below script to prevent TensorFlow from assigning redundant resources to the GPU (which it does tend to do):
## LIMIT GPU USAGE
config = tf.ConfigProto()
config.gpu_options.allow_growth = True # don't pre-allocate memory; allocate as-needed
config.gpu_options.per_process_gpu_memory_fraction = 0.95 # limit memory to be allocated
K.tensorflow_backend.set_session(tf.Session(config=config)) # create sess w/ above settings
The unusual increased usage you observe may be shared memory resources being temporarily accessed due to exhausting other available resources, especially with use_multiprocessing=True - but unsure, could be other causes
There seems to have been a change to the installation method you referenced : https://www.pugetsystems.com/labs/hpc/The-Best-Way-to-Install-TensorFlow-with-GPU-Support-on-Windows-10-Without-Installing-CUDA-1187
It is now much easier and should eliminate the problems you are experiencing.
Important Edit You don't seem to be looking at the actual compute of the GPU, look at the attached image:
read following two pages ,u will get idea to properly setup with GPU
https://medium.com/#kegui/how-do-i-know-i-am-running-keras-model-on-gpu-a9cdcc24f986
https://datascience.stackexchange.com/questions/41956/how-to-make-my-neural-netwok-run-on-gpu-instead-of-cpu
Usually one layer in dnn consists of MatMul, BiasAdd, Relu, cuBlas provides Gemm for MatMul, and we can do BiasAdd and Relu in another kernel for GPU. They are two GPU lanuch calls, is there any way to fuse them all togather and make them just one? I looked into cuBlas, cudnn, but not found anything. I think it's not difficult because BiasAdd and Relu are just element-wise operaions, and fusion makes it more efficient.
Here is the backgroud:
I am working on a online prediction service which is multi dnn model ensemble. By profiling my program, I found out that both my CPU and GPU is not fully utilized, but requests blocks on GPU-related function call (like lanuchKernel). It seems like there's a big lock in libcuda. I am using tensorflow, XLA enabled, so I use nvprof and tensorflow HLO to visialize GPU-call, and there'are only dot and fused(which is biasadd and relu) operations. Although kernel fusion is done, there're still too many lanuchKernel calls, and GPU utilization is only 60%. I tried multi cuda context in one process, the improvement is trivial.
By the way, I am using one single GPU, Tesla P100.
I'm fairly new to Tensorflow in and ML in general and am wondering what strategies I can use to increase performance of an application I am building.
My app is using the Tensorflow C++ interface, with a source compiled TF 0.11 libtensorflow_cc.so (built with bazel build -c opt --copt=-mavx and optionally adding --config=cuda) for either AVX or AVX + CUDA on Mac OS X 10.12.1, on an MacBook Pro 2.8 GHz Intel Core i7 (2 cores 8 threads) with 16GB ram and a Nvidia 750m w/ 2GB VRam)
My application is using Inception V3 model and pulling feature vectors from pool_3 layer. I'm decoding video frames via native API's and passing those in memory buffers to the C++ interface for TF and running them into a session.
I'm not currently batching, but I am caching my session and re-using it for each individual decoded frame / tensor submission. Ive noticed that both CPU and GPU performance is about the same, taking about 40 to 50 seconds to process 222 frames, which seems very slow to me. Ive confirmed CUDA is being invoked, loaded, and the GPU is functioning (or appears so).
Some questions:
In general what should I expect for reasonable performance time wise of TF doing a frame of Inception on a consumer laptop?
How much of a difference does batching make for these operations? For tensors of 1x299x299x3 , I imagine I am doing more PCI transfer waiting than waiting on for meaningful work from the GPU?
if so Is there a good example of batching under C++ for InceptionV3?
Is there operations that cause additional CPU->GPU Syncronization that might otherwise be avoided?
Is there a way to ensure my sessions / graphs share resources ? Can I use nested scopes somehow in this manner? I couldn't quite get that to work but likely missed something.
Any good documentation of general strategies for things to do / avoid?
My code is below:
https://github.com/Synopsis/Synopsis/blob/TensorFlow/Synopsis/TensorFlowAnalyzer/TensorFlowAnalyzer.mm
Thank you very much
For reference, OpenCV analysis using perceptual hash, histogram, dense optical flow, sparse optical flow for point tracking, and simple saliency detection takes 4 to 5 seconds for the same 222 frames using CPU or CPU + OpenCL.
https://github.com/Synopsis/Synopsis/tree/TensorFlow/Synopsis/StandardAnalyzer
Answering your last question first, if there's documentation about performance optimization, yes:
The TensorFlow Performance Guide
The TensorFlow GPU profiling hints
Laptop performance is highly variable, and TF isn't particularly optimized for laptop GPUs. The numbers you're getting (222 frames in 40-50 seconds) ~= 5 fps don't seem crazy on a laptop platform, using the 2016 version of TensorFlow, with inception. With some of the performance improvements outlined in the performance guide above, that should probably be doubled in late 2017.
For batching, yes - the newer example inception model code allows a variable batch size at inference time. This is mostly about whether the model itself was defined to handle a batch size, which is something improved since 2016.
Batching for inference will make a pretty big difference on GPU. Whether it helps on CPU depends a lot -- for example, if you build with MKL-DNN support, batching should be considered mandatory, but basic TensorFlow may not benefit as much.