Skip to content

Latest commit

 

History

History
329 lines (264 loc) · 18 KB

README.md

File metadata and controls

329 lines (264 loc) · 18 KB
Raw

Nvidia backend support

General information

The Nvidia backend for oneDNN can be exposed to the user via the dnnl::engine::kind::gpu engine kind. Currently, for the case when user's system has both Intel and Nvidia GPUs, DNNL_GPU_VENDOR=NVIDIA flag is used in CMake, since the devices are clustered based on the device vendor ID and index pattern can not be used to distinguish between Intel GPU and Nvidia GPU. However, Intel is working on restructuring the engine creation, so that it would be possible to choose engine kind and vendor kind at runtime. Also, it is possible to create oneDNN engines using sycl::device objects corresponding to Nvidia GPUs. The stream in Nvidia backend for oneDNN defines an out-of-order SYCL queue by default. Similar to the existing oneDNN API, user can specify an in-order queue when creating a stream if needed.

Build command

export CC=/path/to/dpcpp/install/bin/clang
export CXX=/path/to/dpcpp/install/bin/clang++
mkdir build
cd build
cmake -DDNNL_CPU_RUNTIME=DPCPP -DDNNL_GPU_RUNTIME=DPCPP \
      -DDNNL_GPU_VENDOR=NVIDIA -G Ninja ..

Memory

Currently, only the buffer-based oneDNN API is supported for Nvidia backend.

Suported Data Types

The following table documents the supported data types.

Data Type Computation Mode
f32 Training, Inference
f16 Inference
s8 Inference (when applicable)

Supported Primitives and Implementation Limitations

cuDNN functions are not necessarily the same as oneDNN primitives due to lack of standard API for DNN. For each primitive the cuDNN equivalent function is added to the Nvidia backend for oneDNN. However, the added backend cannot provide all functionalities supported by oneDNN primitives. The detailed limitations of each cuDNN primitive are explained as follow.

Batch normalization

The closest equivalent to oneDNN batch normalization can be cudnnBatchNormalizationForward and cudnnBatchNormalizationBackward operations. However, there are some difference between cuDNN and oneDNN batch normalization.

Forward direction

  • When global_stats flag is set for batch normalization, the mean and variance are input only parameters. However, cuDNN does not have the option to accept the mean and variance as inputs in the forward training operation. Therefore, cudnnBatchNormalizationForwardInference is used to match the oneDNN feature. Although inference is not supported without global_stats flags set.
  • The cuDNN precision is different from that of oneDNN for Batch Normalization. (e.g fp:0.0170898 dt:0.0170907 diff:8.27014e-07 rdiff:4.83922e-05)
  • The forward training with no flags accepts mean and variance as an output. However, in cuDNN the mean and variance are running mean and variance respectably so they are both input and output variable. Therefore, they are required to have a sensible value (cannot be NaN). Since oneDNN will not set value for the mean and variance when no flag is passed, the NaN can be propagated as a result. To avoid NaN propagation, cudaMemset function is used to initialize the mean and variance with zero.
  • cuDNN always requires the values for scale and shift. When shift and scale are not defined in oneDNN, cudaMemset is used to initialize scale to 1 and shift to 0.
  • For performance reason in the backward pass, cuDNN requires the mean and inverse variance to be saved in the forward pass. Therefore, when Nvidia backend is used for batch normalization, the workspace must be provided to save the mean and inverse variance.
  • When dnnl_fuse_norm_relu flag is set for batch normalization, the cudnnActivationForward operation is called immediately after the batch normalization, since cuDNN does not have a fused batch normalization with RELU. The implementation for element-wise post operations is the same.
  • When dnnl_fuse_norm_relu is used, the intermediate output of batch normalization, which is used as an input to the activation function, is saved in the workspace as well. This is required to compute the backward pass for dnnl_fuse_norm_relu flag.
  • Forward pass supports f32, f16 and s8 data types. Although blocking is not supported for s8.

Backward direction

  • cuDNN uses alpha and beta parameters to blend the dy, shift and scale. Since oneDNN does not have this feature, the alpha and beta values in the backward direction are set to 1 and 0 respectively to avoid blending.
  • Nvidia backend for backward direction requires the workspace as an input containing the mean and inverse variance computed in the forward pass.
  • The Nvidia backend for oneDNN does not support the backward direction for batch normalization when the flag is set to global_stats. This is due to the fact that oneDNN will skip the

since the mean and variance are constant, however, cuDNN does not have an option to skip this operation. * When `dnnl_fuse_norm_relu` flag is set, Nvidia backend requires the intermediate result of the batch normalization saved in the forward pass. This is used to compute the backward direction of the activation function used for `RELU`.

Binary

The cudnnOpTensor is equivalent of oneDNN binary primitives.

  • Only scales attribute is supported. Post-op attribute is not supported.
  • Blocking is only supported for int8 and only in the C dimension with either 4 or 32 block size (same as other cuDNN primitives).

Concat

The concat operation uses the reorder primitive to concatenate tensors over the chosen dimension, so the same limitation as reorder applies here.

Convolution

The cudnnConvolutionForward, cudnnConvolutionBackward and cudnnConvolutionBackwardFilter is used to compute forward, backward by data or backward by weights for a convolution operation.

  • Blocking is only supported for int8 and only in the C dimension with block size of 4. Input and output tensors must have the same data type.
  • For int8 (s8s8s8) with post-ops the operations are performed as s8s8f32 (due to cuDNN limitations) then reordered to s8 at the end which impacts performance.
  • Direct convolution is not supported, so implicit GEMM is used in those cases.
  • "Left" padding must be greater or equal to "right" padding, and the requested spatial output should match the output formula for two "left" padding used.
  • Eltwise post-op limitations are the same as our eltwise limitation as post-ops are not fused.
  • cuDNN requires padding tensors to 4 dimensions, so 1D convolutions are supported but are performed as 2D.

The following table shows the convolution status for the oneDNN Nvidia backend:

Forward direction

Weights Format Winograd Supported Supported Input Format Supported Output Format Supported Data Type Limitations
2D NCHW YES NCHW, NHWC NCHW, NHWC f32, f16 The Winograd algorithm has limitations:
* Filter size must be 3x3 or 5x5.
* Dilation must be zero for all dimensions.
* Horizontal and vertical filter stride must be 1.
2D NHWC NO NHWC NHWC f32, f16, int8 * Dilation must be zero in all dimensions.
* Output feature maps must be multiple of 4 for int8 type.
3D NCHW NO NCHW, NHWC NCHW, NHWC f32, f16

Backward direction

Weights Format Winograd Supported Supported Input Format Supported Output Format Supported Data Type Limitations
2D NCHW YES NCHW, NHWC NCHW f32, f16 1. Dilation must be zero for all dimensions.
2. The Winograd algorithm has limitations:
* Filter size must be 3x3 or 5x5.
* Dilation must be zero for all dimensions.
* Horizontal and vertical filter stride must be 1.
2D NHWC NO NHWC NHWC f32, f16
3D NCHW NO NCHW, NHWC NCHW f32, f16

Deconvolution

Deconvolution primitive is implemented through the convolution with swapped input abd output channels.

  • Currently, there is a bug, likely in this code, which causes crashes in memory_tracking for 3D backward_weights with bias when backward_weights without bias was also a part of the run. Cache interrogation is suspected due to cache-free runs are successful. Switched off in benchdnn until further investigation and the fix.

Eltwise

The cudnnActivationForward and cudnnActivationBackward is the equivalent of eltwise forward and eltwise backward in oneDNN respectively. There are some limitations when using Nvidia backend for eltwise primitive:

  • cuDNN only supports the following operations - RELU, ELU, TANH, LOGISTIC and BRELU.
  • RELU is only supported with alpha = 0.
  • cuDNN expects x, y and dy as inputs to the backward pass, hence, only RELU and BRELU operations are supported in the backward pass. TODO: add ELU_DST, TANH_DST and LOGISTIC_DST support which require dy.
  • Forward pass supports f32, f16 and s8 data types. Although blocking is not supported for s8.
  • Backward pass supports f32 and f16 data types.

Inner product

The inner product primitives is an implementation of matrix multiplication plus bias activation. There are two implementation of inner product in cuDNN backend.

Using GEMM

The default backend for inner product is the gemm backend using cublasGemmEx for forward, backward data, and backward weight and cudnnReduceTensor for backward bias. A function called gemm_consitency_check(), dense_check() is used to see if the gemm backend can be used for inner product. reorder_check() is used when reorder is required. If none of the above condition are met, it falls back to the convolution backend. cudnnActivationForward operation is used for eltwise operation and cudnnAddTensor is used for bias operation. The beta parameter in gemm is used for the sum scale and alpha parameter is used for the output scale.

Using convolution

For the forward direction, this operation can be implemented by using cudnnConvolutionBiasActivation by converting the inner product to 1x1 convolution. For the backward direction the inner product operation will be equivalent of cudnnConvolutionBackwardData, cudnnConvolutionBackwardWeights and cudnnConvolutionBackwardBias when applied. This implementation of inner product has the following restrictions and performance implications:

  • The only blocked layouts are those that are supported in cuDNN - namely that the blocking is done on the C dimension, the block size is 4, and only for int8 inference. The additional requirement is that both the input and filter must be blocked.
  • The ReLU and sum are supported as a fused post-op, for other post-op a separate call to eltwise primitive is performed. So the limitation for the eltwise primitive is applied here.
  • Only mask = 0 case is supported for output scale.
  • The restrictions for the convolution primitive are applied here for input and filter format. When required, the filter is internally reordered to match the convolution restriction.
  • For int8 cuDNN requires both input and output feature maps to be a multiple of 4.

LRN

The local response normalization primitive in the Nvidia backend is implemented with the cudnnLRNForward and cudnnLRNBackward functions for forward and backward propagation respectively.

  • WITHIN algorithm is not supported.
  • There is a difference in the LRN algorithm used in oneDNN and cuDNN which causes a mismatch when the local size is even.
  • cuDNN supports NCHW tensor formats for all valid dimensions. However, it does not support the NHWC tensor format for above 5 dimensions.

Matrix Multiplication

The matrix multiplication primitive in the Nvidia backend is implemented with cublasGemmEx and cublasGemmStridedBatchedEx functions.

  • Zero points support is not provided by cuBLAS and, hence, not supported by the Nvidia backend.
  • Post-ops and output scale limitations are same as for Inner Product.

Pooling

The pooling primitive in the Nvidia backend is implemented with the cudnnPoolingForward and cudnnPoolingBackward functions for forward and backward propagation respectively.

  • cuDNN only allows the use of symmetric padding, i.e. padding at the beginning of a dimension must be the same as the padding at the end of that dimension. oneDNN doesn't have this limitation. Therefore,

    • Configurations where padding in the beginning is larger than padding at the end are supported and work as expected.
    • For configurations where padding at the end is larger than padding in the beginning of any dimension, the primitive returns status::unimplemented.

  • For backward propagation cuDNN requires the parameters x, y, dx and dy, while oneDNN requires only dx, dy and workspace when the MAX algorithm is used. Hence, the workspace is used to store the x and y parameters in the forward pass for the Nvidia backend. Therefore, the workspace is always required when the Nvidia backend is used (except for the forward inference).

Reorder

The cudnnTransform function is the equivalent of oneDNN reorder function. However, there are some limitations when using SYCL_API-DNN reorder on Nvidia GPU:

  • Per dimension scaling is not supported (a single alpha and beta value is accepted by the transform tensor function).
  • Blocking is only permitted for the channel dimension in cuDNN. This primitive currently supports block size of 4.
  • Blocking is only supported when channel dimension is a multiple of the block size and the datatype is int8.

Resampling

The cudnnSpatialTfSamplerForward and cudnnSpatialTfSamplerBackward are used to implement the resampling primitive.

The Nvidia's spatial sampling is based on Spacial Transformer Network where all the data locations are normalized between -1 <= (xi, yi) <= 1.

  • cuDNN backend requires a grid of coordinates that can be sample-up/down based on theta. The grid is generated by cudnnSpatialTfGridGeneratorForward.
  • The theta is a MB * 2 * 3 matrix scaling factor for each coordinate and is used to generate the grid.
  • The grid value must be normalized in range [-1 , 1]. cuDNN clamps the out of bounds coordinate to zero. Therefore, it is needed to manually clamp the out of bound coordinate to edges in order to avoid incorrect result.
  • 3D spatial sampling is not supported in cuDNN.
  • Nearest neighbour algorithm is not supported in cuDNN.
  • Since cuDNN computation is different from that of oneDNN, the error threshold is smaller than other oneDNN implementation, so reduced testing accuracy for fp32 and fp16 data types are required.
  • The backward pass requires an output parameter for d_grid which cannot be nullptr. However, since the grid coordinates are not a tunable parameter in oneDNN, a dummy memory for d_grid is created and is deleted when the destructor of the primitive is called.

Softmax/LogSoftmax

The cudnnSoftmaxForward and cudnnSoftmaxBackward are used to implement the softmax primitive. For logsoftmax primitive the same functions will be used and the algorithm selection in cuDNN for the above mentioned functions will be changed to CUDNN_SOFTMAX_LOG.

  • The softmax axis is supported for only the channel dimension, (i.e., axis=1).
  • There is a bug in cuDNN softmax for 5D tensor with format NHWC. When the channel size is greater than 1, it only applies softmax for a single channel and leave the others untouched.

Sum

The sum operation uses the reorder primitive to sum tensors, so the same limitation as reorder applies here.

Other primitives

Rest primitives not listed above are not supported by Nvidia backend. This is likely due to either missed functionality in cuDNN or cuBLAS, or lack of priority in supporting of such functionality.