Being able to leverage I/O binding for DML and registering `If` for the
DML EP allows us to avoid copying the past/present key/values back and
forth between the CPU and the GPU after every token.
This gives us a 25% performance increase for Dolly V2 with 128 tokens on
an RTX 4090.
### Description
The [ONNX
standard](https://github.com/onnx/onnx/blob/main/docs/Operators.md#type-constraints-181)
permits the `Unique` operator to have `double` input tensor element
type, however this was not supported in onnxruntime. This PR enables
this kernel.
### Motivation and Context
The lack of support for `float64` forces users currently to cast to
`float32` instead. This loss of precision can be severely problematic in
feature engineering pipelines downstream of the `Unique` operator. It
would be good to prevent this by updating ORT to reflect the standard
and support `double` input tensors.
---------
Signed-off-by: Aditya Goel <agoel4512@gmail.com>
This fixes the type lists used to register DML kernels for Microsoft
domain QuantizeLinear and DequantizeLinear. These previously did not
include FP16 and incorrectly used the same type list for both operators.
The new type lists are the same as opset 19 ONNX which aren't
implemented yet in the DML EP.
### Description
The PR implements FloatE4M3FN, FloatE5M2, FloatE4MEFNUZ, FloatE5M2FNUZ
as described in PR https://github.com/onnx/onnx/pull/4805. It uses CUDA
API to cast float/half to float8 if CUDA>=11.8, a custom implementation
if CUDA<11.8.
* It implements, Cast, QuantizeLinear, DequantizeLinear for all types on
CPU, only for types FloatE4M3FN, FloatE5M2 on CUDA.
* It extends the supported types for control flow operator, Shape,
Reshape, Identity, If, Loop, Scan, Reshape
* It implements Equal(19).
* Cast, QuantizeLinear, DequantizeLinear operators now support a
parameter `saturate` only valid for float 8 types. It is true by
default. In that case, any value out of range is converted into the
maximum float 8 value. If false, it is infinite.
* QuantizeLinear, DequantizeLinear now supports multiple scales on CUDA
(and ROCm by extension), scale = 1D tensor with one scale per channel
### Motivation and Context
Supports latest onnx version.
Fixes
[AB#15395](https://aiinfra.visualstudio.com/6a833879-cd9b-44a4-a9de-adc2d818f13c/_workitems/edit/15395)
---------
Co-authored-by: Xavier Dupre <xadupre@microsoft.com@orttrainingdev8.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
Co-authored-by: Randy Shuai <rashuai@microsoft.com>
Co-authored-by: Edward Chen <18449977+edgchen1@users.noreply.github.com>
Co-authored-by: Scott McKay <Scott.McKay@microsoft.com>
### Description
<!-- Describe your changes. -->
Pad18 adds the `axes` input, which is used to indicate what axes the
padding values should be applied to. Add logic to manipulate paddings
into DML padding operator inputs.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Linnea May <linneamay@microsoft.com>
* graph tools update
* cuda kernel update
* operator spec update and implementation update
* greed search bug fix on wrong assumption for cross/self attention
input length
* avoid use of "" name in value info when loading graph which
historically in many model
### Description
<!-- Describe your changes. -->
Add dml registration for bitwise and, or, xor and not added in opset 18.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Linnea May <linneamay@microsoft.com>
Register Split18 for DirectML
Split13 was previously implemented. Split18 adds a new attribute called
"num_outputs" that must be used mutually exclusively with the "split"
input.
The "num_outputs" attribute wil split the tensor evenly (and handles odd
uneven splits). To implement, the DML split tensor just needs to be
overridden in the presence of the num_output attribute.
---------
Co-authored-by: Dwayne Robinson <dwayner@microsoft.com>
### Description
This PR enables Whisper's multitask format and allows a user to use
Whisper for multiple tasks (e.g. transcription, translation) and for
multilingual purposes (e.g. English, Spanish). This PR also removes
`attention_mask` as a required input for Whisper with beam search.
### Usage
Here is an example of how you can use Whisper for English transcription.
```
import numpy as np
import onnxruntime as ort
from datasets import load_dataset
from transformers import AutoConfig, AutoProcessor
model = "openai/whisper-tiny"
config = AutoConfig.from_pretrained(model)
processor = AutoProcessor.from_pretrained(model)
forced_decoder_ids = processor.get_decoder_prompt_ids(language="english", task="transcribe")
# forced_decoder_ids is of the format [(1, 50259), (2, 50359), (3, 50363)] and needs to be
# of the format [50258, 50259, 50359, 50363] where 50258 is the start token id
forced_decoder_ids = [config.decoder_start_token_id] + list(map(lambda token: token[1], forced_decoder_ids))
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
input_features = processor(ds[0]["audio"]["array"], return_tensors="np").input_features
inputs = {
"input_features": np.float32(input_features),
"max_length": np.array([26], dtype=np.int32),
"min_length": np.array([1], dtype=np.int32),
"num_beams": np.array([2], dtype=np.int32),
"num_return_sequences": np.array([1], dtype=np.int32),
"length_penalty": np.array([1.0], dtype=np.float32),
"repetition_penalty": np.array([1.0], dtype=np.float32),
"decoder_input_ids": np.array([forced_decoder_ids], dtype=np.int32),
}
sess = ort.InferenceSession("whisper-tiny_beamsearch.onnx", providers=["CPUExecutionProvider"])
outputs = sess.run(None, inputs)
# Print tokens and decoded output
print(outputs[0][0][0])
print(processor.decode(outputs[0][0][0]))
```
If you don't want to provide specific decoder input ids or you want
Whisper to predict the output language and task, you can set
`forced_decoder_ids = [config.decoder_start_token_id]` instead.
### Motivation and Context
As seen in the figure below from the [OpenAI Whisper
paper](https://cdn.openai.com/papers/whisper.pdf), Whisper can be used
for multiple tasks and languages.
Register CPU OptionalGetElement, OptionalHasElement on DirectML
Graphs with OptionalGetElement and OptionalHasElement should work in a
DML graph without extra memcpy operation on and off the GPU.
CopyCpuTensor is swapped with DataTransferManager.CopyTensor() to make
the CPU operator usable by other providers.
---------
Co-authored-by: Dwayne Robinson <dwayner@microsoft.com>
### Description
<!-- Describe your changes. -->
Add registration for DML RoiAlign-16 and tests for new
coordinate_transform_mode attribute. PR
[7354](https://github.com/microsoft/onnxruntime/pull/7354) is still open
to fix the CPU EP version, which is why there are skipped tests right
now. That will be completed separately so that, for now, we can
officially support opset16 with the next release.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Linnea May <linneamay@microsoft.com>
Co-authored-by: Dwayne Robinson <dwayner@microsoft.com>
### Description
- Update DML version to 1.11.0
- Disable Gemm+Softmax fusion
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
### Description
<!-- Describe your changes. -->
Add registration for DML reduce functions in opset 18.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Linnea May <linneamay@microsoft.com>
### Description
This PR contains fusion-level and kernel-level optimizations for
[OpenAI's Whisper](https://github.com/openai/whisper).
Some of the added optimizations include:
- Pruning of duplicate/unnecessary inputs and outputs
- Fusion support for Whisper models with or without these inputs/outputs
(e.g. with these inputs/outputs if exporting with an older official
Optimum version, without these inputs/outputs if exporting with Optimum
from source)
- Attention fusions
- For Whisper's encoder and decoder
- Modified symbolic shape inference for present output when no past
input exists (for decoder)
- Multi-head attention fusions
- For Whisper's decoder and decoder with past
- Packed MatMul for the 3 MatMuls excluded in multi-head attention
fusion
- Attention kernel changes
- CPU:
- Different Q and KV sequence lengths
- Parallel memset for large sequence lengths
- Convert broadcast add after MatMul of Q and K (add_qk) to element-wise
add
- Separate present key-value output into present key and present value
(for multi-head attention spec)
- CUDA:
- Use memory efficient attention compute kernel with present state (for
decoder)
- Multi-head attention kernel changes
- CPU:
- Introduction of multi-head attention CPU kernel (previously did not
exist)
- Use AddBiasReshape instead of AddBiasTranspose when sequence length =
1 (for decoder with past)
- Different Q, K, V input shapes
- Pass past key and past value directly as key and value
- CUDA:
- Use memory efficient attention compute kernel with past and/or present
state (for decoder with past)
### Usage
To use the optimizations, run the ORT transformer optimizer script as
follows:
```
$ cd onnxruntime/onnxruntime/python/tools/transformers/
$ python3 optimizer.py --input <filename>.onnx --output <filename>.onnx --model_type bart --num_heads <number of attention heads, depends on the size of the whisper model used> --hidden_size <attention hidden size, depends on the size of the whisper model used> --use_external_data_format --use_multi_head_attention
```
Once optimized, here's an example of how to run Whisper with [Hugging
Face's Optimum](https://github.com/huggingface/optimum):
```
from transformers.onnx.utils import get_preprocessor
from optimum.onnxruntime import ORTModelForSpeechSeq2Seq
from optimum.pipelines import pipeline as ort_pipeline
import whisper # Installed from OpenAI's repo - setup instructions at https://github.com/openai/whisper/
directory = './whisper_opt' # Where the optimized ONNX models are located
model_name = 'openai/whisper-tiny'
device = 'cpu'
# Get pipeline
processor = get_preprocessor(model_name)
model = ORTModelForSpeechSeq2Seq.from_pretrained(
directory,
use_io_binding=(device == 'cuda'),
provider='CPUExecutionProvider',
).to(device)
pipe = ort_pipeline(
"automatic-speech-recognition",
model=model,
tokenizer=processor.tokenizer,
feature_extractor=processor.feature_extractor,
device=(-1 if device == 'cpu' else 0),
)
# Load audio file and run pipeline
audio = whisper.load_audio('tests/jfk.flac')
audio = whisper.pad_or_trim(audio)
outputs = pipe([audio])
print(outputs)
```
Note: In order to use these changes with Optimum, it is recommended to
use Optimum from source to have the following changes:
- https://github.com/huggingface/optimum/pull/872
- https://github.com/huggingface/optimum/pull/920
### Motivation and Context
This PR helps the following issues:
- https://github.com/microsoft/onnxruntime/issues/15100
- https://github.com/microsoft/onnxruntime/issues/15235
- https://github.com/huggingface/optimum/issues/869 (work in progress)
This PR can be used with the other currently merged Whisper PRs:
- https://github.com/microsoft/onnxruntime/pull/15247
- https://github.com/microsoft/onnxruntime/pull/15339
- https://github.com/microsoft/onnxruntime/pull/15362
- https://github.com/microsoft/onnxruntime/pull/15365
- https://github.com/microsoft/onnxruntime/pull/15427
This PR uses changes from the following merged PRs:
- https://github.com/microsoft/onnxruntime/pull/14198
- https://github.com/microsoft/onnxruntime/pull/14146
- https://github.com/microsoft/onnxruntime/pull/14201
- https://github.com/microsoft/onnxruntime/pull/14928 (this introduced
the new multi-head attention spec)
### Description
Adjust various code paths to allow Whisper model to function with
BeamSearch op.
Approach: Add a new kModelType enum value in IGenerationParameters as
so:
#### Old: 0 = GPT2, 1 = T5
#### New: 0 = GPT2, 1 = T5, 2 = Whisper
When the user assigns this attribute value to 2, various shape and type
checks are changed to accommodate Whisper inputs.
### Motivation and Context
BeamSearch is currently designed to function with BERT-based models with
inputs as vocab tokens, and needs changes to function with Whisper
inputs (3-D float values processed from audio data).
---------
Co-authored-by: Peter McAughan <petermca@microsoft.com>
### Description
<!-- Describe your changes. -->
As synced offline, rename this op and will create another op for mha
that supports both self and cross attention.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Ubuntu <wy@v100-2.0cdb2e52twzevn1i4fi45bylyg.jx.internal.cloudapp.net>
### Description
<!-- Describe your changes. -->
Transformer models can handle batch of inputs at once. However,
sequences in a batch usually have different length. Then we have to pad
the short one to have same length as the longest. This is not efficient
especially for large batch with high variance.
This PR introduces a PackedAttention operator which can take in packed
sequences (no padding) and also produces output in packing mode.
There will be another PR to use the PackedAttention to implement the
encoder in packing mode.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
### Description
<!-- Describe your changes. -->
1. support optional bias in Attention op (used in T5 encoder)
2. support broadcasting rel_pos_bias in attention_softmax.h
3. add scale in
MHA op's attributes
4. support past_key/past_value and present_key/present_value in MHA
5. UT and parity tests are added
6. fix an issue: https://github.com/microsoft/onnxruntime/issues/14920
note: the fusions will be in another PR since mt5 needs to be tested and
an issue from github will be investigated.
Future works:
1. support shared buffer for past/present
2. enable trt kernels when possible and investigate (trt/cutlass)kernels
with rel_pos_bias)
3. support KV/QKV packing with past/present
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Ubuntu <wy@v100-2.0cdb2e52twzevn1i4fi45bylyg.jx.internal.cloudapp.net>
### Description
Implements the STFT operator for the DirectML execution provider. This
is implemented as a custom op, just like the DFT kernel, because it's
implemented as a composite of two operators (DML Mul/Identity + DFT). As
such, this inherits the same restrictions as the existing DFT kernel
(requires power-of-two window sizes for now).
This change also adds a native FP16 shader to DFT so that both DFT/STFT
kernels support float16 tensors. There is no typed UAV fallback or
emulation path, so the HW _needs_ to support native float16. It also
appears the stockham shader was compiled with all optimizations disabled
and debug symbols (tsk tsk, Sheil), and this has been fixed.
This is passing all existing STFT tests (i.e. all of 1). I'm adding some
additional collateral in the Windows AI conformance tests in parallel to
check some extra cases.
---------
Co-authored-by: Patrice Vignola <vignola.patrice@gmail.com>
Enable Opset11 Sequence Ops on DirectML, and make the CPU
implementations agnostic to backend EP
Opset 11 introduced the following sequence related operators:
- SequenceAt
- SequenceConstruct
- SequenceEmpty
- SequenceLength
- SequenceErase
- SequenceInsert
- ConcatFromSequence
With the exception of ConcatFromSequence, all of the above operators
were implemented with CPU kernels that a) required all of the contained
tensors to also be on CPU, and b) would clone each tensor into a new
sequence as a side effect of each operator. The implementation of
sequences are backend agnostic, as they dont affect actual tensor layout
or manipulate the contents of the tensors. In addition, with the
exception of SequenceAt, the other operators need not make copies of the
underlying referenced tensors.
Consequently, this change does the following:
1) Sequence* operators (except SequenceAt) no longer copies the contents
of a sequence of tensors on every kernel execution.
2) SequenceAt uses the DataTransferManager to copy tensors agnostic to
backend.
3) The internal container implemented by TensorSeq has changed from
onnxruntime::Tensor to OrtValue. This is because onnxruntime::Tensor
does not support copy or assignment construction, so it must have a
singular owner. However, is same tensor participates in multiple
containers it would have multiple container "owners" and this would not
be possible.
4) Other code that accessed values from TensorSeq have associated
changes to extract Tensors from OrtValues now.
In addition, DirectML execution was very slow when the above Sequence
operators were added to a graph, as this caused MemcpyToHost and
MemcpyFromHost kernels to be inserted between the graph and the sequence
operators. To optimize DirectML,
1) The CPU implementations for the Sequence* ops were registered as DML
implementations. Since the above changes also includes making the CPU
kernel implementations EP agnostic, the CPU kernels can be added as is.
2) The ConcatFromSequence operator needed to be implemented on DirectML.
However, there was little DirectML EP operator framework support for
operators that accept/output sequences of tensors. This change has
modified the internal COM interfaces to include new apis to interrogate
for sequence shapes, and extract the needed tensors from TensorSeq.
---------
Co-authored-by: Patrice Vignola <vignola.patrice@gmail.com>
The third part for stable diffusion CUDA optimizations
(1) Add BiasAdd operator to replace two Add (bias and residual); Add
fusion for BiasAdd
(2) Add Attention fusion for VAE decoder.
(3) Update float16 conversion to handle Resize and GroupNorm. This could
reduce two Cast nodes for each Resize op in fp16 model.
(4) Force inputs and outputs to be float16 to avoid data casts in the
pipeline.
(5) Add options --force_fp32_ops, --inspect etc in optimize script so that
user could force some operator to run in float32 to potentially get
better image quality (with cost of performance).
Performance tests show slight improvement in T4. Average latency reduced
0.1 seconds (from 5.35s to 5.25s) for 512x512 in 50 steps.
### Description
<!-- Describe your changes. -->
1. fix a bug in relative position bias kernel where seq_len > 32
2. rename extra_add_qk to relative_position_bias
3. support relative_position_bias in multihead attention (B, N, S, S*)
4. gru_gate support by Lei
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
---------
Co-authored-by: Ubuntu <wy@v100-2.0cdb2e52twzevn1i4fi45bylyg.jx.internal.cloudapp.net>
Co-authored-by: Lei Zhang <zhang.huanning@hotmail.com>
### Description
Reduce the cuda library size by:
1. refactoring beam_search_top_k to reduce template instantiation. It
saves ~56MB
2. opt out TopK for type uint*, int8_t and int16_t. It saves ~50MB.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->
### Description
Add stable diffusion CUDA kernel optimizations.
The following are included:
(1) GroupNorm operator. This kernel is from TensorRT 8.5.
(2) BiasSplitGelu operator. This kernel is modified from SplitGelu of
TensorRT 8.5. We added bias to the SplitGelu.
(3) NhwcConv operator. This adds support of NHWC format (ONNX Conv
operator uses NCHW format).
(3) Update MultiHeadAttention (packed kv and no bias) for cross
attention. This could avoid transpose of kv for TRT fused cross
attention kernel.
(4) Optimization and benchmark script
Not included:
(1) Script to convert Conv to NhwcConv in onnx graph.
(2) Update symbolic shape inference for NhwcConv.
(3) Add SeqLen2Spatial operator
(4) Documents
Limitations: GroupNorm, BiasSplitGelu and NhwcConv kernels are
implemented based on stable diffusion usage. They might not be
applicable to any input size or dimensions. For example, BiasSplitGelu
requires hidden size to be 2560 | 5120 | 10240, and NhwcConv assumes 4D
input/weight.
There is minor increasement of binary size. For SM=75 only, python
package wheel size adds (33757K - 33640K) = 117 KB. It is possible to
move NHWC from template parameter to constructor to reduce binary size
(with slight cost of performance).
Note: for RTX 4090/4080/4070 Ti, need build with CUDA 11.8 and latest
cuDNN to get best performance.
This PR registers ScatterElements-16 to the DML EP
- CPU fallback is added if the reduction attribute is in use, as this is
not yet supported by DML.
---------
Co-authored-by: Numfor Mbiziwo-Tiapo <numform@microsoft.com>
