### Description
reducemax/min have been updated in onnx(20). implement it in ort
### Motivation and Context
this is for ort1.17.0 release
---------
Signed-off-by: Liqun Fu <liqfu@microsoft.com>
### Description
dft is updated in opset20. implement it in ort
### Motivation and Context
this is for ort 1.17.0 release
Fixes#17723
---------
Signed-off-by: Liqun Fu <liqfu@microsoft.com>
### Allow layer-wise recompute
Early, we need users/developers to specify the subgraphs to recompute,
now we introduced a more user-friendly way to enable recompute for all
detected stashed activation recomputation subgraphs. This scarifies
getting the best configs while makes it easier to support user
requirements when they switches from PyTorch per-layer gradient
checkpoint to ORTModule.
`ORTMODULE_MEMORY_OPT_LEVEL` is introduced to control the usage, by
default, it is 0, e.g. `USER_SPECIFIED`, all subgraphs definedin
`ORTMODULE_MEMORY_OPT_CONFIG` will be recomputed. So this is compatible
to existing recompute usage in ORTModule integrated models.
Using `ORTMODULE_MEMORY_OPT_LEVEL=1`, we will enable all recompute plans
detected, so those configs in `ORTMODULE_MEMORY_OPT_CONFIG` will not be
respected any more.
Add Unit Tests using 3 layer blooms.
https://github.com/microsoft/onnxruntime/blob/pengwa/add_aggresive_recompute/docs/Memory_Optimizer.md
Fix a bug that can't create context binary if the model has inputs/outputs with different data type
### Description
Update EPContext op schema to unblock nodes with different data type among inputs & outputs
### Skip module clone for preparing large model export
For LLAMA2 13B, when running with Lora, DeepSpeed stage2 on 8 GPUs . It
failed during preparing outputs which will be used for
torch.onnx.export. The reason, we deep copy all the params including
both big sizes of frozen weights, + a little bit of Lora trainable
weight.
This PR will firstly check whether the GPU memmory is enough for a
cloned module, if not, skip the copy.
Copying the module is to guarantee the fw path run may change the
weight, while this case should be rare. But for now, Not-Able-To-Run is
worse than Runnable-with-A-little-bit-different-initial-weight,
especially for large models.
This PR:
- Remove unused arguments from generated triton code,
- Remove unnecessary mask for symbolic shape case from generated triton
code.
- Add doc for usage of ORTMODULE_TRITON_CONFIG_FILE.
### Description
<!-- Describe your changes. -->
Add bfloat16 support for `MatMulBnb4` contrib op. This is useful for
QLoRA fine-tuning.
- On GPUs with SM80+ (A100, etc), it uses the native cuda bfloat16
dtype, `nv_bfloat16`. On other GPUs, it uses the onnxruntime `BFloat16`
type which uses float for compute.
- I have validated the op in a llama2-7b training scenario. The losses
match pytorch training and the training throughput is better.
- Cannot add a bfloat16 case in the op unit test since casting BFloat16
to and from float multiple times during the test causes the required
tolerances to be unachievable.
The custom autograd function exporter in onnxruntime-training is updated
to support the latest version of bitsandbytes. They changed how the
`quant_state` is stored.
### 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. -->
Enable QLoRA fine-tuning with bfloat16.
### Description
<!-- Describe your changes. -->
change RotaryEmbeddings op implementation, add support for 4D input
tensor that is with shape of [batch, num_heads, seq_len, head_size].
### 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. -->
Current RotaryEmbedding op only support 3d input tensor with shape
[batch, seq_len, hidden_size]
For llamav2 model, when using FusionRotaryEmbeddings to only fuse
RotaryEmbeddings op, there will be a transpose operation for query and
key, and then the input tensor of RotaryEmbeddings becomes 4D [batch,
num_heads, seq_len, head_size].
This scenario can't be supported by current RotaryEmbeddings
implementation. So it needs to support 4D input tensor.
### Description
Implement preliminary version of local (sliding window) attention.
Currently only supported by Flash Attention (sm >= 80, Linux). Currently
only supports sliding attention with a large cached kv.
### Motivation and Context
This change enables to run Mistral and other models which use sliding
window attention.
### Description
<!-- Describe your changes. -->
1. Introduce MoE CUDA op to ORT based on FT implementation.
2. Upgrade cutlass to 3.1.0 to avoid some build failures on Windows.
Remove patch file for cutlass 3.0.0.
3. Sharded MoE implementation will come with another PR
limitation: __CUDA_ARCH__ >= 700
### 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. -->
Registers BFloat16 datatype as valid input type for CUDA Neg Kernel.
### 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. -->
Enabling `meta-llama/Llama-2-70b` to be finetuned with ONNX Runtime
training.
---------
Co-authored-by: Prathik Rao <prathikrao@microsoft.com@orttrainingdev8.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
### Tune logging experience a bit
After last time we update the ORTModule log experience, we found few
issues:
1. `INFO` level output too many things, including PyTorch exporter
verbose logs (tracing graphs) on every ranks. On this level, we only
want to
- Output a little bit more information to Users than `WARNING` level,
for example the memory recomputation recommendations or other
not-fully-ready features.
- Output a little bit more information for a quick diagnostic, collected
on rank-0 only.
2. ONNX Runtime logging filter during graph build, session init
sometimes will hide the issues (for example segement fault), there is no
useful information in `WARNING`/`INFO` for users to report to us. This
is not good!
3. Some of our devs like using `pdb` to debug Python code, but if we add
`import pdb; pdb.set_trace()` in models' code might hang when they use
`INFO` or `WARNING`, where exporter happens and all output got
redirected due to log filtering. The only workaround is to switch to
VERBOSE, which output toooooooooooo many logs.
The corresponding changes proposed here are:
1. For `INFO` logging,
- We only logs rank-0.
- We restricted the ORT backend logging level to be WARNING in this
case, because ORT backend code output way too many logs that should be
under verbose, while we cannot guarantee we can get them cleaned up
immediately once they are added.
- We output the PyTorch exporter verbose log (including tracing graph),
which is useful for a quick diagnostic when an issue happens.
2. Remove all logging filtering on ORT backend, then the segment fault
issue details will not be hidden once it happens again.
3. Introduced a `DEVINFO` logging,
- Log logs on all ranks
- Log ORT backend logging level INFO
- PyTorch exporter logging filtering are all turned OFF (to unblock the
pdb debugging).
4. Currently, to use Memory Optimizer, need use DEVINFO (which will
output ORT backend INFO log). So update memory optimizer document to
reflect this. https://github.com/microsoft/onnxruntime/pull/17481 will
update the requirement back to INFO for show memory optimization infos.
You can check
https://github.com/microsoft/onnxruntime/blob/pengwa/devinfo_level/docs/ORTModule_Training_Guidelines.md#log-level-explanations
for a better view of different log levels.
This PR also extract some changes from a bigger one
https://github.com/microsoft/onnxruntime/pull/17481, to reduce its
complexity for review.
### 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: mindest <30493312+mindest@users.noreply.github.com>
### Description
GQA now only works with Flash Attention with Attention Mask input,
allowing for batched input. Note: This PR Disables Memory Efficient
Attention, only allowing Flash Attention kernel to be used.
### Motivation and Context
Allows GQA to work with batched input.
---------
Co-authored-by: Yufeng Li <liyufeng1987@gmail.com>
This is a graph implementation of RotaryEmbedding since there's no time
to add it to DML before 1.16.2, but it eventually should move into
DirectML since we're bandwidth-bound.
### Description
<!-- Describe your changes. -->
Adds bfloat16 as a valid input parameter type for where node for ONNX
opset 16+.
### 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. -->
Enabling `meta-llama/Llama-2-70b` to be finetuned with ONNX Runtime
training.
---------
Co-authored-by: Prathik Rao <prathikrao@microsoft.com@orttrainingdev8.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
### Optimize 4bit Qlora training
Extent existing `MatmulBnb4bit` to its usage in training scenarios.
The PR includes following changes:
1. Add special `torch.autograd.Function` export logic for
`bitsandbytes.autograd._functions.MatMul4Bit` that is preferred before
common PythonOp exporter.
2. Add `training_mode` optional attribute for op `MatmulBnb4bit`, which
help skip some inference specific logic in implementation.
3. Add `transB` optional attribute, which is by default be 1; setting it
to be 0 is needed by backward usage.
Changing from `PythonOp` to this `MatmulBnb4bit` brings roughly ~2.9%
throughput gains. The reason is:
`bitsandbytes.autograd._functions.MatMul4Bit` has logic
`ctx.save_for_backward`, which would need an additional copy in
PythonOp, otherwise, the tensor might be released by ORT, while backward
op still references it.
Removing the clones also reduce the peak memory consumptions because
`bitsandbytes.autograd._functions.MatMul4Bit` saved tensors that are not
needed in backward compute.
Implement Cutlass Memory Efficient Attention Kernel into Group Query
Attention Operator.
### Motivation and Context
Before this change, Group Query Attention Operator was supported only by
Flash-Attention. While this is the most efficient kernel for the
operation, it only supports sm >= 80. Cutlass Memory Efficient Attention
Kernel supports sm >= 53, allowing us to support a broader range of GPU
hardware.
### Description
Integration to OpenVINO 2023.1
### Motivation and Context
- Alignment with latest OpenVINO Version.
- Device name change from VPUX to NPU and Remove from supported list
until official public support is available.
---------
Co-authored-by: Sahar Fatima <sfatima.3001@gmail.com>
Co-authored-by: Saurabh Kale <saurabh1.kale@intel.com>
Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com>
Co-authored-by: sfatimar <sahar.fatima@intel.com>
* Add a new operator SkipGroupNorm to support skip and bias inputs.
* Update GroupNorm kernel to support number of channels used in SD XLrefiner.
* Add epsilon in kernel
* Add parity and performance test script
* Remove many limitations including max batch size, max number of groups, c % cPerBlock ==0 etc.
### Motivation and Context
Update GroupNorm to support SD XL Refiner and beyond.
### Description
Add support for Gemm with float 8 as a contrib op.
---------
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>
Co-authored-by: Xavier Dupre <xadupre@microsoft.com@orttrainingdev9.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
### Description
Opset 18 apply the "axes as input" change from ReduceSum to all the
other reduce ops. Our cuda kernel actually support it, but we didn't
enable it for opset18. This PR update the reduce ops' kernel
registration to enable the "axes as input" behavior for opset18.
As part of the fix, I also simplify the reduce op kernel registration
part. ORT doesn't require the kernel definition need to be exactly the
same as onnx op definition. For our case, which we share the same kernel
for all the reduce ops (from version 1 to version 18), we don't need to
maintain different version of kernel definitions. we can simplify it by
just using a single kernel definition for multiple versions. Although
for some cases, we might register more types for legacy versions, but it
is harmless. Framework is using schema to validate the graph, not kernel
definition.
---------
Co-authored-by: Cheng Tang <chenta@a100.crj0ad2y1kku1j4yxl4sj10o4e.gx.internal.cloudapp.net>
Co-authored-by: Cheng Tang <chenta@microsoft.com>
### Description
Add a contrib op MatMulBnb4 (FP4 and NF4) and related toolchain to
support quantization on weight.
This PR adds:
- schema for contrib op MatMulBnb4 which can support FP4 (4-bit floating
point) and NF4 (4-bit NormalFloat) quantization on weight.
- a naive implementation for MatMulBnb4 on CPU and GPU, i.e.,
implemented like MatMul(A, Dequantize(B)).
- a special implementation for GemV for MatMulBnb4 and related benchmark
tool.
- tool to quantize model to FP4 or NF4.
### Description
<!-- Describe your changes. -->
Add a contrib op MatMulNBits and related toolchain to support
quantization on weight. This PR only adds support for 4bits. It:
- add schema for contrib op MatMulNBits which can support 1-7 bits
quantization on weight.
- a naive implementation for 4bits MatMulNBits on CPU and GPU, i.e.,
implemented like MatMul(A, Dequantize(B)).
- a special implementation for GemV for 4bits MatMulNBits and related
benchmark tool
- tool to quantization model with 4bits.
Next:
- add general and more efficient kernels for 4bits MatMulNBits on CPU
and GPU
Support cross qk in beam search for whisper model and related features
Make whisper exporting tools support cross qk and some related features,
* extra_decoding_ids
* no_speech_prob
Implement DTW kernel, unfold tensor kernel with unit test Several fix
related with multiple session running parallel, like:
* guard multihead_attention, fused_fp16_runner_
* some memory allocation with stream awareness
* add use_ep_level_unified_stream option
### Description
Improve the QNN context binary cache feature to reduce the memory
overhead and initialization time overhead.
Instead of dumping a Qnn context binary file with metadata as header, we
dump a Onnx format file with metadata inside Onnx node.
### Motivation and Context
reduce the memory overhead and initialization time overhead
### Description
this is for ORT 1.17.0 - make ORT to use ONNX release 1.15.0 branch. Eventually will update to the release tag once ONNX 1.15.0 is released
### Motivation and Context
Prepare for ORT 1.17.0 release. People can start work on new and updated ONNX ops in ORT.
---------
Signed-off-by: Liqun Fu <liqfu@microsoft.com>
### Description
Updated a couple of old links in the technical documentation that where
pointing to files present prior to the migration to
https://onnxruntime.ai/docs.
### Introduce ZeROOffloadSubscriber for ORTModule
As part of the work: integrate ORTModule with DeepSpeed stage3, this PR
mainly focus on moving original PyTorch-based (leveraging hooks) param
partition/offload implementation to ORTModule compatible implementation.
Changes include:
1. Refactor `SubscriberBase`/`SubcriberManager` to support
pre-forward/post_forward hooks.
2. Implement new `ZeROOffloadSubscriber` by re-using DeepSpeed hook
function as much as possible. Since all hook functions are defined in
`DeepSpeedZeRoOffload._register_hooks_recursively` and
`DeepSpeedZeRoOffload.setup_zero_stage3_hooks`, and the good thing is,
the closure is not complex, all hooks are referencing the owning
`DeepSpeedZeRoOffload` instance, so we can create new hook function with
`FunctionType` by binding the owning `DeepSpeedZeRoOffload` instance,
then call the new created function in subscriber's
`pre_forward_module_apply_impl` and `post_forward_module_apply_impl`
interfaces.
3. Monkey patch `DeepSpeedZeRoOffload.setup_zero_stage3_hooks` to
register the `ZeROOffloadSubscriber` for the model, then we don't need
change any code on the DeepSpeed repo (at least so far).
4. Fix the ATen embedding custom symbolic exporter function by
tolerating weights size be (0) (changed by DeepSpeed zero stage 3).
UT will be added once stage3 is fully supported.
### 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. -->
This PR fixes broken hyperlinks in the documentation that should lead
users to Jupyter notebooks. Currently, the hyperlinks are not working as
intended. The PR resolves this issue by updating the hyperlinks to
correctly direct users to the Jupyter notebooks.
### Motivation and Context
<!-- - Why is this change required? What problem does it solve? -->
It fixes broken hyperlinks leading to the Jupyter notebooks.
### Description
Remove the onnxruntime-extensions submodule since it now was used via
cmake FetchContent
### Motivation and Context
The submodule relies on an outdated version of the extensions, and the
build instructions should be updated to eliminate any confusion.
OpenVINO EP ORT 5.1 Branch
Changes for the new API to take in OpenVINO Provider Options
and compatibility with OV 2023.1
### Motivation and Context
The change is required for the new API to take in OpenVINO Provider
Options
and make it seamless.
---------
Signed-off-by: MaajidKhan <n.maajid.khan@intel.com>
Co-authored-by: saurabhintel0 <saurabh1.kale@intel.com>
Co-authored-by: MaajidKhan <n.maajid.khan@intel.com>
Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com>
Co-authored-by: Preetha Veeramalai <preetha.veeramalai@intel.com>
### Use full qualified name for PythonOp export
Originally, when there are duplicate named torch.autograd.Function in
different module, for example:
`a.b.c.Gelu` v.s. `d.e.func.<locals>.Gelu`
We by default will throw exception to let user be aware we cannot
distinguish the two Gelu because during model export, we did not module
path. The workaround is we introduced
`ORTMODULE_SKIPPED_AUTOGRAD_FUNCTIONS` to ignore those duplicated named
Gelu that is not used by model run. This has limitations obviously for
example if two Gelus are both used in training.
This PR finds a way to construct a full qualified name.
`def _export_pt_1_10(g, n, *args, **kwargs):`
1. in exporter function, kwargs contains `name` and `module`, in the
above example:
`a.b.c.Gelu` --> name: `Gelu`, module: `a.b.c`
`d.e.func.<locals>.Gelu` --> name: `Gelu`, module: `d.e`
Using name and module is not enough to get a full qualified name, for
the second case, where `d.e` is the module path, then there is a
function called `func`, in this function, there is a local
auto.grad.Function named `Gelu`. (Many of our UT looks like this). We
can only get `d.e.Gelu`, but this is not the correct full qual name.
The reason for this: `kwargs[name]` or `n.name` only return the class's
name, not the class's full qual name. (be noted kwargs[module]` is
correct).
2. `n` is torch.Node, we can access `pyobj` to get the
torch.autograd.Function's apply method instance, then use `._self` to
get the torch.autograd.Function class. Then we can get the `module` and
`class`'s ful qual name, added together, we get the full qual name.
With the above change, we don't need use `kwargs[name]` and
`kwargs[module]` , and don't need check naming conflicting or
`ORTMODULE_SKIPPED_AUTOGRAD_FUNCTIONS` env var any more.
### Description
Enhanced SkipLayerNorm by implementing broadcasting for both CPU and
CUDA
### Motivation and Context
The input and skip tensors no longer have to be the same size which
means that it can accept data where the skip shape can be the same size
as the input shape, have a shape of {1, sequence_length, hidden_size},
or {sequence_length, hidden_size}.
---------
Co-authored-by: Tianlei Wu <tlwu@microsoft.com>
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
Fixes the issue with IRFFT output dimension calculation as described in
#13236
### Motivation and Context
Please refer to #13236 for detailed description.
Specifically, [this code](https://github.com/microsoft/onnxruntime/blob/main/onnxruntime/contrib_ops/cuda/math/fft_ops.cc#L103) computes the output dimension as:
```
out_dim = in_dim * 2 - 1
```
while it should be this instead:
```
out_dim = 2 * (in_dim - 1)
```
(assuming the original signal has even number of samples, of course).
For example, if the original signal has 4 samples, then the round trip should look something like:
```
4 -> (one-sided RFFT) -> 3 (complex) -> (one-sided IRFFT) -> 4
```
with the current code the output will be a signal with 5 points.
---------
Co-authored-by: Alexey Kamenev <akamenev@nvidia.com>
Co-authored-by: Nick Geneva <nicholasgeneva@gmail.com>
### Description
Remove VS 2019 code.
### 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. -->
This PR adds support to cache the exported training/evaluation ONNX
model in `ORTModule`. On future runs, instead of exporting the model
again, we can pick up the model from a location on disc and run
`ORTModule` training/evaluation.
### 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. -->
ORT Training DRI Contribution
---------
Co-authored-by: root <root@orttrainingdev8.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
Co-authored-by: Prathik Rao <prathikrao@microsoft.com@orttrainingdev8.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
Co-authored-by: Baiju Meswani <bmeswani@microsoft.com>
Co-authored-by: pengwa <pengwa@microsoft.com>
This will remove transposes that are non needed in the DML kernel. To
keep backward compatiblity, the default behavior is to set NHWC when no
attribute is set.
### Description
Disable two PERF* rules in ruff to allow better readability. Rational
commented inline. This change also removes the unused noqa directives
because of the rule change.
### Motivation and Context
Readability
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at
bottom):
* __->__ #16789
Bump ruff to 0.0.278 and fix new lint errors. I added noqa to all
existing RUF012 errors which requires mutable class variables to be
annotated with `ClassVar`, as well as all PERF issues.
Signed-off-by: Justin Chu <justinchu@microsoft.com>
### Description
This PR is includes changes in the documentation of _readmeOV.rst_ file
and also the changes in the dockerfile which enables to build ORT with
latest OpenVINO 2023.0.0
### Motivation and Context
Modified the dockerfile to incorporate the latest version of OpenVINO
(2023.0.0) for building Onnxruntime.
The changes in the PR aim to improve the overall user experience by
providing accurate and up-to-date documentation while leveraging latest
OpenVINO 2023.0.0
### Description
<!-- Describe your changes. -->
This PR adds support for rotary embeddings in decoder masked
self-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
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>
### Description
This PR includes documentation updates, providing step-by-step
instructions on how to implement the ModuleWithLoss wrapper in a
different codebase.
The documentation outlines the necessary code changes and offers
customization options based on specific requirements.
---------
Co-authored-by: Adam Louly <adamlouly@microsoft.com@orttrainingdev9.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
### Manage ORTModule options
Move all env vars that used for feature ON/OFF into runtime options for
consistent managements.
Be noted: the features' switch are assigned in 2 phases: default values,
overwritten by env vars (if specified by users). So env vars take the
highest priority when all 2 phases both given value explicitly for one
feature.
### 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
Optimize compute graph by eliminating padding in embedding.
### Motivation and Context
The computation for padding in nodes after embedding is unnecessary and
waste computation resources.
This pr just add an Optimizer of PaddingElimination to check and
eliminate the padding after embedding automatically by modifying the
graph.
### Implementation:
1. Find and check embedding node in graph.
2. Iterate the subgraph afterward the embedding node and record all the
input nodes and output nodes to this subgraph.
3. Insert 'Reshape + ShrunkenGather' to flatten each input node shape
from [batch_size, seqlen, ...] to [valid_token_without_padding, ...],
and insert 'GatherGrad + Reshape' to unflatten each output node shape
from [valid_token_without_padding, ...] to [batch_size, seqlen, ...]
---------
Co-authored-by: mindest <linminuser@gmail.com>
### Description
This PR enables execution of subgraphs in OVEP and currently, when OVEP
developers install the onnxruntime-openvino package on windows from
pypi, they would have to additionally download OpenVINO windows binaries
and run the setupvars.bat script which sets the environment PATH to
locate the OV dll's. Also this PR fixes issues of OVEP windows io buffer
sample.
### Motivation and Context
Fix: We want to make the user experience easy for OVEP Python developers
on windows platform.
This fix, introduces a function add_openvino_libs_to_path at the
location tools/python/util/add_openvino_win_libs.py.
The above function, can be called by OVEP python users in the
application code and that takes care of setting
the OpenVINO dll's to the path from the OpenVINO pypi packge (openvino)
which was installed.
This change also makes sure that add_openvino_libs_to_path() function is
added to onnxruntime python package
only when it is build for OpenVINO Execution Provider for ONNXRuntime
and not for default ORT python package builds.
New user experience for Python OVEP developers on windows platform:
step 1: pip install onnxruntime-openvino
step 2: pip install openvino
step 3: <Add these 2 lines in the application code>
import onnxruntime.tools.add_openvino_win_libs as utils
utils.add_openvino_libs_to_path()
---------
Signed-off-by: MaajidKhan <n.maajid.khan@intel.com>
Co-authored-by: MaajidKhan <n.maajid.khan@intel.com>
Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.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.
### Enhance StatisticsSubscriber
There are few improvements for `StatisticsSubscriber`:
- Reduce peak memory impact for tensors (having many many many elements,
consuming too much GPU memory, causing original recipe run failed with
OOM), by split the statistics into two phases (split into buckets, and
merge result across buckets).
- Allow dump intermediate tensors. Originally only nn.Module forward()'s
return value are dumped, there are requirements we want to inspect some
specific intermediate tensor in the forward() function, now we support
it.
- Add documents for collecting dumps on multiple ranks
Docs link on this branch for better view:
https://github.com/microsoft/onnxruntime/blob/pengwa/conv_tool_v2/docs/ORTModule_Convergence_Notes.md
---------
Co-authored-by: mindest <30493312+mindest@users.noreply.github.com>
### 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>
### Enable conditional optimization on inputs
Label sparsity based optimization can be enabled depending on the input
inspection result.
So this PR introduce a conditional optimization path for ORTModule,
where we automatically detect data sparsity from label or embedding, and
enable the graph optimization accordingly without any user interaction.
This feature had a new requirement of delaying passing pre_grad graph
transformation config to OrtModuleGraphBuilder, from `Initialize` phase
to its `Build` phase. Because once after `_initialize_graph_builder` we
can detect the input sparsity, and make a decision to enable the
label/embed sparisty based graph optimizations.
Add UT cases for label/embed input runtime inspector.
* 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>
### Description
In some scenarios, the triton written kernels are more performant than
CK or other handwritten kernels, so we implement a framework that
onnxruntime can use these triton written kernels.
This PR is to integrate triton into ort, so that ort can use kernels
that written and compiled by triton.
The main change focus on two part:
1. a build part to compile triton written kernel and combine these
kernels into libonnxruntime_providers_rocm.so
2. a loader and launcher in c++, for loading and launch triton written
kernels.
#### Build
To compile triton written kernel, add a script
`tools/ci_build/compile_triton.py`. This script will dynamic load all
kernel files, compile them, and generate `triton_kernel_infos.a` and
`triton_kernel_infos.h`.
`triton_kernel_infos.a` contains all compiled kernel instructions, this
file will be combined into libonnxruntime_providers_rocm.so, using
--whole-archive flag.
`triton_kernel_infos.h` defines a const array that contains all the
metadata for each compiled kernel. These metadata will be used for load
and launch. So this header file is included by 'triton_kernel.cu' which
defines load and launch functions.
Add a build flag in build.py and CMakeList.txt, when building rocm
provider, it will call triton_kernel build command, and generate all
necessary files.
#### C++ Load and Launch
On c++ part, we implement load and launch functions in triton_kernel.cu
and triton_kernel.h.
These two files located in `providers/cuda`, and when compiling rocm,
they will be hipified. so this part supports both cuda and rocm. But
currently we only call triton kernel in rocm.
We also implement a softmax triton op for example. Because there will
generate many kernels for different input shape of softmax, we use
TunableOp to select the best one.
### 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. -->
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.
### Description
<!-- Describe your changes. -->
V100, b_4_s_128, max_output_len=64, beam=4
before:
t5_small: 101.28ms
t5_base: 200.07ms
after:
t5_small: 87.65ms
t5_base: 174.44ms
### 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>
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
1. Update VERSION_NUMBER for preparing the upcoming release. This PR's
commit will not be included in the 1.15 release branch
2. Delete package/rpm/onnxruntime.spec since it was not used in past
years.
### Motivation and Context
Preparing the release.
Fixed
[AB#15311](https://aiinfra.visualstudio.com/6a833879-cd9b-44a4-a9de-adc2d818f13c/_workitems/edit/15311)
### 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 changes an EmbedLayerNormalization node's mask index output to
be an optional output if a mask input is not provided.
### Motivation and Context
The documentation for EmbedLayerNormalization states
```
The last input mask is optional. If mask is provided, mask index (that is position of first 0 in mask, or number of words) will be calculated.
```
However, if the mask input is not provided, the mask index output is
still calculated and required.
### Description
The PR adds VPU support to OpenVINO Execution Provider
Bug fixes for GPU, CPU.
Changes to OpenVINO Backend in Serialized Model API for faster First
Inference Latency.
Deprecation to HDDL-VADM and MYRIAD, removed code
Support OpenVINO 2023.0
Dynamic Shapes Support for iGPU
### Motivation and Context
- VPU is an upcoming hardware that can provide AI Acceleration for
Client Systems through OpenVINO
- If it fixes an open issue, please link to the issue here. -->
---------
Signed-off-by: MaajidKhan <n.maajid.khan@intel.com>
Co-authored-by: Suryaprakash Shanmugam <suryaprakash.shanmugam@intel.com>
Co-authored-by: MaajidKhan <n.maajid.khan@intel.com>
Co-authored-by: Preetha Veeramalai <preetha.veeramalai@intel.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
Bump ruff version in CI and fixed new lint errors.
- This change enables the flake8-implicit-str-concat rules which helps
detect unintended string concatenations:
https://beta.ruff.rs/docs/rules/#flake8-implicit-str-concat-isc
- Update gitignore to include common python files that we want to
exclude.
### Motivation and Context
Code quality
### Optimize SCE loss compute
Compute optimization based on label data sparsity:
- Insert ShrunkenGather before SCELoss node, to filter out invalid
labels for compute.
- Support ShrunkenGather upstream.
- Added test for the above.
- Added flag to enable label sparsity optimization with env var, by
default disabled now. Will enable after comprehensive benchmarking
later.
- Extract common logic into test_optimizer_utils.h/cc from
core/optimizer/compute_optimzier_test.cc, then the common functions can
be shared by both core/optimizer/compute_optimzier_test.cc and
orttraining/core/optimizer/compute_optimzier_test.cc
- Extract common logic into shared_utils.h/cc: `GetONNXOpSetVersion` and
`Create1DInitializerFromVector`
### 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
Adding 'Add' functionality to FP16 Conv operator. It takes a tensor that
has the same shape of the output tensor, and add it to the result
tensor.
### Motivation and Context
Needed to run Resnet 50
### 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. -->
Add a tool to convert fused BERT like model to 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
`lintrunner` is a linter runner successfully used by pytorch, onnx and
onnx-script. It provides a uniform experience running linters locally
and in CI. It supports all major dev systems: Windows, Linux and MacOs.
The checks are enforced by the `Python format` workflow.
This PR adopts `lintrunner` to onnxruntime and fixed ~2000 flake8 errors
in Python code. `lintrunner` now runs all required python lints
including `ruff`(replacing `flake8`), `black` and `isort`. Future lints
like `clang-format` can be added.
Most errors are auto-fixed by `ruff` and the fixes should be considered
robust.
Lints that are more complicated to fix are applied `# noqa` for now and
should be fixed in follow up PRs.
### Notable changes
1. This PR **removed some suboptimal patterns**:
- `not xxx in` -> `xxx not in` membership checks
- bare excepts (`except:` -> `except Exception`)
- unused imports
The follow up PR will remove:
- `import *`
- mutable values as default in function definitions (`def func(a=[])`)
- more unused imports
- unused local variables
2. Use `ruff` to replace `flake8`. `ruff` is much (40x) faster than
flake8 and is more robust. We are using it successfully in onnx and
onnx-script. It also supports auto-fixing many flake8 errors.
3. Removed the legacy flake8 ci flow and updated docs.
4. The added workflow supports SARIF code scanning reports on github,
example snapshot:
5. Removed `onnxruntime-python-checks-ci-pipeline` as redundant
### 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. -->
Unified linting experience in CI and local.
Replacing https://github.com/microsoft/onnxruntime/pull/14306
---------
Signed-off-by: Justin Chu <justinchu@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. -->
1. upgrade cutlass to 3.0 that containing attn_bias support.
2. extend Attention/MHA to use memory efficient attention when
rel_pos_bias with [1, num_head, s, s*] and 1d mask with [2 * batch_size
+ 1] are present.
new mask format introduction:
MASK_1D_KEY_SEQ_LEN_START,
[3 * batch_size + 2] with [key_len[0], ..., key_len[batch_size - 1],
query_start[0], ..., query_start[batch_size - 1], query_end[batch_size -
1], key_start[0], ..., key_start[batch_size - 1], key_end[batch_size -
1]]
e.g
2D mask with [[1, 1, 1, 0, 0, 0], [1, 1, 1, 1, 1, 0]] converts to this
1D mask is [3, 5, 0, 6, 12, 0, 6, 12]
### 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. -->
It potentially benefits tnlrv6 and t5(encoder)
---------
Co-authored-by: Ubuntu <wy@v100-2.0cdb2e52twzevn1i4fi45bylyg.jx.internal.cloudapp.net>
Co-authored-by: Kunal Vaishnavi <kvaishnavi@microsoft.com>
Co-authored-by: Kunal Vaishnavi <kvaishnavi@microsoft.com@orttrainingdev7.d32nl1ml4oruzj4qz3bqlggovf.px.internal.cloudapp.net>
### Statistics tool for ORTModule convergence parity
As ORTModule get more and more validated, it is pretty fast to
intergrade PyTorch based model with ORT.
The same time, we need make sure once there is convergence issue, we
don't spend months of time to investigate. As part of this efforts, this
PR is introducing a tool to dump activation statistics without much
involvement from users. The dumping results contains only some statistic
numbers plus sampled data, which is not big, compared with dumping all
the tensors, it is much faster and space efficient.
For us to use it, two single lines are needed before wrapping ORTModule.
For baseline run, need also apply the same trick.
```
+ from onnxruntime.training.utils.hooks import SubscriberManager, StatisticsSubscriber
+ SubscriberManager.subscribe(model, [StatisticsSubscriber("pt_out", override_output_dir=True)])
```
Once you run the steps, following command can be used to merge result
into per-step-summary respectively for ORT and baseline runs.
```bash
python -m onnxruntime.training.utils.hooks.merge_activation_summary --pt_dir pt_out --ort_dir ort_out --output_dir /tmp/output
```
Docs is added here as part of this PR [convergence investigation
notes](https://github.com/microsoft/onnxruntime/blob/pengwa/conv_tool/docs/ORTModule_Convergence_Notes.md)
Based on the generated merged files, we can compare them with tools.
### Design and Implementation
This PR introduced a common mechanism registering custom logic for
nn.Module's post forward hooks. And statistics for activation
(StatisticsSubscriber) is one of the implementations. If there is other
needs, we can define another XXSubscriber to do the customized things.
### 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>
### Description
<!-- Describe your changes. -->
I fixed some broken links in the C API documentation, but then did a
quick pass over all of the links I could find and then fixed those.
### 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. -->
I got some 404's when exploring the documentation and wanted to fix it.
### Description
<!-- Describe your changes. -->
### 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>
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>
(1) Support packed QKV format in MultiHeadAttention. This format could
avoid add bias transpose when TRT fused kernel is used.
(2) Add cache for cumulated sequence length computation. For SD, it only
need computed once since sequence length is fixed.
(3) Do not allocate qkv workspace to save memory for packed KV or QKV.
(4) Add unit tests for packed kv and packed qkv format in
MultiHeadAttention
(5) Mark some fusion options for SD only
Performance tests show slight improvement in T4. Average latency reduced
0.15 seconds (from 5.25s to 5.10s) for 512x512 in 50 steps for SD 1.5
models. Memory usage drops from 5.1GB to 4.8GB.
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>
