Port last bucket change #346
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
SiLU memory leak in fwd
habana_main rebase v4
* Re-enable FusedRoPE for Gaudi1 * add fallback impl of rope
* formatting fixes * Upstream CR update
…project#8767) Signed-off-by: darthhexx <[email protected]>
This PR removes debug printouts in INC shutdown method and covers the case where application exits before model is initialized properly.
Fix the issue that warmup sometimes doesn't work because the default cache_size_limit is only 8 . --------- Signed-off-by: zehao-intel <[email protected]> Co-authored-by: Andrzej Kotłowski <[email protected]>
Change default values for decode bucket flags
Support loading https://huggingface.co/collections/neuralmagic/fp8-llms-for-vllm-666742ed2b78b7ac8df13127 Skip cuda checks Use scaled_fp8_quant instead of _scaled_mm Fix weights and weight_scale for guudi2 flot8_e4m3fn range. --------- Co-authored-by: Nir David <[email protected]> Co-authored-by: Konrad Zawora <[email protected]>
…#8811) Co-authored-by: simon-mo <[email protected]> Co-authored-by: Chang Su <[email protected]> Co-authored-by: Simon Mo <[email protected]> Co-authored-by: Roger Wang <[email protected]> Co-authored-by: Roger Wang <[email protected]>
### Issue:
torch.compile recompiles after warmup because `tensor 'L['input_ids']'
dispatch key set mismatch. expected DispatchKeySet(HPU, BackendSelect),
actual DispatchKeySet(HPU, BackendSelect, ADInplaceOrView). `
### Detail:
Run script with `TORCH_LOGS="guards"` and get different dispatch key set
info:
- warmup:
```
TENSOR_MATCH: check_tensor(L['input_ids'], Tensor, DispatchKeySet(HPU, BackendSelect), torch.int64, device=0, requires_grad=False, size=[2, 1], stride=[1, 1]) # masked_input = input_ # ome/zyuwen/workspace/vllm/habana_main_g3_v2/vllm/model_executor/layers/vocab_parallel_embedding.py:358 in forward
```
- after warmup:
```
TENSOR_MATCH: check_tensor(L['input_ids'], Tensor, DispatchKeySet(HPU, BackendSelect, ADInplaceOrView), torch.int64, device=0, requires_grad=False, size=[2, 1], stride=[1, 1]) # masked_input = input_ # ome/zyuwen/workspace/vllm/habana_main_g3_v2/vllm/model_executor/layers/vocab_parallel_embedding.py:358 in forward
```
### Solution:
The difference in dispatch key set is caused by the
'torch.inference_mode()' decoration, and here is a simple example:
```python
import torch
import habana_frameworks.torch as htorch
@torch.inference_mode()
def func():
x = torch.rand(3, 3).to("hpu")
print(torch._C._dispatch_key_set(x))
func()
# output: DispatchKeySet(HPU, AutocastHPU)
```
```python
import torch
import habana_frameworks.torch as htorch
def func():
x = torch.rand(3, 3).to("hpu")
print(torch._C._dispatch_key_set(x))
func()
# output: DispatchKeySet(HPU, ADInplaceOrView, AutogradHPU, AutocastHPU)
```
In vllm-fork, the warmup phase is decorated with
`torch.inference_mode()` in
[habana_model_runner.py#L1487-L1488](https://github.com/HabanaAI/vllm-fork/blob/b62fba85ac03326e9f466d8d37e91ae1b14a6511/vllm/worker/habana_model_runner.py#L1487-L1488),
but the after-warmup phase is not.
So in this PR I add the decorator to `prepare_input_tensors` function to
keep the dispatch key set the same.
---
<details>
<!-- inside this <details> section, markdown rendering does not work, so
we use raw html here. -->
<summary><b> PR Checklist (Click to Expand) </b></summary>
<p>Thank you for your contribution to vLLM! Before submitting the pull
request, please ensure the PR meets the following criteria. This helps
vLLM maintain the code quality and improve the efficiency of the review
process.</p>
<h3>PR Title and Classification</h3>
<p>Only specific types of PRs will be reviewed. The PR title is prefixed
appropriately to indicate the type of change. Please use one of the
following:</p>
<ul>
<li><code>[Bugfix]</code> for bug fixes.</li>
<li><code>[CI/Build]</code> for build or continuous integration
improvements.</li>
<li><code>[Doc]</code> for documentation fixes and improvements.</li>
<li><code>[Model]</code> for adding a new model or improving an existing
model. Model name should appear in the title.</li>
<li><code>[Frontend]</code> For changes on the vLLM frontend (e.g.,
OpenAI API server, <code>LLM</code> class, etc.) </li>
<li><code>[Kernel]</code> for changes affecting CUDA kernels or other
compute kernels.</li>
<li><code>[Core]</code> for changes in the core vLLM logic (e.g.,
<code>LLMEngine</code>, <code>AsyncLLMEngine</code>,
<code>Scheduler</code>, etc.)</li>
<li><code>[Hardware][Vendor]</code> for hardware-specific changes.
Vendor name should appear in the prefix (e.g.,
<code>[Hardware][AMD]</code>).</li>
<li><code>[Misc]</code> for PRs that do not fit the above categories.
Please use this sparingly.</li>
</ul>
<p><strong>Note:</strong> If the PR spans more than one category, please
include all relevant prefixes.</p>
<h3>Code Quality</h3>
<p>The PR need to meet the following code quality standards:</p>
<ul>
<li>We adhere to <a
href="https://google.github.io/styleguide/pyguide.html">Google Python
style guide</a> and <a
href="https://google.github.io/styleguide/cppguide.html">Google C++
style guide</a>.</li>
<li>Pass all linter checks. Please use <a
href="https://github.com/vllm-project/vllm/blob/main/format.sh"><code>format.sh</code></a>
to format your code.</li>
<li>The code need to be well-documented to ensure future contributors
can easily understand the code.</li>
<li>Include sufficient tests to ensure the project to stay correct and
robust. This includes both unit tests and integration tests.</li>
<li>Please add documentation to <code>docs/source/</code> if the PR
modifies the user-facing behaviors of vLLM. It helps vLLM user
understand and utilize the new features or changes.</li>
</ul>
<h3>Notes for Large Changes</h3>
<p>Please keep the changes as concise as possible. For major
architectural changes (>500 LOC excluding kernel/data/config/test), we
would expect a GitHub issue (RFC) discussing the technical design and
justification. Otherwise, we will tag it with <code>rfc-required</code>
and might not go through the PR.</p>
<h3>What to Expect for the Reviews</h3>
<p>The goal of the vLLM team is to be a <i>transparent reviewing
machine</i>. We would like to make the review process transparent and
efficient and make sure no contributor feel confused or frustrated.
However, the vLLM team is small, so we need to prioritize some PRs over
others. Here is what you can expect from the review process: </p>
<ul>
<li> After the PR is submitted, the PR will be assigned to a reviewer.
Every reviewer will pick up the PRs based on their expertise and
availability.</li>
<li> After the PR is assigned, the reviewer will provide status update
every 2-3 days. If the PR is not reviewed within 7 days, please feel
free to ping the reviewer or the vLLM team.</li>
<li> After the review, the reviewer will put an <code>
action-required</code> label on the PR if there are changes required.
The contributor should address the comments and ping the reviewer to
re-review the PR.</li>
<li> Please respond to all comments within a reasonable time frame. If a
comment isn't clear or you disagree with a suggestion, feel free to ask
for clarification or discuss the suggestion.
</li>
</ul>
<h3>Thank You</h3>
<p> Finally, thank you for taking the time to read these guidelines and
for your interest in contributing to vLLM. Your contributions make vLLM
a great tool for everyone! </p>
</details>
Signed-off-by: yuwenzho <[email protected]>
#289) Re-implements following PRs for current habana_main: #102 (Removing div_i32 operations from each layer) #115 (removing scatter for reshape&cache in case of prompt) Accuracy (GSM8K on Llama3.1-8B-Instruct): | Tasks |Version| Filter |n-shot| Metric | |Value | |Stderr| |---------------|------:|----------------|-----:|-----------|---|-----:|---|-----:| |gsm8k_cot_llama| 3|flexible-extract| 8|exact_match|↑ |0.8415|± |0.0101| | | |strict-match | 8|exact_match|↑ |0.8400|± |0.0101| I've benchmarked this change on Llama3.1-8B-Instruct and on average, +2.50% throughput gain (+558.14 tok/s, ~21594 tok/s -> ~22152 tok/s) can be observed across all prefill buckets on G2, with up to +4.40% (+956.79 tok/s, ~25031 -> ~25988 tok/s) throughput increase in compute-bound scenarios.
…th LoRA (#339) This PR has following fixes, - Increase size of indices tensors used to maintain multi-lora state information from max_num_batched_tokens to 3*max_num_batched_tokens. This increase is done to provide buffer for padding done in batch & sequence dimensions. - Move logic to remove padding from lora_logits from execute_model() back to Class LogitsProcessorWithLoRA, this is done to fix race condition caused by updating multi-lora state information directly. FIX #237
you know the drill
FILL IN THE PR DESCRIPTION HERE This PR refer to [vllm-project#7049](vllm-project#7049) to implement Asynchronous Output Processor on HPU. It is open by default, to disable it, please pass the `--disable_async_output_proc` flag. From my local test on latest habana_main branch(commit 29fb5ed), the throughput improves from 3847 TPS to 4011 TPS. **BEFORE SUBMITTING, PLEASE READ THE CHECKLIST BELOW AND FILL IN THE DESCRIPTION ABOVE** --- <details> <!-- inside this <details> section, markdown rendering does not work, so we use raw html here. --> <summary><b> PR Checklist (Click to Expand) </b></summary> <p>Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.</p> <h3>PR Title and Classification</h3> <p>Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:</p> <ul> <li><code>[Bugfix]</code> for bug fixes.</li> <li><code>[CI/Build]</code> for build or continuous integration improvements.</li> <li><code>[Doc]</code> for documentation fixes and improvements.</li> <li><code>[Model]</code> for adding a new model or improving an existing model. Model name should appear in the title.</li> <li><code>[Frontend]</code> For changes on the vLLM frontend (e.g., OpenAI API server, <code>LLM</code> class, etc.) </li> <li><code>[Kernel]</code> for changes affecting CUDA kernels or other compute kernels.</li> <li><code>[Core]</code> for changes in the core vLLM logic (e.g., <code>LLMEngine</code>, <code>AsyncLLMEngine</code>, <code>Scheduler</code>, etc.)</li> <li><code>[Hardware][Vendor]</code> for hardware-specific changes. Vendor name should appear in the prefix (e.g., <code>[Hardware][AMD]</code>).</li> <li><code>[Misc]</code> for PRs that do not fit the above categories. Please use this sparingly.</li> </ul> <p><strong>Note:</strong> If the PR spans more than one category, please include all relevant prefixes.</p> <h3>Code Quality</h3> <p>The PR need to meet the following code quality standards:</p> <ul> <li>We adhere to <a href="https://google.github.io/styleguide/pyguide.html">Google Python style guide</a> and <a href="https://google.github.io/styleguide/cppguide.html">Google C++ style guide</a>.</li> <li>Pass all linter checks. Please use <a href="https://github.com/vllm-project/vllm/blob/main/format.sh"><code>format.sh</code></a> to format your code.</li> <li>The code need to be well-documented to ensure future contributors can easily understand the code.</li> <li>Include sufficient tests to ensure the project to stay correct and robust. This includes both unit tests and integration tests.</li> <li>Please add documentation to <code>docs/source/</code> if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.</li> </ul> <h3>Adding or changing kernels</h3> <p>Each custom kernel needs a schema and one or more implementations to be registered with PyTorch.</p> <ul> <li>Make sure custom ops are registered following PyTorch guidelines: <a href="https://pytorch.org/tutorials/advanced/cpp_custom_ops.html#cpp-custom-ops-tutorial">Custom C++ and CUDA Operators</a> and <a href="https://docs.google.com/document/d/1_W62p8WJOQQUzPsJYa7s701JXt0qf2OfLub2sbkHOaU">The Custom Operators Manual</a></li> <li>Custom operations that return <code>Tensors</code> require meta-functions. Meta-functions should be implemented and registered in python so that dynamic dims can be handled automatically. See above documents for a description of meta-functions.</li> <li>Use <a href="https://pytorch.org/docs/stable/library.html#torch.library.opcheck"><code>torch.libary.opcheck()</code></a> to test the function registration and meta-function for any registered ops. See <code>tests/kernels</code> for examples.</li> <li>When changing the C++ signature of an existing op, the schema must be updated to reflect the changes.</li> <li>If a new custom type is needed, see the following document: <a href="https://docs.google.com/document/d/18fBMPuOJ0fY5ZQ6YyrHUppw9FA332CpNtgB6SOIgyuA">Custom Class Support in PT2</a>. </ul> <h3>Notes for Large Changes</h3> <p>Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with <code>rfc-required</code> and might not go through the PR.</p> <h3>What to Expect for the Reviews</h3> <p>The goal of the vLLM team is to be a <i>transparent reviewing machine</i>. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process: </p> <ul> <li> After the PR is submitted, the PR will be assigned to a reviewer. Every reviewer will pick up the PRs based on their expertise and availability.</li> <li> After the PR is assigned, the reviewer will provide status update every 2-3 days. If the PR is not reviewed within 7 days, please feel free to ping the reviewer or the vLLM team.</li> <li> After the review, the reviewer will put an <code> action-required</code> label on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.</li> <li> Please respond to all comments within a reasonable time frame. If a comment isn't clear or you disagree with a suggestion, feel free to ask for clarification or discuss the suggestion. </li> </ul> <h3>Thank You</h3> <p> Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone! </p> </details>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Port last bucket from v1.18.0