From 0595bf3e9a15807d038e41f09c0f7e1c595d2417 Mon Sep 17 00:00:00 2001
From: dtlzhuangz <139844877+dtlzhuangz@users.noreply.github.com>
Date: Wed, 31 Jan 2024 19:05:49 +0800
Subject: [PATCH] feat: eetq gemv optimization when batch_size <= 4 (#1502)

# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Add TensorRT-LLM weight-only GEMV kernel support. We extract GEMV kernel
from
[TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM/tree/main/cpp/tensorrt_llm/kernels/weightOnlyBatchedGemv)
to accelerate the decode speed of EETQ when batch_size is smaller or
equal to 4.

- Features

1. There is almost no loss of quantization accuracy.
2. The speed of decoding is 13% - 27% faster than original EETQ which
utilizes GEMM kernel.

- Test
Below is our test on 3090. Environment: torch=2.0.1, cuda=11.8, nvidia
driver: 525.78.01
prompt=1024, max_new_tokens=50

![image](https://github.com/huggingface/text-generation-inference/assets/139844877/98e63b23-23cd-452f-91bd-55ccdc9b7021)


![image](https://github.com/huggingface/text-generation-inference/assets/139844877/5c3132ff-fc1c-4b20-a83f-59b3d5f586b7)



## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Did you read the [contributor
guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? Please add a link
      to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes?
Here are the
[documentation
guidelines](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?


## Who can review?

Anyone in the community is free to review the PR once the tests have
passed. Feel free to tag
members/contributors who may be interested in your PR.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->
---
 server/Makefile-eetq | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/server/Makefile-eetq b/server/Makefile-eetq
index 5e8e9830..8c060987 100644
--- a/server/Makefile-eetq
+++ b/server/Makefile-eetq
@@ -1,4 +1,4 @@
-eetq_commit := 323827dd471458a84e9c840f614e4592b157a4b1
+eetq_commit := 71adb5e191bb8290069a580abff0355d7b2dd5c9
 
 eetq:
     # Clone eetq
@@ -6,7 +6,7 @@ eetq:
 	git clone https://github.com/NetEase-FuXi/EETQ.git eetq
 
 build-eetq: eetq
-	cd eetq && git fetch && git checkout $(eetq_commit)
+	cd eetq && git fetch && git checkout $(eetq_commit) && git submodule update --init --recursive
 	cd eetq && python setup.py build
 
 install-eetq: build-eetq