GitLab

Co-authored-by: Michael Goin <michael@neuralmagic.com>

Co-authored-by: Lily Liu <lilyliupku@gmail.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>

Co-authored-by: Danny Guinther <dguinther@neuralmagic.com>

Co-authored-by: Yun Ding <yunding@nvidia.com>
Co-authored-by: Roger Wang <ywang@roblox.com>

Provide an initial support to FP8 computation. This PR is inspired by HuggingFace TGI: huggingface/text-generation-inference#1726

This feature can be enabled with --quantization fp8 or -q fp8 when launching an engine.

Algorithm:
We still load a model checkpoint in FP16/BF16. After the weights are loaded, Fp8LinearMethod calculates the per-tensor scaling factor of weights and quantizes the weights accordingly. The scaling factor will then be stored for future use. Meanwhile, the per-tensor scaling factor for activations is calculated in every forward pass.

Initial Results:
Currently tested Mistral-7B on 1xH100. With prompt length ~5 and decoding length 128:

BF16: 1.47s
FP8: 1.66s
I'll try to use larger models and try to find more performance bottleneck. Meanwhile, you're welcome to try this code.

Co-authored-by: Alexandre Payot <alexandrep@graphcore.ai>

Co-authored-by: Simon Mo <simon.mo@hey.com>

Co-authored-by: Dylan Hawk <dylanwawk@gmail.com>

Co-authored-by: A-Mahla <>

Co-authored-by: Sahil Suneja <sahilsuneja@gmail.com>

Co-authored-by: Dylan Hawk <dylanwawk@gmail.com>

Signed-off-by: Travis Johnson <tsjohnso@us.ibm.com>
Co-authored-by: Nick Hill <nickhill@us.ibm.com>