[Kernel][FP8] Initial support with dynamic per-tensor scaling (#4118)

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.

vllm/model_executor/model_loader/loader.py

@@ -228,6 +228,10 @@ class DefaultModelLoader(BaseModelLoader):
                                                model,
                                                "fall_back_to_pt_during_load",
                                                True)), )
+            for _, module in model.named_modules():
+                linear_method = getattr(module, "linear_method", None)
+                if linear_method is not None:
+                    linear_method.process_weights_after_loading(module)
         return model.eval()