support w8a8 fp8 kernel with CUTLASS (#3047)

Co-authored-by: yych0745 <1398089567@qq.com>

sgl-kernel/benchmark/bench_fp8_gemm.py

+import argparse
+import copy
+import itertools
+
+import torch
+import triton
+from sgl_kernel import fp8_scaled_mm as sgl_scaled_mm
+from vllm._custom_ops import cutlass_scaled_mm as vllm_scaled_mm
+from vllm._custom_ops import scaled_fp8_quant as vllm_scaled_fp8_quant
+
+# Weight Shapes are in the format
+# ([K, N], TP_SPLIT_DIM)
+# Example:
+#  A shape of ([14336, 4096], 0) indicates the following GEMM shape,
+#   - TP1 : K = 14336, N = 4096
+#   - TP2 : K = 7168, N = 4096
+#  A shape of ([4096, 6144], 1) indicates the following GEMM shape,
+#   - TP1 : K = 4096, N = 6144
+#   - TP4 : K = 4096, N = 1536
+
+# TP1 shapes
+WEIGHT_SHAPES = {
+    "meta-llama/Llama-3.1-8B-Instruct": [
+        ([4096, 6144], 1),
+        ([4096, 4096], 0),
+        ([4096, 28672], 1),
+        ([14336, 4096], 0),
+    ],
+    "meta-llama/Llama-3.3-70B-Instruct": [
+        ([8192, 10240], 1),
+        ([8192, 8192], 0),
+        ([8192, 57344], 1),
+        ([28672, 8192], 0),
+    ],
+    "mistralai/Mistral-Large-Instruct-2407": [
+        ([12288, 14336], 1),
+        ([12288, 12288], 0),
+        ([12288, 57344], 1),
+        ([28672, 12288], 0),
+    ],
+    "Qwen/Qwen2.5-7B-Instruct": [
+        ([3584, 4608], 1),
+        ([3584, 3584], 0),
+        ([3584, 37888], 1),
+        ([18944, 3584], 0),
+    ],
+    "Qwen/Qwen2.5-32B-Instruct": [
+        ([5120, 7168], 1),
+        ([5120, 5120], 0),
+        ([5120, 55296], 1),
+        ([27648, 5120], 0),
+    ],
+    "Qwen/Qwen2.5-72B-Instruct": [
+        ([8192, 10240], 1),
+        ([8192, 8192], 0),
+        ([8192, 59136], 1),
+        ([29568, 8192], 0),
+    ],
+    "deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct": [
+        ([2048, 3072], 1),
+        ([2048, 4096], 1),
+        ([2048, 2048], 0),
+        ([2048, 576], 0),
+        ([2048, 21888], 1),
+        ([10944, 2048], 0),
+        ([2048, 2816], 1),
+        ([1408, 2048], 0),
+    ],
+}
+
+
+@triton.testing.perf_report(
+    triton.testing.Benchmark(
+        x_names=["batch_size"],
+        x_vals=[1, 16, 64, 128, 256, 512, 1024, 2048],
+        x_log=False,
+        line_arg="provider",
+        line_vals=[
+            "vllm-fp8-fp16",
+            "vllm-fp8-bf16",
+            "sglang-fp8-fp16",
+            "sglang-fp8-bf16",
+        ],
+        line_names=[
+            "vllm-fp8-fp16",
+            "vllm-fp8-bf16",
+            "sglang-fp8-fp16",
+            "sglang-fp8-bf16",
+        ],
+        styles=[("green", "-"), ("green", "--"), ("blue", "-"), ("blue", "--")],
+        ylabel="GB/s",
+        plot_name="fp8 scaled matmul",
+        args={},
+    )
+)
+def benchmark(batch_size, provider, N, K):
+    # M, N, K = batch_size, 4096, 8192
+    M = batch_size
+    a = torch.ones((M, K), device="cuda") * 5.0
+    b = torch.ones((N, K), device="cuda") * 5.0
+    scale_a = torch.randn((M,), device="cuda", dtype=torch.float32)
+    scale_b = torch.randn((N,), device="cuda", dtype=torch.float32)
+    a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a)
+    b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+    b_fp8 = b_fp8.t()
+    quantiles = [0.5, 0.2, 0.8]
+
+    dtype = torch.float16 if "fp16" in provider else torch.bfloat16
+
+    if "vllm-fp8" in provider:
+        ms, min_ms, max_ms = triton.testing.do_bench(
+            lambda: vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype),
+            quantiles=quantiles,
+        )
+    elif "sglang-fp8" in provider:
+        ms, min_ms, max_ms = triton.testing.do_bench(
+            lambda: sgl_scaled_mm(
+                a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype, bias=None
+            ),
+            quantiles=quantiles,
+        )
+
+    gbps = lambda ms: (2 * M * N * K + M * N) * a.element_size() * 1e-9 / (ms * 1e-3)
+    return gbps(ms), gbps(max_ms), gbps(min_ms)
+
+
+def prepare_shapes(args):
+    KN_model_names = []
+    models_tps = list(itertools.product(args.models, args.tp_sizes))
+    for model, tp_size in models_tps:
+        assert model in WEIGHT_SHAPES
+        for KN, tp_split_dim in copy.deepcopy(WEIGHT_SHAPES[model]):
+            KN[tp_split_dim] = KN[tp_split_dim] // tp_size
+            KN.append(model)
+            KN_model_names.append(KN)
+    return KN_model_names
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--models",
+        nargs="+",
+        type=str,
+        default=["meta-llama/Llama-3.1-8B-Instruct"],
+        help="List of models to benchmark",
+    )
+    parser.add_argument(
+        "--tp-sizes",
+        nargs="+",
+        type=int,
+        default=[1],
+        help="List of tensor parallel sizes",
+    )
+    args = parser.parse_args()
+
+    KN_model_names = prepare_shapes(args)
+    for K, N, model_name in KN_model_names:
+        print(f"{model_name} N={N} K={K}: ")
+        benchmark.run(
+            print_data=True, show_plots=True, save_path="bench_fp8_res", N=N, K=K
+        )
+
+    print("Benchmark finished!")

sgl-kernel/setup.py

@@ -56,6 +56,7 @@ include_dirs = [
     turbomind.resolve(),
     turbomind.resolve() / "src",
 ]
+
 nvcc_flags = [
     "-DNDEBUG",
     f"-DOPERATOR_NAMESPACE={operator_namespace}",
@@ -82,6 +83,7 @@ sources = [
     "src/sgl-kernel/csrc/trt_reduce_kernel.cu",
     "src/sgl-kernel/csrc/moe_align_kernel.cu",
     "src/sgl-kernel/csrc/int8_gemm_kernel.cu",
+    "src/sgl-kernel/csrc/fp8_gemm_kernel.cu",
     "src/sgl-kernel/csrc/lightning_attention_decode_kernel.cu",
     "src/sgl-kernel/csrc/rotary_embedding.cu",
     "3rdparty/flashinfer/csrc/activation.cu",

sgl-kernel/src/sgl-kernel/__init__.py

@@ -2,6 +2,7 @@ from sgl_kernel.ops import (
     bmm_fp8,
     custom_dispose,
     custom_reduce,
+    fp8_scaled_mm,
     fused_add_rmsnorm,
     gelu_and_mul,
     gelu_tanh_and_mul,
@@ -27,6 +28,7 @@ __all__ = [
     "bmm_fp8",
     "custom_dispose",
     "custom_reduce",
+    "fp8_scaled_mm",
     "fused_add_rmsnorm",
     "gelu_and_mul",
     "gelu_tanh_and_mul",

sgl-kernel/src/sgl-kernel/include/sgl_kernels_ops.h

@@ -40,6 +40,11 @@ torch::Tensor int8_scaled_mm(const torch::Tensor& mat_a, const torch::Tensor& ma
                              const torch::Tensor& scales_b, const torch::Dtype& out_dtype,
                              const c10::optional<torch::Tensor>& bias);
 
+// fp8_scaled_mm
+torch::Tensor fp8_scaled_mm(const torch::Tensor& mat_a, const torch::Tensor& mat_b, const torch::Tensor& scales_a,
+                            const torch::Tensor& scales_b, const torch::Dtype& out_dtype,
+                            const c10::optional<torch::Tensor>& bias);
+
 // lightning_attention_decode
 void lightning_attention_decode(const torch::Tensor& q, const torch::Tensor& k, const torch::Tensor& v,
                                 const torch::Tensor& past_kv, const torch::Tensor& slope, torch::Tensor output,

sgl-kernel/src/sgl-kernel/ops/__init__.py

@@ -71,6 +71,17 @@ def int8_scaled_mm(mat_a, mat_b, scales_a, scales_b, out_dtype, bias=None):
     )
 
 
+def fp8_scaled_mm(mat_a, mat_b, scales_a, scales_b, out_dtype, bias=None):
+    return torch.ops.sgl_kernels.fp8_scaled_mm(
+        mat_a,
+        mat_b,
+        scales_a,
+        scales_b,
+        out_dtype,
+        bias,
+    )
+
+
 def lightning_attention_decode(q, k, v, past_kv, slope, output, new_kv):
     torch.ops.sgl_kernels.lightning_attention_decode(
         q, k, v, past_kv, slope, output, new_kv

sgl-kernel/src/sgl-kernel/torch_extension.cc

@@ -34,6 +34,12 @@ TORCH_LIBRARY_EXPAND(sgl_kernels, m) {
       "bias) -> Tensor");
   m.impl("int8_scaled_mm", torch::kCUDA, &int8_scaled_mm);
 
+  // fp8_scaled_mm
+  m.def(
+      "fp8_scaled_mm(Tensor mat_a, Tensor mat_b, Tensor scales_a, Tensor scales_b, ScalarType out_dtype, Tensor? "
+      "bias) -> Tensor");
+  m.impl("fp8_scaled_mm", torch::kCUDA, &fp8_scaled_mm);
+
   // lightning_attention_decode
   m.def(
       "lightning_attention_decode(Tensor q, Tensor k, Tensor v, Tensor past_kv, Tensor slope, Tensor! output, Tensor! "

sgl-kernel/tests/test_fp8_gemm.py

+import unittest
+
+import torch
+from sgl_kernel import fp8_scaled_mm
+
+
+def torch_scaled_mm(a, b, scale_a, scale_b, out_dtype, bias):
+    o = torch.matmul(a.to(torch.float32), b.to(torch.float32))
+
+    o = o.to(torch.float32)
+    temp1 = o * scale_a.view(-1, 1)
+    temp2 = temp1 * scale_b.view(1, -1)
+    final = temp2.to(out_dtype)
+    if bias is not None:
+        final = final + bias.view(1, -1)
+
+    return final
+
+
+class TestFp8Gemm(unittest.TestCase):
+    def _test_accuracy_once(self, M, N, K, with_bias, out_dtype, device):
+        fp8_info = torch.finfo(torch.float8_e4m3fn)
+        fp8_max, fp8_min = fp8_info.max, fp8_info.min
+
+        a_fp32 = (
+            (torch.rand(M, K, dtype=torch.float32, device=device) - 0.5) * 2 * fp8_max
+        )
+        a_fp8 = a_fp32.clamp(min=fp8_min, max=fp8_max).to(torch.float8_e4m3fn)
+
+        b_fp32 = (
+            (torch.rand(N, K, dtype=torch.float32, device=device) - 0.5) * 2 * fp8_max
+        )
+        b_fp8 = b_fp32.clamp(min=fp8_min, max=fp8_max).to(torch.float8_e4m3fn)
+
+        scale_a = torch.randn((M,), device=device, dtype=torch.float32) * 0.001
+        scale_b = torch.randn((N,), device=device, dtype=torch.float32) * 0.001
+        if with_bias:
+            bias = torch.randn((N,), device=device, dtype=out_dtype)
+        else:
+            bias = None
+        o1 = torch.empty((M, N), device=device, dtype=torch.bfloat16)
+        b_fp8 = b_fp8.t()
+        o = torch_scaled_mm(a_fp8, b_fp8, scale_a, scale_b, out_dtype, bias)
+        o1 = fp8_scaled_mm(a_fp8, b_fp8, scale_a, scale_b, out_dtype, bias)
+        rtol = 0.02
+        atol = 1
+        torch.testing.assert_close(o, o1, rtol=rtol, atol=atol)
+        print(f"M={M}, N={N}, K={K}, with_bias={with_bias}, out_dtype={out_dtype}: OK")
+
+    def test_accuracy(self):
+        Ms = [1, 128, 512, 1024, 4096]
+        Ns = [16, 128, 512, 1024, 4096]
+        Ks = [512, 1024, 4096, 8192, 16384]
+        bias_opts = [True, False]
+        out_dtypes = [torch.bfloat16, torch.float16]
+        for M in Ms:
+            for N in Ns:
+                for K in Ks:
+                    for with_bias in bias_opts:
+                        for out_dtype in out_dtypes:
+                            self._test_accuracy_once(
+                                M, N, K, with_bias, out_dtype, "cuda"
+                            )
+
+
+if __name__ == "__main__":
+    unittest.main()