[ROCm][MoE] moe tuning support for rocm (#12049)

Signed-off-by: Divakar Verma <divakar.verma@amd.com>

benchmarks/kernels/benchmark_moe.py

@@ -1,6 +1,7 @@
 import argparse
 import time
 from datetime import datetime
+from itertools import product
 from typing import Any, Dict, List, Tuple, TypedDict
 
 import ray
@@ -11,7 +12,10 @@ from transformers import AutoConfig
 
 from vllm.model_executor.layers.fused_moe.fused_moe import *
 from vllm.platforms import current_platform
-from vllm.utils import FlexibleArgumentParser
+from vllm.utils import FlexibleArgumentParser, is_navi
+
+FP8_DTYPE = torch.float8_e4m3fnuz if current_platform.is_rocm(
+) and not is_navi() else torch.float8_e4m3fn
 
 
 class BenchmarkConfig(TypedDict):
@@ -80,8 +84,8 @@ def benchmark_config(
         a1_scale = torch.randn(1, dtype=torch.float32)
         a2_scale = torch.randn(1, dtype=torch.float32)
 
-        w1 = w1.to(torch.float8_e4m3fn)
+        w1 = w1.to(FP8_DTYPE)
-        w2 = w2.to(torch.float8_e4m3fn)
+        w2 = w2.to(FP8_DTYPE)
 
     input_gating = torch.empty(num_tokens, num_experts, dtype=torch.float32)
 
@@ -141,28 +145,172 @@ def benchmark_config(
     return avg
 
 
-def get_configs_compute_bound() -> List[Dict[str, int]]:
+def get_rocm_tuning_space(use_fp16):
-    # Reduced search space for faster tuning.
+    block_mn_range = [16, 32, 64, 128, 256]
-    # TODO(woosuk): Increase the search space and use a performance model to
+    block_k_range = [16, 32, 64, 128, 256]
-    # prune the search space.
+    if not use_fp16:
+        block_k_range.remove(16)  # BLOCK_K=16 not supported for fp8
+    num_warps_range = [1, 2, 4, 8]
+    group_m_range = [1, 4, 8, 16, 32]
+    num_stage_range = [2]
+    waves_per_eu_range = [0]
+    matrix_instr_nonkdim_range = [16, 32] if use_fp16 else []
+    kpack_range = [1, 2] if use_fp16 else []
+
+    param_ranges = {
+        "BLOCK_SIZE_M": block_mn_range,
+        "BLOCK_SIZE_N": block_mn_range,
+        "BLOCK_SIZE_K": block_k_range,
+        "GROUP_SIZE_M": group_m_range,
+        "num_warps": num_warps_range,
+        "num_stages": num_stage_range,
+        "waves_per_eu": waves_per_eu_range,
+    }
+    if use_fp16:
+        param_ranges["matrix_instr_nonkdim"] = matrix_instr_nonkdim_range
+        param_ranges["kpack"] = kpack_range
+
+    return param_ranges
+
+
+def get_configs_compute_bound(use_fp16) -> List[Dict[str, int]]:
     configs: List[BenchmarkConfig] = []
-    for num_stages in [2, 3, 4, 5]:
+
-        for block_m in [16, 32, 64, 128, 256]:
+    if current_platform.is_rocm():
-            for block_k in [64, 128, 256]:
+        param_ranges = get_rocm_tuning_space(use_fp16)
-                for block_n in [32, 64, 128, 256]:
+    else:
-                    for num_warps in [4, 8]:
+        # Reduced search space for faster tuning.
-                        for group_size in [1, 16, 32, 64]:
+        # TODO(woosuk): Increase the search space and use a performance model to
-                            configs.append({
+        # prune the search space.
-                                "BLOCK_SIZE_M": block_m,
+        block_m_range = [16, 32, 64, 128, 256]
-                                "BLOCK_SIZE_N": block_n,
+        block_n_range = [32, 64, 128, 256]
-                                "BLOCK_SIZE_K": block_k,
+        block_k_range = [64, 128, 256]
-                                "GROUP_SIZE_M": group_size,
+        num_warps_range = [4, 8]
-                                "num_warps": num_warps,
+        group_m_range = [1, 16, 32, 64]
-                                "num_stages": num_stages,
+        num_stage_range = [2, 3, 4, 5]
-                            })
+
+        param_ranges = {
+            "BLOCK_SIZE_M": block_m_range,
+            "BLOCK_SIZE_N": block_n_range,
+            "BLOCK_SIZE_K": block_k_range,
+            "GROUP_SIZE_M": group_m_range,
+            "num_warps": num_warps_range,
+            "num_stages": num_stage_range,
+        }
+
+    keys, values = zip(*param_ranges.items())
+    for config_values in product(*values):
+        config = dict(zip(keys, config_values))
+        configs.append(config)
     return configs
 
 
+def prune_rocm_search_space(num_tokens, shard_intermediate_size, hidden_size,
+                            search_space, is_fp16):
+    N1, K1 = shard_intermediate_size, hidden_size
+    N2, K2 = hidden_size, shard_intermediate_size // 2
+    pruned_space_1 = prune_rocm_configs(num_tokens * 2, N1, K1, search_space,
+                                        is_fp16)
+    pruned_space_2 = prune_rocm_configs(num_tokens * 2, N2, K2, search_space,
+                                        is_fp16)
+    search_space = merge_unique_dicts(pruned_space_1, pruned_space_2)
+    return search_space
+
+
+# The following code is inspired by ROCm/Triton GEMM tuning script:
+# https://github.com/ROCm/triton/blob/triton-mlir/scripts/amd/gemm/tune_gemm.py#L89
+def prune_rocm_configs(M, N, K, configs, is_fp16=True):
+    pruned_configs = []
+    elemBytes_a = 2 if is_fp16 else 1
+    elemBytes_b = 2 if is_fp16 else 1
+
+    mfma = 16 if M < 32 or N < 32 else 32
+
+    # TODO (zhanglx): figure out the boundary between large and small gemms
+    large_gemm = False
+    if M >= 2048 and N >= 2048:
+        large_gemm = True
+
+    for config in configs:
+        BLOCK_SIZE_M = config.get("BLOCK_SIZE_M")
+        BLOCK_SIZE_N = config.get("BLOCK_SIZE_N")
+        BLOCK_SIZE_K = config.get("BLOCK_SIZE_K")
+        num_warps = config.get("num_warps")
+
+        if is_fp16:
+            matrix_instr_nonkdim = config.get("matrix_instr_nonkdim")
+            if matrix_instr_nonkdim > mfma:
+                continue
+        if mfma == 4 and BLOCK_SIZE_K < 64:
+            continue
+        # some layouts could not work properly in case
+        # number elements per thread is less 1
+        if BLOCK_SIZE_M * BLOCK_SIZE_N < 64:
+            continue
+        SPLIT_K = config.get("SPLIT_K", 1)
+        GROUP_M = config.get("GROUP_SIZE_M")
+        if is_fp16:
+            if (matrix_instr_nonkdim > BLOCK_SIZE_M
+                    or matrix_instr_nonkdim > BLOCK_SIZE_N):
+                continue
+            if (matrix_instr_nonkdim >= M
+                    and matrix_instr_nonkdim != BLOCK_SIZE_M):
+                continue
+            if (matrix_instr_nonkdim >= N
+                    and matrix_instr_nonkdim != BLOCK_SIZE_N):
+                continue
+        # Skip BLOCK_SIZE that is too large compare to M/N
+        # unless BLOCK_SIZE is already small enough
+        if M * 2 < BLOCK_SIZE_M and BLOCK_SIZE_M != 16:
+            continue
+        if N * 2 < BLOCK_SIZE_N and BLOCK_SIZE_N != 16:
+            continue
+        # skip large split_k when not necessary
+        if SPLIT_K != 1 and not need_split_k(M, N, K):
+            continue
+        # skip split_k that leads to EVEN_K = false
+        leap = SPLIT_K * BLOCK_SIZE_K
+        modv = K % leap
+        if modv != 0:
+            continue
+        # skip large GROUP_M
+        if GROUP_M * BLOCK_SIZE_M > M and GROUP_M != 1:
+            continue
+        # out of shared memory resource
+        # TODO (zhanglx): This does not consider the LDS usage in the epilogue
+        LDS = (BLOCK_SIZE_K * BLOCK_SIZE_M * elemBytes_a +
+               BLOCK_SIZE_K * BLOCK_SIZE_N * elemBytes_b)
+        if LDS > 65536:
+            continue
+        # Skip small block sizes and num_warps for large gemm
+        # For fp16 and f8, we want to only use BLOCK_SIZE >= 64
+        if large_gemm:
+            if BLOCK_SIZE_M < 64 or BLOCK_SIZE_N < 64:
+                continue
+            if BLOCK_SIZE_K < 64:
+                continue
+            if num_warps < 4:
+                continue
+
+        pruned_configs.append(config)
+
+    return pruned_configs
+
+
+def need_split_k(SIZE_M, SIZE_N, SIZE_K):
+    return (SIZE_M < 64 or SIZE_N < 64) and SIZE_K > 1024
+
+
+def merge_unique_dicts(list1, list2):
+    result = []
+    combined_list = list1.copy()
+    combined_list.extend(list2)
+    for dictionary in combined_list:
+        if dictionary not in result:
+            result.append(dictionary)
+    return result
+
+
 @ray.remote(num_gpus=1)
 class BenchmarkWorker:
 
@@ -170,6 +318,10 @@ class BenchmarkWorker:
         torch.set_default_device("cuda")
         current_platform.seed_everything(seed)
         self.seed = seed
+        # Get the device ID to allocate tensors and kernels
+        # on the respective GPU. This is required for Ray to work
+        # correctly with multi-GPU tuning on the ROCm platform.
+        self.device_id = int(ray.get_gpu_ids()[0])
 
     def benchmark(
         self,
@@ -217,25 +369,33 @@ class BenchmarkWorker:
     ) -> Dict[str, int]:
         best_config = None
         best_time = float("inf")
-        for config in tqdm(search_space):
+        if current_platform.is_rocm():
-            try:
+            is_fp16 = not (use_fp8_w8a8 or use_int8_w8a16)
-                kernel_time = benchmark_config(config,
+            search_space = prune_rocm_search_space(num_tokens,
-                                               num_tokens,
+                                                   shard_intermediate_size,
-                                               num_experts,
+                                                   hidden_size, search_space,
-                                               shard_intermediate_size,
+                                                   is_fp16)
-                                               hidden_size,
-                                               topk,
-                                               dtype,
-                                               use_fp8_w8a8,
-                                               use_int8_w8a16,
-                                               num_iters=10)
-            except triton.runtime.autotuner.OutOfResources:
-                # Some configurations may be invalid and fail to compile.
-                continue
 
-            if kernel_time < best_time:
+        with torch.cuda.device(self.device_id):
-                best_time = kernel_time
+            for config in tqdm(search_space):
-                best_config = config
+                try:
+                    kernel_time = benchmark_config(config,
+                                                   num_tokens,
+                                                   num_experts,
+                                                   shard_intermediate_size,
+                                                   hidden_size,
+                                                   topk,
+                                                   dtype,
+                                                   use_fp8_w8a8,
+                                                   use_int8_w8a16,
+                                                   num_iters=20)
+                except triton.runtime.autotuner.OutOfResources:
+                    # Some configurations may be invalid and fail to compile.
+                    continue
+
+                if kernel_time < best_time:
+                    best_time = kernel_time
+                    best_config = config
         now = datetime.now()
         print(f"{now.ctime()}] Completed tuning for batch_size={num_tokens}")
         assert best_config is not None
@@ -244,12 +404,27 @@ class BenchmarkWorker:
 
 def sort_config(config: BenchmarkConfig) -> BenchmarkConfig:
     return {
-        "BLOCK_SIZE_M": config["BLOCK_SIZE_M"],
+        "BLOCK_SIZE_M":
-        "BLOCK_SIZE_N": config["BLOCK_SIZE_N"],
+        config["BLOCK_SIZE_M"],
-        "BLOCK_SIZE_K": config["BLOCK_SIZE_K"],
+        "BLOCK_SIZE_N":
-        "GROUP_SIZE_M": config["GROUP_SIZE_M"],
+        config["BLOCK_SIZE_N"],
-        "num_warps": config["num_warps"],
+        "BLOCK_SIZE_K":
-        "num_stages": config["num_stages"],
+        config["BLOCK_SIZE_K"],
+        "GROUP_SIZE_M":
+        config["GROUP_SIZE_M"],
+        "num_warps":
+        config["num_warps"],
+        "num_stages":
+        config["num_stages"],
+        **({
+            "waves_per_eu": config["waves_per_eu"]
+        } if "waves_per_eu" in config else {}),
+        **({
+            "matrix_instr_nonkdim": config["matrix_instr_nonkdim"]
+        } if "matrix_instr_nonkdim" in config else {}),
+        **({
+            "kpack": config["kpack"]
+        } if "kpack" in config else {}),
     }
 
 
@@ -294,7 +469,7 @@ def main(args: argparse.Namespace):
         shard_intermediate_size = 2 * intermediate_size // args.tp_size
 
     hidden_size = config.hidden_size
-    dtype = config.torch_dtype
+    dtype = torch.float16 if current_platform.is_rocm() else config.torch_dtype
     use_fp8_w8a8 = args.dtype == "fp8_w8a8"
     use_int8_w8a16 = args.dtype == "int8_w8a16"
 
@@ -322,7 +497,8 @@ def main(args: argparse.Namespace):
         return ray.get(outputs)
 
     if args.tune:
-        search_space = get_configs_compute_bound()
+        is_fp16 = not (use_fp8_w8a8 or use_int8_w8a16)
+        search_space = get_configs_compute_bound(is_fp16)
         print(f"Start tuning over {len(search_space)} configurations...")
 
         start = time.time()