benchmark_machete.py 21.4 KB
Newer Older
1
2
3
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project

laibao's avatar
laibao committed
4
5
6
7
import argparse
import copy
import itertools
import math
8
import os
laibao's avatar
laibao committed
9
10
import pickle as pkl
import time
11
12
from collections.abc import Iterable
from dataclasses import dataclass
laibao's avatar
laibao committed
13
from itertools import product
14
from typing import Callable, Optional
laibao's avatar
laibao committed
15
16
17
18
19
20
21
22
23

import pandas as pd
import torch
import torch.utils.benchmark as TBenchmark
from torch.utils.benchmark import Measurement as TMeasurement
from weight_shapes import WEIGHT_SHAPES

from vllm import _custom_ops as ops
from vllm.model_executor.layers.quantization.utils.marlin_utils import (
24
25
26
27
28
    GPTQ_MARLIN_MAX_PARALLEL,
    GPTQ_MARLIN_MIN_THREAD_N,
    marlin_permute_scales,
    marlin_zero_points,
)
laibao's avatar
laibao committed
29
from vllm.model_executor.layers.quantization.utils.marlin_utils_test import (
30
31
    MarlinWorkspace,
)
laibao's avatar
laibao committed
32
from vllm.model_executor.layers.quantization.utils.quant_utils import (
33
34
35
    pack_rows,
    quantize_weights,
)
laibao's avatar
laibao committed
36
37
38
39
40
41
42
from vllm.scalar_type import ScalarType, scalar_types
from vllm.utils import FlexibleArgumentParser

DEFAULT_MODELS = ["meta-llama/Llama-3-8b", "meta-llama/Llama-2-70b-hf"]
DEFAULT_BATCH_SIZES = [1, 16, 32, 64, 128, 256, 512, 1024]
DEFAULT_TP_SIZES = [1]

43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
NVTX_PROFILE = os.environ.get("NVTX_PROFILE", False)

if NVTX_PROFILE:
    import nvtx


def terse_type_name(dt):
    return {
        torch.bfloat16: "bf16",
        torch.float16: "fp16",
        torch.int8: "int8",
        torch.float8_e4m3fn: "fp8",
        torch.float: "float",
        torch.int: "int",
    }[dt]


@dataclass
class BenchmarkTensors:
    w_ref: torch.Tensor
    a: torch.Tensor

    w_q: torch.Tensor
    group_size: Optional[int]
    wtype: ScalarType
    w_g_s: torch.Tensor
    w_g_zp: Optional[torch.Tensor]
    w_ch_s: Optional[torch.Tensor]
    w_tok_s: Optional[torch.Tensor]


@dataclass
class TypeConfig:
    act_type: torch.dtype
    weight_type: ScalarType
    output_type: Optional[torch.dtype]
    group_scale_type: Optional[torch.dtype]
    group_zero_type: Optional[torch.dtype]
    channel_scale_type: Optional[torch.dtype]
    token_scale_type: Optional[torch.dtype]


def rand_data(shape, dtype=torch.float16, scale=1):
    if dtype.is_floating_point:
        return (scale * torch.rand(shape, device="cuda") - 0.3).to(dtype)
    else:
        return torch.randint(-15, 15, shape, dtype=dtype, device="cuda")


def quantize_and_pack(
    atype: torch.dtype,
    w: torch.Tensor,
    wtype: ScalarType,
    stype: Optional[torch.dtype],
    group_size: Optional[int],
    zero_points: bool = False,
):
    assert wtype.is_integer(), "TODO: support floating point weights"

    w_ref, w_q, w_s, w_zp = quantize_weights(
        w,
        wtype,
        group_size=group_size,
        zero_points=zero_points,
        # to match how the kernel applies zps
        ref_zero_points_after_scales=True,
    )
laibao's avatar
laibao committed
110
111

    w_q = pack_rows(w_q, wtype.size_bits, *w_q.shape)
112
    return w_ref, w_q, w_s, w_zp
laibao's avatar
laibao committed
113
114


115
116
117
118
def create_bench_tensors(
    shape: tuple[int, int, int], types: TypeConfig, group_size: Optional[int]
) -> list[BenchmarkTensors]:
    m, n, k = shape
laibao's avatar
laibao committed
119
120
121
122

    # we want to make sure that weights don't fit into L2 cache between runs so
    #  we construct enough weights to exceed L2 cache, which is 50mb on a H100
    #  so we target total weight size > 2*50mb
123
124
125
    num_weights = math.ceil(
        2 * 50 * 1024**2 * 8 / (k * n * types.weight_type.size_bits)
    )
laibao's avatar
laibao committed
126

127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
    a = rand_data((m, k), types.act_type, scale=5)

    benchmark_tensors: list[BenchmarkTensors] = []
    for _ in range(num_weights):
        w = rand_data((k, n), types.act_type, scale=5)

        if types.group_scale_type is not None:
            w = w.to(types.group_scale_type)
        if w.dtype.itemsize == 1:
            w = w.to(torch.float16)

        w_ref, w_q_packed, w_s, w_zp = quantize_and_pack(
            a.dtype,
            w,
            types.weight_type,
            types.group_scale_type,
            group_size,
            types.group_zero_type is not None,
        )

        if not a.dtype.is_floating_point:
            aiinfo = torch.iinfo(a.dtype)
            w_ref = w_ref.round().clamp(aiinfo.min, aiinfo.max)

        w_ref = w_ref.to(torch.float32)

        w_ch_s = (
            None
            if types.channel_scale_type is None
            else rand_data((n,), types.channel_scale_type)
        )
        w_tok_s = (
            None
            if types.token_scale_type is None
            else rand_data((m,), types.token_scale_type)
        )

        benchmark_tensors.append(
            BenchmarkTensors(
                w_ref=w_ref,
                a=a,
                w_q=w_q_packed,
                wtype=types.weight_type,
                w_g_s=w_s,
                w_g_zp=w_zp,
                group_size=group_size,
                w_ch_s=w_ch_s,
                w_tok_s=w_tok_s,
            )
        )

    return benchmark_tensors


def torch_matmul_f16_create_bench_fn(bt: BenchmarkTensors) -> Callable:
    a = bt.a
    w = bt.w_ref.to(bt.a.dtype)  # use float reference tensor
    if a.dtype not in [torch.float16, torch.bfloat16]:
        a = a.to(torch.float16)
        w = w.to(torch.float16)
    return lambda: torch.matmul(a, w)


def cutlass_scaled_mm_create_bench_fn(bt: BenchmarkTensors) -> Callable:
    if bt.w_ch_s is not None and bt.w_tok_s is not None:
        scale_a = bt.w_tok_s.to(torch.float32)
        scale_b = bt.w_ch_s.to(torch.float32)
    else:
        scale_a = torch.tensor(1.0, dtype=torch.float32, device=bt.a.device)
        scale_b = torch.tensor(1.0, dtype=torch.float32, device=bt.a.device)
    w_col_major = bt.w_ref.to(bt.a.dtype).t().contiguous().t()
    return lambda: ops.cutlass_scaled_mm(
        bt.a, w_col_major, scale_a, scale_b, out_dtype=torch.float16
    )
laibao's avatar
laibao committed
201
202


203
204
def marlin_create_bench_fn(bt: BenchmarkTensors) -> Callable:
    device = bt.a.device
laibao's avatar
laibao committed
205

206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
    workspace = MarlinWorkspace(
        bt.w_ref.shape[1], GPTQ_MARLIN_MIN_THREAD_N, GPTQ_MARLIN_MAX_PARALLEL
    )

    if bt.w_g_zp is None:
        w_zp = torch.empty(0, dtype=torch.int, device=device)
    else:
        w_zp = marlin_zero_points(
            bt.w_g_zp, bt.w_ref.shape[0], bt.w_ref.shape[1], bt.wtype.size_bits
        )

    if bt.group_size is None:
        w_s = torch.tensor([], device="cuda", dtype=torch.half)
    else:
        w_s = marlin_permute_scales(
            bt.w_g_s, bt.w_ref.shape[0], bt.w_ref.shape[1], bt.group_size
        )

    sort_indices = torch.empty(0, dtype=torch.int, device=device)
    g_idx = torch.empty(0, dtype=torch.int, device=device)
    w_q = ops.gptq_marlin_repack(
        bt.w_q, sort_indices, bt.w_ref.shape[0], bt.w_ref.shape[1], bt.wtype.size_bits
    )

    if bt.a.dtype.is_floating_point:
        assert bt.w_ch_s is None
        assert bt.w_tok_s is None
        assert bt.group_size is not None

        fn = lambda: ops.gptq_marlin_gemm(
            a=bt.a,
            c=None,
            b_q_weight=w_q,
            b_scales=w_s,
            global_scale=None,
            b_zeros=w_zp,
            g_idx=g_idx,
            perm=sort_indices,
            workspace=workspace.scratch,
            b_q_type=bt.wtype,
            size_m=bt.a.shape[0],
            size_n=bt.w_ref.shape[1],
            size_k=bt.w_ref.shape[0],
            is_k_full=True,
            is_zp_float=False,
        )
    else:
        assert bt.a.dtype == torch.int8
        assert bt.wtype == scalar_types.uint4b8

        if bt.w_ch_s is not None:
            s_ch = bt.w_ch_s.to(torch.float32)
        else:
            s_ch = torch.ones(bt.w_ref.shape[1], dtype=torch.float32, device=device)

        if bt.w_tok_s is not None:
            s_tok = bt.w_tok_s.to(torch.float32)
        else:
            s_tok = torch.ones(bt.a.shape[0], dtype=torch.float32, device=device)

        fn = lambda: ops.marlin_qqq_gemm(
            a=bt.a,
            b_q_weight=w_q,
            s_group=w_s,
            s_tok=s_tok,
            s_ch=s_ch,
            workspace=workspace.scratch,
            size_m=bt.a.shape[0],
            size_n=bt.w_ref.shape[1],
            size_k=bt.w_ref.shape[0],
        )

    return fn


def machete_create_bench_fn(
    bt: BenchmarkTensors, out_type=torch.dtype, schedule=None
) -> Callable:
    w_q = bt.w_q.t().contiguous().t()  # make col major
    w_q = ops.machete_prepack_B(
        w_q, bt.a.dtype, bt.wtype, None if bt.w_g_s is None else bt.w_g_s.dtype
    )

    w_g_zp = bt.w_g_zp
    if w_g_zp is not None:
        w_g_zp = -1 * bt.w_g_s * (w_g_zp.to(bt.w_g_s.dtype))

    return lambda: ops.machete_mm(
        a=bt.a,
        b_q=w_q,
        b_type=bt.wtype,
        b_group_scales=bt.w_g_s,
        b_group_zeros=w_g_zp,
        b_group_size=bt.group_size,
        b_channel_scales=bt.w_ch_s,
        a_token_scales=bt.w_tok_s,
        out_type=out_type,
        schedule=schedule,
    )
laibao's avatar
laibao committed
305
306


307
308
# impl

laibao's avatar
laibao committed
309
# bench
310
311
312
313
314
315
316
317
318
319


def bench_fns(label: str, sub_label: str, description: str, fns: list[Callable]):
    min_run_time = 1 if not NVTX_PROFILE else 0.1
    res = TBenchmark.Timer(
        stmt="""
        for fn in fns:
            fn()
        """,
        globals={"fns": fns},
laibao's avatar
laibao committed
320
321
322
323
324
        label=label,
        sub_label=sub_label,
        description=description,
    ).blocked_autorange(min_run_time=min_run_time)

325
326
327
328
329
330
    if NVTX_PROFILE:
        with (
            nvtx.annotate("mm-bench"),
            nvtx.annotate(f"{label}|{sub_label}|{description}"),
        ):
            fns[0]()
laibao's avatar
laibao committed
331

332
    return res
laibao's avatar
laibao committed
333
334
335
336
337
338


_SWEEP_SCHEDULES_RESULTS: Optional[pd.DataFrame] = None
_SWEEP_SCHEDULES_RESULTS_CSV: Optional[str] = None


339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
def bench(
    types: TypeConfig,
    group_size: int,
    m: int,
    k: int,
    n: int,
    label: str,
    sub_label: str,
    sweep_schedules: bool = True,
) -> list[TMeasurement]:
    benchmark_tensors = create_bench_tensors((m, n, k), types, group_size)
    sub_label += f", L={len(benchmark_tensors)}"

    name_type_string = f"W{types.weight_type}" + f"-A{terse_type_name(types.act_type)}"
    if types.group_scale_type is not None:
        name_type_string += f"-GS{terse_type_name(types.group_scale_type)}"
    if types.group_zero_type is not None:
        name_type_string += f"-GZ{terse_type_name(types.group_zero_type)}"
    if group_size is not None:
        name_type_string += f"-G{group_size}"
    if types.channel_scale_type is not None:
        name_type_string += f"-CS{terse_type_name(types.channel_scale_type)}"
    if types.token_scale_type is not None:
        name_type_string += f"-TS{terse_type_name(types.token_scale_type)}"
laibao's avatar
laibao committed
363
364
365
366

    timers = []
    # pytorch impl
    timers.append(
367
368
369
370
371
372
373
        bench_fns(
            label,
            sub_label,
            "torch.matmul (fp16)",
            [torch_matmul_f16_create_bench_fn(bt) for bt in benchmark_tensors],
        )
    )
laibao's avatar
laibao committed
374

375
    if types.act_type == torch.int8 or types.act_type == torch.float8_e4m3fn:
laibao's avatar
laibao committed
376
        timers.append(
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
            bench_fns(
                label,
                sub_label,
                f"cutlass_scaled_mm ({terse_type_name(types.act_type)})",
                [cutlass_scaled_mm_create_bench_fn(bt) for bt in benchmark_tensors],
            )
        )

    if types.act_type != torch.float8_e4m3fn:
        timers.append(
            bench_fns(
                label,
                sub_label,
                f"marlin ({name_type_string})",
                [marlin_create_bench_fn(bt) for bt in benchmark_tensors],
            )
        )
laibao's avatar
laibao committed
394
395
396

    # machete
    timers.append(
397
398
399
400
401
402
403
404
405
406
        bench_fns(
            label,
            sub_label,
            f"machete ({name_type_string})",
            [
                machete_create_bench_fn(bt, out_type=types.output_type)
                for bt in benchmark_tensors
            ],
        )
    )
laibao's avatar
laibao committed
407
408

    if sweep_schedules:
409
410
        global _SWEEP_SCHEDULES_RESULTS

laibao's avatar
laibao committed
411
412
413
        print("Finding best schedule for machete")
        best = None
        best_schedule = None
414
415
416
417
418
419
420
421
422
423
424
425
426
        schedules = ops.machete_supported_schedules(
            a_type=types.act_type,
            b_type=types.weight_type,
            group_scales_type=types.group_scale_type,
            group_zeros_type=types.group_zero_type,
            token_scales_type=types.token_scale_type,
            channel_scales_type=types.channel_scale_type,
            out_type=types.output_type,
        )

        if schedules is None or len(schedules) == 0:
            raise ValueError("No schedules found to sweep")

laibao's avatar
laibao committed
427
428
429
430
431
432
433
        for schedule in reversed(schedules):
            schedule_M = int(schedule.split("_")[0].split("x")[1])

            # Prune known bad schedules
            if schedule_M >= 2 * max(m, 16) or schedule_M < m // 4:
                continue

434
435
436
437
438
439
440
441
442
443
444
            res = bench_fns(
                label,
                sub_label,
                "machete_best",
                [
                    machete_create_bench_fn(
                        bt, out_type=types.output_type, schedule=schedule
                    )
                    for bt in benchmark_tensors
                ],
            )
laibao's avatar
laibao committed
445
446
447
448
449
450
451
452
453
454

            results_row = {
                "M": m,
                "K": k,
                "N": n,
                "group_size": group_size,
                "schedule": schedule,
                "median": res.median,
            }
            if _SWEEP_SCHEDULES_RESULTS is None:
455
456
                _SWEEP_SCHEDULES_RESULTS = pd.DataFrame(columns=results_row.keys())
            _SWEEP_SCHEDULES_RESULTS.loc[len(_SWEEP_SCHEDULES_RESULTS)] = results_row
laibao's avatar
laibao committed
457
458
459
460
461
462
463
464
465
466
467
468

            print(f"  {res.median:5.5} ", schedule)
            if not best or res.median < best.median:
                best = res
                best_schedule = schedule
        print("Best schedule:", best_schedule)
        timers.append(best)

    return timers


# runner
469
def print_timers(timers: list[TMeasurement]):
laibao's avatar
laibao committed
470
471
472
473
    compare = TBenchmark.Compare(timers)
    compare.print()


474
475
476
477
478
479
480
481
482
483
484
485
def run(args, MKNs: Iterable[tuple[int, int, int]]) -> Iterable[TMeasurement]:
    types = TypeConfig(
        act_type=args.act_type,
        weight_type=scalar_types.uint4b8
        if args.group_zero_type is None
        else scalar_types.uint4,
        output_type=args.out_type,
        group_scale_type=args.group_scale_type,
        group_zero_type=args.group_zero_type,
        channel_scale_type=args.channel_scale_type,
        token_scale_type=args.token_scale_type,
    )
laibao's avatar
laibao committed
486

487
    results: list[TMeasurement] = []
laibao's avatar
laibao committed
488
    for m, k, n in MKNs:
489
490
491
492
493
494
495
496
497
498
        timers = bench(
            types,
            args.group_size,
            m,
            k,
            n,
            f"{args.act_type}-gemm",
            f"MKN=({m}x{k}x{n})",
            sweep_schedules=args.sweep_schedules,
        )
laibao's avatar
laibao committed
499
500
501
502
503
504
505
506
        print_timers(timers)
        results.extend(timers)

    return results


# output makers
def make_output(
507
508
    data: list[TMeasurement],
    MKNs: Iterable[tuple[int, int, int]],
laibao's avatar
laibao committed
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
    base_description: str,
    timestamp=None,
):
    print(f"== All Results {base_description} ====")
    print_timers(data)

    # pickle all the results
    timestamp = int(time.time()) if timestamp is None else timestamp
    with open(f"{base_description}-{timestamp}.pkl", "wb") as f:
        pkl.dump(data, f)


# argparse runners


def run_square_bench(args):
525
    dim_sizes = list(range(args.dim_start, args.dim_end + 1, args.dim_increment))
laibao's avatar
laibao committed
526
527
528
529
530
531
532
    MKNs = list(zip(dim_sizes, dim_sizes, dim_sizes))
    data = run(args.dtype, args.sweep_schedules, MKNs)

    make_output(data, MKNs, f"square_bench-{args.dtype}")


def run_range_bench(args):
533
534
535
536
537
    m_start, k_start, n_start = (int(x) for x in args.dim_start.split(","))
    m_end, k_end, n_end = (int(x) for x in args.dim_end.split(","))
    m_increment, k_increment, n_increment = (
        int(x) for x in args.dim_increment.split(",")
    )
laibao's avatar
laibao committed
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
    Ms = list(range(m_start, m_end + 1, m_increment))
    Ks = list(range(k_start, k_end + 1, k_increment))
    Ns = list(range(n_start, n_end + 1, n_increment))
    MKNs = list(product(Ms, Ks, Ns))

    data = run(args.dtype, args.sweep_schedules, MKNs)

    make_output(data, MKNs, f"range_bench-{args.dtype}")


def run_model_bench(args):
    print("Benchmarking models:")
    for i, model in enumerate(args.models):
        print(f"[{i}]  {model}")

553
    def model_shapes(model_name: str, tp_size: int) -> list[tuple[int, int]]:
laibao's avatar
laibao committed
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
        KNs = []
        for KN, tp_split_dim in copy.deepcopy(WEIGHT_SHAPES[model_name]):
            KN[tp_split_dim] = KN[tp_split_dim] // tp_size
            KNs.append(KN)
        return KNs

    model_bench_data = []
    models_tps = list(itertools.product(args.models, args.tp_sizes))
    for model, tp_size in models_tps:
        Ms = args.batch_sizes
        KNs = model_shapes(model, tp_size)
        MKNs = []
        for m in Ms:
            for k, n in KNs:
                MKNs.append((m, k, n))

570
        data = run(args, MKNs)
laibao's avatar
laibao committed
571
572
        model_bench_data.append(data)

573
574
    type_string = f"{args.act_type}"

laibao's avatar
laibao committed
575
576
577
    # Print all results
    for data, model_tp in zip(model_bench_data, models_tps):
        model, tp_size = model_tp
578
        print(f"== Results {type_string} {model}-TP{tp_size} ====")
laibao's avatar
laibao committed
579
580
        print_timers(data)

581
    timestr = time.strftime("%Y%m%d-%H%M%S")
laibao's avatar
laibao committed
582

583
    all_results = []
laibao's avatar
laibao committed
584
    for d in model_bench_data:
585
586
        all_results.extend(d)

laibao's avatar
laibao committed
587
    # pickle all data
588
589
590
591
592
593
594
595
596
597
    with open(f"model_bench-{type_string}-{timestr}.pkl", "wb") as f:
        args_dict = vars(args)
        args_dict.pop("func")
        pkl.dump(
            {
                "args": args_dict,
                "results": all_results,
            },
            f,
        )
laibao's avatar
laibao committed
598
599
600
601
602


if __name__ == "__main__":

    def to_torch_dtype(dt):
603
604
605
606
607
608
609
610
611
612
613
614
        return {
            "bfloat16": torch.bfloat16,
            "float16": torch.float16,
            "int8": torch.int8,
            "float8_e4m3fn": torch.float8_e4m3fn,
            "int": torch.int,
            "float": torch.float,
        }[dt]

    class ToTorchDtype(argparse.Action):
        def __call__(self, parser, namespace, values, option_string=None):
            setattr(namespace, self.dest, to_torch_dtype(values))
laibao's avatar
laibao committed
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634

    parser = FlexibleArgumentParser(
        description="""
Benchmark Machete GEMM.

    To run square GEMMs:
        python3 ./benchmarks/kernels/benchmark_machete.py --dtype float16 square_bench --dim-start 128 --dim-end 512 --dim-increment 64
    
    To run constant N and K and sweep M:
        python3 ./benchmarks/kernels/benchmark_machete.py --dtype float16 range_bench --dim-start 128 --dim-end 512 --dim-increment 64 --n-constant 16384 --k-constant 16384
    
    To run dimensions from a model:
        python3 ./benchmarks/kernels/benchmark_machete.py --dtype float16 model_bench --models meta-llama/Llama-2-7b-hf --batch-sizes 16 --tp-sizes 1
    
    Output:
        - a .pkl file, that is a list of raw torch.benchmark.utils.Measurements for the pytorch and cutlass implementations for the various GEMMs.
            """,  # noqa: E501
        formatter_class=argparse.RawTextHelpFormatter,
    )
    parser.add_argument(
635
636
        "--act-type",
        action=ToTorchDtype,
laibao's avatar
laibao committed
637
        required=True,
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
        choices=["bfloat16", "float16", "int8", "float8_e4m3fn"],
    )
    parser.add_argument(
        "--group-scale-type",
        action=ToTorchDtype,
        choices=["bfloat16", "float16"],
    )
    parser.add_argument(
        "--group-zero-type",
        type=to_torch_dtype,
        choices=["bfloat16", "float16"],
    )
    parser.add_argument(
        "--channel-scale-type",
        action=ToTorchDtype,
        choices=["float"],
    )
    parser.add_argument(
        "--token-scale-type",
        action=ToTorchDtype,
        choices=["float"],
    )
    parser.add_argument(
        "--out-type",
        action=ToTorchDtype,
        choices=["bfloat16", "float16"],
    )
    parser.add_argument(
        "--group-size",
        type=int,
        help="Available options are ['None', '-1', '128'], default=128",
        default=128,
laibao's avatar
laibao committed
670
671
672
673
674
675
    )
    parser.add_argument(
        "--sweep-schedules",
        action="store_true",
        help="Run a sweep over all supported schedules",
    )
676
677
678
679
680
    parser.add_argument(
        "--sweep-csv-out",
        help="CSV to store sweep results",
        default="sch_sweep_results.csv",
    )
laibao's avatar
laibao committed
681
682
683
684
685
686
687
688
689
690
691
692
693
    subparsers = parser.add_subparsers(dest="cmd", required=True)

    square_parser = subparsers.add_parser("square_bench")
    square_parser.add_argument("--dim-start", type=int, required=True)
    square_parser.add_argument("--dim-end", type=int, required=True)
    square_parser.add_argument("--dim-increment", type=int, required=True)
    square_parser.set_defaults(func=run_square_bench)

    range_parser = subparsers.add_parser("range_bench")
    range_parser.add_argument(
        "--dim-start",
        type=str,
        required=True,
694
695
        help="Start value for M,K,N as common separated list",
    )
laibao's avatar
laibao committed
696
697
698
699
    range_parser.add_argument(
        "--dim-end",
        type=str,
        required=True,
700
701
        help="End value (inclusive) for M,K,N as common separated list",
    )
laibao's avatar
laibao committed
702
703
704
705
    range_parser.add_argument(
        "--dim-increment",
        type=str,
        required=True,
706
707
        help="Increment value for M,K,N as common separated list",
    )
laibao's avatar
laibao committed
708
709
710
711
712
713
714
715
716
717
    range_parser.set_defaults(func=run_range_bench)

    model_parser = subparsers.add_parser("model_bench")
    model_parser.add_argument(
        "--models",
        nargs="+",
        type=str,
        default=DEFAULT_MODELS,
        choices=WEIGHT_SHAPES.keys(),
    )
718
719
720
721
722
723
    model_parser.add_argument(
        "--tp-sizes", nargs="+", type=int, default=DEFAULT_TP_SIZES
    )
    model_parser.add_argument(
        "--batch-sizes", nargs="+", type=int, default=DEFAULT_BATCH_SIZES
    )
laibao's avatar
laibao committed
724
725
726
727
728
729
730
731
732
    model_parser.set_defaults(func=run_model_bench)

    args = parser.parse_args()

    _SWEEP_SCHEDULES_RESULTS_CSV = args.sweep_csv_out
    args.func(args)

    if _SWEEP_SCHEDULES_RESULTS is not None:
        _SWEEP_SCHEDULES_RESULTS.to_csv(_SWEEP_SCHEDULES_RESULTS_CSV)