biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[ Misc ] Refactor Marlin Python Utilities (#6082)

Browse Source 
Co-authored-by: Robert Shaw <rshaw@neuralmagic.com>
This commit is contained in:
Robert Shaw
2024-07-11 11:40:11 -04:00
committed by GitHub
parent 55f692b46e
commit b675069d74
12 changed files with 690 additions and 728 deletions
Download Patch File Download Diff File Expand all files Collapse all files

415
vllm/model_executor/layers/quantization/utils/marlin_utils.py
View File

@@ -1,21 +1,9 @@
"""This file is used for /tests and /benchmarks"""
import random
from typing import Optional
from typing import List, Optional, Tuple
import numpy
import torch
from vllm import _custom_ops as ops
from vllm.model_executor.layers.quantization.utils.format_24 import (
mask_creator, sparse_semi_structured_from_dense_cutlass)
from vllm.model_executor.layers.quantization.utils.marlin_24_perms import (
marlin_24_perm, marlin_24_scale_perm, marlin_24_scale_perm_single)
from vllm.model_executor.layers.quantization.utils.marlin_perms import (
marlin_perm, marlin_scale_perm, marlin_scale_perm_single)
from vllm.model_executor.layers.quantization.utils.quant_utils import (
get_pack_factor, quantize_weights, sort_weights)
from vllm.platforms import current_platform
from vllm.utils import print_warning_once
GPTQ_MARLIN_TILE = 16
GPTQ_MARLIN_MIN_THREAD_N = 64
@@ -25,135 +13,110 @@ GPTQ_MARLIN_MAX_PARALLEL = 16
GPTQ_MARLIN_SUPPORTED_NUM_BITS = [4, 8]
GPTQ_MARLIN_SUPPORTED_GROUP_SIZES = [-1, 32, 64, 128]
GPTQ_MARLIN_SUPPORTED_SYM = [True]
GTPQ_MARLIN_UNSUPPORTED_GROUP_SIZE_ACT_ORDER = [-1]
def is_marlin_supported():
capability = current_platform.get_device_capability()
return capability[0] >= 8
def check_marlin_supported(num_bits: int, group_size: int, is_sym: bool,
min_capability: int) -> bool:
# If the capability of the device is too low, cannot convert.
major, minor = current_platform.get_device_capability()
device_capability = major * 10 + minor
if device_capability < min_capability:
return False
return (device_capability >= min_capability
and num_bits in GPTQ_MARLIN_SUPPORTED_NUM_BITS
and group_size in GPTQ_MARLIN_SUPPORTED_GROUP_SIZES
and is_sym in GPTQ_MARLIN_SUPPORTED_SYM)
def apply_fp8_marlin_linear(
input: torch.Tensor,
weight: torch.Tensor,
weight_scale: torch.Tensor,
workspace: torch.Tensor,
size_n: int,
size_k: int,
bias: Optional[torch.Tensor],
) -> torch.Tensor:
# For GPUs that lack FP8 hardware support, we can leverage the
# Marlin kernel for fast weight-only FP8 quantization
def verify_marlin_supported(num_bits: int, group_size: Optional[int],
is_sym: bool) -> None:
reshaped_x = input.reshape(-1, input.shape[-1])
out_shape = input.shape[:-1] + (size_n, )
output = ops.fp8_marlin_gemm(
a=reshaped_x,
b_q_weight=weight,
b_scales=weight_scale,
workspace=workspace,
num_bits=8,
size_m=reshaped_x.shape[0],
size_n=size_n,
size_k=size_k,
)
if bias is not None:
output.add_(bias) # In-place add
return output.reshape(out_shape)
if num_bits not in GPTQ_MARLIN_SUPPORTED_NUM_BITS:
raise ValueError(
f"Marlin does not support weight_bits = {num_bits}. "
f"Only weight_bits = {GPTQ_MARLIN_SUPPORTED_NUM_BITS} "
"are supported.")
if (group_size is None
or group_size not in GPTQ_MARLIN_SUPPORTED_GROUP_SIZES):
raise ValueError(
f"Marlin does not support group_size = {group_size}. "
f"Only group_sizes = {GPTQ_MARLIN_SUPPORTED_GROUP_SIZES} "
"are supported.")
if is_sym not in GPTQ_MARLIN_SUPPORTED_SYM:
raise ValueError(
f"Marlin does not support is_sym = is_sym. "
f"Only sym = {GPTQ_MARLIN_SUPPORTED_SYM} are supported.")
def prepare_fp8_layer_for_marlin(layer: torch.nn.Module) -> None:
print_warning_once(
"Your GPU does not have native support for FP8 computation but "
"FP8 quantization is being used. Weight-only FP8 compression will "
"be used leveraging the Marlin kernel. This may degrade "
"performance for compute-heavy workloads.")
def verify_marlin_supports_shape(output_size_per_partition: int,
input_size_per_partition: int,
input_size: int, group_size: int) -> None:
part_size_n = layer.output_size_per_partition
part_size_k = layer.input_size_per_partition
# Validate output_size_per_partition
if output_size_per_partition % GPTQ_MARLIN_MIN_THREAD_N != 0:
raise ValueError(f"Weight output_size_per_partition = "
f"{output_size_per_partition} is not divisible by "
f" min_thread_n = {GPTQ_MARLIN_MIN_THREAD_N}. "
"Consider reducing tensor_parallel_size or running "
"with --quantization gptq.")
device = layer.weight.device
# Validate input_size_per_partition
if input_size_per_partition % GPTQ_MARLIN_MIN_THREAD_K != 0:
raise ValueError(f"Weight input_size_per_partition = "
f"{input_size_per_partition} is not divisible "
f"by min_thread_k = {GPTQ_MARLIN_MIN_THREAD_K}. "
"Consider reducing tensor_parallel_size or running "
"with --quantization gptq.")
# WEIGHTS
# Repack weights to gptq format (packed int32 elements)
packed_gptq_qweight = pack_fp8_to_int32(layer.weight)
if (group_size < input_size
and input_size_per_partition % group_size != 0):
raise ValueError(
f"Weight input_size_per_partition = {input_size_per_partition}"
f" is not divisible by group_size = {group_size}."
"Consider reducing tensor_parallel_size or running "
"with --quantization gptq.")
# Repack weights to marlin format
marlin_qweight = ops.gptq_marlin_repack(
b_q_weight=packed_gptq_qweight,
perm=torch.empty(0, dtype=torch.int, device=device),
size_k=part_size_k,
size_n=part_size_n,
num_bits=8,
)
layer.weight = torch.nn.Parameter(marlin_qweight, requires_grad=False)
# WEIGHT SCALES
# Currently Marlin doesn't support per-tensor scales, so we
# expand it to channelwise
scales = layer.weight_scale.repeat(1, part_size_n).to(
layer.orig_dtype).to(device)
# Permute scales
num_bits = 8
marlin_scales = marlin_permute_scales(
s=scales,
size_k=part_size_k,
size_n=part_size_n,
group_size=-1,
scale_perm=marlin_scale_perm[num_bits],
scale_perm_single=marlin_scale_perm_single[num_bits])
layer.weight_scale = torch.nn.Parameter(marlin_scales, requires_grad=False)
# Allocate marlin workspace
max_workspace_size = (part_size_n //
def marlin_make_workspace(output_size_per_partition: int,
device: torch.device) -> torch.Tensor:
max_workspace_size = (output_size_per_partition //
GPTQ_MARLIN_MIN_THREAD_N) * GPTQ_MARLIN_MAX_PARALLEL
workspace = torch.zeros(max_workspace_size,
dtype=torch.int,
device=device,
requires_grad=False)
layer.workspace = workspace
return torch.zeros(max_workspace_size,
dtype=torch.int,
device=device,
requires_grad=False)
def marlin_permute_weights(q_w, size_k, size_n, perm, tile=GPTQ_MARLIN_TILE):
assert q_w.shape == (size_k, size_n)
assert size_k % tile == 0, f"size_k = {size_k}, tile = {tile}"
assert size_n % tile == 0, f"size_k = {size_n}, tile = {tile}"
# Permute weights to 16x64 marlin tiles
q_w = q_w.reshape((size_k // tile, tile, size_n // tile, tile))
q_w = q_w.permute((0, 2, 1, 3))
q_w = q_w.reshape((size_k // tile, size_n * tile))
q_w = q_w.reshape((-1, perm.numel()))[:, perm].reshape(q_w.shape)
return q_w
def marlin_make_empty_g_idx(device: torch.device) -> torch.Tensor:
return torch.nn.Parameter(torch.empty(0, dtype=torch.int, device=device),
requires_grad=False)
def marlin_weights(q_w, size_k, size_n, num_bits, perm):
# Permute
q_w = marlin_permute_weights(q_w, size_k, size_n, perm)
# Pack
pack_factor = get_pack_factor(num_bits)
orig_device = q_w.device
q_w = q_w.cpu().numpy().astype(numpy.uint32)
q_packed = numpy.zeros((q_w.shape[0], q_w.shape[1] // pack_factor),
dtype=numpy.uint32)
for i in range(pack_factor):
q_packed |= q_w[:, i::pack_factor] << num_bits * i
q_packed = torch.from_numpy(q_packed.astype(numpy.int32)).to(orig_device)
return q_packed
def marlin_sort_g_idx(
g_idx: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
g_idx_sort_indices = torch.argsort(g_idx).to(torch.int)
return g_idx[g_idx_sort_indices], g_idx_sort_indices
def marlin_permute_scales(s, size_k, size_n, group_size, scale_perm,
scale_perm_single):
def get_scale_perms():
scale_perm: List[int] = []
for i in range(8):
scale_perm.extend([i + 8 * j for j in range(8)])
scale_perm_single: List[int] = []
for i in range(4):
scale_perm_single.extend(
[2 * i + j for j in [0, 1, 8, 9, 16, 17, 24, 25]])
return scale_perm, scale_perm_single
def marlin_permute_scales(s: torch.Tensor, size_k: int, size_n: int,
group_size: int) -> torch.Tensor:
scale_perm, scale_perm_single = get_scale_perms()
if group_size < size_k and group_size != -1:
s = s.reshape((-1, len(scale_perm)))[:, scale_perm]
else:
@@ -163,180 +126,44 @@ def marlin_permute_scales(s, size_k, size_n, group_size, scale_perm,
return s
def marlin_quantize(
w: torch.Tensor,
num_bits: int,
group_size: int,
act_order: bool,
):
size_k, size_n = w.shape
# Normalize group_size
if group_size == -1:
group_size = size_k
assert group_size <= size_k
# Quantize (and apply act_order if provided)
w_ref, q_w, s, g_idx, rand_perm = quantize_weights(w, num_bits, group_size,
act_order)
# For act_order, sort the "weights" and "g_idx" so that group ids are
# increasing
sort_indices = torch.empty(0, dtype=torch.int, device=w.device)
if act_order:
q_w, g_idx, sort_indices = sort_weights(q_w, g_idx)
# Reformat to marlin
marlin_q_w = marlin_weights(q_w, size_k, size_n, num_bits,
marlin_perm[num_bits])
marlin_s = marlin_permute_scales(s, size_k, size_n, group_size,
marlin_scale_perm[num_bits],
marlin_scale_perm_single[num_bits])
# Create result
res_list = [w_ref, marlin_q_w, marlin_s, g_idx, sort_indices, rand_perm]
for i in range(len(res_list)):
res_list[i] = res_list[i].to(w.device)
return res_list
# Newly generated tensors need to replace existing tensors that are
# already registered as parameters by vLLM (and won't be freed)
def replace_tensor(layer: torch.nn.Module, name: str,
new_t: torch.Tensor) -> None:
# It is important to use resize_() here since it ensures
# the same buffer is reused
getattr(layer, name).resize_(new_t.shape)
getattr(layer, name).copy_(new_t)
del new_t
def inject_24(w, size_k, size_n):
assert w.shape == (size_k, size_n)
def apply_marlin_linear(input: torch.Tensor,
weight: torch.Tensor,
weight_scale: torch.Tensor,
g_idx: torch.Tensor,
g_idx_sort_indices: torch.Tensor,
workspace: torch.Tensor,
num_bits: int,
output_size_per_partition: int,
input_size_per_partition: int,
is_k_full: bool,
bias: Optional[torch.Tensor] = None) -> torch.Tensor:
reshaped_x = input.reshape(-1, input.shape[-1])
out_shape = input.shape[:-1] + (output_size_per_partition, )
mask = mask_creator(w.t()).t().cuda().bool()
output = ops.gptq_marlin_gemm(reshaped_x,
weight,
weight_scale,
g_idx,
g_idx_sort_indices,
workspace,
num_bits,
size_m=reshaped_x.shape[0],
size_n=output_size_per_partition,
size_k=input_size_per_partition,
is_k_full=is_k_full)
return (mask * w).contiguous(), mask.contiguous()
if bias is not None:
output.add_(bias) # In-place add
def check_24(w, num_rows_to_sample=50, _verbose=False):
BLOCK_SIZE = 4
MAX_NON_ZEROS = 2
w = w.t().contiguous()
print("check_24: w.shape = {}".format(w.shape))
num_rows, num_cols = w.shape
sampled_row_idxs = random.choices(range(num_rows), k=num_rows_to_sample)
if _verbose:
print(f"Sampled row idxs = {sampled_row_idxs}")
total_segments = 0
non_24_segments = 0
for i in sampled_row_idxs:
for j in range(0, num_cols - BLOCK_SIZE, BLOCK_SIZE):
total_segments += 1
block = w[i, j:j + BLOCK_SIZE]
num_nonzero = torch.count_nonzero(block)
if num_nonzero > MAX_NON_ZEROS:
print("i = {} j = {} block = {}".format(i, j, block))
non_24_segments += 1
print(f"{non_24_segments} / {total_segments} do not have 2:4 structure.")
def compress_quantized_24_weight(q_24, size_k, size_n, num_bits):
assert q_24.shape == (size_k, size_n)
# Remove zp to normalize over 0
max_q_val = (1 << num_bits) - 1
zp = (max_q_val + 1) // 2
q_24_no_zp = q_24 - zp
# Compress
q_24_no_zp = q_24_no_zp.t().contiguous()
q_24_no_zp_comp, meta = sparse_semi_structured_from_dense_cutlass(
q_24_no_zp)
q_24_no_zp_comp = q_24_no_zp_comp.t().contiguous()
# Restore zp
q_24_comp = q_24_no_zp_comp + zp
# Resize meta to its actual shape (without moving any data)
meta = meta.resize_(meta.shape[1] // 2, meta.shape[0] * 2)
return q_24_comp, meta
def marlin_24_quantize(
w: torch.Tensor,
num_bits: int,
group_size: int,
):
size_k, size_n = w.shape
# Normalize group_size
if group_size == -1:
group_size = size_k
assert group_size <= size_k
# Inject 2:4 sparsity
w_24, mask_24 = inject_24(w, size_k, size_n)
# Quantize
w_24_ref, q_w_24, s, g_idx, rand_perm = quantize_weights(w_24,
num_bits,
group_size,
act_order=False)
# Compress quantized weight
q_w_24_comp, meta = compress_quantized_24_weight(q_w_24, size_k, size_n,
num_bits)
size_k_comp = size_k // 2
# Reformat to marlin
marlin_24_q_w_comp = marlin_weights(q_w_24_comp, size_k_comp, size_n,
num_bits, marlin_24_perm[num_bits])
marlin_24_s = marlin_permute_scales(s, size_k, size_n, group_size,
marlin_24_scale_perm[num_bits],
marlin_24_scale_perm_single[num_bits])
# Create result
res_list = [w_24_ref, marlin_24_q_w_comp, meta, marlin_24_s]
for i in range(len(res_list)):
res_list[i] = res_list[i].to(w.device)
return res_list
def compute_max_diff(output, output_ref):
return torch.mean(torch.abs(output - output_ref)) / torch.mean(
torch.abs(output_ref))
class MarlinWorkspace:
def __init__(self, out_features, min_thread_n, max_parallel):
assert (out_features % min_thread_n == 0), (
"out_features = {} is undivisible by min_thread_n = {}".format(
out_features, min_thread_n))
max_workspace_size = ((out_features // min_thread_n) * max_parallel)
self.scratch = torch.zeros(max_workspace_size,
dtype=torch.int,
device="cuda")
def pack_fp8_to_int32(fp8_tensor: torch.Tensor) -> torch.Tensor:
"""
Repack FP8 weights to gptq format (packed int32 elements)
"""
assert fp8_tensor.dtype == torch.float8_e4m3fn
assert fp8_tensor.shape[0] % 4 == 0
# Reshape to prepare for packing
reshaped = fp8_tensor.reshape(-1, 4, *fp8_tensor.shape[1:])
# Convert fp8 to uint8 (byte) representation
byte_tensor = reshaped.view(torch.uint8)
# Pack 4 uint8 values into one int32
packed = (byte_tensor[:, 0].to(torch.int32) |
(byte_tensor[:, 1].to(torch.int32) << 8) |
(byte_tensor[:, 2].to(torch.int32) << 16) |
(byte_tensor[:, 3].to(torch.int32) << 24))
return packed.view(fp8_tensor.shape[0] // 4,
*fp8_tensor.shape[1:]).contiguous()
return output.reshape(out_shape)