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

[Kernel][Quantization] add w4a8 support for marlin kernel (#24722)

Browse Source 
Signed-off-by: Jinzhen Lin <jinzhen.ljz@antgroup.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Michael Goin <mgoin@redhat.com>
This commit is contained in:
Jinzhen Lin
2025-11-29 23:19:33 +08:00
committed by GitHub
parent fa59fe417f
commit 1656ad3704
46 changed files with 4371 additions and 2240 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
csrc/quantization/gptq_marlin/kernel.h
View File

@@ -11,17 +11,19 @@
const int4 *__restrict__ A, const int4 *__restrict__ B, \
int4 *__restrict__ C, int4 *__restrict__ C_tmp, \
const int4 *__restrict__ b_bias_ptr, \
const float *__restrict__ a_scales_ptr, \
const int4 *__restrict__ scales_ptr, \
const uint16_t *__restrict__ scale2_ptr, \
const uint16_t *__restrict__ global_scale_ptr, \
const int4 *__restrict__ zp_ptr, const int *__restrict__ g_idx, \
int num_groups, int prob_m, int prob_n, int prob_k, int lda, int *locks, \
bool has_bias, bool use_atomic_add, bool use_fp32_reduce, \
int max_shared_mem
namespace MARLIN_NAMESPACE_NAME {
template <typename scalar_t, // compute dtype, half or nv_float16
const vllm::ScalarTypeId w_type_id, // weight ScalarType id
const vllm::ScalarTypeId s_type_id, // weight ScalarType id
template <const vllm::ScalarTypeId a_type_id, // A ScalarType id
const vllm::ScalarTypeId b_type_id, // B ScalarType id
const vllm::ScalarTypeId c_type_id, // C ScalarType id
const vllm::ScalarTypeId s_type_id, // B_SCALE ScalarType id
const int threads, // number of threads in a threadblock
const int thread_m_blocks, // number of 16x16 blocks in the m
// dimension (batchsize) of the