From 60a948bf4495dfdac083df88588bfe16435b43ab Mon Sep 17 00:00:00 2001 From: Andrew Keen Chan Date: Tue, 20 May 2025 16:55:45 +0000 Subject: [PATCH] delete unused and noncompiling mmvq functions --- ggml/src/ggml-cuda/iqk_mmvq.cu | 12 ------------ ggml/src/ggml-cuda/iqk_mmvq.cuh | 4 ---- 2 files changed, 16 deletions(-) diff --git a/ggml/src/ggml-cuda/iqk_mmvq.cu b/ggml/src/ggml-cuda/iqk_mmvq.cu index 97addc4a..6a2db725 100644 --- a/ggml/src/ggml-cuda/iqk_mmvq.cu +++ b/ggml/src/ggml-cuda/iqk_mmvq.cu @@ -590,11 +590,6 @@ __device__ __forceinline__ float vec_dot_iq2_ks_q8_1( + __low2float(bq8_1[4*(i4/4)+3].ds) * sumi4); } -__device__ __forceinline__ float vec_dot_iq2_kt_q8_1( - const void * __restrict__ vbq, const block_q8_1 * __restrict__ bq8_1, const int & kbx, const int & iqs) { - return 0.f; -} - #define VDR_IQ3_K_Q8_1_MMVQ 4 #define VDR_IQ3_K_Q8_1_MMQ 4 @@ -829,13 +824,6 @@ void mul_mat_vec_iq2_ks_q8_1_cuda( iqk_mul_mat_vec_q_cuda(vx, vy, dst, ids_data, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, ne2, nb02, nb12, nb2, ids_nb0, stream); } -void mul_mat_vec_iq2_kt_q8_1_cuda( - const void * vx, const void * vy, float * dst, - const int ncols_x, const int nrows_x, const int nrows_y, const int ncols_y, const int nrows_dst, cudaStream_t stream) { - - iqk_mul_mat_vec_q_cuda(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream); -} - void mul_mat_vec_iq5_k_q8_1_cuda( const void * vx, const void * vy, float * dst, const char * ids_data, const int ncols_x, const int nrows_x, const int nrows_y, const int ncols_y, const int nrows_dst, diff --git a/ggml/src/ggml-cuda/iqk_mmvq.cuh b/ggml/src/ggml-cuda/iqk_mmvq.cuh index 67ab2b6b..b81d2114 100644 --- a/ggml/src/ggml-cuda/iqk_mmvq.cuh +++ b/ggml/src/ggml-cuda/iqk_mmvq.cuh @@ -51,10 +51,6 @@ void mul_mat_vec_iq2_ks_q8_1_cuda( const int ncols_x, const int nrows_x, const int nrows_y, const int ncols_y, const int nrows_dst, const int ne2, const uint64_t nb02, const uint64_t nb12, const uint64_t nb2, const int64_t ids_nb0, cudaStream_t stream); -void mul_mat_vec_iq2_kt_q8_1_cuda( - const void * vx, const void * vy, float * dst, - const int ncols_x, const int nrows_x, const int nrows_y, const int ncols_y, const int nrows_dst, cudaStream_t stream); - void mul_mat_vec_iq1_bn_q8_1_cuda( const void * vx, const void * vy, float * dst, const char * ids_data, const int ncols_x, const int nrows_x, const int nrows_y, const int ncols_y, const int nrows_dst,