summaryrefslogtreecommitdiff
path: root/ggml/src/ggml-cuda/mmvq.cu
diff options
context:
space:
mode:
authorIwan Kawrakow <iwan.kawrakow@gmail.com>2024-08-01 16:08:32 +0300
committerKawrakow <48489457+ikawrakow@users.noreply.github.com>2024-08-09 16:00:31 +0200
commitcfb0410067be051dd8ea76c08280a3b04d5a5188 (patch)
tree64ffc06e18f75a8cfc4d04b98875f68cd7fa99d4 /ggml/src/ggml-cuda/mmvq.cu
parenta9f302ebe2373321c12b01d8760904901aa064a4 (diff)
iq6_k: WIP (nothing works)
Diffstat (limited to 'ggml/src/ggml-cuda/mmvq.cu')
-rw-r--r--ggml/src/ggml-cuda/mmvq.cu3
1 files changed, 3 insertions, 0 deletions
diff --git a/ggml/src/ggml-cuda/mmvq.cu b/ggml/src/ggml-cuda/mmvq.cu
index 428d822f..9eb3fa4f 100644
--- a/ggml/src/ggml-cuda/mmvq.cu
+++ b/ggml/src/ggml-cuda/mmvq.cu
@@ -447,6 +447,9 @@ void ggml_cuda_op_mul_mat_vec_q(
case GGML_TYPE_IQ5_K:
mul_mat_vec_iq5_k_q8_1_cuda(src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, src1_padded_row_size, src1_ncols, nrows_dst, stream);
break;
+ case GGML_TYPE_IQ6_K:
+ mul_mat_vec_iq6_k_q8_1_cuda(src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, src1_padded_row_size, src1_ncols, nrows_dst, stream);
+ break;
case GGML_TYPE_IQ3_S:
mul_mat_vec_iq3_s_q8_1_cuda(src0_dd_i, src1_ddq_i, dst_dd_i, ne00, row_diff, src1_padded_row_size, src1_ncols, nrows_dst, stream);
break;