vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-15 19:39:42 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixed wrong sized struct from legacy q8_1, fixed opencl varsize arrays

Browse source
This commit is contained in:
Concedo 2023-05-13 23:56:08 +08:00
parent c9eb2ba1c5
commit e05455f852
7 changed files with 25 additions and 16 deletions
Download patch file Download diff file Expand all files Collapse all files

15
ggml.c
View file

@ -3102,11 +3102,12 @@ static const int GGML_BLCK_SIZE[GGML_TYPE_COUNT] = {
[GGML_TYPE_Q5_1] = QK5_1,
[GGML_TYPE_Q8_0] = QK8_0,
[GGML_TYPE_Q8_1] = QK8_1,
[GGML_TYPE_Q8_1B] = QK8_1,
[GGML_TYPE_I8] = 1,
[GGML_TYPE_I16] = 1,
[GGML_TYPE_I32] = 1,
};
static_assert(GGML_TYPE_COUNT == 13, "GGML_BLCK_SIZE is outdated");
static_assert(GGML_TYPE_COUNT == 14, "GGML_BLCK_SIZE is outdated");
static const size_t GGML_TYPE_SIZE[GGML_TYPE_COUNT] = {
[GGML_TYPE_F32] = sizeof(float),
@ -3119,11 +3120,12 @@ static const size_t GGML_TYPE_SIZE[GGML_TYPE_COUNT] = {
[GGML_TYPE_Q5_1] = sizeof(block_q5_1),
[GGML_TYPE_Q8_0] = sizeof(block_q8_0),
[GGML_TYPE_Q8_1] = sizeof(block_q8_1),
[GGML_TYPE_Q8_1B] = sizeof(block_q8_1_v2),
[GGML_TYPE_I8] = sizeof(int8_t),
[GGML_TYPE_I16] = sizeof(int16_t),
[GGML_TYPE_I32] = sizeof(int32_t),
};
static_assert(GGML_TYPE_COUNT == 13, "GGML_TYPE_SIZE is outdated");
static_assert(GGML_TYPE_COUNT == 14, "GGML_TYPE_SIZE is outdated");
static const char * GGML_TYPE_NAME[GGML_TYPE_COUNT] = {
@ -3137,11 +3139,12 @@ static const char * GGML_TYPE_NAME[GGML_TYPE_COUNT] = {
[GGML_TYPE_Q5_1] = "q5_1",
[GGML_TYPE_Q8_0] = "q8_0",
[GGML_TYPE_Q8_1] = "q8_1",
[GGML_TYPE_Q8_1B] = "q8_1b",
[GGML_TYPE_I8] = "i8",
[GGML_TYPE_I16] = "i16",
[GGML_TYPE_I32] = "i32",
};
static_assert(GGML_TYPE_COUNT == 13, "GGML_TYPE_NAME is outdated");
static_assert(GGML_TYPE_COUNT == 14, "GGML_TYPE_NAME is outdated");
static bool GGML_IS_QUANTIZED[GGML_TYPE_COUNT] = {
[GGML_TYPE_F32] = false,
@ -3154,11 +3157,12 @@ static bool GGML_IS_QUANTIZED[GGML_TYPE_COUNT] = {
[GGML_TYPE_Q5_1] = true,
[GGML_TYPE_Q8_0] = true,
[GGML_TYPE_Q8_1] = true,
[GGML_TYPE_Q8_1B] = true,
[GGML_TYPE_I8] = false,
[GGML_TYPE_I16] = false,
[GGML_TYPE_I32] = false,
};
static_assert(GGML_TYPE_COUNT == 13, "GGML_IS_QUANTIZED is outdated");
static_assert(GGML_TYPE_COUNT == 14, "GGML_IS_QUANTIZED is outdated");
static const char * GGML_OP_LABEL[GGML_OP_COUNT] = {
"NONE",
@ -8041,6 +8045,7 @@ static void ggml_compute_forward_mul_mat(
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q8_0:
case GGML_TYPE_Q8_1:
case GGML_TYPE_Q8_1B:
{
ggml_compute_forward_mul_mat_q_f32(params, src0, src1, dst);
} break;
@ -8273,6 +8278,7 @@ static void ggml_compute_forward_get_rows(
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q8_0:
case GGML_TYPE_Q8_1:
case GGML_TYPE_Q8_1B:
{
ggml_compute_forward_get_rows_q(params, src0, src1, dst);
} break;
@ -8599,6 +8605,7 @@ static void ggml_compute_forward_alibi(
case GGML_TYPE_Q5_1:
case GGML_TYPE_Q8_0:
case GGML_TYPE_Q8_1:
case GGML_TYPE_Q8_1B:
case GGML_TYPE_I8:
case GGML_TYPE_I16:
case GGML_TYPE_I32: