vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Apply ggerganov's fixes for test-backend-ops

Browse source
This commit is contained in:
Justine Tunney 2024-04-26 02:40:30 -07:00
parent bc278c8a0e
commit 2741a99791
No known key found for this signature in database
GPG key ID: 52965314629936D4
4 changed files with 10 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml-metal.m
View file

@ -803,7 +803,7 @@ static bool ggml_metal_supports_op(const struct ggml_metal_context * ctx, const
case GGML_OP_DIAG_MASK_INF:
case GGML_OP_GET_ROWS:
{
return op->ne[3] == 1;
return op->src[0]->type != GGML_TYPE_BF16 && op->ne[3] == 1;
}
default:
return false;

5
ggml.c
View file

@ -19765,7 +19765,10 @@ struct ggml_cplan ggml_graph_plan(const struct ggml_cgraph * cgraph, int n_threa
case GGML_OP_CPY:
case GGML_OP_DUP:
{
if (ggml_is_quantized(node->type)) {
if (ggml_is_quantized(node->type) ||
// F16 -> BF16 and BF16 -> F16 copies go through intermediate F32
(node->src[0]->type == GGML_TYPE_F16 && node->src[1] && node->src[1]->type == GGML_TYPE_BF16) ||
(node->src[0]->type == GGML_TYPE_BF16 && node->src[1] && node->src[1]->type == GGML_TYPE_F16)) {
cur = ggml_type_size(GGML_TYPE_F32) * node->ne[0] * n_tasks;
}
} break;

2
gguf-py/gguf/constants.py
View file

@ -863,7 +863,6 @@ QK_K = 256
GGML_QUANT_SIZES = {
GGMLQuantizationType.F32: (1, 4),
GGMLQuantizationType.F16: (1, 2),
GGMLQuantizationType.BF16: (1, 2),
GGMLQuantizationType.Q4_0: (32, 2 + 16),
GGMLQuantizationType.Q4_1: (32, 2 + 2 + 16),
GGMLQuantizationType.Q5_0: (32, 2 + 4 + 16),
@ -890,6 +889,7 @@ GGML_QUANT_SIZES = {
GGMLQuantizationType.I64: (1, 8),
GGMLQuantizationType.F64: (1, 8),
GGMLQuantizationType.IQ1_M: (256, QK_K // 8 + QK_K // 16 + QK_K // 32),
GGMLQuantizationType.BF16: (1, 2),
}

6
tests/test-backend-ops.cpp
View file

@ -50,7 +50,7 @@ static void init_tensor_uniform(ggml_tensor * tensor, float min = -1.0f, float m
if (tensor->type == GGML_TYPE_F32 || tensor->type == GGML_TYPE_I32) {
ggml_backend_tensor_set(tensor, data.data(), 0, size * sizeof(float));
} else if (ggml_is_quantized(tensor->type) || tensor->type == GGML_TYPE_F16) {
} else if (ggml_is_quantized(tensor->type) || tensor->type == GGML_TYPE_F16 || tensor->type == GGML_TYPE_BF16) {
GGML_ASSERT(size % ggml_blck_size(tensor->type) == 0);
std::vector<uint8_t> dataq(ggml_row_size(tensor->type, size));
std::vector<float> imatrix(tensor->ne[0], 1.0f); // dummy importance matrix
@ -92,6 +92,8 @@ static std::vector<float> tensor_to_float(const ggml_tensor * t) {
size_t i = i3*t->nb[3] + i2*t->nb[2] + i1*t->nb[1] + i0/bs*t->nb[0];
if (t->type == GGML_TYPE_F16) {
tv.push_back(ggml_fp16_to_fp32(*(ggml_fp16_t*)&buf[i]));
} else if (t->type == GGML_TYPE_BF16) {
tv.push_back(ggml_bf16_to_fp32(*(ggml_bf16_t*)&buf[i]));
} else if (t->type == GGML_TYPE_F32) {
tv.push_back(*(float *) &buf[i]);
} else if (t->type == GGML_TYPE_I32) {
@ -1898,7 +1900,7 @@ static bool test_backend(ggml_backend_t backend, test_mode mode, const char * op
std::default_random_engine rng(0);
const ggml_type all_types[] = {
GGML_TYPE_F32, GGML_TYPE_F16,
GGML_TYPE_F32, GGML_TYPE_F16, GGML_TYPE_BF16,
GGML_TYPE_Q4_0, GGML_TYPE_Q4_1,
GGML_TYPE_Q5_0, GGML_TYPE_Q5_1,
GGML_TYPE_Q8_0,