vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 09:34:37 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'master' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	Makefile
#	examples/quantize/quantize.cpp
#	ggml.c
#	pocs/vdot/vdot.cpp
#	scripts/build-info.cmake
#	scripts/build-info.h.in
#	scripts/build-info.sh
#	tests/test-opt.cpp
#	tests/test-quantize-fns.cpp
#	tests/test-quantize-perf.cpp
#	tests/test-sampling.cpp
This commit is contained in:
Concedo 2023-09-16 12:14:19 +08:00
commit c96fb3984d
39 changed files with 1098 additions and 340 deletions
Download patch file Download diff file Expand all files Collapse all files

12
ggml-alloc.c
View file

@ -131,6 +131,10 @@ static bool ggml_allocr_is_own(struct ggml_allocr * alloc, const struct ggml_ten
return ptr >= alloc->data && (char *)ptr < (char *)alloc->data + alloc->max_size;
}
static bool ggml_is_view(struct ggml_tensor * t) {
return t->view_src != NULL;
}
void ggml_allocr_alloc(struct ggml_allocr * alloc, struct ggml_tensor * tensor) {
#ifdef GGML_ALLOCATOR_DEBUG
GGML_ASSERT(!ggml_is_view(tensor)); // views generally get data pointer from one of their sources
@ -338,8 +342,8 @@ static void free_vmem(void * base_addr, size_t size) {
// allocate uncommitted virtual memory to measure the size of the graph
static void alloc_measure_vmem(void ** base_addr, size_t * size) {
// 1TB for 64-bit, 1GB for 32-bit
*size = sizeof(void *) == 4 ? 1ULL<<30 : 1ULL<<40;
// 128GB for 64-bit, 1GB for 32-bit
*size = sizeof(void *) == 4 ? 1ULL<<30 : 1ULL<<37;
do {
*base_addr = alloc_vmem(*size);
if (*base_addr != NULL) {
@ -399,10 +403,6 @@ bool ggml_allocr_is_measure(struct ggml_allocr * alloc) {
//////////// compute graph allocator
static bool ggml_is_view(struct ggml_tensor * t) {
return t->view_src != NULL;
}
static bool ggml_are_same_layout(const struct ggml_tensor * a, const struct ggml_tensor * b) {
if (a->type != b->type) {
return false;