root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-27 20:04:35 +00:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

metal : disable buffer allocation logs

Browse Source
This commit is contained in:
Georgi Gerganov 2024-01-29 18:12:24 +02:00
parent 4794821a31
commit abeaf0d90e
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 15 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
ggml-metal.m
View File

@ -2421,10 +2421,13 @@ GGML_CALL static const char * ggml_backend_metal_buffer_type_get_name(ggml_backe
UNUSED(buft);
}
static void ggml_backend_metal_log_allocated_size(id<MTLDevice> device) {
static void ggml_backend_metal_log_allocated_size(id<MTLDevice> device, size_t size_aligned) {
#ifndef GGML_METAL_NDEBUG
#if TARGET_OS_OSX || (TARGET_OS_IOS && __clang_major__ >= 15)
if (@available(macOS 10.12, iOS 16.0, *)) {
GGML_METAL_LOG_INFO(", (%8.2f / %8.2f)",
GGML_METAL_LOG_INFO("%s: allocated buffer, size = %8.2f MiB, (%8.2f / %8.2f)",
__func__,
size_aligned / 1024.0 / 1024.0,
device.currentAllocatedSize / 1024.0 / 1024.0,
device.recommendedMaxWorkingSetSize / 1024.0 / 1024.0);
@ -2434,10 +2437,15 @@ static void ggml_backend_metal_log_allocated_size(id<MTLDevice> device) {
GGML_METAL_LOG_INFO("\n");
}
} else {
GGML_METAL_LOG_INFO(", (%8.2f)\n", device.currentAllocatedSize / 1024.0 / 1024.0);
GGML_METAL_LOG_INFO("%s: allocated buffer, size = %8.2f MiB, (%8.2f)\n",
__func__,
size_aligned / 1024.0 / 1024.0,
device.currentAllocatedSize / 1024.0 / 1024.0);
}
#endif
#endif
UNUSED(device);
UNUSED(size_aligned);
}
GGML_CALL static ggml_backend_buffer_t ggml_backend_metal_buffer_type_alloc_buffer(ggml_backend_buffer_type_t buft, size_t size) {
@ -2471,8 +2479,7 @@ GGML_CALL static ggml_backend_buffer_t ggml_backend_metal_buffer_type_alloc_buff
return NULL;
}
GGML_METAL_LOG_INFO("%s: allocated buffer, size = %8.2f MiB", __func__, size_aligned / 1024.0 / 1024.0);
ggml_backend_metal_log_allocated_size(device);
ggml_backend_metal_log_allocated_size(device, size_aligned);
return ggml_backend_buffer_init(buft, ggml_backend_metal_buffer_i, ctx, size);
}
@ -2549,7 +2556,7 @@ GGML_CALL ggml_backend_buffer_t ggml_backend_metal_buffer_from_ptr(void * data,
return false;
}
GGML_METAL_LOG_INFO("%s: allocated buffer, size = %8.2f MiB", __func__, size_aligned / 1024.0 / 1024.0);
ggml_backend_metal_log_allocated_size(device, size_aligned);
++ctx->n_buffers;
} else {
@ -2572,7 +2579,8 @@ GGML_CALL ggml_backend_buffer_t ggml_backend_metal_buffer_from_ptr(void * data,
return false;
}
GGML_METAL_LOG_INFO("%s: allocated buffer, size = %8.2f MiB, offs = %12ld", __func__, size_step_aligned / 1024.0 / 1024.0, i);
ggml_backend_metal_log_allocated_size(device, size_step_aligned);
if (i + size_step < size) {
GGML_METAL_LOG_INFO("\n");
}
@ -2581,8 +2589,6 @@ GGML_CALL ggml_backend_buffer_t ggml_backend_metal_buffer_from_ptr(void * data,
}
}
ggml_backend_metal_log_allocated_size(device);
return ggml_backend_buffer_init(ggml_backend_metal_buffer_type(), ggml_backend_metal_buffer_i, ctx, size);
}