sdgoij/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-03-17 16:44:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

reenable transfer queue if graphics queue is not used

Browse Source
This commit is contained in:
Ruben Ortlam
2026-03-16 09:17:44 +01:00
parent 824aa149ca
commit 840dfe4a7a
1 changed files with 4 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
ggml/src/ggml-vulkan/ggml-vulkan.cpp
View File

@@ -5444,11 +5444,14 @@ static vk_device ggml_vk_get_device(size_t idx) {
ggml_vk_load_shaders(device);
// Only use transfer queue on AMD non-GCN, when the graphics queue is not enabled
const bool prefers_transfer_queue = device->vendor_id == VK_VENDOR_ID_AMD && device->architecture != AMD_GCN && !allow_graphics_queue;
if (!device->single_queue) {
const uint32_t transfer_queue_index = compute_queue_family_index == transfer_queue_family_index ? 1 : 0;
ggml_vk_create_queue(device, device->transfer_queue, transfer_queue_family_index, transfer_queue_index, { vk::PipelineStageFlagBits::eTransfer }, true);
device->async_use_transfer_queue = (getenv("GGML_VK_ASYNC_USE_TRANSFER_QUEUE") != nullptr);
device->async_use_transfer_queue = prefers_transfer_queue || (getenv("GGML_VK_ASYNC_USE_TRANSFER_QUEUE") != nullptr);
} else {
// TODO: Use pointer or reference to avoid copy
device->transfer_queue.copyFrom(device->compute_queue);