happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge 56b8f27fab into 18ddaea2ae

This commit is contained in:
汞齐 2026-01-03 06:37:36 +08:00 committed by GitHub
parent 18ddaea2ae 56b8f27fab
commit 4ce06e6756
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 26 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
ggml/src/ggml-vulkan/ggml-vulkan.cpp
View File

@ -90,6 +90,7 @@ static bool is_pow2(uint32_t x) { return x > 1 && (x & (x-1)) == 0; }
#define VK_VENDOR_ID_AMD 0x1002
#define VK_VENDOR_ID_APPLE 0x106b
#define VK_VENDOR_ID_ARM 0x13B5
#define VK_VENDOR_ID_INTEL 0x8086
#define VK_VENDOR_ID_NVIDIA 0x10de
@ -2531,7 +2532,11 @@ static vk_buffer ggml_vk_create_buffer_device(vk_device& device, size_t size) {
vk::MemoryPropertyFlagBits::eDeviceLocal});
} else if (device->uma) {
// Fall back to host memory type
buf = ggml_vk_create_buffer(device, size, {vk::MemoryPropertyFlagBits::eDeviceLocal,
// 1. Prefer Device Local AND Host Visible (Best for UMA)
// 2. Fallback to Device Local
// 3. Fallback to Host Visible
buf = ggml_vk_create_buffer(device, size, {vk::MemoryPropertyFlagBits::eDeviceLocal | vk::MemoryPropertyFlagBits::eHostVisible | vk::MemoryPropertyFlagBits::eHostCoherent,
vk::MemoryPropertyFlagBits::eDeviceLocal,
vk::MemoryPropertyFlagBits::eHostVisible | vk::MemoryPropertyFlagBits::eHostCoherent});
} else if (device->disable_host_visible_vidmem) {
if (device->allow_sysmem_fallback) {
@ -2543,6 +2548,9 @@ static vk_buffer ggml_vk_create_buffer_device(vk_device& device, size_t size) {
} else {
// use rebar if available, otherwise fallback to device only visible memory
if (device->allow_sysmem_fallback) {
// 1. Prefer Device Local AND Host Visible (ReBAR)
// 2. Fallback to Device Local
// 3. Fallback to Host Visible
buf = ggml_vk_create_buffer(device, size, {vk::MemoryPropertyFlagBits::eDeviceLocal | vk::MemoryPropertyFlagBits::eHostVisible | vk::MemoryPropertyFlagBits::eHostCoherent,
vk::MemoryPropertyFlagBits::eDeviceLocal,
vk::MemoryPropertyFlagBits::eHostVisible | vk::MemoryPropertyFlagBits::eHostCoherent});
@ -2933,7 +2941,15 @@ static void ggml_vk_load_shaders(vk_device& device) {
s_warptile_mmqid_int_k = { mul_mat_subgroup_size_32, 32, 32, 32, 32, 32, 1, 2, 1, 1, mul_mat_subgroup_size_16 };
// chip specific tuning
if ((device->architecture == AMD_GCN) && (device->driver_id != vk::DriverId::eAmdProprietary)) {
if (device->vendor_id == VK_VENDOR_ID_ARM) {
m_warptile_mmq = m_warptile_mmq_int = { 64, 64, 64, 16, 16, 32, 2, 2, 2, 1, 16 };
m_warptile = { 64, 64, 64, 16, 16, 32, 2, 2, 2, 1, 16 };
m_warptile_id = m_warptile_mmqid = { 64, 64, 64, 16, 16, 32, 2, 2, 2, 1, 16 };
l_warptile_mmq = l_warptile_mmq_int = m_warptile_mmq;
l_warptile = m_warptile;
l_warptile_id = l_warptile_mmqid = m_warptile_id;
} else if ((device->architecture == AMD_GCN) && (device->driver_id != vk::DriverId::eAmdProprietary)) {
m_warptile_mmq = m_warptile_mmq_int = { 256, 64, 64, 32, 16, 16, 2, 2, 2, 1, 16 };
m_warptile_mmqid = m_warptile_mmqid_int = { 256, 64, 64, 32, 16, 16, 2, 2, 2, 1, 16 };
}
@ -4503,6 +4519,11 @@ static vk_device ggml_vk_get_device(size_t idx) {
device->vendor_id = device->properties.vendorID;
device->driver_id = driver_props.driverID;
if (device->vendor_id == VK_VENDOR_ID_ARM) {
// Previously forced FP32 here due to poor FP16 performance on some ARM GPUs.
// With adjusted l_warptile (below), FP16 is now performant and preferred.
}
// Implementing the async backend interfaces seems broken on older Intel HW,
// see https://github.com/ggml-org/llama.cpp/issues/17302.
device->support_async = (device->vendor_id != VK_VENDOR_ID_INTEL ||
@ -4537,6 +4558,9 @@ static vk_device ggml_vk_get_device(size_t idx) {
if (GGML_VK_SUBALLOCATION_BLOCK_SIZE != nullptr) {
device->suballocation_block_size = std::stoull(GGML_VK_SUBALLOCATION_BLOCK_SIZE);
} else if (device->vendor_id == VK_VENDOR_ID_ARM) {
// Limit batching of allocations to 256MB on Mali GPUs to avoid fragmentation issues
device->suballocation_block_size = 256 * 1024 * 1024;
} else {
// Limit batching of allocations to 1GB by default to avoid fragmentation issues
device->suballocation_block_size = 1024*1024*1024;