Skip to content

Improve AMD GCN detection fallback #4

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
rasbid wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Improve AMD GCN detection fallback #4

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
65 changes: 55 additions & 10 deletions ggml/src/ggml-vulkan/ggml-vulkan.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -237,7 +237,24 @@ enum vk_device_architecture {
NVIDIA_PRE_TURING,
};

static vk_device_architecture get_device_architecture(const vk::PhysicalDevice& device) {
static const char * vk_device_architecture_name(vk_device_architecture arch) {
switch (arch) {
case vk_device_architecture::AMD_GCN: return "AMD GCN";
case vk_device_architecture::AMD_RDNA1: return "AMD RDNA1";
case vk_device_architecture::AMD_RDNA2: return "AMD RDNA2";
case vk_device_architecture::AMD_RDNA3: return "AMD RDNA3";
case vk_device_architecture::INTEL_XE2: return "Intel Xe2";
case vk_device_architecture::NVIDIA_PRE_TURING: return "NVIDIA pre-Turing";
case vk_device_architecture::OTHER: return "Other";
}
return "Other";
}

static vk_device_architecture get_device_architecture(const vk::PhysicalDevice& device, bool *missing_subgroup_size_control = nullptr) {
if (missing_subgroup_size_control != nullptr) {
*missing_subgroup_size_control = false;
}

vk::PhysicalDeviceProperties props = device.getProperties();

if (props.vendorID == VK_VENDOR_ID_AMD) {
Expand All @@ -257,7 +274,7 @@ static vk_device_architecture get_device_architecture(const vk::PhysicalDevice&
}
}

if (!amd_shader_core_properties || !integer_dot_product || !subgroup_size_control) {
if (!amd_shader_core_properties) {
return vk_device_architecture::OTHER;
}

Expand All @@ -267,24 +284,48 @@ static vk_device_architecture get_device_architecture(const vk::PhysicalDevice&
vk::PhysicalDeviceSubgroupSizeControlPropertiesEXT subgroup_size_control_props;

props2.pNext = &shader_core_props_amd;
shader_core_props_amd.pNext = &integer_dot_props;
integer_dot_props.pNext = &subgroup_size_control_props;
shader_core_props_amd.pNext = nullptr;

device.getProperties2(&props2);
VkBaseOutStructure *last_struct = (VkBaseOutStructure *)&shader_core_props_amd;

if (subgroup_size_control_props.maxSubgroupSize == 64 && subgroup_size_control_props.minSubgroupSize == 64) {
return vk_device_architecture::AMD_GCN;
if (integer_dot_product) {
integer_dot_props.pNext = nullptr;
last_struct->pNext = (VkBaseOutStructure *)&integer_dot_props;
last_struct = (VkBaseOutStructure *)&integer_dot_props;
}
if (subgroup_size_control_props.maxSubgroupSize == 64 && subgroup_size_control_props.minSubgroupSize == 32) {

if (subgroup_size_control) {
subgroup_size_control_props.pNext = nullptr;
last_struct->pNext = (VkBaseOutStructure *)&subgroup_size_control_props;
last_struct = (VkBaseOutStructure *)&subgroup_size_control_props;
} else {
last_struct->pNext = nullptr;
}

device.getProperties2(&props2);

if (subgroup_size_control) {
if (subgroup_size_control_props.maxSubgroupSize == 64 && subgroup_size_control_props.minSubgroupSize == 64) {
return vk_device_architecture::AMD_GCN;
}
if (subgroup_size_control_props.maxSubgroupSize == 64 && subgroup_size_control_props.minSubgroupSize == 32) {
// RDNA
if (shader_core_props_amd.wavefrontsPerSimd == 20) {
return vk_device_architecture::AMD_RDNA1;
}
if (integer_dot_props.integerDotProduct4x8BitPackedMixedSignednessAccelerated) {
if (integer_dot_product && integer_dot_props.integerDotProduct4x8BitPackedMixedSignednessAccelerated) {
return vk_device_architecture::AMD_RDNA3;
}
return vk_device_architecture::AMD_RDNA2;
}
}

if (shader_core_props_amd.wavefrontSize == 64) {
if (missing_subgroup_size_control != nullptr && !subgroup_size_control) {
*missing_subgroup_size_control = true;
}
return vk_device_architecture::AMD_GCN;
}
} else if (props.vendorID == VK_VENDOR_ID_INTEL) {
const std::vector<vk::ExtensionProperties> ext_props = device.enumerateDeviceExtensionProperties();

Expand Down Expand Up @@ -415,6 +456,7 @@ struct vk_device_struct {
int32_t mmvq_mode;

bool subgroup_size_control;
bool subgroup_size_control_missing;
uint32_t subgroup_min_size;
uint32_t subgroup_max_size;
bool subgroup_require_full_support;
Expand Down Expand Up @@ -3685,7 +3727,10 @@ static vk_device ggml_vk_get_device(size_t idx) {
device->physical_device = physical_devices[dev_num];
const std::vector<vk::ExtensionProperties> ext_props = device->physical_device.enumerateDeviceExtensionProperties();

device->architecture = get_device_architecture(device->physical_device);
device->architecture = get_device_architecture(device->physical_device, &device->subgroup_size_control_missing);
GGML_LOG_INFO("ggml_vulkan: detected architecture %s%s\n",
vk_device_architecture_name(device->architecture),
device->subgroup_size_control_missing ? " (VK_EXT_subgroup_size_control unavailable)" : "");

const char* GGML_VK_PREFER_HOST_MEMORY = getenv("GGML_VK_PREFER_HOST_MEMORY");
device->prefer_host_memory = GGML_VK_PREFER_HOST_MEMORY != nullptr;
Expand Down