fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-07 00:38:48 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

panvk: Use per-queue shader core count for CSF group creation

Browse source 
Extract VkDeviceQueueShaderCoreControlCreateInfoARM from the queue
create info pNext chain and use shaderCoreCount to limit
max_compute_cores and max_fragment_cores in the panthor group create
ioctl. The core masks remain unchanged, letting the kernel pick which
cores to schedule on.

Signed-off-by: Christian Gmeiner <cgmeiner@igalia.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40063>
This commit is contained in:
Christian Gmeiner 2026-02-24 09:14:16 +01:00 • committed by Marge Bot
parent a8ba682919
commit fff1410c8a
1 changed files with 18 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

22
src/panfrost/vulkan/csf/panvk_vX_gpu_queue.c
View file

@ -615,7 +615,8 @@ err_cleanup_queue:
static VkResult
create_group(struct panvk_gpu_queue *queue,
enum drm_panthor_group_priority group_priority)
enum drm_panthor_group_priority group_priority,
uint32_t shader_core_count)
{
const struct panvk_device *dev = to_panvk_device(queue->vk.base.device);
const struct panvk_physical_device *phys_dev =
@ -639,12 +640,20 @@ create_group(struct panvk_gpu_queue *queue,
},
};
uint8_t max_compute_cores = util_bitcount64(phys_dev->compute_core_mask);
uint8_t max_fragment_cores = util_bitcount64(phys_dev->fragment_core_mask);
if (shader_core_count) {
max_compute_cores = MIN2(shader_core_count, max_compute_cores);
max_fragment_cores = MIN2(shader_core_count, max_fragment_cores);
}
struct drm_panthor_group_create gc = {
.compute_core_mask = phys_dev->compute_core_mask,
.fragment_core_mask = phys_dev->fragment_core_mask,
.tiler_core_mask = 1,
.max_compute_cores = util_bitcount64(phys_dev->compute_core_mask),
.max_fragment_cores = util_bitcount64(phys_dev->fragment_core_mask),
.max_compute_cores = max_compute_cores,
.max_fragment_cores = max_fragment_cores,
.max_tiler_cores = 1,
.priority = group_priority,
.queues = DRM_PANTHOR_OBJ_ARRAY(ARRAY_SIZE(qc), qc),
@ -1373,7 +1382,12 @@ panvk_per_arch(create_gpu_queue)(struct panvk_device *dev,
if (result != VK_SUCCESS)
goto err_destroy_syncobj;
result = create_group(queue, get_panthor_group_priority(create_info));
const VkDeviceQueueShaderCoreControlCreateInfoARM *core_ctrl =
vk_find_struct_const(create_info->pNext,
DEVICE_QUEUE_SHADER_CORE_CONTROL_CREATE_INFO_ARM);
result = create_group(queue, get_panthor_group_priority(create_info),
core_ctrl ? core_ctrl->shaderCoreCount : 0);
if (result != VK_SUCCESS)
goto err_cleanup_tiler;