fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-04 01:08:03 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

intel: Fix calculation of max_scratch_ids on fused devices
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
Details
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Details

Browse source 
The subslice IDs provided by the SR0.0 EU register are not adjusted to account
for fusing, so the upper bound max_scratch_ids can vary from device to device
depending on what specific slices were fused during manufacturing.

Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38689>
This commit is contained in:
Calder Young 2025-11-26 20:21:57 -08:00 • committed by Marge Bot
parent d86f5f6bcb
commit c0d809820f
1 changed files with 2 additions and 29 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

31
src/intel/dev/intel_device_info.c
View file

@ -1698,37 +1698,10 @@ intel_device_info_adjust_memory(struct intel_device_info *devinfo)
static void
init_max_scratch_ids(struct intel_device_info *devinfo)
{
/* Determine the max number of subslices that potentially might be used in
/* Determine the max subslice that potentially might be used in
* scratch space ids.
*
* For, Gfx11+, scratch space allocation is based on the number of threads
* in the base configuration.
*
* For Gfx9, devinfo->subslice_total is the TOTAL number of subslices and
* we wish to view that there are 4 subslices per slice instead of the
* actual number of subslices per slice. The documentation for 3DSTATE_PS
* "Scratch Space Base Pointer" says:
*
* "Scratch Space per slice is computed based on 4 sub-slices. SW
* must allocate scratch space enough so that each slice has 4
* slices allowed."
*
* According to the other driver team, this applies to compute shaders
* as well. This is not currently documented at all.
*
* For Gfx8 and older we user devinfo->subslice_total.
*/
unsigned subslices;
if (devinfo->verx10 == 125)
subslices = 32;
else if (devinfo->ver == 12)
subslices = (devinfo->platform == INTEL_PLATFORM_DG1 || devinfo->gt == 2 ? 6 : 2);
else if (devinfo->ver == 11)
subslices = 8;
else if (devinfo->ver >= 9 && devinfo->ver < 11)
subslices = 4 * devinfo->num_slices;
else
subslices = devinfo->subslice_total;
unsigned subslices = intel_device_info_dual_subslice_id_bound(devinfo);
assert(subslices >= devinfo->subslice_total);
unsigned scratch_ids_per_subslice;