OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

anv: Bump the number of update-after-bind descriptors to 1M

Browse Source 
It's a bit hard to exactly map our implementation to the limits
described by Vulkan.  The bindless surface handle in the extended
message descriptors is 20 bits and it's an index into the table of
RENDER_SURFACE_STATE structs that starts at bindless surface base
address.  This means that we can have at must 1M surface states
allocated at any given time.  Since most image views take two
descriptors, this means we have a limit of about 500K image views.

However, since we allocate surface states at vkCreateImageView time,
this means our limit is actually something on the order of 500K image
views allocated at any time.  The actual limit describe by Vulkan, on
the other hand, is a limit of how many you can have in a descriptor set.
Assuming anyone using 1M descriptors will be using the same image view
twice a bunch of times (or a bunch of null descriptors), we can safely
advertise a larger limit.  1M is what's required by D3D12, so let's
advertise that.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3335
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7180>
This commit is contained in:
Jason Ekstrand
2020-10-15 20:33:12 -05:00
committed by Marge Bot
parent 3c0ba7b180
commit dd1971c9d8
1 changed files with 12 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
src/intel/vulkan/anv_device.c
View File

@@ -1744,17 +1744,22 @@ anv_get_physical_device_properties_1_2(struct anv_physical_device *pdevice,
p->shaderSignedZeroInfNanPreserveFloat64 = true;
/* It's a bit hard to exactly map our implementation to the limits
* described here. The bindless surface handle in the extended
* described by Vulkan. The bindless surface handle in the extended
* message descriptors is 20 bits and it's an index into the table of
* RENDER_SURFACE_STATE structs that starts at bindless surface base
* address. Given that most things consume two surface states per
* view (general/sampled for textures and write-only/read-write for
* images), we claim 2^19 things.
* address. This means that we can have at must 1M surface states
* allocated at any given time. Since most image views take two
* descriptors, this means we have a limit of about 500K image views.
*
* For SSBOs, we just use A64 messages so there is no real limit
* there beyond the limit on the total size of a descriptor set.
* However, since we allocate surface states at vkCreateImageView time,
* this means our limit is actually something on the order of 500K image
* views allocated at any time. The actual limit describe by Vulkan, on
* the other hand, is a limit of how many you can have in a descriptor set.
* Assuming anyone using 1M descriptors will be using the same image view
* twice a bunch of times (or a bunch of null descriptors), we can safely
* advertise a larger limit here.
*/
const unsigned max_bindless_views = 1 << 19;
const unsigned max_bindless_views = 1 << 20;
p->maxUpdateAfterBindDescriptorsInAllPools = max_bindless_views;
p->shaderUniformBufferArrayNonUniformIndexingNative = false;
p->shaderSampledImageArrayNonUniformIndexingNative = false;