Since GDI doesn't support this format, we need a fallback path to
get contents on-screen if we're not using DXGI. For that scenario,
we allocate a proxy display target and blit during frontbuffer flush.
Once we have that fallback in place, we can override the sw winsys
format support check for that format.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27512>
When the frontend asks for the front buffer, we return an offscreen
surface, when the frontend asks us to flush that surface, we copy it
to the swapchain, and then present remains unchanged.
This means the only reason we need to avoid using a swapchain is if
the app asks for GDI compatibility.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27512>
We already implement all of the relevant bits of this extension, which
is just handling WGL_TYPE_RGBA_FLOAT_ARB for WGL_PIXEL_TYPE_ARB in
wglGetPixelFormatAttrib[i/f]vARB and wglChoosePixelFormatARB. We just
didn't have any float formats enumerated so it was never seen in
practice.
Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Reviewed-by: Jose Fonseca <jose.fonseca@broadcom.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27512>
Some apps don't look at what kind of PFDthey get, and if they
get a single-buffered one, they only ever call swap and never flush,
so nothing shows up on-screen.
Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Reviewed-by: Jose Fonseca <jose.fonseca@broadcom.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27512>
It seems like an L3 setting is different between DG2 & MTL, breaking
the L3 coherency with CS.
We can apply the same tile cache flush as we do on TGL for now until
we figure out how to configure L3 properly to avoid this.
Tested with :
$ query_clear_with_blorp_threshold=0 query_copy_with_shader_threshold=0 ./deqp-vk -n dEQP-VK.query_pool.*
to force the shader optimizations.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27597>
Turns out we have to do it regardless as opts after explicit_types could
get rid of even more scratch and shared memory.
Fixes: ea023ff5cd ("rusticl/kernel: no need to reset the scratch size anymore")
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27634>
Abuse void* here so there's one less file dependent on specific brw
types. We can revisit if there's a need later, if we ever end up with
a common type for prog_data.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27581>
Allocate the prog_data instead of expecting one from the user, also
explicitly return both kernel and prog_data size, so that the
plumbing code isn't required to use the exact prog_data type.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27581>
Remove the brw_ prefix from some blorp functions and structs to avoid
confusion with what's related to brw and what's not. Some of those will
become independent of brw in upcoming patches.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27581>
Make interface less dependent on brw types. If we care, later
might make sense to add a tagged union for the possible types here.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27581>
Even though we set .lower_flrp32 = true there is no actuall flrp
lowering in nir_opt_algebraic. Mesa does not produce any but nine does,
so lower it explicitly to fix nine.
Fixes: f8a5cba3b4
Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Reviewed-by: Filip Gawin <filip.gawin@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27467>
And move them inside the compiler since they (especially asm) rely on
a bunch of internal types.
Acked-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27579>
HLSL docs say WaveReadLaneAt is undefined if the target lane is inactive. This makes
sense since the target lane may need to *send* the data, rather than it being pulled
by the calling lane. So don't early-out on the loop, iterate through the whole wave
on all threads and read the cross-lane data before branching.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27624>
We should never get here currently as the parser should not even
process float16_t without half float enabled. However it seems
like a good idea to add this for completeness.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27585>
1. avoids cluttering shaders,internal print
2. reduces screen create overhead. this cuts average "glxinfo" runtime by about
25%, and seems to shave maybe 0.5% off the CTS job.
3. enforces helper program builds successfully as a build-time assertion
4. reduces the libagx generated blob for now, since g13 binaries are more
compact than serialized NIR
Closes https://gitlab.freedesktop.org/asahi/mesa/-/issues/37
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27616>
