58d8dc9543
r600/sfn: dump the lowered shader when translation fails
...
So far we printed the variant before the final lowering was done,
this is usually not that helpful. For this the code to dump the
shader has to go into r600_shader_from_nir.
Signed-off-by: Gert Wollny <gert.wollny@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34999 >
2025-05-16 08:45:58 +00:00
2ce2c1f835
zink/ci: Skip flaky trace on TGL
...
Signed-off-by: Valentine Burley <valentine.burley@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35016 >
2025-05-16 08:00:11 +00:00
cc4fd7cc9d
vulkan: fix random tabs to spaces
...
Only the vk_time_max_deviation tab usage requires a manual fix to make
sense. Others are boring replacements.
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35011 >
2025-05-16 03:57:31 +00:00
360d252888
amd/vpelib: Fix CodeQL issues Pt1
...
[WHY]
Redundant / inconsisten nullptr checks
Reviewed-by: Jude Shih <Jude.Shih@amd.com >
Reviewed-by: Evan Damphousse <Evan.Damphousse@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
134b5bede7
amd/vpelib: Update headers
...
Update new headers
Reviewed-by: Jude Shih <Jude.Shih@amd.com >
Reviewed-by: Ricky Lin <Ricky.Lin@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Tomson Chang <tomson.chang@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
ffa5aadd2f
amd/vpelib: fix doxgen warnings
...
What does the change do after sanization if any.
Delete this section if not applicable.
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Roy Chan <roy.chan@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
80af9d4756
amd/vpelib: Fix VPELIB Build Warnings
...
Fix build warnings.
Signed-off-by: Iswara Nagulendran <Iswara.Nagulendran@amd.com >
Reviewed-by: Roy Chan <Roy.Chan@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
71b285d091
amd/vpelib: Resolve query, predication and timestamp operations
...
[HOW]
Some clients of vpelib requires additional operations to be implemented.
Implemented support for new firmware operations.
Reviewed-by: Roy Chan <Roy.Chan@amd.com >
Reviewed-by: Jesse Agate <Jesse.Agate@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Okenczyc, Andrzej <andrzej.okenczyc@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
f10244d8d9
amd/vpelib: Setup cost profiling support
...
Add logs to profile registers writes.
Reviewed-by: Roy Chan <Roy.Chan@amd.com >
Reviewed-by: Jesse Agate <Jesse.Agate@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Nagulendran, Iswara <Iswara.Nagulendran@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
8bdbd12508
amd/vpelib: add TF_BT709
...
[HOW]
Add VPE_SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb -> VPE_TF_BT709 in
vpe_visual_confirm.c
Reviewed-by: Tomson Chang <Tomson.Chang@amd.com >
Reviewed-by: Min-Hsuan You <Min-Hsuan.You@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Han, Szu Chih <SzuChih.Han@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
3bd03b3cb0
amd/vpelib: add VPE_TF_SRGB
...
[HOW]
Add VPE_SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb -> VPE_TF_SRGB in
vpe_visual_confirm.c
Co-authored-by: Mike Han <szuchhan@amd.com >
Reviewed-by: Tomson Chang <Tomson.Chang@amd.com >
Reviewed-by: Min-Hsuan You <Min-Hsuan.You@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Han, Szu Chih <SzuChih.Han@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
7b68174bf0
amd/vpelib: Alpha Fill add mode
...
Add destination and source stream mode
[WHY]
In the new version, we add the two alpha fill mode
[HOW]
Add the two enum first
Reviewed-by: Brendan Steve Leder <BrendanSteve.Leder@amd.com >
Reviewed-by: Tomson Chang <Tomson.Chang@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Jude Shih <shenshih@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
cb0b8d58a0
amd/vpelib: remove sensitive keywords
...
[WHY]
The KWS hit the words in the vpelib as forbidden
[HOW]
replace it with proper ones
Reviewed-by: Tomson Chang <Tomson.Chang@amd.com >
Reviewed-by: Tiberiu Visan <Tiberiu.Visan@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Jude Shih <shenshih@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
c40ac921aa
amd/vpelib: VPE descriptor added
...
[WHY]
- Need to implement code for new version
- There are some format changes in new version
[HOW]
- Add plane descriptor
- Add plane descriptor patch
- Add command builder
- Restructure the code base to apply for multiple version
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com >
Reviewed-by: Tomson Chang <Tomson.Chang@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Jude Shih <shenshih@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
f80a69d756
amd/vpelib: Add macro to perform ceil division
...
Add macro to perform ceil division
Reviewed-by: Brendan Steve Leder <BrendanSteve.Leder@amd.com >
Reviewed-by: Navid Assadian <Navid.Assadian@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Agate, Jesse <Jesse.Agate@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
7b8c912c22
amd/vpelib: G2084 STUDIO output gamma fix
...
Made output gamma a hook function
Reviewed-by: Jesse Agate <Jesse.Agate@amd.com >
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Zhao, Jiali <Jiali.Zhao@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
fc5973429f
amd/vpelib: Update comment and fix some code alignment issues.
...
Update comment and fix some code alignment issues.
Reviewed-by: Tomson Chang <Tomson.Chang@amd.com >
Reviewed-by: Evan Damphousse <Evan.Damphousse@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
96b7e52b57
amd/vpelib: feature augment
...
[WHY]
support future feature augment
Co-authored-by: Chan, Roy <Roy.Chan@amd.com >
Reviewed-by: Navid Assadian <Navid.Assadian@amd.com >
Reviewed-by: Jesse Agate <Jesse.Agate@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Visan, Tiberiu <Tiberiu.Visan@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
f6144116c9
amd/vpelib: Remove RGBE_ALPHA
...
No such DXGI_FORMAT, nor is there space to squeeze in alpha.
Reviewed-by: Navid Assadian <Navid.Assadian@amd.com >
Reviewed-by: Jesse Agate <Jesse.Agate@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
f2e65c02e6
amd/vpelib: Fix memory leak from segment ctx
...
[WHY]
There is potential memory leak from vpe_alloc_segment_ctx.
This memory leak occurs only in multi-frame VPE tests where
between vpe_create and vpe_destroy, multiple calls are made
to vpe_check_support that allocates new segment context without
releasing the old one.
[HOW]
Allocate segment_ctx only when it is not already allocated. If it is
already allocated, check whether re-allocation is needed. If not, skip
the allocation.
Signed-off-by: Phoebe Chen <phoebe.chen@amd.com >
Reviewed-by: Roy Chan <Roy.Chan@amd.com >
Reviewed-by: Jesse Agate <Jesse.Agate@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
c1a3d021f3
amd/vpelib: Make BG gen stream idx adjustable
...
Update BKGR API
Change bg generation code so bg gen isn't hard-coded to stream 0, as
certain cases result in bg being generated at different stream.
Reviewed-by: Roy Chan <Roy.Chan@amd.com >
Reviewed-by: Evan Damphousse <Evan.Damphousse@amd.com >
Acked-by: ChuanYu Tseng <ChuanYu.Tseng@amd.com >
Signed-off-by: Leder, Brendan Steve <BrendanSteve.Leder@amd.co >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35012 >
2025-05-16 11:33:08 +08:00
5b34895e81
nak/dce: Use BitSet for live phis and SSA values
...
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
52dcb1c9c1
nak: Add a Phi struct type
...
Instead of passing raw u32's arround, this adds a new Phi wrapper struct
which is treated as opaque by most of the rest of the compiler. This is
similar to what we're already doing with Label and SSAValue. This also
gives us the opportunity to properly document NAK's phi model.
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
531070cf91
nak: Use op instead of phi when referring to OpPhiSrcs or OpPhiDsts
...
This will reduce name conflicts in the next patch
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
5f5ce00410
nak/repair_ssa: Rename Phi to PhiTracker
...
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
fc54e6e1d2
nak/liveness: Use typed bitsets
...
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
afb11a5a9e
nak/liveness: Use an SSA value as the last_use HashMap key
...
We're not saving ourselves anything by using just the index.
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
56f06a072c
compiler/rust/bitset: Make BitSet a generic container type
...
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:28:31 -04:00
323769d3bb
compiler/rust,nak: Rename BitSet::get() to contains()
...
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:14:12 -04:00
20d247d754
nak: Use Default::default() for BitSet constructors
...
Reviewed-by: Mel Henning <mhenning@darkrefraction.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34994 >
2025-05-15 22:14:12 -04:00
ed658a2d3a
compiler/rust/cfg: Add a hash func type parameter
...
and switch nak over to rustc-hash for CFGBuilder
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:05 +00:00
f5231e0677
nak: Also use rustc-hash for UnionFind
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:04 +00:00
43c3f5a8db
nak: Switch most Hash{Set,Map} uses to rustc-hash
...
We shouldn't need to worry about HashDoS attacks because:
1. All of these keys are compiler-assigned IDs which makes it very
difficult to force collitions, and
2. Anyone who can hit the NAK compiler backend can already use CPU
power by spamming shader compiles or using O(n^2) behavior in
shader opt loops
As a result, we can afford to use a weaker hash function without
randomization.
This decreases total compile times by around 12% on shaderdb
(from 8113.74 user to 7115.47 user) on my machine.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:04 +00:00
c1f979d84a
nak: Add a dependency on rustc-hash
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:04 +00:00
b4b557c9ef
nak: s/HashSet::new()/Default::default()/
...
wherever this doesn't result in type inference failing.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:04 +00:00
7a47f29d6d
nak: s/HashMap::new()/Default::default()/
...
wherever this doesn't result in type inference failing.
Using default() makes it easier to swap out the underlying type.
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:04 +00:00
bb62044836
nak: rustfmt ir_proc.rs
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34865 >
2025-05-16 01:40:04 +00:00
bfe8340296
nouveau/headers: Ignore PermissionError in rustfmt
...
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13172
Fixes: 591b5da4https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35008 >
2025-05-16 01:08:07 +00:00
da22094593
nouveau/headers: Run rustfmt after file is closed
...
If we run a subprocess while the file is still open, we may not have
flushed the file contents to disk.
Fixes: 591b5da4https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35008 >
2025-05-16 01:08:07 +00:00
20222cd956
anv: Use the new nir_opt_acquire_release_barriers pass
...
Improves performance of Phasmophobia with the "Eye Adaptation" video
setting enabled on Arc B570 by about 9.5%.
fossil-db results on Battlemage:
Totals:
Instrs: 148797922 -> 148797865 (-0.00%)
Send messages: 7066341 -> 7066317 (-0.00%)
Cycle count: 21459978352 -> 21459975048 (-0.00%)
Totals from 8 (0.00% of 574410) affected shaders:
Instrs: 4633 -> 4576 (-1.23%)
Send messages: 479 -> 455 (-5.01%)
Cycle count: 611886 -> 608582 (-0.54%)
Observed to cut 15% of sends in a Phasmophobia shader, 8.3% in a Far Cry
New Dawn shader, 7% in a Borderlands 3 DX11 shader, and 3.4-3.7% of
sends in a few Witcher 3 and Dark Souls 3 shaders.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33504 >
2025-05-16 00:29:13 +00:00
deb1d47155
nir: Add a new optimization for acquire/release atomics & barriers
...
Some shaders contain back-to-back atomic accesses in SPIR-V with
AcquireRelease semantics. In NIR, we translate these to a release
memory barrier, the atomic, then an acquire memory barrier.
This results in a lot of unnecessary memory barriers in the middle
of the sequence of atomics:
0. Release barrier
1. Atomic
2. Acquire barrier
3. Release barrier
4. Atomic
5. Acquire barrier
6. Release barrier
7. Atomic
8. Acquire barrier
In the absence of loads/stores, and when the atomic destinations are
unused, these barriers in-between atomics shouldn't be required.
This optimization pass would drop them (lines 2-3 and 5-6 above) while
leaving the first and last barriers (0 and 8), so the sequence remains
synchronized against other access elsewhere in the program.
One common example where this occurs is a sequence of min and max
atomics to clamp a certain memory location's value within a range.
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33504 >
2025-05-16 00:29:13 +00:00
65e18a8494
freedreno: Fix shader-clock when kernel exposes UCHE_TRAP_BASE
...
Fixes: 4b1b4ee10crobdclark@chromium.org >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35010 >
2025-05-15 22:27:17 +00:00
089e2cb6f9
radeonsi: Disable av1 cdef_channel_strength for VCN4
...
VCN4 hardware doesn't support this feature, it can only be supported in VCN5.
Signed-off-by: Yinjie Yao <yinjie.yao@amd.com >
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35004 >
2025-05-15 18:52:08 +00:00
10fad5081d
nouveau: implement Default for Push
...
By convention, a struct with a `new()` method which has no parameters
should have a `Default` impl which calls `new()`.
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
f4f4b25d25
nak,nil: style cleanup
...
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
adecea4af9
nak,nouveau: adjust function/method signatures to better match convention
...
v2: restore `to_cssa()` naming
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
e559c63fd8
nak,nil: elide lifetimes where possible
...
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
e4f045df58
nak,nil: avoid explicit returns at the end of functions
...
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
e32c82d0f5
nak: use standard methods and macros to improve readability
...
v2: Leave `Op::is_branch()` and `Op::no_scoreboard()` matches alone
v3: Revert additional changes with unclear readability
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
ba2b9345e8
nak: use Option propagation instead of explicit let-else clauses
...
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34372 >
2025-05-15 17:52:32 +00:00
Gert Wollny