203415 commits

Author	SHA1	Message	Date
Alyssa Rosenzweig	edccd06a0b	nir/lower_subgroups: fix boolean clustered reductions ... It is legal to have a cluster size larger than the subgroup/ballot size, but our lowering would blow up in this case due to the nir_ishl_imm overflowing in the lowering. Fortunately, this is easy to handle. Fixes sub_group_clustered_reduce_logical_and() Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Reviewed-by: Karol Herbst <kherbst@redhat.com> Reviewed-by: Georg Lehmann <dadschoorse@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40224>	2026-03-09 14:50:37 +00:00
Aitor Camacho	97756219c4	kk: Set command buffer state to 0 when reset ... nvk and hk were setting the state to 0, but we weren't which led to issues with the state being carried over after resetting. Signed-off-by: Aitor Camacho <aitor@lunarg.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40193>	2026-03-09 13:37:09 +00:00
Eric R. Smith	8521051cfa	pco: fix a typo in the check for optimization looping ... The count isn't incremented anywhere else. Signed-off-by: Eric R. Smith <eric.smith@collabora.com> Reviewed-by: Simon Perretta <simon.perretta@imgtec.com> Fixes: f1b24267d2 ("pco: rework nir processing and passes") Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40240>	2026-03-09 11:27:27 +00:00
Pavel Ondračka	ce33f82f83	r300: disable clip-discard watermark for triangles ... Commit 0d4aa5f55f introduced the watermark to optimize the guardband state changes and always computed new_distance as MAX2(distance, watermark). That is correct for point/line paths where distance > 0, but it keeps a non-zero discard distance alive when the next draw sets distance = 0 (triangles). This leaks wide point/line clip-discard state into later triangle draws and can clip away large parts of geometry (as observed in Sauerbraten). Only apply the watermark when distance > 0 and reset it to zero otherwise so triangle draws disable clip-discard as intended. Fixes: 0d4aa5f55f ("r300: pop-free clipping") Reviewed-by: Marek Olšák <marek.olsak@amd.com> Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/14959 Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40223>	2026-03-09 10:18:51 +00:00
Samuel Pitoiset	8f10e2886c	radv: dump the PS epilog in the GPU hang report ... Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40219>	2026-03-09 09:53:27 +00:00
Samuel Pitoiset	4c7958ae6c	radv: apply the 1D workgroup optimization for mesh/task shaders too ... Suggested-by: Georg Lehmann <dadschoorse@gmail.com> Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40194>	2026-03-09 09:29:35 +00:00
Samuel Pitoiset	3c4cb16159	radv: fix local invocation index for mesh/task and quad derivatives on GFX12 ... It must be lowered. This fixes dEQP-VK.spirv_assembly.instruction.compute.compute_shader_derivatives.{mesh,task}.*. Cc: mesa-stable Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40194>	2026-03-09 09:29:34 +00:00
Samuel Pitoiset	4428541c54	radv/meta: fix HTILE fixup after copying depth/stencil image copies ... Typo, it should be false because it's after the copy. Fixes: 4f41818194 ("radv/meta: add a function to fixup HTILE metadata for copies on compute queue") Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40270>	2026-03-09 09:07:09 +00:00
Samuel Pitoiset	fff16a9748	radv: replace radv_sdma_surf by ac_sdma_surf ... Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:47 +00:00
Samuel Pitoiset	c40225e490	radv: tidy up radv_sdma_surf ... Adjust few things before replacing it by ac_sdma_surf. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:47 +00:00
Samuel Pitoiset	0616fd22a5	radv: simplify getting bpe for SDMA surfaces ... Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:46 +00:00
Samuel Pitoiset	9893ac3674	radv: remove unnecessary radv_sdma_surf::{blk_w,blk_h} ... Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:46 +00:00
Samuel Pitoiset	94acb7edd5	radv: simplify computing offset/extent of SDMA surfaces ... By computing in elements earlier. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:45 +00:00
Samuel Pitoiset	5923a7b8c6	radv: use vk_image_buffer_copy_layout() for SDMA buf layout ... For consistency with non-SDMA paths. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:45 +00:00
Samuel Pitoiset	02d047099e	radv: simplify 96-bit copies with SDMA ... By adjusting offset/extent earlier. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:45 +00:00
Samuel Pitoiset	6f3b9a62b3	radv: remove redundant radv_sdma_surf::is_linear ... is_linear is never used for buffers. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:44 +00:00
Samuel Pitoiset	dba9809e0c	radv: remove redundant radv_sdma_surf::is_3d ... Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40186>	2026-03-09 08:40:44 +00:00
Samuel Pitoiset	1a00587c44	radv: fix a GPU hang with PS epilogs and secondary command buffers ... If the secondary changes the fragment output state and if the same PS epilog used before ExecuteCommands() is re-bind immediately after that call, the PS epilog state wouldn't be re-emitted. Apply the same change for VS prologs, although the logic is slightly different and the bug shouldn't occur. The whole logic of secondaries should be completely rewritten because it's definitely not robust. This fixes a GPU hang in Where Winds Meet, see https://github.com/doitsujin/dxvk/issues/5436. Cc: mesa-stable Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40221>	2026-03-09 08:16:49 +00:00
Samuel Pitoiset	ac3fd06987	radv: always enable DISABLE_CONSERVATIVE_ZPASS_COUNTS on GFX11 ... This might cause incorrect occlusion queries count. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40235>	2026-03-09 07:26:25 +00:00
Kenneth Graunke	952bf55483	nir: Fix divergence of Intel URB input/output handle intrinsics ... Tessellation evaluation shaders have a single convergent URB handle (for the common patch data) used by all lanes. Every other stage's IO handles have separate handles in each lane. Thanks to Alyssa Rosenzweig for catching this bug. Reviewed-by: Caio Oliveira <caio.oliveira@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40280>	2026-03-09 02:38:59 +00:00
Connor Abbott	6e3d805735	freedreno: Rename afuc to QRisc ... In [1] the AQE is called the "Application QRisc Engine." Thus the real name of afuc is QRisc. Rename everything. [1] a698ebd321 Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40287>	2026-03-08 22:32:39 +00:00
Connor Abbott	554eec159b	freedreno/afuc: Update cread/cwrite syntax in README ... We now print actual modifiers instead of mysterious flags. Remove the remaining ones. Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40287>	2026-03-08 22:32:39 +00:00
Mel Henning	1371c53e6a	nvk: Expose VK_KHR_depth_clamp_zero_one ... Promoted from EXT Reviewed-By: Thomas H.P. Andersen <phomes@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39812>	2026-03-08 17:16:26 -04:00
Mel Henning	8e2707950b	nvk: Use the MME for cond rendering on Turing+ ... We can avoid the stalls from subc switches by avoiding using the copy engine during vkCmdBeginConditionalRenderingEXT. Implement this by loading the cond render value using the MME, since the hardware doesn't have a suitable 32-bit comparison itself. This brings the Sascha Willems conditionalrender demo from from 1661 to 8334 fps on my blackwell system with all meshes disabled. Reviewed-by: Mary Guillemard <mary@mary.zone> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40277>	2026-03-08 17:31:32 +00:00
Mel Henning	905557ab31	nvk: Use SET_GLOBAL_RENDER_ENABLE ... This brings the Sascha Willems conditionalrender demo from 927 to 1661 fps on my blackwell system with all meshes disabled. Reviewed-by: Mary Guillemard <mary@mary.zone> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40277>	2026-03-08 17:31:32 +00:00
Eric Engestrom	1c14a7f283	etnaviv/ci: fix expectation ... Fixes: 3c0aa7c633 ("etnaviv/ci: update expectations") Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40286>	2026-03-07 23:35:55 +00:00
Karol Herbst	65c5c4e1a1	nvk: run nir_opt_large_constants before nir_lower_load_const_to_scalar ... nir_opt_large_constants isn't able to deal with complex derefs and nir_lower_load_const_to_scalar e.g. splits up vectors to scalars. This prevented nir_opt_large_constants from extracting large constants in shaders that e.g. use a array of vector constant table. Totals: CodeSize: 9460341008 -> 9443435056 (-0.18%); split: -0.20%, +0.02% Number of GPRs: 47363466 -> 47300498 (-0.13%); split: -0.13%, +0.00% SLM Size: 5409320 -> 1202912 (-77.76%) Static cycle count: 6130972462 -> 6121193466 (-0.16%); split: -0.20%, +0.04% Spills to reg: 184840 -> 184828 (-0.01%); split: -0.01%, +0.01% Fills from reg: 223889 -> 223874 (-0.01%); split: -0.01%, +0.00% Max warps/SM: 50637796 -> 50641540 (+0.01%); split: +0.01%, -0.00% Totals from 32429 (2.79% of 1163204) affected shaders: CodeSize: 824883920 -> 807977968 (-2.05%); split: -2.25%, +0.20% Number of GPRs: 2413077 -> 2350109 (-2.61%); split: -2.61%, +0.00% SLM Size: 4437016 -> 230608 (-94.80%) Static cycle count: 1208715713 -> 1198936717 (-0.81%); split: -1.02%, +0.21% Spills to reg: 11934 -> 11922 (-0.10%); split: -0.20%, +0.10% Fills from reg: 14118 -> 14103 (-0.11%); split: -0.14%, +0.04% Max warps/SM: 1035736 -> 1039480 (+0.36%); split: +0.37%, -0.01% Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/14993 Reviewed-by: Mary Guillemard <mary@mary.zone> Reviewed-by: Mel Henning <mhenning@darkrefraction.com> Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com> Tested-by: Thomas H.P. Andersen <phomes@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40282>	2026-03-07 23:21:40 +00:00
Karol Herbst	faea742c3a	nouveau/drm-shim: implement get_zcull_info ... Reviewed-by: Mary Guillemard <mary@mary.zone> Reviewed-by: Mel Henning <mhenning@darkrefraction.com> Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com> Tested-by: Thomas H.P. Andersen <phomes@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40282>	2026-03-07 23:21:40 +00:00
Lucas Fryzek	844c5a1ae6	lvp: Export -1 as sync fd ... If the gallium context does not support `native_fence_fd`, we can still support sync fd export/import by exporting -1 as sync_fd in vulkan. Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40211>	2026-03-07 22:21:13 +00:00
Lucas Fryzek	420b934494	lvp: Mark opaque FD and dmabuf as compatible is supported ... If dmabuf export is supported we can now mark them as compatible handle types. Additionally we can always store the backed_fd for export. v2 (zzyiwei): hide opaque fd compat with dmabuf export behind udmabuf Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40211>	2026-03-07 22:21:13 +00:00
Yiwei Zhang	848da336dd	lvp: hide import-only dmabuf support from zink ... Zink has assumed both import and export when dmabuf extension is advertised, so lavapipe has to hide the extension for zink without supporting both. Together with the prior commit, now zink-on-lvp in the CI env without udmabuf will no longer test against fake dmabuf support. Reviewed-by: Lucas Fryzek <lfryzek@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40211>	2026-03-07 22:21:13 +00:00
Yiwei Zhang	5ab8c8a439	lvp: avoid advertising dmabuf support for kms_swrast ... Lavapipe relies on true udmabuf support for dmabuf export allocation. This changes aligns the behavior with both llvmpipe_allocate_memory_fd and llvmpipe_import_memory_fd. Fixes: 7d0a631f20 ("llvmpipe: export dmabuf caps for kms_swrast") Reviewed-by: Lucas Fryzek <lfryzek@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40211>	2026-03-07 22:21:12 +00:00
Mel Henning	bfde63e4d8	driconf: force_vk_vendor on No Man's Sky + NVK ... Cc: mesa-stable Reviewed-by: Mary Guillemard <mary@mary.zone> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40278>	2026-03-07 15:55:08 +00:00
Georg Lehmann	406935c6fe	radv: use nir_opt_fp_math_ctrl ... Foz-DB Navi21: Totals from 10833 (9.63% of 112497) affected shaders: Instrs: 10090308 -> 10043030 (-0.47%); split: -0.48%, +0.01% CodeSize: 53681564 -> 53556756 (-0.23%); split: -0.25%, +0.01% VGPRs: 511568 -> 511296 (-0.05%); split: -0.08%, +0.03% SpillSGPRs: 2442 -> 2438 (-0.16%); split: -0.20%, +0.04% Latency: 58989785 -> 58935280 (-0.09%); split: -0.18%, +0.09% InvThroughput: 15142587 -> 15067217 (-0.50%); split: -0.52%, +0.02% VClause: 200588 -> 200410 (-0.09%); split: -0.20%, +0.11% SClause: 257273 -> 257262 (-0.00%); split: -0.20%, +0.19% Copies: 741430 -> 741397 (-0.00%); split: -0.22%, +0.22% Branches: 211023 -> 211020 (-0.00%); split: -0.00%, +0.00% PreSGPRs: 491752 -> 491663 (-0.02%); split: -0.02%, +0.00% PreVGPRs: 418558 -> 418089 (-0.11%); split: -0.12%, +0.01% VALU: 7064149 -> 7017847 (-0.66%); split: -0.66%, +0.01% SALU: 1227287 -> 1226639 (-0.05%); split: -0.13%, +0.07% SMEM: 449268 -> 449343 (+0.02%) Reviewed-by: Rhys Perry <pendingchaos02@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40098>	2026-03-07 08:16:29 +01:00
Georg Lehmann	7c217e540c	nir: add a pass to optimize fp_math_ctrl ... Reviewed-by: Rhys Perry <pendingchaos02@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40098>	2026-03-07 08:16:27 +01:00
Georg Lehmann	042ee8dafc	panvk/ci: document new crashes on bifrost ... Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:45 +00:00
Georg Lehmann	f474e9853e	nir: add fp class analysis tests ... Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:45 +00:00
Georg Lehmann	4885e5cf3a	nir: remove more fsat using range analysis ... Foz-DB Navi48: Totals from 3018 (3.65% of 82636) affected shaders: MaxWaves: 69274 -> 69280 (+0.01%) Instrs: 7165414 -> 7157581 (-0.11%); split: -0.12%, +0.01% CodeSize: 38890212 -> 38823132 (-0.17%); split: -0.18%, +0.00% VGPRs: 228672 -> 228624 (-0.02%) Latency: 64789026 -> 64784877 (-0.01%); split: -0.01%, +0.00% InvThroughput: 11805156 -> 11802642 (-0.02%); split: -0.02%, +0.00% VClause: 136900 -> 136886 (-0.01%); split: -0.03%, +0.02% SClause: 150135 -> 150130 (-0.00%); split: -0.01%, +0.01% Copies: 574690 -> 574894 (+0.04%); split: -0.03%, +0.06% Branches: 187169 -> 187086 (-0.04%); split: -0.04%, +0.00% PreSGPRs: 190074 -> 190067 (-0.00%); split: -0.00%, +0.00% PreVGPRs: 189564 -> 189538 (-0.01%); split: -0.02%, +0.00% VALU: 3955188 -> 3949411 (-0.15%); split: -0.15%, +0.00% SALU: 1114659 -> 1114729 (+0.01%); split: -0.02%, +0.03% SMEM: 231080 -> 231077 (-0.00%); split: -0.00%, +0.00% VOPD: 116150 -> 116180 (+0.03%); split: +0.04%, -0.02% Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:45 +00:00
Georg Lehmann	506bb5a609	nir/search_helpers: use fp class analysis more ... Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:45 +00:00
Georg Lehmann	a9e75d8ee4	nir: remove nir_analyze_fp_range ... Use fp class analysis instead. Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	eb431efc19	nir/search_helpers: switch to fp class analysis ... Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	58799c4e7c	nir/gather_tcs_info: use nir_analyze_fp_class directly ... The information around positive one helps in theory. Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	08cac48170	aco/isel: skip min/max for SALU fsat if possible ... Foz-DB Navi48: Totals from 789 (0.95% of 82636) affected shaders: Instrs: 4144156 -> 4141345 (-0.07%); split: -0.07%, +0.00% CodeSize: 23345212 -> 23333960 (-0.05%); split: -0.05%, +0.00% Latency: 22988205 -> 22986666 (-0.01%); split: -0.01%, +0.00% InvThroughput: 4378321 -> 4377874 (-0.01%); split: -0.01%, +0.00% Copies: 302311 -> 302313 (+0.00%); split: -0.00%, +0.00% SALU: 647622 -> 645901 (-0.27%); split: -0.27%, +0.00% Reviewed-by: Daniel Schürmann <daniel@schuermann.dev> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	0ecf2c322e	nir: add fp class analysis for fround_even ... Foz-DB Navi48: Totals from 383 (0.33% of 114655) affected shaders: MaxWaves: 9806 -> 9808 (+0.02%) Instrs: 502508 -> 501762 (-0.15%); split: -0.16%, +0.01% CodeSize: 2711404 -> 2707604 (-0.14%); split: -0.15%, +0.01% VGPRs: 24360 -> 24348 (-0.05%) Latency: 2068105 -> 2066817 (-0.06%); split: -0.07%, +0.01% InvThroughput: 370962 -> 370081 (-0.24%) VClause: 7045 -> 7041 (-0.06%) SClause: 10551 -> 10559 (+0.08%); split: -0.08%, +0.15% Copies: 29135 -> 29117 (-0.06%); split: -0.12%, +0.05% Branches: 17333 -> 17328 (-0.03%) PreSGPRs: 21511 -> 21510 (-0.00%) PreVGPRs: 18555 -> 18545 (-0.05%) VALU: 274445 -> 273874 (-0.21%); split: -0.21%, +0.00% SALU: 78819 -> 78779 (-0.05%); split: -0.07%, +0.02% VMEM: 10918 -> 10913 (-0.05%) SMEM: 17662 -> 17656 (-0.03%) Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	7509b4a199	nir: add fp class analysis for fsub ... Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	d8734e5453	nir: add fp class analysis for shadow compare ... Foz-DB Navi48: Totals from 145 (0.18% of 82636) affected shaders: Instrs: 280871 -> 280729 (-0.05%) CodeSize: 1545724 -> 1545488 (-0.02%); split: -0.02%, +0.00% Latency: 10840265 -> 10840216 (-0.00%); split: -0.00%, +0.00% InvThroughput: 2093707 -> 2093646 (-0.00%) SClause: 4483 -> 4481 (-0.04%) VALU: 188142 -> 188039 (-0.05%) SALU: 22238 -> 22236 (-0.01%) Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	6d3a279a3b	nir: add fp class analysis for some intrinsics ... I also tried ddx/ddy, but that was not worth it. Foz-DB Navi48: Totals from 1019 (1.23% of 82636) affected shaders: Instrs: 516459 -> 515700 (-0.15%); split: -0.17%, +0.02% CodeSize: 2712428 -> 2707008 (-0.20%); split: -0.21%, +0.01% VGPRs: 70152 -> 70140 (-0.02%) Latency: 1799198 -> 1795926 (-0.18%); split: -0.19%, +0.00% InvThroughput: 233497 -> 232628 (-0.37%); split: -0.37%, +0.00% VClause: 15315 -> 15346 (+0.20%); split: -0.11%, +0.31% Copies: 30009 -> 30035 (+0.09%); split: -0.06%, +0.14% VALU: 305519 -> 304727 (-0.26%); split: -0.27%, +0.01% SALU: 45855 -> 45854 (-0.00%) Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	73bce23f65	nir: add fp class analysis for flog2 ... Foz-DB Navi48: Totals from 230 (0.28% of 82636) affected shaders: Instrs: 599005 -> 598615 (-0.07%); split: -0.09%, +0.02% CodeSize: 3110528 -> 3103136 (-0.24%); split: -0.24%, +0.00% Latency: 3661526 -> 3663241 (+0.05%); split: -0.01%, +0.05% InvThroughput: 526561 -> 526487 (-0.01%); split: -0.01%, +0.00% Copies: 33735 -> 33820 (+0.25%); split: -0.06%, +0.31% VALU: 378034 -> 377904 (-0.03%); split: -0.03%, +0.00% SALU: 65156 -> 65045 (-0.17%); split: -0.19%, +0.02% Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	81e272aa1d	nir: add fp class analysis for sin/cos ... Foz-DB Navi48: Totals from 264 (0.32% of 82636) affected shaders: CodeSize: 1688676 -> 1688672 (-0.00%) Latency: 510773 -> 510772 (-0.00%) InvThroughput: 138569 -> 138568 (-0.00%) Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00
Georg Lehmann	5a298f3560	nir: rewrite fp range analysis as a fp class analysis ... Knowing if a value is not larger than one helps proving finite results of fmul/fadd and will allow skipping/creating more fsat. Knowing that a value is larger than one helps proving non zero results of fmul. Separating positive and negative zero also has advantages when signed zero correctness is required. Foz-DB Navi48: Totals from 1344 (1.63% of 82636) affected shaders: Instrs: 5319389 -> 5312280 (-0.13%); split: -0.14%, +0.01% CodeSize: 29702516 -> 29665684 (-0.12%); split: -0.13%, +0.01% Latency: 40694344 -> 40694545 (+0.00%); split: -0.01%, +0.02% InvThroughput: 7481192 -> 7480403 (-0.01%); split: -0.02%, +0.01% VClause: 121947 -> 121946 (-0.00%); split: -0.00%, +0.00% SClause: 104972 -> 104923 (-0.05%); split: -0.05%, +0.00% Copies: 371098 -> 371092 (-0.00%); split: -0.02%, +0.02% Branches: 122929 -> 122919 (-0.01%); split: -0.01%, +0.00% PreSGPRs: 82506 -> 82510 (+0.00%); split: -0.00%, +0.01% PreVGPRs: 79175 -> 79168 (-0.01%) VALU: 2906718 -> 2904777 (-0.07%); split: -0.07%, +0.00% SALU: 726256 -> 723454 (-0.39%); split: -0.39%, +0.00% VMEM: 205021 -> 205016 (-0.00%) SMEM: 163972 -> 163916 (-0.03%) VOPD: 303354 -> 303298 (-0.02%); split: +0.02%, -0.04% Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39987>	2026-03-07 05:01:44 +00:00