215413 commits

This branch

This branch

All branches

Author	SHA1	Message	Date
Georg Lehmann	37d3c63a12	aco/optimizer: add new helpers for applying output modifiers ... To replace the old instr_mod_labels. Foz-DB Navi21: Totals from 683 (0.70% of 97591) affected shaders: Instrs: 3341288 -> 3340447 (-0.03%); split: -0.03%, +0.00% CodeSize: 18522460 -> 18520212 (-0.01%); split: -0.01%, +0.00% Latency: 34359519 -> 34358772 (-0.00%); split: -0.00%, +0.00% InvThroughput: 9229621 -> 9229494 (-0.00%); split: -0.00%, +0.00% Copies: 368383 -> 368260 (-0.03%); split: -0.04%, +0.00% PreSGPRs: 48060 -> 48061 (+0.00%) SALU: 543991 -> 543150 (-0.15%); split: -0.16%, +0.00% Changes are caused by optimizing not(salu) without killed scc. Reviewed-by: Daniel Schürmann <daniel@schuermann.dev> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38658>	2025-11-29 08:27:56 +00:00
Georg Lehmann	fc29821d3b	aco/optimizer: move med3 -> add_clamp opt later ... Soon we will apply omod later, when the combine_instruction reaches the multiplication with constant. Reviewed-by: Daniel Schürmann <daniel@schuermann.dev> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38658>	2025-11-29 08:27:55 +00:00
Georg Lehmann	39a61502e5	aco/opt_postRA: allow v_cmpx to clobber exec before nop split/create vector ... Kind of ugly, but I really hate seeing this in every rt traversal loop: image_bvh64_intersect_ray v[56:59], [v40, v41, v42, v47, v48, v49, v50, v51, v52, v53, v54, v55], s[44:47] v_cmp_class_f32_e64 s57, 0xff800000, v12 s_and_b32 exec_lo, s57, exec_lo s_cbranch_execz BB219 Foz-DB Navi21: Totals from 3394 (3.48% of 97591) affected shaders: Instrs: 9536259 -> 9533592 (-0.03%) CodeSize: 51657072 -> 51640120 (-0.03%); split: -0.03%, +0.00% Latency: 109493553 -> 109513317 (+0.02%); split: -0.01%, +0.02% InvThroughput: 29125525 -> 29131876 (+0.02%); split: -0.00%, +0.02% Copies: 815888 -> 818219 (+0.29%); split: -0.01%, +0.30% Branches: 277451 -> 277449 (-0.00%) SALU: 1217642 -> 1214976 (-0.22%) Reviewed-by: Daniel Schürmann <daniel@schuermann.dev> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38697>	2025-11-29 08:02:24 +00:00
Marek Olšák	1f2d129bfa	gallium: add a flag to finalize_nir to allow drivers to skip NIR opts ... This could help achieve better compile times. Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38600>	2025-11-29 07:29:05 +00:00
Marek Olšák	9294448fe1	nir/recompute_io_bases: report progress only if anything was changed ... also preserve all metadata because it doesn't add/remove any instructions Reviewed-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Iago Toral Quiroga <itoral@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38599>	2025-11-29 05:00:40 +00:00
Marek Olšák	e6499fa73e	nir/recompute_io_bases: move color input bases after all other inputs ... This is related to the FS prolog. It should have no effect on other drivers. v2: make it optional via io_options Reviewed-by: Timur Kristóf <timur.kristof@gmail.com> (v1) Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38599>	2025-11-29 05:00:40 +00:00
Marek Olšák	18a338066b	nir/recompute_io_bases: don't use safe iterators ... the pass doesn't remove anything Reviewed-by: Timur Kristóf <timur.kristof@gmail.com> Reviewed-by: Iago Toral Quiroga <itoral@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38599>	2025-11-29 05:00:40 +00:00
Faith Ekstrand	4711e5954e	nir: Always use sysvals in lower_input_attachments() ... The last holdouts of the var options are gone so we can just emit the system values. This is overall simpler as it confines all the sysval to var logic to nir_lower_sysvals_to_varyings(). Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Reviewed-by: Emma Anholt <emma@anholt.net> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38562>	2025-11-29 00:50:34 +00:00
Faith Ekstrand	5bbbf5cf9b	tu: Set use_layer_id_sysval for nir_lower_input_attachments ... We can just use nir_lower_sysvals_to_varyings instead. Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Reviewed-by: Emma Anholt <emma@anholt.net> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38562>	2025-11-29 00:50:33 +00:00
Faith Ekstrand	b02a98d7d8	microsof: Run lower_sysvals_to_varyings after lower_input_attachments ... This lets us request system values from lower_input_attachments and just lower them ourselves instead of asking it to create variables. Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38562>	2025-11-29 00:50:32 +00:00
Faith Ekstrand	82280a7e86	nir: Support sysval intrinsics in lower_sysvals_to_varyings() ... Since this is a downgrade path for drivers, it's useful to support both forms of these common sysvals. Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Reviewed-by: Emma Anholt <emma@anholt.net> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38562>	2025-11-29 00:50:32 +00:00
Faith Ekstrand	0c36c39103	spirv: Emit SYSTEM_VALUE_LAYER_ID for fragment shaders ... We have nir_lower_sysvals_to_varyings() so we can just have that lower it for the drivers who don't want a sysval. Most have to support the sysval version anyway for various lowering so making them all have to support both is pretty annoying. Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Reviewed-by: Emma Anholt <emma@anholt.net> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38562>	2025-11-29 00:50:32 +00:00
Faith Ekstrand	701a9c269e	nir: Add LAYER_ID and VIEW_INDEX to nir_lower_sysvals_to_varyings() ... Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Reviewed-by: Emma Anholt <emma@anholt.net> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38562>	2025-11-29 00:50:31 +00:00
Marek Olšák	fa0bea5ff8	nir: remove nir_io_add_const_offset_to_base ... nir_opt_constant_folding does it now. Acked-by: Emma Anholt <emma@anholt.net> Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38277>	2025-11-29 00:16:38 +00:00
Marek Olšák	726bbb352e	nir/opt_constant_folding: add nir_io_add_const_offset_to_base behavior ... We almost always call both passes next to each other. The code is copied from nir_io_add_const_offset_to_base. No changes. Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38277>	2025-11-29 00:16:38 +00:00
Marek Olšák	9a56672f56	nir: add shader_info::disable_input/output_offset_src_constant_folding ... and set it where needed to prevent nir_opt_constant_folding from breaking those drivers. Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38277>	2025-11-29 00:16:38 +00:00
Marek Olšák	7330bca9db	nir: handle load_fs_input_interp_deltas in nir_is_input_load ... for nir_opt_constant_folding Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38277>	2025-11-29 00:16:37 +00:00
Marek Olšák	ffcbbeb54a	nir/validate: don't require offset src to be 0 if constant ... nir_opt_constant_folding does the folding, so this can be non-zero before that. Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38277>	2025-11-29 00:16:36 +00:00
Eric Engestrom	b87b83d15e	broadcom/ci: update device count in ci-tron farm ... Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38719>	2025-11-28 23:50:01 +01:00
Eric Engestrom	c09550e3c0	broadcom/ci: apply "Cannot open root device" reboot workaround to all rpi boards ... The problem has been observed on rpi4 and rpi5 as well. Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38719>	2025-11-28 23:43:07 +01:00
Marek Olšák	21cdbfa223	ac,radv: move opt_vectorize_callback to common code ... radeonsi will use it. Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38603>	2025-11-28 20:16:10 +00:00
Marek Olšák	2c9995a94f	ac/nir: move aco_nir_op_supports_packed_math_16bit here ... aco_nir_op_supports_packed_math_16bit currently can't be used by amd/common because tests don't link with ACO, so linking would fail, but we want to move the nir_opt_vectorize callback here that uses it. Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38603>	2025-11-28 20:16:10 +00:00
Juan A. Suarez Romero	d95b43e07b	broadcom/ci: remove ci-tron- prefix from nightly jobs ... All the nighly jobs are run with CI-Tron, so no need to prefix them. Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38715>	2025-11-28 18:09:09 +01:00
Juan A. Suarez Romero	50ba2a0e34	broadcom/ci: remove all baremetal nightly jobs ... All those jobs will be executed using CI-Tron. Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38715>	2025-11-28 18:09:09 +01:00
Yiwei Zhang	a6ade961b2	venus: implement VK_EXT_map_memory_placed ... Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38706>	2025-11-28 16:38:26 +00:00
Yiwei Zhang	8adfdc3304	venus: add renderer support for placed mapping ... Prepare for VK_EXT_map_memory_placed support. Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38706>	2025-11-28 16:38:25 +00:00
David Rosca	38090d5be0	radv/video: Drop casts from vk_find_struct* ... The macro itself does the cast. Reviewed-by: Benjamin Cheng <benjamin.cheng@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38521>	2025-11-28 15:35:26 +00:00
David Rosca	32a02720a8	radv/video: Init session and update rate control in ControlVideoCoding ... This eliminates the last state we kept in encode video session. Also fixes changing encode resolution without reset. Reviewed-by: Benjamin Cheng <benjamin.cheng@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38521>	2025-11-28 15:35:26 +00:00
David Rosca	a7fe0188d4	radv/video: Remove tile config and skip mode from video session state ... Reviewed-by: Benjamin Cheng <benjamin.cheng@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38521>	2025-11-28 15:35:25 +00:00
David Rosca	5d0d00e5f8	radv/video: Use radv_enc_aligned_coded_extent for session params overrides ... Reviewed-by: Benjamin Cheng <benjamin.cheng@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38521>	2025-11-28 15:35:25 +00:00
David Rosca	0fc4ead36f	radv/video: Remove enc_session from video session state ... It was only used to store aligned picture size. Add helper function to get the aligned size and use it when needed. Reviewed-by: Benjamin Cheng <benjamin.cheng@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38521>	2025-11-28 15:35:25 +00:00
Samuel Pitoiset	c3420ca932	Revert "radv: remove the workaround for DISPATCH_TASKMESH_INDIRECT_MULTI_ACE on GFX10.3" ... This reverts commit 0391902eb5. Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38711>	2025-11-28 15:34:53 +01:00
Georg Lehmann	653716b745	nir/opt_algebraic: create more bit test ... Helps hackends with has_bit_test more (i.e. ACO), but it shouldn't hurt others either. Foz-DB Navi21: Totals from 1138 (1.17% of 97591) affected shaders: Instrs: 5478747 -> 5476055 (-0.05%); split: -0.05%, +0.00% CodeSize: 29850188 -> 29853140 (+0.01%); split: -0.04%, +0.05% SpillSGPRs: 1406 -> 1401 (-0.36%) Latency: 42324245 -> 42325921 (+0.00%); split: -0.01%, +0.01% InvThroughput: 11396940 -> 11394048 (-0.03%); split: -0.04%, +0.01% VClause: 142294 -> 142309 (+0.01%); split: -0.00%, +0.01% SClause: 124412 -> 124411 (-0.00%); split: -0.00%, +0.00% Copies: 572696 -> 572749 (+0.01%); split: -0.02%, +0.03% Branches: 199932 -> 199929 (-0.00%) PreSGPRs: 73372 -> 74970 (+2.18%) PreVGPRs: 79514 -> 79511 (-0.00%) VALU: 3628764 -> 3625744 (-0.08%); split: -0.08%, +0.00% SALU: 818258 -> 818475 (+0.03%); split: -0.03%, +0.06% Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38700>	2025-11-28 13:25:24 +00:00
Valentine Burley	f9ef7e0f64	Revert "anv/ci: Run vkd3d job in parallel" ... With the new vkd3d-proton uprev, a random crash has appeared when running in parallel. This reverts commit 45c9c61ad3. Signed-off-by: Valentine Burley <valentine.burley@collabora.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38652>	2025-11-28 11:44:28 +00:00
Samuel Pitoiset	92a468f8f2	ci: uprev vkd3d ... vkd3d-proton had an issue with its runner and few tests were excluded by accident. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38652>	2025-11-28 11:44:28 +00:00
Erik Faye-Lund	60e115dedf	mesa/st: do not drop binding prematurely ... While it's true that we currently need to eventually fall back to checking without the render-target binding, this should really be the last resort. Because otherwise we might end up picking a format that isn't possible to render to for a color-renderable internalformat. In the long run, this code should be rewritten to check *properly* if the internalformat is color-renderable or not *up front*, and not even try to fall back. But we're currently missing proper helpers for this, and reworking what we have is a fair bit of work. So for now, let's just do what we currently do, but shuffle around the order of testing things so we don't end up dropping unless we absolutely have to. Tested-by: Christian Gmeiner <cgmeiner@igalia.com> Reviewed-by: Marek Olšák <marek.olsak@amd.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38673>	2025-11-28 10:46:53 +00:00
Samuel Pitoiset	0391902eb5	radv: remove the workaround for DISPATCH_TASKMESH_INDIRECT_MULTI_ACE on GFX10.3 ... Only very old MEC firmwares are concerned, so let's remove it and disable mesh shaders with those firmwares. Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38691>	2025-11-28 10:21:30 +00:00
Christoph Pillmayer	262a427a51	pan/bi: Add missing 8bit widen swizzles ... Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38109>	2025-11-28 09:52:11 +00:00
Tapani Pälli	ba89826b75	anv: add furmark workaround layer ... Cc: mesa-stable Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/14274 Signed-off-by: Tapani Pälli <tapani.palli@intel.com> Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38410>	2025-11-28 09:26:41 +00:00
Timothy Arceri	d10036362f	util/driconf: Add linux version of Penumbra fixes ... Cc: mesa-stable Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38563>	2025-11-28 08:49:55 +00:00
Job Noorman	233b77878d	ir3/ra: try to allocate overlapping regs for shared subreg movs ... This was implemented for vector RA but not for shared RA yet. Totals from 1361 (0.77% of 176279) affected shaders: Instrs: 1175437 -> 1170238 (-0.44%); split: -0.45%, +0.01% CodeSize: 2300656 -> 2290258 (-0.45%) NOPs: 221042 -> 220527 (-0.23%); split: -0.48%, +0.25% MOVs: 30645 -> 30643 (-0.01%); split: -0.01%, +0.00% COVs: 47425 -> 47016 (-0.86%) (ss): 35953 -> 35890 (-0.18%); split: -0.21%, +0.03% (sy): 20174 -> 20168 (-0.03%) (ss)-stall: 124094 -> 123625 (-0.38%); split: -0.38%, +0.00% (sy)-stall: 806166 -> 805832 (-0.04%); split: -0.06%, +0.02% Preamble Instrs: 173151 -> 171299 (-1.07%) Cat0: 250836 -> 250321 (-0.21%); split: -0.43%, +0.22% Cat1: 78738 -> 78327 (-0.52%); split: -0.52%, +0.00% Cat2: 386528 -> 382255 (-1.11%) Signed-off-by: Job Noorman <jnoorman@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38573>	2025-11-28 08:00:42 +00:00
Job Noorman	1fc49eb120	ir3/ra: try to allocate subreg movs earlier ... Successful subreg allocations allow us to remove the instruction so it makes sense to try this first before trying other allocation strategies. Totals from 72 (0.04% of 176279) affected shaders: Instrs: 144346 -> 144277 (-0.05%); split: -0.06%, +0.01% CodeSize: 312174 -> 312182 (+0.00%); split: -0.01%, +0.01% NOPs: 32438 -> 32443 (+0.02%); split: -0.07%, +0.09% MOVs: 5923 -> 5934 (+0.19%) COVs: 3039 -> 3000 (-1.28%) (ss): 2967 -> 2968 (+0.03%) (sy): 1831 -> 1830 (-0.05%) (ss)-stall: 9113 -> 9128 (+0.16%) (sy)-stall: 45844 -> 45858 (+0.03%); split: -0.03%, +0.06% Cat0: 36136 -> 36141 (+0.01%); split: -0.06%, +0.08% Cat1: 9010 -> 8982 (-0.31%); split: -0.37%, +0.06% Cat2: 53533 -> 53487 (-0.09%) Signed-off-by: Job Noorman <jnoorman@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38573>	2025-11-28 08:00:42 +00:00
Samuel Pitoiset	5fd7af9e42	ac/surface: do not use tile swizzle for replayable/aliased FMASK surfaces ... Otherwise the VA might change. Fixes: 2bbc7d1db6 ("radv: move more surf_index logic to use_tile_swizzle") Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38696>	2025-11-28 07:39:33 +00:00
Emma Anholt	2d441c10af	ir3: Make the debug-print block numbers be the NIR block numbers. ... We still have to fall back to our pointers-as-indices for things like the preamble's block or old streamout (I'm presuming here that pointers will always be much greater than the NIR block count, which seems safe enough for debug). This gives us nice printouts in debugoptimized builds, and helps you correlate your ir3 back to NIR (which was helpful in the hundreds of blocks in the shader I fixed in the previous commit). Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38666>	2025-11-28 07:06:22 +00:00
Emma Anholt	a35f26a983	ir3: Fix incorrect use of predicated ifs on getlast. ... The getlast lowering will generate new branches, violating the assumptions of prede_sched(). Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38666>	2025-11-28 07:06:21 +00:00
Job Noorman	435db6fabe	ir3: merge rpt groups after postsched ... It often happens that postsched puts rpt groups back into an order that allows them to be merged into a repeated instruction. Breaking up these rpt groups before postsched loses this opportunity. To fix this, change ir3_cleanup_rpt to not take the ip into account and call ir3_merge_rpt after postsched. Totals from 129238 (73.31% of 176279) affected shaders: MaxWaves: 1834226 -> 1834248 (+0.00%); split: +0.00%, -0.00% Instrs: 46484782 -> 46382869 (-0.22%); split: -0.69%, +0.48% CodeSize: 95513914 -> 93871848 (-1.72%); split: -2.24%, +0.52% NOPs: 8018516 -> 7939362 (-0.99%); split: -3.28%, +2.30% MOVs: 1391770 -> 1408039 (+1.17%); split: -4.39%, +5.56% COVs: 776518 -> 776182 (-0.04%); split: -0.06%, +0.02% Full: 1473903 -> 1489694 (+1.07%); split: -0.76%, +1.83% (ss): 1143180 -> 1146977 (+0.33%); split: -3.07%, +3.40% (sy): 552487 -> 562122 (+1.74%); split: -1.83%, +3.57% (ss)-stall: 4292082 -> 4259946 (-0.75%); split: -3.95%, +3.20% (sy)-stall: 16573976 -> 17151457 (+3.48%); split: -2.41%, +5.89% STPs: 16131 -> 16157 (+0.16%); split: -0.10%, +0.26% LDPs: 19583 -> 19634 (+0.26%); split: -0.02%, +0.28% Preamble Instrs: 9889595 -> 9887178 (-0.02%); split: -0.23%, +0.21% Early Preamble: 103194 -> 103646 (+0.44%); split: +0.51%, -0.07% Cat0: 8850422 -> 8769964 (-0.91%); split: -3.00%, +2.09% Cat1: 2212326 -> 2226425 (+0.64%); split: -2.90%, +3.54% Cat2: 17452525 -> 17448724 (-0.02%); split: -0.02%, +0.00% Cat6: 501182 -> 501263 (+0.02%); split: -0.00%, +0.02% Cat7: 1293844 -> 1262010 (-2.46%); split: -4.17%, +1.71% Signed-off-by: Job Noorman <jnoorman@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38576>	2025-11-28 06:41:32 +00:00
Job Noorman	e8dbed2be4	ir3: don't use list_head for rpt groups ... To link together instructions in a rpt group we currently (ab)use list_head. This is a bit of a hack because we don't actually have a list_head that points to the first instruction without being embedded in an instruction itself (the way list_head is supposed to be used). Instead, the list_head embedded in the first instruction of a rpt group also serves as the one pointing to the list. In order make a distinction between the first and last instruction (for which the main list_head would usually be used), we rely on the fact that (currently) instructions in a rpt group are emitted in order which means that later instructions have a larger serialno than earlier ones. In order to make all this less hacky, and to lift the restriction of needing instructions to be emitted in order, replace the list_head with explicit rpt_next/rpt_prev pointers which link the instructions together in a doubly but non-circular linked list. Signed-off-by: Job Noorman <jnoorman@igalia.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38576>	2025-11-28 06:41:32 +00:00
Emma Anholt	2cf0ba35bc	ir3: Drop the vector splitting and simplify ir3_nir_lower_64b_global(). ... There's no need to generate the separate memory accesses, when nir_lower_mem_access_bit_sizes() has already done so. Also, the way the 64b address is handled changed in 2490ecf5fc ("ir3: ingest global addresses as 64b values from NIR") Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38688>	2025-11-28 06:14:00 +00:00
Emma Anholt	997c500cc4	ir3: Drop ir3_nir_lower_64b_intrinsics ... Our 64-bit memory load/stores are already split to 32 bits by nir_lower_mem_access_bit_sizes. Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38688>	2025-11-28 06:14:00 +00:00
Emma Anholt	f8901bddac	ir3: Drop use of nir_lower_wrmasks(). ... It gets done by nir_lower_mem_access_bit_sizes that's called right after. Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38688>	2025-11-28 06:13:59 +00:00