summaryrefslogtreecommitdiff
AgeCommit message (Expand)AuthorFilesLines
39 hoursradeonsi: iterate from draw 1 for total/min_direct_count computationMarek Olšák1-2/+3
39 hoursradeonsi: enable accidentally disabled fast launch with non-indexed tri stripsMarek Olšák1-5/+9
39 hoursradeonsi: skip some code for ALLOW_PRIM_DISCARD_CS if tess or GS is enabledMarek Olšák1-3/+5
39 hoursradeonsi: rename SI_SGPR_RW_BUFFERS to SI_SGPR_INTERNAL_BINDINGSMarek Olšák18-75/+77
39 hoursradeonsi: move if (sctx->vertex_buffers_dirty) into the upload functionMarek Olšák1-82/+84
39 hoursradeonsi: don't set vertex buffer dirty flags when they don't do anythingMarek Olšák3-8/+16
39 hoursradeonsi: move y_inverted out of si_viewportsMarek Olšák4-7/+7
39 hoursradeonsi: inline si_blend_color and si_clip_state structuresMarek Olšák3-21/+13
39 hoursradeonsi: simplify determining whether render condition is enabled at draw timeMarek Olšák6-10/+11
39 hoursradeonsi: add internal blitter_running flagMarek Olšák6-10/+13
39 hoursradeonsi: don't use rasterizer_discard to validate draws, only check ps_shaderMarek Olšák1-2/+1
39 hoursradeonsi: optimize translating index_size to index_typeMarek Olšák1-18/+13
39 hoursradeonsi: don't mark NULL states as dirty in si_pm4_reset_emittedMarek Olšák1-1/+5
39 hoursradeonsi: clear dirty_states if si_pm4_bind_state is unbinding or no-opMarek Olšák1-1/+4
39 hoursradeonsi: add new possibly faster command submission helpersMarek Olšák19-395/+593
3 daysradeonsi: Use util_writes_stencil() helperRob Clark1-9/+1
3 daysradeonsi: allow instance_count == 0 on chips that handle it correctlyMarek Olšák1-1/+1
3 daysradeonsi: don't validate inlinable uniforms at draw timeMarek Olšák4-16/+1
3 daysradeonsi: move variables closer to their use in most draw state functionsMarek Olšák1-38/+36
3 daysradeonsi: clear dirty_atoms and dirty_states only if we entered the emit loopMarek Olšák1-11/+17
3 daysradeonsi: enable the GS tri strip adj workaround with primitive_restartMarek Olšák1-5/+4
3 daysradeonsi: evaluate si_get_vs in si_draw_vbo at compile timeMarek Olšák2-9/+17
3 daysradeonsi: inline the last use of si_get_vs_stateMarek Olšák2-10/+7
3 daysradeonsi: evaluate sh_base in si_emit_vs_state at compile timeMarek Olšák1-7/+8
3 daysradeonsi: add si_get_user_data_base selecting user data registersMarek Olšák2-50/+88
3 daysradeonsi: don't set context_roll for non-gfx9 in templated functionsMarek Olšák1-4/+5
3 daysradeonsi: don't pass pipe_draw_info into si_emit_draw_registersMarek Olšák1-8/+9
3 daysradeonsi: unify uploaders on APUs tooMarek Olšák1-2/+8
5 daysradeonsi: inhibit clockgating when using SQTTPierre-Eric Pelloux-Prayer3-7/+12
6 daysradeonsi: trim the size of si_vgt_param_key and si_vgt_stages_keyMarek Olšák3-38/+38
6 daysradeonsi: don't use si_get_vs_state in most placesMarek Olšák3-11/+11
6 daysradeonsi: rearrange condition for streamout workaround on gfx7 and gfx8Marek Olšák1-2/+2
6 daysradeonsi: get out of si_emit_vs_state early for blit vertex shadersMarek Olšák1-5/+5
6 daysradeonsi: remove MRT-draw-calls, spill-draw-calls, spill-compute-callsMarek Olšák5-33/+0
6 daysradeonsi: make sctx->vertex_elements always non-NULLMarek Olšák7-27/+35
6 daysradeonsi: add a specialized function for CP DMA L2 prefetchMarek Olšák5-14/+43
6 daysradeonsi: make cik_emit_prefetch_L2 templated and move it to si_state_draw.cppMarek Olšák4-137/+142
6 daysradeonsi: fix si_num_prims_for_vertices for PIPE_PRIM_POLYGONMarek Olšák1-1/+2
6 daysradeonsi: don't compute average vertex count in si_draw_vboMarek Olšák1-5/+3
6 daysradeonsi: don't pass pipe_draw_info into si_emit_derived_tess_stateMarek Olšák1-3/+4
6 daysradeonsi: translate pipe_prim_type only when it changesMarek Olšák1-3/+4
6 daysradeonsi: don't pass pipe_draw_info into si_emit_ia_multi_vgt_paramMarek Olšák1-14/+15
6 daysradeonsi: don't pass pipe_draw_info into si_emit_vs_stateMarek Olšák1-3/+3
6 daysradeonsi: move emit_cache_flush functions into si_gfx_cs.cMarek Olšák4-451/+449
6 daysradeonsi: don't clear unaligned bits when unbinding vertex buffersMarek Olšák1-3/+1
6 daysradeonsi: constant buffer cleanupsMarek Olšák3-10/+4
10 daysradeonsi: invalidate compute sgprs in si_rebind_bufferPierre-Eric Pelloux-Prayer1-6/+16
11 daysdri: Restrict glthread for CS:GO to radeonsiNanley Chery1-4/+1
11 daysradeonsi: Only set modifier creation function for GFX9+ & with kernel support.Bas Nieuwenhuizen3-5/+14
13 daysdri: enable glthread + radeonsi workaround for CS:GOPierre-Eric Pelloux-Prayer1-0/+5