Commit graph

1426 commits

Author SHA1 Message Date
Bas Nieuwenhuizen
6bc42855f9 radv: enable GS on GFX9
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 07:14:00 +01:00
Bas Nieuwenhuizen
73749caf0e radv: calculate and emit GFX9 GS registers to pipeline state.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:23:47 +01:00
Bas Nieuwenhuizen
9961ae2447 ac/nir: Fix up GS input vgprs.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:23:37 +01:00
Bas Nieuwenhuizen
d4ecc3c929 ac/nir: Add loading from LDS for merged GS.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:23:29 +01:00
Bas Nieuwenhuizen
ec53e52742 ac/nir: Add ES output to LDS for GFX9.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:23:18 +01:00
Bas Nieuwenhuizen
3e77333030 ac/nir: Add merged GS function.
[airlied: merged fixup + and fixed up a couple more bits].

Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:23:14 +01:00
Bas Nieuwenhuizen
f82797b56d radv: Only emit TES when it exists.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:14:14 +01:00
Bas Nieuwenhuizen
6e21b7a294 radv: Use control shader presence for detecting tess.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:11:10 +01:00
Dave Airlie
5bc5e07d81 radv: fixup tess eval shader when combined.
This fixes some access to the tess eval shader when it's combined
with geometry on gfx9.

This is a review of Bas's commit:
radv: Prevent crashing by accessing TES for VGT reuse depth.

Signed-off-by: Dave Airlie <airlied@redhat.com>
2017-10-20 06:11:10 +01:00
Bas Nieuwenhuizen
e6acc20b6a radv: Set VGT_GS_MODE properly for gfx9
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 05:55:11 +01:00
Dave Airlie
99281c1e8f radv: ensure correct outinfo is picked.
This struct used to rely on being in a union, it isn't anymore,
so we have to pick the correct outinfo struct now.

This should fix a regression since the union became a struct.

dEQP-VK.tessellation.geometry_interaction.point_size.vertex_set_geometry_set

Fixes: 6078a3bd51 (ac/nir: Allow ac_shader_variant_info to contain info about multiple stages.)
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2017-10-20 14:44:09 +10:00
Bas Nieuwenhuizen
ffaf4d608a radv: Enable tessellation shaders for GFX9.
It mostly works now.

Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 01:50:43 +02:00
Dave Airlie
1dda214d9c ac/nir: init full exec mask for merged shaders.
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-20 01:50:40 +02:00
Dave Airlie
14978a1c3b radv: drop unused r600_htile_info.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2017-10-20 00:38:57 +01:00
Dave Airlie
c8eb3558cc radv: fix CLEAR_STATE packet length.
Looking at shader traces I noticed some registers were missing,
one of them was being eaten by the wrong clear state length.

Fixes: 4f42ea4dc (radv: use CLEAR_STATE for initializing some registers)
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2017-10-19 23:56:48 +01:00
Timothy Arceri
087e010b2b radv: copy indirect lowering settings from radeonsi
It looks the original indirect mask was probably copied from
ANV.

Sascha Willems demo results:

tessellation ~4000 -> ~4200 fps

V2: continue lowering local indirects due to llvm deficiencies.

Tested-by: Alex Smith <asmith@feralinteractive.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-20 08:01:26 +11:00
Timothy Arceri
5549b47d7b radv: stop redundant setting of active_stages
We already set it when above in the nir compilation loop.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
2017-10-20 08:01:26 +11:00
Timothy Arceri
bebfeb7e1c ac: move some code out of loop in store_tcs_output()
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
2017-10-20 08:01:26 +11:00
Bas Nieuwenhuizen
228325f4b7 radv: Modify rsrc1/rsrc2 generation for merged tess.
No OC_LDS_EN for HS, and the included LS vgpr_comp_cnt is at
a different offset.

Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:44 +02:00
Bas Nieuwenhuizen
8250efb90a radv: Set correct registers for merged shader rings.
We need different regs to end up in s0/s1.

Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:39 +02:00
Bas Nieuwenhuizen
6a074f87be radv: Add GFX9 HS emitting code.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:34 +02:00
Bas Nieuwenhuizen
b096245030 radv: Remove remaining hard coded references to VS.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:31 +02:00
Bas Nieuwenhuizen
91b033f4f6 radv: Update GFX9 user data regs for GS/tess.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:27 +02:00
Bas Nieuwenhuizen
ce03c119ce radv: Add code to compile merged shaders.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:23 +02:00
Bas Nieuwenhuizen
640f2c458f ac/nir: Add LS-HS input VGPR workaround.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:19 +02:00
Bas Nieuwenhuizen
0a182e73d9 ac/nir: Compile the bodies of multiple shaders.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:15 +02:00
Bas Nieuwenhuizen
56d8af1ec5 ac/nir: Expand user SGPR descriptions a bit.
To prevent VS/TCS collisions in merged shaders.

Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:07 +02:00
Bas Nieuwenhuizen
25efef40d2 ac/nir: Don't write to the dynamic HS word on GFX9.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:04 +02:00
Bas Nieuwenhuizen
d8bd693d03 ac/nir: Add function creation for merged LS+HS.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:25:00 +02:00
Bas Nieuwenhuizen
0cdc8b26f8 ac/nir: Make scan_shader_output_decl less dependent on the context.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:24:56 +02:00
Bas Nieuwenhuizen
6078a3bd51 ac/nir: Allow ac_shader_variant_info to contain info about multiple stages.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:24:51 +02:00
Bas Nieuwenhuizen
a996ed1f9b ac/nir: Change interface to allow multiple source shaders.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:24:47 +02:00
Bas Nieuwenhuizen
872b21487c ac/nir: Add HS calling convention.
Needed for GFX9 merged shaders.

Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:24:42 +02:00
Bas Nieuwenhuizen
163a4bf386 ac: Parse the new HS RSRC1 register.
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-19 22:24:20 +02:00
Samuel Pitoiset
535aa43df0 radv: reset dirty flags after flushing all states
Move it to radv_cmd_buffer_flush_state() because if
rasterizerDiscardEnable is true, the flags are not cleared.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 21:21:48 +02:00
Samuel Pitoiset
966d66f28f radv: do not re-emit the index buffer for every draw call
It can only be changed when CmdBindIndexBuffer() is called
or when a secondary buffer is used. Though not always, but
let's re-emit the packets in this situation for now.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 21:21:43 +02:00
Samuel Pitoiset
e5480be0d1 radv: remove useless mask operation in radv_cs_emit_draw_indexed_packet()
This saves few CPU cycles when CmdDrawIndexed() is used a lot.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 21:21:30 +02:00
Bas Nieuwenhuizen
fa226e9933 radv: Do not read from the disk cache with RADV_DEBUG=nocache.
Otherwise the flag is borderline useless.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
2017-10-18 20:37:10 +02:00
Alex Smith
2cccc74f56 radv: Set active_stages after getting cached shaders
Fixes: 7d45d22fdd ("radv: switch to using radv_create_shaders()")
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 20:37:10 +02:00
Alex Smith
f557673237 radv: Don't free NIR shaders if tracing
Fixes a crash while generating a hang report.

Fixes: 7d45d22fdd ("radv: switch to using radv_create_shaders()")
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 20:37:10 +02:00
Timothy Arceri
e5e9e21e9f radv: don't create dummy fs when compiling compute stage
Fixes: d1c9f30d7f "radv: add radv_create_shaders() helper"

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 22:47:53 +11:00
Samuel Pitoiset
e6b9abf294 radv: use the dispatch initiator for indirect dispatches
Missed that when I allowed waves to be launched out-of-order.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 11:22:41 +02:00
Samuel Pitoiset
095e709717 radv: remove XtoY_temps structs
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 11:22:39 +02:00
Bas Nieuwenhuizen
06f05040eb radv: Link shaders.
Here we make use of NIR the linking helpers to remove unused
varyings.

Sascha Willems demo results:

computecullandlod 39 -> 41 fps
pipelines ~6100 -> ~6200 fps

Signed-off-by: Bas Nieuwenhuizen <basni@google.com>
Signed-off-by: Timothy Arceri <tarceri@itsqueeze.com>
Acked-by: Dave Airlie <airlied@redhat.com>
2017-10-18 09:19:35 +11:00
Timothy Arceri
dbbf10541b radv: reuse the multiple shader store & load functions for gs copy variant
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 09:19:35 +11:00
Timothy Arceri
351f9dde60 radv: remove some now unused shader compile code
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 09:19:35 +11:00
Timothy Arceri
7d45d22fdd radv: switch to using radv_create_shaders()
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
2017-10-18 09:19:35 +11:00
Bas Nieuwenhuizen
d1c9f30d7f radv: add radv_create_shaders() helper
This is a combined shader creation helper than will help us to
create the shaders for each stage at once. This will allow us to
do some link time optimisations.

Signed-off-by: Timothy Arceri <tarceri@itsqueeze.com>
Acked-by: Dave Airlie <airlied@redhat.com>
2017-10-18 09:19:35 +11:00
Bas Nieuwenhuizen
ed9218f154 radv: add radv_hash_shaders() helper
This will be used to create a hash of the combined shaders in the
pipeline.

Signed-off-by: Timothy Arceri <tarceri@itsqueeze.com>
Acked-by: Dave Airlie <airlied@redhat.com>
2017-10-18 09:19:35 +11:00
Bas Nieuwenhuizen
7f29055751 radv: Add multiple shader cache store & load functions.
Signed-off-by: Timothy Arceri <tarceri@itsqueeze.com>
Acked-by: Dave Airlie <airlied@redhat.com>
2017-10-18 09:19:35 +11:00