Change the terminology around the post-RA optimizer, primarily this
changes the use of "clobbered" to "overwritten" to avoid confusion,
and it removes some redundant states.
Proposed for backporting to stable, to make sure it is easy to
backport further fixes (if any) on top of this.
Fossil DB stats unaffected on Navi 21.
Cc: mesa-stable
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18488>
Change reset_block() so it only considers the logical
predecessors for VGPRs. Relevant for some optimizations
across loops.
This commit fixes an assertion failure which was triggered
by Zink in a piglit test.
Fossil DB stats unaffected on Navi 21.
Fixes: 2e56e23420
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18488>
This assumption is no longer true since the post-RA optimizer
can work across blocks. It is now possible that some control
flow paths overwrite some but not all registers of an operand.
This commit may prevent invalid optimizations and/or assertion
failures (on debug builds).
Fossil DB stats unaffected on Navi 21.
Fixes: 0e4747d3fb
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18488>
Affects blocks whose some (but not all) predecessors overwrite a register.
This commit fixes glitches in some games which regressed because of the
improved SCC no-compare optimization.
Fossil DB stats on Navi 21:
Totals from 2816 (2.09% of 134906) affected shaders:
CodeSize: 24224276 -> 24241580 (+0.07%)
Instrs: 4570595 -> 4574921 (+0.09%)
Latency: 53680256 -> 53693655 (+0.02%); split: -0.00%, +0.02%
InvThroughput: 9829289 -> 9830573 (+0.01%)
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7257
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7305
Fixes: 2e56e23420
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18488>
These are intended to make sure that the post-RA optimizer works
correctly across control flow. The new tests emit a divergent
if-else branch (with full logical+linear CFG).
- dpp_across_cf:
Simple case of DPP optimizable across control flow. Should pass.
- dpp_across_cf_overwritten:
Similar case but the DPP source register is overwritten in CF.
This shows a bug so the test fails now (will be fixed).
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18488>
This is actually not necessary and generates a lot of superfluous
instructions at every phi (setting the value to zero).
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18566>
Prevent nir_lower_scratch accidentally turning these dead variables
into scratch. This can especially happen to arrayed outputs of
eg. tess control or mesh shaders, which become large shader temps.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18566>
This was accidentally left at the maximum possible value.
However I now tested this with the cadscene demo app and there is
hardly any benefit from going above 64. Set it to 64 for now.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18566>
When the main FS needs sample positions and the number of samples
isn't known at compile time with GPL.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18677>
Use two different paths (static vs dynamic) to avoid running more NIR
pass to remove dead CF code when static is used.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18677>
With GPL, it's possible to build the main FS without the multisample
state, but the number of rasterization samples is required for
lowering interpolateAtSample(). In this rare situation, the number of
samples will be passed through a new user SGPR.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18677>
AMDGPU pstate is per-device, not per Vulkan logical devices. The same
AMDGPU device is shared accross logical devices because the driver
creates only one winsys per fd. The kernel only allows one context
at a time per AMDGPU device, otherwise it returns -EBUSY.
Fixes this by acquiring pstate on-demand to avoid this multiple
logical device problem.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17712>
Setting pstate is used for RGP captures and performance counters, so
this name is more generic. Also make it non static.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17712>
Warning:
aco_register_allocation.cpp(383): warning C4819: The file contains a character that cannot be represented in the current code page (0). Save the file in Unicode format to prevent data loss
This warning was treated as error with compiling with msvc
u8 is belongs to c11 standard so it's safe to use it
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18682>
We can't detect color attachment without exports when compiling a PS
epilog, so we can't compact MRTs.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18514>
This was used to distinguish definitions fixed before and during RA, but
it seems it isn't used anymore.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18547>
Corner cases like implicit gl_PrimitiveID are currently broken and
will be fixed later, but the general case should work.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18516>
With GPL, it's possible to build the pre-rasterization stages separately
from the fragment stage. Implicit IO (like gl_PrimitiveID) between the
last pre-rast stage and the FS will be addressed later.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18516>
fossils-db (NAVI21):
Totals from 158 (0.12% of 134913) affected shaders:
CodeSize: 569456 -> 568824 (-0.11%)
Only Control seems affected.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18615>
Otherwise, there might be redundant barycentric_at_sample intrinsics
that will be lowered and this will increase code size.
No fossils-db changes.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18615>
../mesa/src/amd/common/ac_surface.c:2324:48: warning: implicit conversion from enumeration type 'AddrResourceType' (aka 'enum _AddrResourceType') to different enumeration type 'enum gfx9_resource_type' [-Wenum-conversion]
surf->u.gfx9.resource_type = AddrSurfInfoIn.resourceType;
~ ~~~~~~~~~~~~~~~^~~~~~~~~~~~
../mesa/src/amd/common/ac_surface.c:3046:38: warning: implicit conversion from enumeration type 'const enum gfx9_resource_type' to different enumeration type 'AddrResourceType' (aka 'enum _AddrResourceType') [-Wenum-conversion]
input.resourceType = surf->u.gfx9.resource_type;
~ ~~~~~~~~~~~~~^~~~~~~~~~~~~
../mesa/src/amd/common/ac_surface.c:3069:38: warning: implicit conversion from enumeration type 'const enum gfx9_resource_type' to different enumeration type 'AddrResourceType' (aka 'enum _AddrResourceType') [-Wenum-conversion]
input.resourceType = surf->u.gfx9.resource_type;
The enums are compatible so lets just add some casts.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18694>
Fix defect reported by Coverity Scan.
Useless call (USELESS_CALL)
side_effect_free: Calling count_tes_user_sgprs(key) is only useful for
its return value, which is ignored.
Fixes: 8253ec3855 ("radv: add shader arguments for dynamic patch control points")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18659>
The current stack size is a significant limiter for occupancy, and
hence we need smaller stacks in LDS.
Rhys earlier had a patch that just put the N entries closest to the
root in LDS and the rest in scratch. However, this is not ideal for
performance as most of the activity is happening away from the root,
near the leaves. Of course we can't just switch it around, as the
leaf activity likely isn't happening all the way at the end of the
stack.
So what we do is make the LDS stack kinda a ringbuffer by always
accessing it using the stack index modulo the buffer size (always
a power of two so we can efficiently mask). If we then do not have
free space in this buffer we evict the entries closest to the root
to scratch and if we hit the "bottom" of the LDS space we load from
scratch.
Some rough perf numbers for indication with Q2RTX:
| evicting | LDS entries | perf |
|----------|-------------|------|
| no | 76 | 55% |
| no | 32 | 100% |
| no | 24 | 105% |
| yes | 32 | 95% |
| yes | 16 | 100% |
| yes | 8 | 90% |
| yes | 4 | 75% |
(For the case with 4 entries we need to do some extra accounting as
a full batch may not be available to evict)
So an obvious choice is to use a stack of 16 entries.
One might wonder if Q2RTX perf is mainly good due to BVHs with very
little geometry and hence low depth, so I also did some profiling
with control. This is done with RGP instruction timing, so this is
instructions executed not weighted for enabled masks, i.e. divergence
effects included.
| game | LDS entries | scratch action | fraction of iterations |
|---------|-------------|----------------|------------------------|
| Control | 8 | store | 10.3% |
| Control | 8 | load | 34.8% |
| Control | 16 | store | 0.58% |
| Control | 16 | load | 2.62% |
| Q2RTX | 16 | store | 1.00% |
| Q2RTX | 16 | load | 3.07% |
So Q2RTX doesn't seem like an unreasonably good case for this
algorithm.
On the implementation side, we can always place the scratch stack at
address 0 by just reserving the scratch space, and in the case of fixed
callstack size moving that up. In the dynamic case the dynamic stack
base already takes any reserved scratch space into account.
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18541>
This reverts commit 3750663c72.
Doing things with derefs adds extra instructions for multiplying the
index with the element size, e.g.
BBF0_13:
s_waitcnt vmcnt(0)
v_mov_b32_e32 v27, v55
s_mov_b32 s23, exec_lo
v_cmpx_ne_i32_e32 -1, v27
s_cbranch_execz _L14
BBF0_14:
v_lshlrev_b32_e32 v48, 2, v46 <--
ds_write_b32 v48, v27
v_add_nc_u32_e32 v46, 32, v46
_L14:
s_mov_b32 exec_lo, s23
v_mov_b32_e32 v27, v54
s_mov_b32 s23, exec_lo
v_cmpx_ne_i32_e32 -1, v27
s_cbranch_execz _L15
BBF0_15:
v_lshlrev_b32_e32 v48, 2, v46 <--
ds_write_b32 v48, v27
v_add_nc_u32_e32 v46, 32, v46
On Q2RTC indirect lighting this saves about 2.3 VALU instructions
per loop iteration, which is ~4% of VALU instructions (we're at
58 per iteration now according to RGP).
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18538>
