| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On at least Sky Lake, ce0 does not contain the full story as far as enabled
channels goes. It is possible to have completely disabled channels where
the corresponding bits in ce0 are 1. In order to get the correct execution
mask, you have to mask off those channels which were disabled from the
beginning by taking the AND of ce0 with either sr0.2 or sr0.3 depending on
the shader stage. Failure to do so can result in FIND_LIVE_CHANNEL
returning a completely dead channel.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Cc: Francisco Jerez <currojerez@riseup.net>
[ Francisco Jerez: Fix a couple of typos, add mask register type
assertion, clarify reason why ce0 can have bits set for disabled
channels, clarify that this may only be a problem when thread
dispatch doesn't pack channels tightly in the SIMD thread. Apply
same treatment to Align16 path. ]
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
| |
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Generated by:
sed -i -e 's/brw_device_info/gen_device_info/g' src/intel/**/*.c
sed -i -e 's/brw_device_info/gen_device_info/g' src/intel/**/*.h
sed -i -e 's/brw_device_info/gen_device_info/g' **/i965/*.c
sed -i -e 's/brw_device_info/gen_device_info/g' **/i965/*.cpp
sed -i -e 's/brw_device_info/gen_device_info/g' **/i965/*.h
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Without this, we would pass over the instructions in the SIMD8 program
(which is located earlier in the buffer) when brw_set_uip_jip() is
called to handle the SIMD16 program.
The assertion about compacted control flow was bogus: halt, cont, break
cannot be compacted because they have both JIP and UIP. Instead, we
should never see a compacted instruction in this code at all.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
|
|
|
|
| |
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
| |
brw_set_dp_read_message() was setting the data cache as send message
SFID on Gen7+ hardware, ignoring the target cache specified by the
caller. Some of the callers were passing a bogus target cache value
as argument relying on brw_set_dp_read_message not to take it into
account. Fix them too.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
According to HSW's PRM, vol02b, the DIM instruction has the following
restriction:
"Restriction : src0 must be immediate. src0 must specify the :f (F, Float)
type encoding but is an immediate 64-bit DF (Double Float) value. dst
must have type DF."
This commit allows to upload the immediate 64-bit DF value to the source
of a DIM instruction even when it is of float type encoding.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
|
|
|
|
|
|
|
|
|
|
| |
v2 (Matt):
- Take a DF source argument for the DIM instruction emission
in the visitors.
- Indentation.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit c1107cec44ab030c7fcc97c67baa12df1cc9d7b5.
Apparently the hardware spec text I quoted in the commit message was
outright lying about scalar source math being supported on SNB, the
hardware seems to load 32 contiguous bits of data for each channel
regardless of the regioning mode. Fixes regressions in the following
CTS tests (which we didn't catch early due to CTS being temporarily
disabled in our CI system):
es2-cts.gtf.gl.atan.atan_vec3_frag_xvary
es2-cts.gtf.gl.cos.cos_vec2_frag_xvary
es2-cts.gtf.gl.atan.atan_vec2_frag_xvary
es2-cts.gtf.gl.pow.pow_vec2_frag_xvary_yconsthalf
es2-cts.gtf.gl.cos.cos_float_frag_xvary
es2-cts.gtf.gl.pow.pow_float_frag_xvary_yconsthalf
es2-cts.gtf.gl.atan.atan_vec3_frag_xvaryyvary
es2-cts.gtf.gl.pow.pow_vec3_frag_xvary_yconsthalf
es2-cts.gtf.gl.cos.cos_vec3_frag_xvary
es2-cts.gtf.gl.atan.atan_vec2_frag_xvaryyvary
Cc: mesa-stable@lists.freedesktop.org
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96346
Reported-by: Mark Janes <mark.a.janes@intel.com>
Acked-by: Matt Turner <mattst88@gmail.com>
|
|
|
|
|
|
|
|
|
|
| |
Among other thigs, fix a gpu hang when using INTEL_DEBUG=shader_time
for any shader.
Signed-off-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I haven't found any evidence that this isn't supported by the
hardware, in fact according to the SNB hardware spec:
"The supported regioning modes for math instructions are align16,
align1 with the following restrictions:
- Scalar source is supported.
[...]
- Source and destination offset must be the same, except the case of
scalar source."
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Alternatively we could have extended the current semantics to 32-wide
mode by changing brw_broadcast() to emit multiple indexed MOV
instructions in the generator copying the selected value to all
destination registers, but it seemed rather silly to waste EU cycles
unnecessarily copying the exact same value 32 times in the GRF.
The vstride change in the Align16 path is required to avoid assertions
in validate_reg() since the change causes the execution size of the
MOV and SEL instructions to be equal to the source region width.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
|
|
|
| |
This makes FIND_LIVE_CHANNEL behave like a normal instruction for
non-zero quarter control. On Gen8+ we just leave the quarter control
field of the emitted FBL instruction set to the default value so the
hardware applies the expected shift to the execution mask signals. On
Gen7 we apply the offset manually by specifying a non-zero subregister
offset in the source region of the FBL instruction.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
|
|
|
|
| |
FIND_LIVE_CHANNEL.
Due to a Gen7-specific hardware bug native 32-wide instructions get
the lower 16 bits of the execution mask applied incorrectly to both
halves of the instruction, so the MOV trick we currently use wouldn't
work. Instead emit multiple 16-wide MOV instructions in 32-wide mode
in order to cover the whole execution mask.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
| |
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
| |
Gen7 hardware expects the block size field in the message descriptor
to be the number of registers minus one instead of the log2 of the
number of registers.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
| |
We don't want to emit a 32-wide send message in 32-wide programs. The
memory fence message should have the same effect regardless of the
execution size (as long as it's valid) so just set it to one.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
|
|
|
| |
In SIMD32 programs the compiler is responsible for providing the
appropriate half of the sample mask in the message header, so the
first and third quarters both map to the first slot group of the
provided 16-bit half, while the second and fourth quarters map to the
second slot group -- IOW they should be equivalent to 1Q and 2Q modulo
two.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
| |
Most of these are bugs because the intended execution size of an
instruction and the dispatch width of the shader aren't necessarily
the same (especially in SIMD32 programs).
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
|
|
| |
generation.
This was kind of an abuse of p->compressed, dataport send message
instructions are always uncompressed. Use the current execution size
instead since p->compressed is on its way out.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
|
| |
instructions.
p->compressed won't work for SIMD32, we should just be using the
execution size value specified via p->current instead.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
|
|
|
|
|
| |
Most of these were resetting quarter control to zero incorrectly even
though everything they needed to do was disable instruction
compression -- The brw_SAMPLE() case was doing the right thing but it
can be simplified slightly by using the new compression control
interface.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
|
|
|
|
|
| |
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We've apparently always been botching JIP for sequences such as:
do
cmp.f0.0 ...
(+f0.0) break
...
do
...
while
...
while
Because the "do" instruction doesn't actually exist, the inner "while"
is at the same depth as the "break". brw_find_next_block_end() thus
mistook the inner "while" as the end of the loop containing the "break",
and set the "break" to point to the wrong place.
Only "while" instructions that jump before our instruction are relevant.
We need to ignore the rest, as they're sibling control flow nodes (or
children, but this was already handled by the depth == 0 check).
See also commit 1ac1581f3889d5f7e6e231c05651f44fbd80f0b6.
This prevents channel masks from being screwed up, and fixes GPU
hangs(*) in dEQP-GLES31.functional.shaders.multisample_interpolation.
interpolate_at_sample.centroid_qualified.multisample_texture_16.
The test ended up executing code with no channels enabled, and that
code contained FIND_LIVE_CHANNEL, which returned 8 (out of range for
a SIMD8 program), which then was used in indirect GRF addressing,
which randomly got a boolean value (0xFFFFFFFF), interpreted it as
a sample ID, OR'd it into an indirect send message descriptor,
which corrupted the message length, sending a pixel interpolator
message with mlen 15, which is illegal. Whew :)
(*) Technically, the test doesn't GPU hang currently, but only
because another bug prevents it from issuing pixel interpolator
messages entirely...with that fixed, it hangs.
Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
| |
I need to use this in an additional place.
Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
| |
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
|
|
|
|
|
|
|
|
|
| |
v2 (Sam):
- Remove 'however' from the comment (Topi)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
|
|
| |
v2:
- set 3src_src_type for BRW_REGISTER_TYPE_DF (Connor)
Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I noticed some heap corruption running virgl tests, and valgrind
helped me to track it down to the following error:
==29272== Invalid write of size 4
==29272== at 0x90283D4: push_loop_stack (brw_eu_emit.c:1307)
==29272== by 0x9029A7D: brw_DO (brw_eu_emit.c:1750)
==29272== by 0x90554B0: fs_generator::generate_code(cfg_t const*, int) (brw_fs_generator.cpp:1999)
==29272== by 0x904491F: brw_compile_fs (brw_fs.cpp:5685)
==29272== by 0x8FC5DC5: brw_codegen_wm_prog (brw_wm.c:137)
==29272== by 0x8FC7663: brw_fs_precompile (brw_wm.c:638)
==29272== by 0x8FA4040: brw_shader_precompile(gl_context*, gl_shader_program*) (brw_link.cpp:51)
==29272== by 0x8FA4A9A: brw_link_shader (brw_link.cpp:260)
==29272== by 0x8DEF751: _mesa_glsl_link_shader (ir_to_mesa.cpp:3006)
==29272== by 0x8C84325: _mesa_link_program (shaderapi.c:1042)
==29272== by 0x8C851D7: _mesa_LinkProgram (shaderapi.c:1515)
==29272== by 0x4E4B8E8: add_shader_program (vrend_renderer.c:880)
==29272== Address 0xf2f3cb0 is 0 bytes after a block of size 112 alloc'd
==29272== at 0x4C2AA98: calloc (vg_replace_malloc.c:711)
==29272== by 0x8ED11F7: ralloc_size (ralloc.c:113)
==29272== by 0x8ED1282: rzalloc_size (ralloc.c:134)
==29272== by 0x8ED14C0: rzalloc_array_size (ralloc.c:196)
==29272== by 0x9019C7B: brw_init_codegen (brw_eu.c:291)
==29272== by 0x904F565: fs_generator::fs_generator(brw_compiler const*, void*, void*, void const*, brw_stage_prog_data*, unsigned int, bool, gl_shader_stage) (brw_fs_generator.cpp:124)
==29272== by 0x9044883: brw_compile_fs (brw_fs.cpp:5675)
==29272== by 0x8FC5DC5: brw_codegen_wm_prog (brw_wm.c:137)
==29272== by 0x8FC7663: brw_fs_precompile (brw_wm.c:638)
==29272== by 0x8FA4040: brw_shader_precompile(gl_context*, gl_shader_program*) (brw_link.cpp:51)
==29272== by 0x8FA4A9A: brw_link_shader (brw_link.cpp:260)
==29272== by 0x8DEF751: _mesa_glsl_link_shader (ir_to_mesa.cpp:3006)
if_depth_in_loop is an array of size p->loop_stack_array_size, and
push_loop_stack() will access if_depth_in_loop[p->loop_stack_depth+1],
thus the condition to grow the array should be
p->loop_stack_array_size <= (p->loop_stack_depth + 1) (it's currently
off by 2...)
This can be reproduced by running the following test with virgl test
server:
LIBGL_ALWAYS_SOFTWARE=y GALLIUM_DRIVER=virpipe bin/shader_runner
./tests/shaders/glsl-fs-unroll-explosion.shader_test -auto
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This code in brw_set_dest adjusts the execution size of any instruction
with a dst.width < 8. However, we don't want to do this with instructions
operating on doubles, since these will have a width of 4, but still
need an execution size of 8 (for SIMD8). Unfortunately, we can't just check
the size of the operands involved to detect if we are doing an operation on
doubles, because we can have instructions that do operations on double
operands interpreted as UD, operating on any of its 2 32-bit components.
Previous commits have made it so we never emit instructions with a horizontal
width of 4 that don't have the correct execution size set for gen6+, so
we can skip it in this case, avoiding the conflicts with fp64 requirements.
Expanding the same fix to other hardware generations requires many more
changes but since we are not targetting fp64 support on them
wer don't really care for now.
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
|
|
|
|
| |
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
|
|
|
|
| |
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
|
|
|
|
|
|
|
| |
v2: NOP should have an execsize of 1 (Matt)
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
|
|
|
|
|
|
|
|
| |
And replace brw_swizzle1() with brw_swizzle(). Seems slightly cleaner
and will allow reusing brw_swizzle() in the vec4 back-end more easily.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Without this, on SIMD 16 the send instruction destination will appear
to write more than one destination register, causing the simulator to
report an error.
Of course, the send instruction can actually write more than one
destination register regardless of the type set for the destination,
so this is a bit strange.
Suggested-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
While most align16 instructions only support a SubRegNum of 0 or 4
(using swizzling to control the other channels), 3-src instructions
actually support arbitrary SubRegNums. When the RepCtrl bit is set,
we believe it ignores the swizzle and uses the equivalent of a <0,1,0>
region from the subnr.
In the past, we adopted a vec4-centric approach of specifying subnr of
0 or 4 and a swizzle, then having brw_eu_emit.c convert that to a proper
SubRegNum. This isn't a great fit for the scalar backend, where we
don't set swizzles at all, and happily set subnrs in the range [0, 7].
This patch changes brw_eu_emit.c to use subnr and swizzle directly,
relying on the higher levels to set them sensibly.
This should fix problems where scalar sources get copy propagated into
3-src instructions in the FS backend. I've only observed this with
TES push model inputs, but I suppose it could happen in other cases.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
|
|
|
|
|
|
|
|
| |
I want to use this directly from brw_vec4_generator.cpp.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We've apparently always been botching JIP for sequences such as:
do
cmp.f0.0 ...
(+f0.0) break
...
if
...
else
...
endif
...
while
Normally, UIP is supposed to point to the final destination of the jump,
while in nested control flow, JIP is supposed to point to the end of the
current nesting level. It essentially bounces out of the current nested
control flow, to an instruction that has a JIP which bounces out another
level, and so on.
In the above example, when setting JIP for the BREAK, we call
brw_find_next_block_end(), which begins a search after the BREAK for the
next ENDIF, ELSE, WHILE, or HALT. It ignores the IF and finds the ELSE,
setting JIP there.
This makes no sense at all. The break is supposed to skip over the
whole if/else/endif block entirely. They have a sibling relationship,
not a nesting relationship.
This patch fixes brw_find_next_block_end() to track depth as it does
its search, and ignore anything not at depth 0. So when it sees the
IF, it ignores everything until after the ENDIF. That way, it finds
the end of the right block.
I noticed this while reading some assembly code. We believe jumping
earlier is harmless, but makes the EU walk through a bunch of disabled
instructions for no reason. I noticed that GLBenchmark Manhattan had
a shader that contained a BREAK with a bogus JIP, but didn't measure
any performance improvement (it's likely miniscule, if there is any).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The thread scratch space is thread-local so using the full IA-coherent
stateless surface index (255 since Gen8) is unnecessary and
potentially expensive. On Gen8 and early steppings of Gen9 this is
not a functional change because the kernel already sets bit 4 of
HDC_CHICKEN0 which overrides all HDC memory access to be non-coherent
in order to workaround a hardware bug.
This happens to fix a full system hang when running any spilling code
on a pre-production SKL GT4e machine I have on my desk (forcing all
HDC access to non-coherent from the kernel up to stepping F0 might be
a good idea though regardless of this patch), and improves performance
of the OglPSBump2 SynMark benchmark run with INTEL_DEBUG=spill_fs by
33% (11 runs, 5% significance) on a production SKL GT2 (on which HDC
IA-coherency is apparently functional so it wouldn't make sense to
disable globally).
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
|
|
|
|
|
| |
Reviewed-by: Emil Velikov <emil.velikov@collabora.co.uk>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
| |
Reviewed-by: Emil Velikov <emil.velikov@collabora.co.uk>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Generated by
sed -i -e 's/\.bits\././g' *.c *.h *.cpp
sed -i -e 's/dw1\.//g' *.c *.h *.cpp
and then reverting changes to comments in gen7_blorp.cpp and
brw_fs_generator.cpp.
There wasn't any utility offered by forcing the programmer to list these
to access their fields. Removing them will reduce churn in future
commits.
This is C11 (and gcc has apparently supported it for sometime
"compatibility with other compilers")
See https://gcc.gnu.org/onlinedocs/gcc/Unnamed-Fields.html
Reviewed-by: Emil Velikov <emil.velikov@collabora.co.uk>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
| |
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
|
|
|
|
|
|
| |
This has been wrong since the initial import of the i965 driver.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
|
| |
We always set the mask to 0xffff, which is what it defaults to when no
header is present. Let's drop the header instead.
v2: Only remove header for untyped reads. Typed reads always need the
header.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Signed-off-by: Kristian Høgsberg Kristensen <krh@bitplanet.net>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If a non-const sample number is given to interpolateAtSample it will
now generate an indirect send message with the sample ID similar to
how non-const sampler array indexing works. Previously non-const
values were ignored and instead it ended up using a constant 0 value.
The generator will try to determine if the sample ID is dynamically
uniform via nir_src_is_dynamically_uniform. If not it will query the
pixel interpolator in a loop, once for each different live sample
number. The next live sample number is found using emit_uniformize. If
multiple live channels have the same sample number then they will be
handled in a single iteration of the loop. The loop is necessary
because the indirect send message doesn't seem to have a way to
specify a different value for each fragment.
This fixes the following two Piglit tests:
arb_gpu_shader5-interpolateAtSample-nonconst
arb_gpu_shader5-interpolateAtSample-dynamically-nonuniform
v2: Handle dynamically non-uniform sample ids.
v3: Remove the BREAK instruction and predicate the WHILE directly.
Make the tokens arrays const. (Matt Turner)
v4: Iterate over the live channels instead of each possible sample
number.
v5: Don't special case immediate values in
brw_pixel_interpolator_query. Make a better wrapper for the
function to set up the PI send instruction. Ensure that the SHL
instructions are scalar. (Francisco Jerez).
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
|
|
|
|
|
|
|
|
|
|
| |
compr4 is represented by setting the high bit on the MRF number.
We need to mask it out before sanity checking the register number.
Fixes ~8000 assert fails on Ironlake and G45.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=92066
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are some bug reports about shaders failing to compile in gen6
because MRF 14 is used when we need to spill. For example:
https://bugs.freedesktop.org/show_bug.cgi?id=86469
https://bugs.freedesktop.org/show_bug.cgi?id=90631
Discussion in bugzilla pointed to the fact that gen6 might actually have
24 MRF registers available instead of 16, so we could use other MRF
registers and avoid these conflicts (we still need to investigate why
some shaders need up to MRF 14 anyway, since this is not expected).
Notice that the hardware docs are not clear about this fact:
SNB PRM Vol4 Part2's "Table 5-4. MRF Registers Available in Device
Hardware" says "Number per Thread" - "24 registers"
However, SNB PRM Vol4 Part1, 1.6.1 Message Register File (MRF) says:
"Normal threads should construct their messages in m1..m15. (...)
Regardless of actual hardware implementation, the thread should
not assume th at MRF addresses above m15 wrap to legal MRF registers."
Therefore experimentation was necessary to evaluate if we had these extra
MRF registers available or not. This was tested in gen6 using MRF
registers 21..23 for spilling and doing a full piglit run (all.py) forcing
spilling of everything on the FS backend. It was also tested by doing
spilling of everything on both the FS and the VS backends with a piglit run
of shader.py. In both cases no regressions were observed. In fact, many of
these tests where helped in the cases where we forced spilling, since that
triggered the same underlying problem described in the bug reports. Here are
some results using INTEL_DEBUG=spill_fs,spill_vec4 for a shader.py run on
gen6 hardware:
Using MRFs 13..15 for spilling:
crash: 2, fail: 113, pass: 6621, skip: 5461
Using MRFs 21..23 for spilling:
crash: 2, fail: 12, pass: 6722, skip: 5461
This patch sets the ground for later patches to implement spilling
using MRF registers 21..23 in gen6.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In a later patch we will make BRW_MAX_MRF return a different value depending
on the hardware generation, but it is inconvenient to add a gen parameter
to the brw_reg functions only for the assertions, so move these to places where
we have the hardware generation available.
Ken suggested to add the asserts to brw_set_src0 and brw_set_dest since that
would make sure that we catch all uses of MRF registers, even those coming
from modules that generate native code directly, like blorp. Unfortunately,
this is very late in the process which can make things harder to debug, so add
asserts to the generator as well.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
|