Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Cc: Eric Anholt <eric@anholt.net>
Cc: Paul Berry <stereotype441@gmail.com>
Search and replace:
][0] -> ].x
][1] -> ].y
][2] -> ].z
][3] -> ].w
Fixes piglit tests inverse-mat[234].{vert,frag}. These tests call the
inverse function with constant parameters and expect proper constant
folding to happen. My suspicion is that this patch papers over some bug
in constant propagation involving array accesses.
Either way, all of these accesses eventually get lowered to swizzles.
This cuts out the middle man (saving a trivial amount of CPU).
NOTE: This is a candidate for the 9.1 branch.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Cc: Eric Anholt <eric@anholt.net>
Cc: Paul Berry <stereotype441@gmail.com>
Since the case was missing bec4->get_scalar_type() would return bvec4,
but vec4->get_scalar_type() would return float.
NOTE: This is a candidate for stable branches.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
"discard" instructions generate HALT instructions which jump to a final
HALT near the end of the shader. Previously, fs_generator created this
final jump target when it saw the first FS_OPCODE_FB_WRITE, causing it
to jump right before the FB write epilogue. This is normally good.
However, INTEL_DEBUG=shader_time also has an epilogue section which
records the final timestamp. The frontend emits IR for this just before
FS_OPCODE_FB_WRITE. Unfortunately, this led to the following ordering:
1. Shader Time Epilogue
2. Final HALT (where discards jump)
3. Framebuffer Write Epilogue
This meant that discarded pixels completely skipped the shader time
epilogue, causing no ending timestamp to be written. This obviously
led to inaccurate results.
This patch adds a new FS_OPCODE_PLACEHOLDER_HALT in the IR stream just
before any epilogue sections. This is where the final HALT should be
generated, and makes it easy to ensure the correct ordering:
1. Final HALT
2. Shader Time Epilogue
3. Framebuffer Write Epilogue
For shaders that don't discard, this opcode compiles away to nothing.
The scheduler adds barrier dependencies to make sure that it doesn't
get moved above any FS_OPCODE_DISCARD_JUMP instructions.
One 8-wide shader in GLBenchmark 2.7 dropped from 2291.67 Gcycles to
a mere 5.13 Gcycles.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
I'd previously added the minimum names to understand my dumps, but this
makes dumps in general much easier to read.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
v2 [mattst88]:
- Rebase.
- #define GL_ARB_texture_query_lod to 1.
- Remove comma after ir_lod in ir.h for MSVC.
- Handled ir_lod in ir_hv_accept.cpp, ir_rvalue_visitor.cpp,
opt_tree_grafting.cpp.
- Rename textureQueryLOD to textureQueryLod, see
https://www.khronos.org/bugzilla/show_bug.cgi?id=821
- Fix ir_reader of (lod ...).
v3 [mattst88]:
- Rename textureQueryLod to textureQueryLOD, pending resolution of
Khronos 821.
- Add ir_lod case to ir_to_mesa.cpp.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
x before
+ after
+------------------------------------------------------------------------------+
| x x + |
| xx ++ x + |
| xx ++ + xx ++ |
|x xxx x+++++ + xxx x*x+*+++ + x +|
| |_____|____________A______A____M____M_|_______| |
+------------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 23 8083.78 8287.83 8205.55 8162.7461 68.307951
+ 23 8107.56 8358.74 8224.33 8186.1765 71.506301
No difference proven at 95.0% confidence
Reviewed-by: Eric Anholt <eric@anholt.net>
Set its latency to what happens to be the default floating-point
instruction latency. One day we may want to handle latency based on
register bank information.
Reviewed-by: Eric Anholt <eric@anholt.net>
Similar enough that we can try to use shared code.
v2: fix a stupid bug using wrong variable causing mayhem with Inf and NaNs.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com
Previously at least i915 failed to provide an implementation, but
exposed ARB_internalformat_query anyway, leading to crashes when
QueryInternalformativ was called.
Default implementation just returns 1 for everything, so is suitable for
any driver which does not support multisampling.
V2: - Move from intel to core mesa.
Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
There's more, but this only adds (most) of the counters that are
handled directly by the shader processors.
The other counter domains are not handled on the multiprocessor and
there are no FIFO object methods for configuring them.
Instead, they have to be programmed by the kernel via PCOUNTER, and
the interface for this isn't in place yet.
We can use %-6s%-6s rather than manually counting characters, resulting
in much more readable code.
This necessitates a small secondary change: using "total fs16" and ""
now causes the "" string to be padded out to 6 characters, resulting in
too much whitespace. Splitting it into "total" and "fs16" produces the
same output as before.
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
This much more accurately reflects the cost of the vertex shader, since
the payload setup is often a significant fraction of the instructions in
the VS.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This incidentally also teaches it a bit about gen6 math -- we now allow
unswizzled, unmodified GRF temps as the sources for math.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Previously, if you just wrote a constant color to the render target, no
time got noted at all. This is convenient for doing single-instruction
timings, but not so much for actual program analysis.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This avoids conflicts between shader_time and FB writes, so we can include
more of the program under our profiling. This does mean hiding more of
the message setup from the optimizer, which doesn't have a way to handle
multi-reg sends from GRFs.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Ken asked me the other day what -1 vs 0 vs 3 vs other meant in our shader
names, and I realized that it was really unclear. I'd like to do even
better, like noting which one is the clear shader, but that would require
exposing the metaops struct to the driver.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
We weren't correctly propagating the samplers and sampler views
when they were related to geometry shaders.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
We were allocating the output buffer but using the input
primitives. We need to allocate that buffer using the
maximum number of output, not input, primitives.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Required by more modern examples. Like BRK but with a condition.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
TGSI semantics currently require an implicit endprim at the end
of GS if an ending primitive hasn't been emitted.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
This commits implements code generation of the geometry shaders in
the SOA paths. All the code is there but bugs are likely present.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Allows executing gs on up to 4 primitives at a time. Will also be
required by the llvm code because there we definitely don't want
to flush with just a single primitive.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
To be able to add llvm paths later on we need to have some common
interface for them.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
The member was never used and we'll need to handle it differently
because gs will also need samplers/textures setup.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
A few tests were missing this crucial property.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Same as on r600, trace cs execution by writting cs offset after each
states, this allow to pin point lockup inside command stream and
narrow down the scope of lockup investigation.
v2: Use WRITE_DATA packet instead of WRITE_MEM
v3: Remove useless nop packet
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Fixes various test fallout from 90b5a2425a on Pineview, which claims to
support ARB_internalformat_query but doesn't actually provide the
driverfunc.
That driver is still broken [GetInternalformativ will still segfault!]
but it was silly to be going through the sample count logic in the
nonmultisampling case at all.
Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Ian Romanick
Kenneth Graunke
Eric Anholt
Matt Turner
Dave Airlie
Roland Scheidegger
Chris Forbes
Christoph Bumiller
Brian Paul
Marek Olšák
Zack Rusin
Jerome Glisse
Christian König