Jordan added this in commit 741782b594 for
Gen7 platforms. I missed this when adding the Broadwell code.
Fixes Piglit's spec/arb_gpu_shader5/invocation-id-{basic,in-separate-gs}
with MESA_EXTENSION_OVERRIDE=GL_ARB_gpu_shader5 set.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Cc: mesa-stable@lists.freedesktop.org
While working on NIR's memory allocation model, I realized the GLSL IR
memory model was broken.
During glCompileShader, we allocate everything out of the
_mesa_glsl_parse_state context, and reparent it to gl_shader at the end.
During glLinkProgram, we allocate everything out of a temporary context,
then reparent it to the exec_list containing the linked IR.
But during brw_link_shader - the driver's final opportunity to do
lowering and optimization - we just allocated everything out of the
permanent context given to us by the linker. That memory stayed
forever.
Notably, passes like brw_fs_channel_expressions cause us to churn the
majority of the code, so we really want to free dead IR here.
Saves 125MB of memory when replaying a Dota 2 trace on Broadwell.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
This allows SIMD16 mode to work for a lot more programs. Texturing is
also more efficient in SIMD16 mode than SIMD8. Several messages don't
actually exist in SIMD8 mode, so we did SIMD16 messages and threw away
half of the data. Now we compute real data in both halves.
Also, the SIMD16 "sample" message doesn't require all three coordinate
components to exist (like the SIMD8 one), so we can shorten the message
lengths, cutting register usage a bit.
I chose to implement the visitor functionality in a separate function,
since mixing true SIMD16 with SIMD8 code that uses SIMD16 fallbacks
seemed like a mess. The new code bails on a few cases where we'd
have to do two SIMD8 messages - we just fall back to SIMD8 for now.
Improves performance in "Shadowrun: Dragonfall - Director's Cut" by
about 20% on GM45 (measured with LIBGL_SHOW_FPS=1 while standing around
in the first mission).
v2: Add ir_txf to the has_lod case (caught by Jordan Justen).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Gen5+ systems allow you to specify multiple shader programs - both SIMD8
and SIMD16 - and the hardware will automatically dispatch to the most
appropriate one, given the number of subspans to be processed.
However, that is not the case on Gen4. Instead, you program a single
shader. If you enable multiple dispatch modes (SIMD8 and SIMD16), the
shader is supposed to contain a series of jump instructions at the
beginning. The hardware will launch the shader at a small offset,
hitting one of the jumps.
We've always thought that sounds like a pain, and weren't clear how it
affected performance - is it worth having multiple shader types? So,
we never bothered with SIMD16 until now.
This patch takes a simpler approach: try and compile a SIMD16 shader.
If possible, set the no_8 flag, telling the hardware to just use the
SIMD16 variant all the time.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
This flag means to ignore the SIMD8 program and only use the SIMD16 one.
It was originally meant for repdata clear shaders, but I plan to use it
for other things on Gen4 as well.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
I've no idea why this was 4. It certainly seems wrong.
Prevents assertion failures in fp-incomplete-tex with some upcoming
patches of mine.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
The CSE algorithm will continuously allocate new ae_entry objects. As
each new basic block is exited, all of the previously allocated objects
are dumped. Instead, put them in a free list and re-use them in the
next basic block. Reduce, reuse, recycle!
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
These were added in commit f2616e56, presumably in preparation for
translating ARB vp/fp into GLSL IR. That never happened, and neither did
a lowering pass that actually generated these instructions.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
From GLSL 3.30 and GLSL ES 1.00 on, after processing the line
directive (including its new-line), the implementation should
behave as if it is compiling at the line number passed as
argument. In previous versions, it behaved as if compiling
at the passed line number + 1.
Partially fixes https://bugs.freedesktop.org/show_bug.cgi?id=88815
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
From GLSL 1.30.10, section 3.3 (Preprocessor):
"#line line source-string-number ... After processing this directive
(including its new-line), the implementation will behave as if it is
compiling at ... source string number source-string-number. Subsequent
source strings will be numbered sequentially, until another #line
directive overrides that numbering."
In the previous implementation the source number was always zero.
Subsequent source strings are still not numbered sequentially, because
in the glShaderSource implementation we are concatenating the source code
strings into one long string.
Partially fixes https://bugs.freedesktop.org/show_bug.cgi?id=88815
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Even if they only have one slice, otherwise textureSize() won't
produce correct results for the depth value.
Fixes 10 dEQP tests in this category:
dEQP-GLES3.functional.shaders.texture_functions.texturesize.sampler2darray*
Reviewed-by: Mark Janes <mark.a.janes at intel.com>
The NIR compiler frontend is an alternative to the TGSI f/e, producing
the same ir3 IR and using the same backend passes for scheduling, etc.
It is not enabled by default yet, as there are still some regressions.
To enable, use 'FD_MESA_DEBUG=nir'. It is enough to use with, for
example, xonotic or supertuxkart.
With the NIR f/e, scalarizing and a number of other lowering steps
happen in NIR, so we don't have to do them in ir3. Which simplifies the
f/e and allows the lowered instructions to pass through other
optimization stages.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Use the correct sprite replacement depending on the flip of the coord
mode, using either T or 1-T depending on whether we have an upper-left or
lower-left coordinate origin. This fixes all the point sprite piglits.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Copies nouveau_buffer and radeon_buffer. This allows a write to proceed
to an uninitialized part of a buffer even when the GPU is using the
previously-initialized portions.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Waiting on a bo being ready is handled in fd_bo_cpu_prep. No need to
keep separate timestamps around.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
A resource flush is an upload of a hypothetically-staging texture to the
GPU. For a UMA system, this will largely be a no-op or
cache-maintenance. Move the render flush logic into transfer_map where
it belongs, and clear out the transfer_flush function.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
pipe_sampler_view already contains a texture, remove the redundant
tex_resource member which pointed at the same thing.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Since NIR f/e currently encodes immediates in instructions (rather than
passing via const), we need to ensure that when const's are used the get
initialized to the proper values. Otherwise comparing NIR to TGSI
compiler, it will use proper immediate values in one case, and randomly
initialize values in the other. Which confuses ir3test.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Be smarter about propagating copies from const or immed, or with abs/neg
modifiers. Also, realize that absneg.s and absneg.f are really "fancy"
mov instructions.
This opens up the possibility to remove more copies. It helps the TGSI
frontend a bit, but will be really needed for the NIR f/e which builds
everything up in SSA form (ie. will *always* insert a mov from const or
immediate).
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Even though in the end, they map to the same bits, the backend will need
to be able to differentiate float abs/neg vs integer abs/neg. Rather
than making the backend figure it out based on instruction opcode (which
when combined with mov/absneg instructions, can be awkward), just split
out different flags for each so the frontend can signal it's intentions
more clearly. Also, since (neg) for bitwise op's is actually a bitwise-
not, split it out into bnot flag.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Add helpers for constructing SSA forms of instructions.
Only partial cat5/cat6 coverage.. but we can add stuff as needed.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
We need to pull in libnir.la and it's dependency libglsl_util.la. Also,
_mesa_error_no_memory() must be defined.
Fortunately with libnir.la (vs pulling in all of libglsl.la) we don't
also need libstdc++.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
If we want to use NIR from state trackers that don't already pull in the
whole of glsl (ie. anything other than mesa state tracker), we need a
separate more minimal libnir. Possibly NIR should be better split out
from glsl, but for now, generate a second smaller libnir.la for those
who just want NIR but not all of glsl.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Based on the algo from NV50LegalizeSSA::handleDIV() and handleMOD().
See also trans_idiv() in freedreno/ir3/ir3_compiler.c (which was an
adaptation of the nv50 code from Ilia Mirkin).
A python/numpy script which implements the same algorithm (and is
possibly useful for debugging or analysis) can be found here:
http://people.freedesktop.org/~robclark/div-lowering.py
I've tested this on i965 hacked up to insert the idiv lowering pass,
and on freedreno with NIR frontend.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Tested-by: Eric Anholt <eric@anholt.net> (vc4)
v2: discovered that i2b/b2i are also confused
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
In freedreno these get implemented as the matching f* instruction plus a
u2f to convert the result to float 1.0/0.0. But less lines of code to
just let nir_opt_algebraic handle this for us, plus opens up some small
window for other opt passes to improve (ie. if some shader ended up with
both a flt and slt with same src args, for example).
v2: use b2f rather than u2f
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Switch between the two clip space definitions already available
in hardware. Update winding order dependent state according
to the clip control state.
This change did not introduce new piglit quick.test regressions on
an Ivybridge Mobile and a GM45 Express chipset.
Also it enables and passes the clip-control and clip-control-depth-precision
tests on these two chipsets.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Mathias Froehlich <Mathias.Froehlich@web.de>
The _WindowMap can be dropped from gl_viewport_attrib now.
Simplify gl_viewport_attrib handling where possible.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Mathias Froehlich <Mathias.Froehlich@web.de>
This is the only real user of _WindowMap which has the depth
buffer scaling multiplied in. Maintain the _WindowMap of the
one and only viewport inside TNLcontext.
v2:
Remove unneeded parentheses.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Mathias Froehlich <Mathias.Froehlich@web.de>
Instead of _WindowMap just use the translation and scale
of the viewport transform directly. Thereby avoid dividing by
_DepthMaxF again.
v2:
Change order of assignments.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Mathias Froehlich <Mathias.Froehlich@web.de>
Instead of _WindowMap just use the translation and scale
of the viewport transform directly. Thereby avoid dividing by
_DepthMaxF again.
Reviewed-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Mathias Froehlich <Mathias.Froehlich@web.de>
Commit 1a170980a0 started writing to q->data[4]/[5] but kept the
per-query space at 16, which meant that in some cases we would write
past the end of the buffer. Rotate by 32, like nvc0 does. This ensures
that we always have 32 bytes in front of us, and the data writes will go
within the allocated space.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89679
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Nick Tenney <nick.tenney@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Cc: "10.4 10.5" <mesa-stable@lists.freedesktop.org>
As of da5ec2a, we allocate instruction sources out of the instruction
itself. When we realloc the texture sources we need to use the right
memory context or ralloc will get angry and assert-fail
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
