In the Vulkan driver we use a single flat input instead of a uniform
because setting up push constants is more disruptive to the pipeline than
setting up another vertex input. This uses the number of uniforms as a key
to keep it working for the GL driver.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
It's used by brw_compile_gs in brw_vec4_gs_visitor.cpp so it needs to be in
a file that's linked into libi965_compiler.la.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Note that old mesa + new LLVM or new mesa + old LLVM breaks
with this change and the corresponding LLVM change (D18559).
For LLVM version <= 3.8 we use the old method, but we can't detect
people using a post 3.8 svn version that is still too old.
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
With this change we create the UBO and SSBO arrays separately from the
beginning rather than putting them into a combined array and splitting
it apart later.
A bug is with UBO and SSBO stage reference querying is also fixed as
we now use the block index to lookup the references in the separate arrays
not the combined buffer block array.
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
We had an implicit assumption that the phi src was assigned in it's
source (pred) block leading into the phi. But this is not true with
NIR, so we can't just ignore the source block specified in the
nir_phi_src. Insert an extra mov in the source block. If it is not
required the CP pass will take it back out again.
Fixes:
./tests/spec/glsl-1.10/execution/vs-call-in-nested-loop.shader_test
./tests/spec/glsl-1.10/execution/vs-inner-loop-modifies-outer-loop-var.shader_test
and probably others.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
The frontend inserts (abs) and (neg)'s to convert between NIR boolean
(~0/0) and native boolean (1/0). So we'd end up with things like:
cmps.s.ge r1.x, ...
absneg.s r1.x, (neg)r1.x
absneg.s r1.x, (abs)r1.x
sel.b32 r2.x, r0.x, r1.x, r0.y
The (neg) already gets collapsed due to the following (abs). Now by
realizing that r1.x comes from a cmps.s instruction, we can drop the
(abs) as well.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Instead of failing an assertion, disable DCC and CMASK on the first export
that needs it, and merge the external usage flags.
v2: clear the EXPLICIT_FLUSH flag if it's not set; whitespace fixes
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
This patch enables an EGL extension, EGL_KHR_reusable_sync.
This new extension basically provides a way for multiple APIs or
threads to be excuted synchronously via a "reusable sync"
primitive shared by those threads/API calls.
This was implemented based on the specification at
https://www.khronos.org/registry/egl/extensions/KHR/EGL_KHR_reusable_sync.txt
v2
- use thread functions defined in C11/threads.h instead of
using direct pthread calls
- make the timeout set with reference to CLOCK_MONOTONIC
- cleaned up the way expiration time is calculated
- (bug fix) in dri2_client_wait_sync, case EGL_SYNC_CL_EVENT_KHR
has been added.
- (bug fix) in dri2_destroy_sync, return from cond_broadcast
call is now stored in 'err' intead of 'ret' to prevent 'ret'
from being reset to 'EGL_FALSE' even in successful case
- corrected minor syntax problems
v3
- dri2_egl_unref_sync now became 'void' type. No more error check
is needed for this function call as a result.
- (bug fix) resolved issue with duplicated unlocking of display in
eglClientWaitSync when type of sync is "EGL_KHR_REUSABLE_SYNC"
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Otherwise we end up with funny things like:
mov.f32f32 r0.x, r1.y
mov.f32f32 r0.x, r1.y
(It doesn't happen as much after fixing the problem w/ CP into phi src,
but it can still happen since we aren't too clever about generating phi
sources in the first place.)
Signed-off-by: Rob Clark <robclark@freedesktop.org>
The block defining a phi source might not have been executed. If we
allow copy propagation, we could end up pointing to a src instruction in
the wrong block.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Been on my TODO list for a while. If nothing else this will make gdb
properly grok the opc_t enum.
This first step preserves ir3_instruction::category (with an added
assert that category matches what is encoded in opc_t). Next step is
to drop the category field (and arg to ir3_instr_create()), but that
is split into next commit for bisectability and so that we can run
piglit in the intermediate state to flush out any problems.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
They are compute-shader only and that's where the code for doing atomics on
shared variables lives so it seemes to make sense.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Our hardware requires an LOD for all texelFetch commands even if they are
on buffer textures. GLSL IR gives us an LOD of 0 in that case, but the LOD
is really rather meaningless. This commit allows other NIR producers to be
more lazy and not provide one at all.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
%ld and %lu aren't the right format specifiers for int64_t and uint64_t
on 32-bit (x86) systems. They're %zu on Linux and %Iu on Windows.
Use the standard C99 macros in hopes that they work everywhere.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
There may not be a previous block. In this case, there's no real work
to do, so just continue on to the next one.
v2: Update for bblock->prev() API change.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
The bblock_t::prev/prev_const/next/next_const API returns bblock_t
pointers, rather than exec_nodes. So it's a bit surprising.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
lower_variable_index_to_cond_assign() did not handle system values.
gl_SampleMaskIn[] is a system value, and also an array. Accessing it
with a variable index would trigger an unreachable() assert.
Rather than adding a new EmitNoIndirectSystemValues flag, we simply
lower unconditionally. There is exactly one case where this occurs,
and for all current drivers, lowering produces optimal code. Even
for future drivers with 32x MSAA, it produces reasonable code.
Fixes Piglit's new samplemaskin-indirect test. Also fixes many ES31-CTS
tests when OES_sample_variables is enabled.
Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
In the first pass of implementing exact handling, I made a mistake with
search-and-replace. In particular, we only reallly handled exact/inexact
on the root of the tree. Instead, we need to check every node in the tree
for an exact/inexact match. As an example of this, consider the following
GLSL code
precise float a = b + c;
if (a < 0) {
do_stuff();
}
In that case, only the add will be declared "exact" and an expression that
looks for "b + c < 0" will still match and replace it with "b < -c" which
may yield different results. The solution is to simply bail if any of the
values are exact when matching an inexact expression.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
The SIN and COS instructions on Intel hardware can produce values
slightly outside of the [-1.0, 1.0] range for a small set of values.
Obviously, this can break everyone's expectations about trig functions.
According to an internal presentation, the COS instruction can produce
a value up to 1.000027 for inputs in the range (0.08296, 0.09888). One
suggested workaround is to multiply by 0.99997, scaling down the
amplitude slightly. Apparently this also minimizes the error function,
reducing the maximum error from 0.00006 to about 0.00003.
When enabled, fixes 16 dEQP precision tests
dEQP-GLES31.functional.shaders.builtin_functions.precision.
{cos,sin}.{highp,mediump}_compute.{scalar,vec2,vec4,vec4}.
at the cost of making every sin and cos call more expensive (about
twice the number of cycles on recent hardware). Enabling this
option has been shown to reduce GPUTest Volplosion performance by
about 10%.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
See commit 3b0279a69 - this restriction is documented in the "Surface
Format" field of RENDER_SURFACE_STATE.
Looking at newer documentation, this restriction appears to exist on
Haswell, but no longer applies on Gen8+.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
