When an argument for a structure constructor or initializer doesn't
match the expected type, only Section 4.1.10 “Implicit Conversions”
are allowed to try to match that expected type.
From page 32 (page 38 of the PDF) of the GLSL 1.20 spec:
" The arguments to the constructor will be used to set the structure's
fields, in order, using one argument per field. Each argument must
be the same type as the field it sets, or be a type that can be
converted to the field's type according to Section 4.1.10 “Implicit
Conversions.”"
From page 35 (page 41 of the PDF) of the GLSL 4.20 spec:
" In all cases, the innermost initializer (i.e., not a list of
initializers enclosed in curly braces) applied to an object must
have the same type as the object being initialized or be a type that
can be converted to the object's type according to section 4.1.10
"Implicit Conversions". In the latter case, an implicit conversion
will be done on the initializer before the assignment is done."
v2: Remove also the now redundant constant conversion, the
constant_record_constructor helper and the replacement code
(Timothy).
Fixes GL44-CTS.shading_language_420pack.initializer_list_negative
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Signed-off-by: Andres Gomez <agomez@igalia.com>
v2: Refactor also the conversion to constant and replacement code
(Timothy).
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Signed-off-by: Andres Gomez <agomez@igalia.com>
Implicit conversions were added in the GLSL 1.20 spec version.
v2: Join the checks for GLSL 1.10 and ESSL (Timothy).
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Signed-off-by: Andres Gomez <agomez@igalia.com>
Previously, for every input, we moved the dispatch mask to the flag
register, then emitted two predicated PLN instructions, one with
centroid barycentric coordinates (for normal pixels), and one with
pixel barycentric coordinates (for unlit helper pixels).
Instead, we can simply emit a set of predicated MOVs at the top of
the program which copy the pixel barycentric coordinates over the
centroid ones for unlit helper pixel channels. Then, we can just
use normal PLNs.
On Sandybridge:
total instructions in shared programs: 7538470 -> 7534500 (-0.05%)
instructions in affected programs: 101268 -> 97298 (-3.92%)
helped: 705
HURT: 9 (all of which are SIMD16 programs)
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Sync now uses a callback to ensure that it's called by the last
thread moving past a DC. This will help with the new counter
handling.
Signed-off-by: Tim Rowley <timothy.o.rowley@intel.com>
By allocating and initializing the matrices at context creation, the OS
couldn't even overcommit the pages. This saves about 63k (out of 946k) of
maximum memory size according to massif on simulated vc4
glsl-algebraic-add-add-1. It also means we could potentially relax the
maximum stack sizes, but that should be a separate commit.
v2: Drop redundant Top update, explain why the stack is small at init
time.
Reviewed-by: Brian Paul <brianp@vmware.com>
It's only used for rarely-used deprecated GL features
(feedback/rasterpos), so we can skip the memory allocation and
initialization for it most of the time.
Saves about 659k (out of 1605k) of maximum memory size according to massif
on simulated vc4 glsl-algebraic-add-add-1
Reviewed-by: Brian Paul <brianp@vmware.com>
This works out to be a wash in terms of memory usage: We use more memory
to store the separate ALU instructions, but we optimize out a lot of code
as well. The main result, though, is that we do more of our work at link
time rather than draw time.
We don't want to bake the whole array into the FS key, because of the
hashing overhead. But we can keep a set of the arrays seen, and use a
pointer to the copy in as the array's proxy.
Between this and the previous patch, gl-1.0-blend-func now passes on
hardware, where previously it was filling the 256MB CMA area with shaders
and OOMing.
Drops 712 shaders from shader-db.
I found a shader in Tales of Maj'Eyal that contains:
if ssa_21 {
block block_1:
/* preds: block_0 */
...instructions that prevent the select peephole...
vec1 32 ssa_23 = imov ssa_4
vec1 32 ssa_24 = imov ssa_4.y
vec1 32 ssa_25 = imov ssa_4.z
/* succs: block_3 */
} else {
block block_2:
/* preds: block_0 */
vec1 32 ssa_26 = imov ssa_4
vec1 32 ssa_27 = imov ssa_4.y
vec1 32 ssa_28 = imov ssa_4.z
/* succs: block_3 */
}
block block_3:
/* preds: block_1 block_2 */
vec1 32 ssa_29 = phi block_1: ssa_23, block_2: ssa_26
vec1 32 ssa_30 = phi block_1: ssa_24, block_2: ssa_27
vec1 32 ssa_31 = phi block_1: ssa_25, block_2: ssa_28
Here, copy propagation will bail because phis cannot perform swizzles,
and CSE won't do anything because there is no dominance relationship
between the imovs. By making nir_opt_remove_phis handle identical moves,
we can eliminate the phis and rewrite everything to use ssa_4 directly,
so all the moves become dead and get eliminated.
I don't think we need to check "exact" - just the alu sources.
Presumably phi sources should match in their exactness.
On Broadwell:
total instructions in shared programs: 11639872 -> 11638535 (-0.01%)
instructions in affected programs: 134222 -> 132885 (-1.00%)
helped: 338
HURT: 0
v2: Fix return value to be NULL, not false (caught by Iago).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Previously, we allocated a new VGRF for every undefined definition.
Instead, this patch makes us allocate a new VGRF for every use of an
undefined definition. This makes sure that undefined values are
fully independent of one another, and have live ranges limited to
their single use. This allows register coalescing to combine the
source and destination of MOVs from undefined sources, eliminating
the MOV altogether.
On Broadwell:
total instructions in shared programs: 11641187 -> 11640214 (-0.01%)
instructions in affected programs: 70199 -> 69226 (-1.39%)
helped: 213
HURT: 1
v2: Add a comment (based on Iago's suggested one).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Without this, the X server may accumulate stale Present event contexts
if a client performs several video decoding sessions using the same
window.
v2: Based on Chris Wilson's review:
* Use xcb_discard_reply() instead of free(xcb_request_check())
Reviewed-and-Tested-by: Leo Liu <leo.liu@amd.com>
Without this, the X server may accumulate stale Present event contexts
if a client ends up creating and destroying DRI drawables for the same
window.
v2: Based on Chris Wilson's review:
* Use xcb_present_select_input_checked so that protocol errors
generated by old X servers can be handled gracefully
* Use xcb_discard_reply() instead of free(xcb_request_check())
We were baking in the LOD of the source level to each shader. Instead,
pass it in as a uniform -- this requires storing it to a temp register,
but that's better than compiling a ton of separate shaders:
total instructions in shared programs: 115032 -> 115036 (0.00%)
instructions in affected programs: 96 -> 100 (4.17%)
LOST: 572
This helps in debugging memory pressure. It would be nice if we could
tell valgrind about it all the way from allocation time to destroy, but we
need a pointer to hand to VALGRIND_MALLOCLIKE_BLOCK.
The ranges are in units of bytes, not dwords. This wasn't caught by
piglit tests because ttn tends to make one big uniform file, so we only
had one UBO range with a src and dst offset of 0.
nir_opt_peephole_select has the job of removing IF statements with no side
effects. However, if the IF statement's successor didn't have any
instructions in it, we were skipping it, which occurred in mupen64 on vc4
with glsl_to_nir enabled:
instructions in affected programs: 6134 -> 4120 (-32.83%)
total uniforms in shared programs: 38268 -> 38219 (-0.13%)
No changes on Haswell shader-db.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
v_fma_f32 runs at FP64 rate (= slow). Alien Isolation and F1 2015 seem
to use fma for all d3d multiply-add instructions, which is silly.
This tries to restore performance for those games.
The main difference between v_mad_f32 and v_fma_f32 is that v_mad doesn't
support denormals, which we don't enable anyway, because they are slow too.
Also, there is code size reduction:
Totals from affected shaders:
VGPRS: 109796 -> 109808 (0.01 %)
Spilled SGPRs: 29995 -> 30022 (0.09 %)
Spilled VGPRs: 12 -> 13 (8.33 %) <-- it's just one shader going from 12 to 13
Code Size: 6667596 -> 6476356 (-2.87 %) bytes
Max Waves: 26931 -> 26899 (-0.12 %)
I've not actually tested real performance.
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Jason Ekstrand
Andres Gomez
Kenneth Graunke
Tim Rowley
Eric Anholt
Michel Dänzer
Ben Widawsky
Jan Ziak
Marek Olšák