Francisco Jerez [Mon, 16 Feb 2015 11:38:39 +0000 (13:38 +0200)]
ra: Disable round-robin strategy for optimistically colorable nodes.
The round-robin allocation strategy is expected to decrease the amount
of false dependencies created by the register allocator and give the
post-RA scheduling pass more freedom to move instructions around. On
the other hand it has the disadvantage of increasing fragmentation and
decreasing the number of equally-colored nearby nodes, what increases
the likelihood of failure in presence of optimistically colorable
nodes.
This patch disables the round-robin strategy for optimistically
colorable nodes. These typically arise in situations of high register
pressure or for registers with large live intervals, in both cases the
task of the instruction scheduler shouldn't be constrained excessively
by the dense packing of those nodes, and a spill (or on Intel hardware
a fall-back to SIMD8 mode) is invariably worse than a slightly less
optimal scheduling.
Shader-db results on the i965 driver:
total instructions in shared programs:
5488539 ->
5488489 (-0.00%)
instructions in affected programs: 1121 -> 1071 (-4.46%)
helped: 1
HURT: 0
GAINED: 49
LOST: 5
v2: Re-enable round-robin already for the lowest one of the nodes
pushed optimistically onto the sack (Connor).
v3: Use UINT_MAX instead of ~0, open-code MIN2 (Jason, Connor).
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Francisco Jerez [Thu, 5 Feb 2015 10:20:03 +0000 (12:20 +0200)]
i965/fs: Fix lower_load_payload() not to use an incorrect half for immediates and uniforms.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Francisco Jerez [Fri, 6 Feb 2015 13:06:05 +0000 (15:06 +0200)]
i965/fs: Fix lower_load_payload() to take into account non-zero reg_offset.
Fixes metadata guess when instructions in the program specify a
destination register with non-zero reg_offset and when the payload of
a LOAD_PAYLOAD spans several registers.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Francisco Jerez [Wed, 4 Feb 2015 17:49:32 +0000 (19:49 +0200)]
i965/fs: Remove logic to keep track of MRF metadata in lower_load_payload().
MRFs cannot be read from anyway so they cannot possibly be a valid
source of LOAD_PAYLOAD.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Francisco Jerez [Sat, 17 Jan 2015 12:12:34 +0000 (14:12 +0200)]
i965/fs: Less broken handling of force_writemask_all in lower_load_payload().
It's perfectly fine to read the second half of a register written with
force_writemask_all from a first half MOV instruction or vice versa, and
lower_load_payload shouldn't mark the whole MOV as belonging to the second
half in that case. Replicate the same metadata to both halves of the
destination when writemasking is disabled.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Matt Turner [Mon, 23 Feb 2015 04:25:28 +0000 (20:25 -0800)]
mesa/vbo: Use unreachable to silence uninitialized var warning.
Reviewed-by: Eric Anholt <eric@anholt.net>
Matt Turner [Sat, 21 Feb 2015 04:27:06 +0000 (20:27 -0800)]
mesa: Move START/END_FAST_MATH macros to their only use.
Reviewed-by: Eric Anholt <eric@anholt.net>
Matt Turner [Sat, 21 Feb 2015 04:23:26 +0000 (20:23 -0800)]
mesa: Remove definition of NULL.
If your stdlib.h doesn't define this you should fix your stdlib.h.
Reviewed-by: Eric Anholt <eric@anholt.net>
Matt Turner [Sat, 21 Feb 2015 04:18:47 +0000 (20:18 -0800)]
mesa: Use assert() instead of ASSERT wrapper.
Acked-by: Eric Anholt <eric@anholt.net>
Matt Turner [Sat, 21 Feb 2015 04:12:14 +0000 (20:12 -0800)]
mesa: Remove CHECK macro.
There's some commentary about how it's defined by other "modules", and
maybe that was true in 2000 when the code was added.
Reviewed-by: Eric Anholt <eric@anholt.net>
Matt Turner [Sat, 21 Feb 2015 04:08:51 +0000 (20:08 -0800)]
mesa: Remove dead CAPI define.
Reviewed-by: Eric Anholt <eric@anholt.net>
Matt Turner [Sat, 21 Feb 2015 03:46:21 +0000 (19:46 -0800)]
gallium: Use util_cpu_to_le{16,32} in many more places.
... and util_le{16,32}_to_cpu. I think I've used the right ones for
describing the actual operation performed (even though they're both just
"byte-swap this if I'm on big-endian").
The Linux Kernel has typedefs __le32/__be32 and friends that static
analysis tools can use to check that byte-orderings are correct. It
might be interesting to apply that here as well.
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Matt Turner [Sat, 21 Feb 2015 02:46:43 +0000 (18:46 -0800)]
gallium/util: Use HAVE___BUILTIN_* macros.
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Matt Turner [Sat, 21 Feb 2015 02:32:05 +0000 (18:32 -0800)]
mesa: Move C99 MSVC compatibility code from u_math.h to c99_compat.h.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Matt Turner [Sun, 22 Feb 2015 06:41:24 +0000 (22:41 -0800)]
i965: Link test programs with gtest before pthreads.
Cc: "10.5" <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.gentoo.org/show_bug.cgi?id=540962
Brian Paul [Mon, 23 Feb 2015 15:39:11 +0000 (08:39 -0700)]
osmesa: add gallium include dirs to Makefile.am
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89260
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Brian Paul [Mon, 23 Feb 2015 15:57:32 +0000 (08:57 -0700)]
util: move pipe_prim_names array into u_prim_name()
Also, wrapping the array in #ifdef DEBUG / #endif doesn't seem necessary.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Brian Paul [Mon, 23 Feb 2015 15:57:32 +0000 (08:57 -0700)]
util: rewrite debug_print_transfer_flags() using debug_dump_flags()
Add add missing PIPE_TRANSFER_PERSISTENT, PIPE_TRANSFER_COHERENT flags.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eduardo Lima Mitev [Mon, 23 Feb 2015 12:03:56 +0000 (13:03 +0100)]
mesa: Adds missing error condition in _mesa_check_sample_count()
This corrects a trivial error introduced in commit
19252fee46b835cb4f6b1cce18d7737d62b64a2e. That patch was merged recently
and omits one condition (that 'samples' is greater than zero) in one of
the error checks. That error will definitely cause regressions.
Also corrects the reference to the specification above the error check,
which was wrongly quoting OpenGL instead of OpenGL-ES.
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>
Marek Olšák [Mon, 23 Feb 2015 10:42:59 +0000 (11:42 +0100)]
radeonsi: fix a warning caused by previous commit
Cc: 10.5 10.4 <mesa-stable@lists.freedesktop.org>
Marek Olšák [Mon, 23 Feb 2015 10:39:53 +0000 (11:39 +0100)]
radeonsi: fix point sprites
Broken by
a27b74819ad375e8c0bc88e13f42c951d2b5cd6a.
This fix is critical and should be ported to stable ASAP.
Cc: 10.5 10.4 <mesa-stable@lists.freedesktop.org>
Ben Widawsky [Thu, 19 Feb 2015 23:49:34 +0000 (15:49 -0800)]
i965/skl: Use 1 register for uniform pull constant payload
When under dispatch_width=16 the previous code would allocate 2 registers for
the payload when only one is needed. This manifested itself through bugs on SKL
which needs to mess with this instruction.
Ken though this might impact shader-db, but apparently it doesn't
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89118
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=88999
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Tested-by: Timo Aaltonen <timo.aaltonen@canonical.com>
Eric Anholt [Fri, 20 Feb 2015 08:00:27 +0000 (00:00 -0800)]
nir: Generalize the optimization of subs of subs from 0.
I initially wrote this based on the "(('fneg', ('fneg', a)), a)" above,
but we can generalize it and make it more potentially useful. In the
specific original case of a 0 for our new 'a' argument, it'll get further
algebraic optimization once the 0 is an argument to the new add.
No shader-db effects.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Eric Anholt [Fri, 20 Feb 2015 09:20:34 +0000 (01:20 -0800)]
nir: Collapse repeated bcsels on the same argument.
vc4 results:
total instructions in shared programs: 39881 -> 39794 (-0.22%)
instructions in affected programs: 6302 -> 6215 (-1.38%)
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Eric Anholt [Fri, 20 Feb 2015 09:18:46 +0000 (01:18 -0800)]
nir: When faced with a csel on !condition, just flip the arguments.
total NIR instructions in shared programs: 39426 -> 39411 (-0.04%)
NIR instructions in affected programs: 3748 -> 3733 (-0.40%)
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Eric Anholt [Fri, 20 Feb 2015 08:57:04 +0000 (00:57 -0800)]
nir: Allow nir_opt_algebraic to see booleanness through &&, ||, ^, !.
We have some useful optimizations to drop things like 'ine a, 0' on a
boolean argument, but if 'a' came from logical operations on bools, it
couldn't tell. These kinds of constructs appear as a result of TGSI->NIR
quite frequently (at least with if flattening), so being a little more
aggressive in detecting booleans can pay off.
v2: Add ixor as a booleanness-preserving op (Suggestion by Connor).
vc4 results:
total instructions in shared programs: 40207 -> 39881 (-0.81%)
instructions in affected programs: 6677 -> 6351 (-4.88%)
Reviewed-by: Matt Turner <mattst88@gmail.com> (v1)
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Eric Anholt [Thu, 29 Jan 2015 23:50:18 +0000 (15:50 -0800)]
nir: Add a couple of simplifications of csel operations.
vc4 was already cleaning these up, but it does shave 4 NIR instructions in
shader-db.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Ilia Mirkin [Sat, 21 Feb 2015 17:53:43 +0000 (12:53 -0500)]
glsl: ensure that enter/leave record get a record type
May make life easier for tools like Coverity.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ilia Mirkin [Sat, 21 Feb 2015 17:44:05 +0000 (12:44 -0500)]
tgsi: avoid returning pointer to local var, make it static
Spotted by Coverity.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Rob Clark [Sat, 21 Feb 2015 18:55:37 +0000 (13:55 -0500)]
freedreno/a4xx: set PC_PRIM_VTX_CNTL.VAROUT properly
Fixes xonotic, some webgl stuff, and really pretty much anything with
more than 4 varyings.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sat, 21 Feb 2015 18:50:52 +0000 (13:50 -0500)]
freedreno: update generated headers
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sat, 21 Feb 2015 18:39:06 +0000 (13:39 -0500)]
freedreno/a4xx: bit of cleanup
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sun, 15 Feb 2015 06:59:17 +0000 (01:59 -0500)]
loader: not having a pci-id should not be a warn
If there is no pci-id, which is valid for vc4 and freedreno, just emit
an info msg. Keep malformed but existing pci-id's as a warning.
Mostly just to clean up a warning that confuses users for the non-pci
devices.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sun, 15 Feb 2015 05:04:57 +0000 (00:04 -0500)]
freedreno: implement fence
I never actually implemented the stubbed out fence stuff back in the
early days. Fix that.
We'll need a few libdrm_freedreno changes to handle timeout properly,
so ignore that for now to avoid a libdrm_freedreno dependency bump.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Tue, 3 Feb 2015 20:52:53 +0000 (15:52 -0500)]
freedreno/a2xx: fix increment in assert
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=88883
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Jordan Justen [Fri, 20 Feb 2015 20:12:25 +0000 (12:12 -0800)]
i965/fs: Use fs_reg for CS/VS atomics pixel mask immediate data
The brw_imm_ud will yield a HW_REG which then will introduce a barrier
for certain optimization opportunities.
No piglit regressions seen with gen8 (simd8vs).
Suggested-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jordan Justen [Mon, 22 Sep 2014 01:31:45 +0000 (18:31 -0700)]
i965/fs: Set pixel/sample mask for compute shaders atomic ops
For fragment programs, we pull this mask from the payload header. The same
mask doesn't exist for compute shaders, so we set all bits to enabled.
Previously we were setting 0xff to support SIMD8 VS, but with CS we
support SIMD16, and therefore we change this to 0xffff.
Related commits for SIMD8 VS:
commit
d9cd982d556be560af3bcbcdaf62b6b93eb934a5
Author: Ben Widawsky <benjamin.widawsky@intel.com>
Date: Sun Feb 15 20:06:59 2015 -0800
i965/simd8vs: Fix SIMD8 atomics
commit
4a95be9772a255776309f23180519a4a8560f2dd
Author: Jordan Justen <jordan.l.justen@intel.com>
Date: Tue Feb 17 09:57:35 2015 -0800
i965/simd8vs: Fix SIMD8 atomics (read-only)
Note: this mask is ANDed with the execution mask, so some channels may not end
up issuing the atomic operation.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Chia-I Wu [Fri, 20 Feb 2015 16:35:49 +0000 (00:35 +0800)]
ilo: R32G32B32_FLOAT need no special care on Gen8+
Gen8+ must use VALIGN_4. Unlike prior Gens, R32G32B32_FLOAT should supposedly
support VALIGN_4.
Chia-I Wu [Fri, 20 Feb 2015 07:27:14 +0000 (15:27 +0800)]
ilo: 128 BPP formats can use TiledY on Gen7.5+
The restriction is lifted.
Ilia Mirkin [Thu, 24 Jul 2014 02:32:55 +0000 (22:32 -0400)]
nvc0: enable double support
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Mon, 7 Jul 2014 05:53:52 +0000 (01:53 -0400)]
nvc0/ir: remove merge/split pairs to allow normal propagation to occur
Because the TGSI interface creates merges for each instruction source
and then splits them back out, there are a lot of unnecessary
merge/split pairs which do essentially nothing. The various modifier/etc
propagation doesn't know how to walk though those, so just remove them
when they're unnecessary.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Mon, 7 Jul 2014 03:39:59 +0000 (23:39 -0400)]
nvc0/ir: add support for new TGSI double opcodes
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Mon, 7 Jul 2014 06:32:45 +0000 (02:32 -0400)]
nvc0/ir: handle zero and negative sqrt arguments
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Mon, 7 Jul 2014 03:39:38 +0000 (23:39 -0400)]
nvc0/ir: no instruction can load a double immediate
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Mon, 7 Jul 2014 04:04:19 +0000 (00:04 -0400)]
nvc0/ir: fix lowering of RSQ/RCP/SQRT/MOD to work with F64
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Fri, 26 Sep 2014 06:38:53 +0000 (02:38 -0400)]
gm107/ir: fix F2F flipped stype/dtype flags
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Fri, 26 Sep 2014 06:35:51 +0000 (02:35 -0400)]
gm107/ir: fix DSET boolean float flag
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Fri, 26 Sep 2014 06:21:55 +0000 (02:21 -0400)]
gm107/ir: fix DMUL opcode encoding
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sat, 19 Jul 2014 02:20:42 +0000 (22:20 -0400)]
gk110/ir: add emission of dadd/dmul/dmad opcodes
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Mon, 7 Jul 2014 03:36:32 +0000 (23:36 -0400)]
nvc0/ir: add emission of dadd/dmul/dmad opcodes, fix minmax
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Roland Scheidegger [Sat, 14 Feb 2015 15:34:04 +0000 (16:34 +0100)]
mesa: don't enable NV_fragment_program_option with swrast
Since dropping some NV_fragment_program opcodes (commits
868f95f1da74cf6dd7468cba1b56664aad585ccb,
a3688d686f147f4252d19b298ae26d4ac72c2e08)
we can no longer parse all opcodes necessary for this extension, leading
to bugs (https://bugs.freedesktop.org/show_bug.cgi?id=86980).
Hence don't announce support for it in swrast (no other driver enabled it).
(Note that remnants of some NV_fp/vp extensions remain, they could be
dropped but are required as hacks for getting viewperf11 catia to run.)
Brian Paul [Fri, 20 Feb 2015 20:19:50 +0000 (13:19 -0700)]
drivers/x11: add gallium include dirs to Makefile.am
Fixes xlib driver build after
e8c5cbfd921680c.
Acked-by: Matt Turner <mattst88@gmail.com>
Marek Olšák [Fri, 20 Feb 2015 19:17:39 +0000 (20:17 +0100)]
vbo: fix an unitialized-variable warning
It looks like a bug to me.
Cc: 10.5 10.4 10.3 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
Marek Olšák [Fri, 20 Feb 2015 19:17:20 +0000 (20:17 +0100)]
gallium/sw/kms: fix a type-mismatch warning
Reviewed-by: Brian Paul <brianp@vmware.com>
Marek Olšák [Fri, 20 Feb 2015 19:15:59 +0000 (20:15 +0100)]
gallium/sw/kms: don't redefine DEBUG
Reviewed-by: Brian Paul <brianp@vmware.com>
Marek Olšák [Fri, 20 Feb 2015 19:15:35 +0000 (20:15 +0100)]
targets/d3dadapter9: remove an unused variable
Reviewed-by: Brian Paul <brianp@vmware.com>
Marek Olšák [Fri, 20 Feb 2015 19:15:05 +0000 (20:15 +0100)]
tgsi: fix type-mismatch warning
Reviewed-by: Brian Paul <brianp@vmware.com>
Marek Olšák [Fri, 20 Feb 2015 19:14:33 +0000 (20:14 +0100)]
gallivm: fix uninitialized-variable warnings
Reviewed-by: Brian Paul <brianp@vmware.com>
Matt Turner [Fri, 20 Feb 2015 20:41:46 +0000 (12:41 -0800)]
mesa: Have configure define NDEBUG, not mtypes.h.
mtypes.h had been defining NDEBUG (used by assert) if DEBUG was not
defined. Confusing and bizarre that you don't get NDEBUG if you don't
include mtypes.h.
... which is just what happened in commit
bef38f62e.
Let's let configure define this for us if not using --enable-debug.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Kenneth Graunke [Fri, 20 Feb 2015 20:31:31 +0000 (12:31 -0800)]
nir: Fix the Mesa build without -DDEBUG.
With -DDEBUG -UNDEBUG, this assert uses reg_state::stack_size, which
doesn't exist, breaking the build:
assert(state->states[index].index < state->states[index].stack_size);
Switch it to ifndef NDEBUG, so the field will exist if the assertion
actually generates code.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Eric Anholt [Wed, 11 Feb 2015 23:08:02 +0000 (15:08 -0800)]
nir: Drop dependency on mtypes.h for core NIR.
One less new directory necessary for gallium code that wants to interact
with NIR.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eric Anholt [Wed, 11 Feb 2015 23:21:37 +0000 (15:21 -0800)]
glsl: Only include mtypes from glsl_types.h for the C++ code that needs it.
It's used in one of the methods, not in the structure definitions.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eric Anholt [Wed, 11 Feb 2015 23:05:06 +0000 (15:05 -0800)]
util: Move Mesa's bitset.h to util/.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eric Anholt [Wed, 11 Feb 2015 22:57:55 +0000 (14:57 -0800)]
mesa: Make bitset.h not rely on Mesa-specific types and functions.
Note that we can't use u_math.h's align() because it's a function instead
of a macro, while BITSET_DECLARE needs a constant expression for nouveau's
usage in global declarations.
v2: Stick some parens around the bits macro argument usage (review by Jose).
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eric Anholt [Wed, 11 Feb 2015 22:24:33 +0000 (14:24 -0800)]
mesa: Use u_math.h from macros.h
This avoids duplication of some macros and other definitions across the
tree.
Note that COPY_4FV switches from a memcpy-based implementation to an
assignment of 4 floats.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eric Anholt [Wed, 11 Feb 2015 22:28:44 +0000 (14:28 -0800)]
gallium/util: Don't include unused debug functions from u_math.h
It introduces references to gallium util/ symbols which means we don't get
to include it from outside-of-gallium code.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Eric Anholt [Wed, 11 Feb 2015 22:18:50 +0000 (14:18 -0800)]
mesa: Add gallium include dirs to more parts of the tree.
v2: Try to patch up the scons bits.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Marek Olšák [Fri, 20 Feb 2015 19:13:44 +0000 (20:13 +0100)]
gallium/radeon: fix an uninitialized-variable warning
Ilia Mirkin [Fri, 20 Feb 2015 04:30:36 +0000 (23:30 -0500)]
gallium: add new double-related shader caps to all the getters
Missed a few drivers in the earlier changes, this should fix up all the
ones that print unknown caps or don't have a default statement.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Brian Paul [Fri, 20 Feb 2015 15:09:36 +0000 (08:09 -0700)]
svga: add missing _DROUND,DFRACEXP_DLDEXP_SUPPORTED switch cases
To silence unhandled switch case warnings.
Marek Olšák [Thu, 19 Feb 2015 12:03:54 +0000 (13:03 +0100)]
radeonsi: don't use SQC_CACHES to flush ICACHE and KCACHE on SI
This reverts
73c2b0d18c51459697d8ec194ecfc4438c98c139.
It doesn't seem to be reliable. It's probably missing a wait packet or
something, because it's just a register write and doesn't wait for anything.
SURFACE_SYNC at least seems to wait until the flush is done. Just guessing.
Let's not complicate things and revert this.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=88561
Cc: 10.5 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Iago Toral Quiroga [Fri, 20 Feb 2015 07:21:25 +0000 (08:21 +0100)]
i965/gen6: Fix GL_GEOMETRY_SHADER_PRIMITIVES_EMITTED_ARB
In gen6 we need to compute the primitive count in the generated GS program.
The current implementation only counts full primitives, that is, if the
output primitive type is a triangle strip, it won't count individual
triangles in the strip, only complete strips.
If we want to count basic primitives instead we have two options: rework
the assembly code we generate for strip primitives or simply use
CL_INVOCATION_COUNT to resolve the query and let the hardware do that work
for us. This patch implements the latter approach.
Fixes the following piglit test:
bin/arb_pipeline_statistics_query-geom -auto
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89210
Tested-by: Mark Janes <mark.a.janes@intel.com>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Eduardo Lima Mitev [Fri, 20 Feb 2015 08:32:42 +0000 (09:32 +0100)]
mesa: Check that draw buffers are valid for glDrawBuffers on GLES3
Section 4.2 (Whole Framebuffer Operations) of the OpenGL 3.0 specification
says:
"Each buffer listed in bufs must be BACK, NONE, or one of the values from
table 4.3 (NONE, COLOR_ATTACHMENTi)".
Fixes 1 dEQP test:
* dEQP-GLES3.functional.negative_api.buffer.draw_buffers
Reviewed-by: Matt Turner <mattst88@gmail.com>
Samuel Iglesias Gonsalvez [Tue, 25 Nov 2014 13:03:05 +0000 (14:03 +0100)]
glsl: don't allow invariant qualifiers for interface blocks
GLSL 1.50 and GLSL 4.40 specs, they both say the same in
"Interface Blocks" section:
"If optional qualifiers are used, they can include interpolation qualifiers,
auxiliary storage qualifiers, and storage qualifiers and they must declare
an input, output, or uniform member consistent with the interface qualifier
of the block"
From GLSL ES 3.0, chapter 4.3.7 "Interface Blocks", page 38:
"GLSL ES 3.0 does not support interface blocks for shader inputs or outputs."
and from GLSL ES 3.0, chapter 4.6.1 "The invariant qualifier", page 52.
"Only variables output from a shader can be candidates for invariance."
This patch fixes the following dEQP tests:
dEQP-GLES3.functional.shaders.declarations.invalid_declarations.invariant_uniform_block_2_vertex
dEQP-GLES3.functional.shaders.declarations.invalid_declarations.invariant_uniform_block_2_fragment
No piglit regressions.
Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
v2:
- Enable this check for GLSL.
Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Thu, 19 Feb 2015 21:22:31 +0000 (13:22 -0800)]
vc4: Keep an array of pointers to instructions defining the temps around.
The optimization passes are always regenerating it and throwing it away,
but it's not hard to keep track of.
Eric Anholt [Thu, 19 Feb 2015 20:19:44 +0000 (12:19 -0800)]
vc4: Move qir_uniform() and the constant-value versions to vc4_qir.c/h.
I may want them in optimization passes, and they're not really particular
to the program translation stage.
Eric Anholt [Thu, 19 Feb 2015 20:58:53 +0000 (12:58 -0800)]
vc4: Enforce one-uniform-per-instruction after optimization.
This lets us more intelligently decide which uniform values should be put
into temporaries, by choosing the most reused values to push to temps
first.
total uniforms in shared programs: 13457 -> 13433 (-0.18%)
uniforms in affected programs: 1524 -> 1500 (-1.57%)
total instructions in shared programs: 40198 -> 40019 (-0.45%)
instructions in affected programs: 6027 -> 5848 (-2.97%)
I noticed this opportunity because with the NIR work, some programs were
happening to make different uniform copy propagation choices that
significantly increased instruction counts.
Eric Anholt [Thu, 19 Feb 2015 20:16:25 +0000 (12:16 -0800)]
vc4: Rename add_uniform() to qir_uniform().
Eric Anholt [Fri, 20 Feb 2015 07:34:37 +0000 (23:34 -0800)]
vc4: Shut up runtime warnings about new pipe caps.
Matt Turner [Thu, 12 Feb 2015 01:42:43 +0000 (01:42 +0000)]
i965/vec4: Add and use byte-MOV instruction for unpack 4x8.
Previously we were using a B/UB source in an Align16 instruction, which
is illegal. It for some reason works on all platforms, except Broadwell.
Cc: "10.5" <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=86811
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Tue, 10 Feb 2015 06:54:51 +0000 (22:54 -0800)]
i965/blorp: Emit MADs.
Low hanging fruit: cuts a couple of instructions.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Tue, 10 Feb 2015 05:26:14 +0000 (21:26 -0800)]
i965/blorp: Optimize clamping tex coords.
Each emit_cond_mov() emits a CMP of its first to arguments using the
specified conditional mod, followed by a predicated MOV of the fifth
argument into the fourth. In all four cases here, it was just
implementing MIN/MAX which we can do in a single SEL instruction.
Also reorder the instructions for a slightly better schedule.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Tue, 10 Feb 2015 05:11:46 +0000 (21:11 -0800)]
i965: Use greater-equal cmod to implement maximum.
The docs specifically call out SEL with .l and .ge as the
implementations of MIN and MAX respectively. Among other things, SEL
with these conditional mods are commutative.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Tue, 10 Feb 2015 06:21:21 +0000 (22:21 -0800)]
i965: Don't emit saturates for instructions without destinations.
We were special casing OPCODE_END but no other instructions that have no
destination, like OPCODE_KIL, leading us to emitting MOVs with null
destinations.
total instructions in shared programs:
5705243 ->
5701539 (-0.06%)
instructions in affected programs: 124104 -> 120400 (-2.98%)
helped: 904
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Wed, 11 Feb 2015 00:25:47 +0000 (16:25 -0800)]
i965/fs: Consider MOV.SAT to interfere if it has a source modifier.
The saturate propagation pass recognizes that the second instruction
below does not interfere with an attempt to propagate the saturate
modifier from instruction 3 to 1.
1: add(8) dst0 src0 src1
2: mov.sat(8) dst1 dst0
3: mov.sat(8) dst2 dst0
Unfortunately, we did not consider the case of instruction 2 having a
source modifier on dst0. Take for instance:
1: add(8) dst0 src0 src1
2: mov.sat(8) dst1 -dst0
3: mov.sat(8) dst2 dst0
Consider such an instruction to interfere. Increase instruction counts
in Anomaly 2, which could be a bug fix depending on the values the first
instruction produces.
instructions in affected programs: 53228 -> 53934 (1.33%)
HURT: 360
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Wed, 28 Jan 2015 06:43:28 +0000 (22:43 -0800)]
i965/fs: Use fs_inst::overwrites_reg() in saturate propagation.
This is safer and matches the conditional_mod propagation pass.
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Matt Turner [Tue, 10 Feb 2015 21:38:07 +0000 (13:38 -0800)]
i965/fs: Add unit tests for saturate propagation pass.
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Timothy Arceri [Thu, 19 Feb 2015 10:32:21 +0000 (21:32 +1100)]
glsl: Use the without_array predicate
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ilia Mirkin [Wed, 18 Feb 2015 08:35:23 +0000 (03:35 -0500)]
nv50: add PIPELINE_STATISTICS query support, based on nvc0
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Nick Tenney <nick.tenney@gmail.com>
Ilia Mirkin [Fri, 20 Feb 2015 01:15:28 +0000 (20:15 -0500)]
svga: add missing :
Fixes: 924ee3f408 ("gallium: add shader cap for dldexp/dfracexp support")
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Jason Ekstrand [Tue, 10 Feb 2015 04:18:44 +0000 (20:18 -0800)]
nir/GCM: Pull unpinned instructions out of blocks while pinning
This lets us be slightly more efficient by not walking the CFG extra times.
Also, it may make it easier to ensure that GVN happens on only unpinned
instructions.
Reviewed-by: Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Mon, 9 Feb 2015 22:58:12 +0000 (14:58 -0800)]
nir/GCM: Use pass_flags instead of bitsets for tracking visited/pinned
Reviewed-by: Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Tue, 3 Feb 2015 18:11:23 +0000 (10:11 -0800)]
nir: Add a global code motion (GCM) pass
v2 Jason Ekstrand <jason.ekstrand@intel.com>:
- Use nir_dominance_lca for computing least common anscestors
- Use the block index for comparing dominance tree depths
- Pin things that do partial derivatives
Reviewed-by: Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Mon, 9 Feb 2015 22:41:10 +0000 (14:41 -0800)]
nir/instr: Change "live" to a more generic "pass_flags" field
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 5 Feb 2015 05:22:45 +0000 (21:22 -0800)]
nir: Make nir_[cf_node/instr]_[prev/next] return null if at the end
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Thu, 5 Feb 2015 05:38:28 +0000 (21:38 -0800)]
nir/from_ssa: Don't try to read an invalid instruction
Right now, the nir_instr_prev function function blindly looks up the
previous element in the exec list and casts it to an instruction even if
it's the tail sentinel. The next commit will change this to return null if
it's the first instruction. Making this change first avoids getting a
segfault between commits. The only reason we never noticed is that, thanks
to the way things are laid out in nir_block, the casted instruction's type
was never parallal_copy.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Wed, 4 Feb 2015 22:01:51 +0000 (14:01 -0800)]
nir/validate: Validate SSA defs the same way we do for registers
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Wed, 4 Feb 2015 21:58:12 +0000 (13:58 -0800)]
nir/validate: Validate if_uses on registers
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Wed, 4 Feb 2015 05:39:56 +0000 (21:39 -0800)]
nir: Properly clean up CF nodes when we remove them
Previously, if you remved a CF node that still had instructions in it, none
of the use/def information from those instructions would get cleaned up.
Also, we weren't removing if statements from the if_uses of the
corresponding register or SSA def. This commit fixes both of these
problems
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Wed, 4 Feb 2015 05:04:57 +0000 (21:04 -0800)]
nir: use nir_foreach_ssa_def for indexing ssa defs
This is both simpler and more correct. The old code didn't properly index
load_const instructions.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>