This enables GL4.1 for radeonsi, and updates the
docs in the correct places.
v2: enable only for llvm 3.7 which has fixes in place.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This is the final piece for ARB_gpu_shader5,
The code is based on the r600 code from Glenn Kennard,
and myself.
While developing this, I'm not 100% sure of all the calculations
made in the GS registers, this is why the max_stream is worked
out there and used to limit the changes in registers. Otherwise
my initial attempts either regressed GS texelFetch tests
or primitive-id-restart. The current code has no regressions
in piglit.
This commit doesn't enable ARB_gpu_shader5, since that just
bumps the glsl level to 4.00, so I'll just do a separate patch
for 4.10.
v1.1: fix bug introduced in rebase.
v2: Address Marek's review comments,
remove my llvm stream code for simpler C,
move gsvs_ring and gs_next_vertex to arrays.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Fixes following compiler warning:
brw_cs.cpp:386:27: warning: comparison between signed and unsigned
integer expressions [-Wsign-compare]
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
This is easily accomplished by moving simd16 3src to GEN9_FEATURES.
v2: small cleanup to make it more similar to GEN8_FEATURES
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Otherwise the final dri module will have (additional) unresolved
symbols.
Cc: Brian Paul <brianp@vmware.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviwed-by: Jose Fonseca <jfonseca@vmware.com>
There isn't a single instance in mesa that
mentions HAVE_SYS_TYPES_H, other than this file.
Cc: Jose Fonseca <jfonseca@vmware.com>
Acked-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
There are a couple of unrelated changes in t_vb_lighttmp.h that I hope
you'll excuse -- there's a block of code that's duplicated modulo a few
trivial differences that I took the liberty of fixing.
Literals without an f/F suffix are of type double, and implicit
conversion rules specify that the float in (float op double) be
converted to a double before the operation is performed. I believe float
execution was intended (in nearly all cases) or is sufficient (in the
case of gen7_urb.c).
Removes a lot of float <-> double conversion instructions and replaces
many double instructions with float instructions which are cheaper.
text data bss dec hex filename
4928659 195160 26192 5150011 4e953b i965_dri.so before
4928315 195152 26192 5149659 4e93db i965_dri.so after
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
ARB_viewport_array specifies that DEPTH_RANGE consists of double-
precision parameters (corresponding commit d4dc35987), and a preparatory
commit (6340e609a) added _mesa_get_viewport_xform() which returned
double-precision scale[3] and translate[3] vectors, even though X, Y,
Width, and Height were still floats.
All users of _mesa_get_viewport_xform() immediately convert the double
scale and translation vectors into floats (which were floats originally,
but were converted to doubles in _mesa_get_viewport_xform(), sigh).
i965 at least cannot consume doubles (see SF_CLIP_VIEWPORT). If we want
to pass doubles to hardware, we should have a different function that
does that.
Acked-by: Mathias Froehlich <Mathias.Froehlich@web.de>
Not a typo. Replace the default builder with one of bogus width to
catch cases in which optimization passes assume that the default
dispatch width is good enough. The execution controls of instructions
emitted during optimization should in general match the original code
that is being manipulated. Many of the problems fixed in this series
were caught by the assertions introduced in this patch.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
This could have led to somewhat increased bandwidth usage for lowered
texturing instructions on Gen4 (which is the only case in which
lower_width may be greater than inst->exec_size). After the previous
patches the invariant mentioned in the comment should no longer be
assumed by any of the other optimization and lowering passes, so the
exec_all() call shouldn't be necessary anymore.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Instead of relying on the default one. This shouldn't lead to any
functional changes because DEP_RESOLVE_MOV overrides the execution
size of the instruction anyway and other execution controls are
irrelevant.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
The global CSE pass stinks and is unable to pull this out. Easy enough
to handle it here and avoid generating unnecessary special register
loads (which can allegedly be quite slow).
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
PFETCH retrieves the address for incoming vertices, not output vertices
in TCS. For output vertices, we must use the laneid as a base.
Fixes barrier piglit test, which was failing for entirely non-barrier
reasons, but rather that it was (a) trying to draw multiple patches and
(b) the incoming patch size was not the same as the outgoing patch size.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
This reverts commit a27ec5dc46. It
breaks the intended behaviour of pipe_loader_probe() with ndev==0 as
relied upon by clover to query the number of devices available to the
pipe loader in the system.
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>
opt_sampler_eot() was relying on the default builder to have the same
width as the sampler and FB write opcodes it was eliminating, the
channel selects didn't matter because the builder was only being used
to allocate registers, no new instructions were being emitted with it.
A future commit will change the width of the default builder what will
break this assumption, so initialize it explicitly here.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
This wasn't taking into account the execution controls of the original
instruction, but it was most likely not a bug because control flow
instructions are typically full width.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Emit the SELs and MOVs with the same execution controls as the
original MOVs, and the CMP with the same execution controls as the IF.
Also explicitly check that the execution controls of any pair of MOVs
being folded into a SEL are compatible (which is almost always going
to be the case), since otherwise it would seem wrong to initialize the
builder object below from the then_mov instruction only.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
lower_integer_multiplication() was ignoring the execution controls of
the original MUL instruction. Fix it by using the new fs_builder
constructor.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
demote_pull_constants() was ignoring the execution size and channel
selects of the instruction that wanted the constant, which doesn't
matter for uniform pull constant loads because all channels get the
same scalar value, but it might for varying pull constant loads. Fix
it by using the new fs_builder() constructor that takes care of
setting execution controls compatible with the instruction passed as
argument.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
The execution size was being left equal to the default of 8/16, which
AFAICT would have overwritten components other than the one we wanted
to initialize and could potentially have corrupted other registers.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
We have a number of optimization passes that repeat the same pattern
before inserting new instructions into the program based on some
previous instruction: They point the default builder at the original
instruction, then call exec_all() and group() to select the same
execution controls the original instruction had, and then maybe call
annotate() to clone the debug annotation from the original
instruction.
In fact an optimization pass missing any of these steps is likely to
be broken if the intention was to emit new code based on a preexisting
instruction, so let's make it easy for passes to do the right thing by
having an fs_builder constructor that automates the task of setting up
a builder to emit a given instruction provided as argument.
The following patches fix all cases I've found in which we weren't
explicitly initializing the execution controls of the emitted
instructions, and clean-up optimization passes which were already
doing the right thing to use the new constructor.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Images take up zero uniform slots in the nir_shader::num_uniforms
calculation, but nir_setup_uniforms needs to be executed even if the
program has no non-image uniforms so the driver-specific image
parameters are uploaded. nir_setup_uniforms is a no-op if there are
really no uniforms, so checking the num_uniform count is useless in
any case.
The nir_setup_inputs and _outputs changes shouldn't lead to any
functional change, they are just meant to preserve the symmetry
between them and nir_setup_uniforms.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Image variables need to allocate additional uniform slots over
nir_shader::num_uniforms. nir_setup_uniforms() overwrites the values
imported from the SIMD8 visitor and then exits early before entering
the nir_shader::uniforms loop, so image uniforms are never re-created.
Instead leave the imported values alone, they *must* be the same for
the uniform layout of both runs to be compatible.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Rewrite the NIR atomic counter intrinsics translation code making use
of the recently introduced surface builder. This will allow the
removal of some of the functionality duplicated between the visitor
and surface builder.
v2: Drop VEC4 suport.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
