This doesn't make any real difference now, but future work (not in this
series) will add a LOT of ffma patterns. Having to duplicate all of
them for ffma(a, b, c) and ffma(b, a, c) is just terrible.
No shader-db changes on any Intel platform.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
v2: Instead of handling 3 sources as a special case, generalize with
loops to N sources. Suggested by Jason.
v3: Further generalize by only checking that number of sources is >= 2.
Suggested by Jason.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
The meaning of the new name is that the first two sources are
commutative. Since this is only currently applied to two-source
operations, there is no change.
A future change will mark ffma as 2src_commutative.
It is also possible that future work will add 3src_commutative for
opcodes like fmin3.
v2: s/commutative_2src/2src_commutative/g. I had originally considered
this, but I discarded it because I did't want to deal with identifiers
that (should) start with 2. Jason suggested it in review, so we decided
that _2src_commutative would be used in nir_opcodes.py. Also add some
comments documenting what 2src_commutative means. Also suggested by
Jason.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Instead of re-building the interference graph every time we spill, we
modify it in place so we can avoid recalculating liveness and the whole
O(n^2) interference graph building process. We make a simplifying
assumption in order to do so which is that all spill/fill temporary
registers live for the entire duration of the instruction around which
we're spilling. This isn't quite true because a spill into the source
of an instruction doesn't need to interfere with its destination, for
instance. Not re-calculating liveness also means that we aren't
adjusting spill costs based on the new liveness. The combination of
these things results in a bit of churn in spilling. It takes a large
cut out of the run-time of shader-db on my laptop.
Shader-db results on Kaby Lake:
total instructions in shared programs: 15311224 -> 15311360 (<.01%)
instructions in affected programs: 77027 -> 77163 (0.18%)
helped: 11
HURT: 18
total cycles in shared programs: 355544739 -> 355830749 (0.08%)
cycles in affected programs: 203273745 -> 203559755 (0.14%)
helped: 234
HURT: 190
total spills in shared programs: 12049 -> 12042 (-0.06%)
spills in affected programs: 2465 -> 2458 (-0.28%)
helped: 9
HURT: 16
total fills in shared programs: 25112 -> 25165 (0.21%)
fills in affected programs: 6819 -> 6872 (0.78%)
helped: 11
HURT: 16
Total CPU time (seconds): 2469.68 -> 2360.22 (-4.43%)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This is slightly less convenient in some places but it will make it much
easier when we want to start adding nodes dynamically.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
The old code was arranged by the type of interference being added. It
would set up payload registers and then add payload interference for all
VGRFs. It would set up MRFs and add MRF interference for all VGRFs.
This commit re-arranges things to be organized differently. It first
creates and sets up all RA nodes and then groups interference into two
new categories: live range and instruction interference. Once all the
RA nodes have been set up, it walks the list of VGRFs and sets up their
live range interference and then walks the list of instructions and sets
up instruction interference. This new arrangement will be advantageous
for a future patch but, at the moment, it cuts 2% off the run-time of
shader-db on my laptop.
Shader-db results on Kaby Lake:
total instructions in shared programs: 15311224 -> 15311224 (0.00%)
instructions in affected programs: 0 -> 0
helped: 0
HURT: 0
total cycles in shared programs: 355544739 -> 355544739 (0.00%)
cycles in affected programs: 0 -> 0
helped: 0
HURT: 0
Total CPU time (seconds): 2523.45 -> 2469.68 (-2.13%)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
The only use of the MRF hack these days is for spilling and there we
don't need the precise MRF usage information. If we're spilling then we
know pretty well how many MRFs are going to be used. It is possible if
the only things that are spilled have fewer SIMD channels than the
dispatch width of the shader that this may be more MRFs than needed.
That's a risk we're willing to takd.
Shader-db results on Kaby Lake:
total instructions in shared programs: 15311100 -> 15311224 (<.01%)
instructions in affected programs: 16664 -> 16788 (0.74%)
helped: 1
HURT: 5
total cycles in shared programs: 355543197 -> 355544739 (<.01%)
cycles in affected programs: 731864 -> 733406 (0.21%)
helped: 3
HURT: 6
The hurt shaders are all SIMD32 compute shaders where we reserve enough
space for a 32-wide spill/fill but don't need it.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This accomplishes two things. First, it makes interfaces which are
really private to RA private to RA. Second, it gives us a place to
store some common stuff as we go through the algorithm.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
There's no reason why we need to use the calculated payload_node_count
value which is just first_non_payload_grf aligned up. The grf_used
value will be aligned up to 16 anyway (which is a much bigger alignment)
before being handed off to hardware.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
We only have one node per VGRF so this was adding way too much
interference. No idea how we didn't catch this before.
Shader-db results on Kaby Lake:
total instructions in shared programs: 15311100 -> 15311100 (0.00%)
instructions in affected programs: 0 -> 0
helped: 0
HURT: 0
total cycles in shared programs: 355468050 -> 355543197 (0.02%)
cycles in affected programs: 2472492 -> 2547639 (3.04%)
helped: 17
HURT: 20
Fixes: 014edff0d2 "intel/fs: Add interference between SENDS sources"
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
We want to be able to call ra_allocate() and, when it fails, mutate the
graph and try again rather than re-building the graph from scratch.
This commit moves all the scratch bits except the final register
allocation (which is really an out value not scratch) into sub-structs
named "tmp" to make it clear which things are scratch. It also adds
bits to the ra_select() initialization loop to initialize things (since
we can't trust rzalloc anymore) and copy q_test and forced_reg over.
Reviewed-by: Eric Anholt <eric@anholt.net>
Unfortunately, we can't quite follow the standard C conventions for
these because ralloc doesn't know the sizes of pointers.
Reviewed-by: Eric Anholt <eric@anholt.net>
In the last phase of the schedule and RA loop, the RA call is redundant
if we spill. Immediately afterwards, we're going to see that we
couldn't allocate without spilling and call back into RA and tell it to
go ahead and spill. We've known about it for a while but we've always
brushed over it on the theory that, if you're going to spill, you'll be
calling RA a bunch anyway and what does one extra RA hurt? As it turns
out, it hurts more than you'd expect. Because the RA interference graph
gets sparser with each spill and the RA algorithm is more efficient on
sparser graphs, the RA call that we're duplicating is actually the most
expensive call in the RA-and-spill loop.
There's another extra RA call we do that's a bit harder to see which
this also removes. If we try to compile a shader that isn't the minimum
dispatch width and it fails to allocate without spilling we call fail()
to set an error but then go ahead and do the first spilling RA pass and
only after that's complete do we detect the fail and bail out. By
making minimum dispatch widths part of the spill condition, we side-step
this problem.
Getting rid of these extra spills takes the compile time of a nasty
Aztec Ruins shader from about 28 seconds to about 26 seconds on my
laptop. It also makes shader-db 1.5% faster
Shader-db results on Kaby Lake:
total instructions in shared programs: 15311100 -> 15311100 (0.00%)
instructions in affected programs: 0 -> 0
helped: 0
HURT: 0
total cycles in shared programs: 355468050 -> 355468050 (0.00%)
cycles in affected programs: 0 -> 0
helped: 0
HURT: 0
Total CPU time (seconds): 2524.31 -> 2486.63 (-1.49%)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
The most expensive part of register allocation is the ra_simplify step
which is a fixed-point algorithm with a worst-case complexity of O(n^2)
which adds the registers to a stack which we then use later to do the
actual allocation. This commit uses bit sets and changes the core loop
of ra_simplify to first walk 32-node chunks and then walk each chunk.
This lets us skip whole 32-node chunks in one go based on bit operations
and compute the minimum q value potentially 32x as fast. Of course, the
algorithm still has the same fundamental O(n^2) run-time but the
constant is now much lower.
In the nasty Aztec Ruins compute shader, this shaves a full four seconds
off the 30s compile time for a release build of mesa. In a debug build
(needed for accurate stack traces), perf says that ra_select takes 20%
of runtime before this patch and only 5-6% of runtime after this patch.
It also makes shader-db runs faster.
Shader-db results on Kaby Lake:
total instructions in shared programs: 15311100 -> 15311100 (0.00%)
instructions in affected programs: 0 -> 0
helped: 0
HURT: 0
total cycles in shared programs: 355468050 -> 355468050 (0.00%)
cycles in affected programs: 0 -> 0
helped: 0
HURT: 0
Total CPU time (seconds): 2602.37 -> 2524.31 (-3.00%)
Reviewed-by: Eric Anholt <eric@anholt.net>
We only use q_total if the reg is not assigned so there's no point in
updating it if the reg is not assigned. This has no known perf benefit
but it will reduce churn in a future commit.
Reviewed-by: Eric Anholt <eric@anholt.net>
This shaves about half a second off the 30 second compile time of one of
the compute shaders in Aztec ruins.
Reviewed-by: Eric Anholt <eric@anholt.net>
Both apps and we (see virgl_buffer_transfer_flush_region) might
flush regions that are unmodified. We have to read back for those
flushes.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
It currently has no user and is probably incorrect (resource_wait is
required in some more cases). Remove it so that we can focus on
transfers first.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>
No surface requires an auxiliary surface to operate correctly. Fall back
to an uncompressed surface if mesa fails to create and allocate an
auxiliary surface. This enables adding more restrictions to ISL without
having to update iris.
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
intel_tiling_supports_hiz() and intel_miptree_supports_hiz() duplicate
much the work done by isl_surf_get_hiz_surf(). Replace them with simple
expressions.
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Import some restrictions from intel_tiling_supports_hiz() and
intel_miptree_supports_hiz().
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
intel_tiling_supports_ccs() and intel_miptree_supports_ccs() duplicate
much the work done by isl_surf_get_ccs_surf(). Drop them both and index
a boolean array to choose CCS_D in intel_miptree_choose_aux_usage().
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
This function duplicates much the work done by isl_surf_get_mcs_surf().
Replace it with a simple expression.
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Import some restrictions from intel_miptree_supports_mcs() and don't
assume that the caller knows which device generations are supported.
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
No surface requires an auxiliary surface to operate correctly. Fall back
to an uncompressed surface if mesa fails to create and allocate an
auxiliary surface. This enables adding more restrictions to ISL without
having to update i965.
Reviewed-by: Rafael Antognolli <rafael.antognolli@intel.com>
Instead of checking the API variant on entry of set_varying_vp_inputs
to check if we can ever be interrested in fixed function processing
or not, we can check if we are actually fixed function processing.
To check this we can use the immediately updated
gl_context::VertexProgram._VPMode value that tells us if we have a
user provided shader program or if we are in fixed function processing
either through an internal TNL shader of directly through hardware.
When doing so, we also need to recheck the varying_vp_inputs variable
at the time gl_context::VertexProgram._VPMode is set to VP_MODE_FF.
Put asserts at the consumers of gl_context::varying_vp_inputs to make
sure gl_context::VertexProgram._VPMode is set to VP_MODE_FF. By that
gl_context::varying_vp_inputs should be up to date then.
By not looking at the opengl api for this decision we should actually
catch more cases where we can avoid setting a state change flag, including
the ones where we cannot get into VP_MODE_FF by the choice of the api.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Mathias Fröhlich <Mathias.Froehlich@web.de>
The precondition stated in the comment is not true. The values mentioned are
only set from _mesa_update_state which in turn may not yet be called.
For now set the _NEW_VARYING_VP_INPUTS flag a bit more often, we will narrow
that down to a minimum again in a later patch.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Mathias Fröhlich <Mathias.Froehlich@web.de>
Ian Romanick
Jason Ekstrand
Chia-I Wu
Nanley Chery
Mathias Fröhlich