When an ir_call is encountered that invokes a builtin, it will now try
to generate a lowered version of the builtin. This only happens if all
of the arguments to the function are lowerable. Previously the builtin
would be inlined before the lowering pass is invoked and then the
implementation would be lowered as a consequence of the pass. However
this causes problems if the builtin has multiple arguments and the
implementation has operations on only a few of the arguments before
combining it with the others. In that case the entire builtin should
only be lowered if all of the arguments are lower precision. The
previous approach would end up lowering only parts of the
implementation.
The lowered implementations are cached in a hash table in case they can
be reused.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3885>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3885>
Previously, the ir_call functions for builtin functions were replaced
with the inline implementation immediately after being added to the
instruction list. This patch replaces that with a separate pass that
lowers them after the conversion from AST to IR is complete. This will
be useful to be able to insert some handling for the precision lowering
pass before the inlining. This needs to happen because the precision
of the operations in the inlined implementation depends on the highest
precision of all of the arguments to the call.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3885>
The pass lowers 1-bit booleans produced by NIR to the native bitsize
of the operations that produce them.
v2: change on lower_load_const_instr after upstream changes. Added
TODO2 to explain it, as it was not properly tested yet (see
already existing TODO) (Neil)
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3885>
This works by finding the first rvalue that it can lower using an
ir_rvalue_visitor. In that case it adds a conversion to float16
after each rvalue and a conversion back to float before storing
the assignment.
Also it uses a set to keep track of rvalues that have been
lowred already. The handle_rvalue method of the rvalue visitor doesn’t
provide any way to stop iteration. If we handle a value in
find_precision_visitor we want to be able to stop it from descending into
the lowered rvalue again.
Additionally this pass disallows converting nodes containing non-float.
The can_lower_rvalue function explicitly excludes any branches
that have non-float types except bools. This avoids the need to have
special handling for functions that convert to int or double.
Co-authored-by: Hyunjun Ko <zzoon@igalia.com>
v2. Adds lowering for texture samples
v3. Instead of checking whether each node can be lowered while walking the
tree, a separate tree walk is now done to check all of the nodes in a
single pass. The lowerable nodes are added to a set which is checked
during find_precision_visitor instead of calling can_lower_rvalue.
v4. Move the special case for temporaries to find_lowerable_rvalues. This
needs to be handled while checking for lowerable rvalues so that any
later dereferences of the variable will see the right precision.
v5. Add an override to visit ir_call instructions and apply the same
technique to override the precision of the temporary variable in the
same way as done for builtin temporaries and ir_assignment calls.
v6. Changes the pass so that it doesn’t need to lower an entire subtree in
order do perform a lowering. Instead, certain instructions can be
marked as being indepedent of their child instructions. For example,
this is the case with array dereferences. The precision of the array
index doesn’t have any bearing on whether things using the result of
the array deref can be lowered.
Now, only toplevel lowerable nodes are added to the lowerable_rvalues
instead instead of additionally adding all of the subnodes.
It now also only needs one hash table instead of two.
v7. Don’t try to lower sampler types. Instead, the sample instruction is
now treated as an independent point where the result of the sample can
be used in a lowered section. The precision of the sampler type
determines the precision of the sample instruction. This also means
the coordinates to the sampler can be lowered.
v8. Use f2fmp instead of f2f16.
v9. Disable lowering derivatives calcualtions, which might not work
properly on some hw backends.
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3885>
Adds ir_dereference::precision(). For a normal variable dereference,
the precision comes from the variable. For a record member it comes
from the field within the record. For an array it can come from
either, depending on where the underlying array is stored. The method
recursively walks the derefs until it finds one of the first two.
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3885>
This function has code like:
if (0x7FD <= zExp) {
if ((0x7FD < zExp) ||
((zExp == 0x7FD) &&
(0x001FFFFFu == zFrac0 && 0xFFFFFFFFu == zFrac1) &&
increment)) {
...
return ...;
}
if (zExp < 0) {
I saw that, and I thought, "Uh... what? Dead code?" I thought it was a
bit fishy, so I grabbed the Berkeley SoftFloat Library 3e code, and
there is similar code in softfloat_roundPackToF64
(source/s_roundPackToF64.c), but it has an extra (uint16_t) cast in the
first comparison. This is basicially a shortcut for
if (zExp < 0 || zExp >= 0x7FD) {
So, having the nesting kind of makes sense. On a CPU, nesting the flow
control can be an optimization. On a GPU, it's just fail. Split the
block so that we don't need the uint16_t cast magic.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 683638 -> 658127 (-3.73%)
instructions in affected programs: 666839 -> 641328 (-3.83%)
helped: 92
HURT: 0
helped stats (abs) min: 26 max: 2456 x̄: 277.29 x̃: 144
helped stats (rel) min: 3.21% max: 4.22% x̄: 3.79% x̃: 3.90%
95% mean confidence interval for instructions value: -345.84 -208.75
95% mean confidence interval for instructions %-change: -3.86% -3.73%
Instructions are helped.
total cycles in shared programs: 5458858 -> 5344600 (-2.09%)
cycles in affected programs: 5360114 -> 5245856 (-2.13%)
helped: 92
HURT: 0
helped stats (abs) min: 126 max: 10300 x̄: 1241.93 x̃: 655
helped stats (rel) min: 1.71% max: 2.37% x̄: 2.12% x̃: 2.17%
95% mean confidence interval for cycles value: -1539.93 -943.94
95% mean confidence interval for cycles %-change: -2.16% -2.08%
Cycles are helped.
Fixes: f111d72596 ("glsl: Add "built-in" functions to do add(fp64, fp64)")
Reviewed-by: Matt Turner <mattst88@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
The previous two commits were just setting the scene for this change.
The mix(..., __propagateFloat64NaN(a, b), propagate) statements are not
identical in the two halves, but they are equivalent. The first clause
of the mix in the else-branch is trivally ±Inf. The first clause in the
then-branch __packFloat64(aSign, aExp, aFracHi, aFracLo). The
preceeding conditions prove that aExp=0x7ff, aFracHi=0, and aFracLo=0.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 819560 -> 787258 (-3.94%)
instructions in affected programs: 757737 -> 725435 (-4.26%)
helped: 74
HURT: 0
helped stats (abs) min: 43 max: 3545 x̄: 436.51 x̃: 296
helped stats (rel) min: 3.54% max: 6.16% x̄: 4.52% x̃: 4.36%
95% mean confidence interval for instructions value: -548.42 -324.61
95% mean confidence interval for instructions %-change: -4.68% -4.37%
Instructions are helped.
total cycles in shared programs: 6817254 -> 6483227 (-4.90%)
cycles in affected programs: 6385272 -> 6051245 (-5.23%)
helped: 74
HURT: 0
helped stats (abs) min: 430 max: 33271 x̄: 4513.88 x̃: 3047
helped stats (rel) min: 4.28% max: 7.45% x̄: 5.48% x̃: 5.31%
95% mean confidence interval for cycles value: -5610.46 -3417.30
95% mean confidence interval for cycles %-change: -5.65% -5.32%
Cycles are helped.
total spills in shared programs: 591 -> 553 (-6.43%)
spills in affected programs: 591 -> 553 (-6.43%)
helped: 1
HURT: 0
total fills in shared programs: 1353 -> 1307 (-3.40%)
fills in affected programs: 1353 -> 1307 (-3.40%)
helped: 1
HURT: 0
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
In one branch we know that expDiff is already positive.
In the other branch we know the expDiff is negative. Previously in that
branch the code was -(expDiff + 1). This is equvialent to (-expDiff) -
1, and since expDiff is negative, abs(expDiff) - 1.
The main purpose of this commit is to prepare for "soft-fp64/fadd: Move
common code out of both branches of an if-statement".
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 818246 -> 819560 (0.16%)
instructions in affected programs: 756423 -> 757737 (0.17%)
helped: 1
HURT: 73
helped stats (abs) min: 1205 max: 1205 x̄: 1205.00 x̃: 1205
helped stats (rel) min: 1.36% max: 1.36% x̄: 1.36% x̃: 1.36%
HURT stats (abs) min: 2 max: 149 x̄: 34.51 x̃: 27
HURT stats (rel) min: 0.14% max: 1.09% x̄: 0.41% x̃: 0.30%
95% mean confidence interval for instructions value: -16.56 52.07
95% mean confidence interval for instructions %-change: 0.30% 0.47%
Inconclusive result (value mean confidence interval includes 0).
total cycles in shared programs: 6816686 -> 6817254 (<.01%)
cycles in affected programs: 6384704 -> 6385272 (<.01%)
helped: 37
HURT: 37
helped stats (abs) min: 30 max: 5790 x̄: 289.05 x̃: 102
helped stats (rel) min: 0.04% max: 0.86% x̄: 0.29% x̃: 0.31%
HURT stats (abs) min: 2 max: 1020 x̄: 304.41 x̃: 232
HURT stats (rel) min: <.01% max: 1.58% x̄: 0.55% x̃: 0.43%
95% mean confidence interval for cycles value: -165.37 180.72
95% mean confidence interval for cycles %-change: <.01% 0.27%
Inconclusive result (value mean confidence interval includes 0).
total spills in shared programs: 705 -> 591 (-16.17%)
spills in affected programs: 705 -> 591 (-16.17%)
helped: 1
HURT: 0
total fills in shared programs: 1501 -> 1353 (-9.86%)
fills in affected programs: 1501 -> 1353 (-9.86%)
helped: 1
HURT: 0
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
Exchanging aFracHi / bFracHi and aFracLo / bFracLo should not affect the
result of the later call to __add64.
The main purpose of this commit is to prepare for "soft-fp64/fadd: Move
common code out of both branches of an if-statement".
v2: Fix a typo in a comment. Noticed by Matt.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 812094 -> 818246 (0.76%)
instructions in affected programs: 750271 -> 756423 (0.82%)
helped: 0
HURT: 74
HURT stats (abs) min: 7 max: 520 x̄: 83.14 x̃: 59
HURT stats (rel) min: 0.52% max: 1.48% x̄: 0.89% x̃: 0.84%
95% mean confidence interval for instructions value: 63.96 102.31
95% mean confidence interval for instructions %-change: 0.83% 0.95%
Instructions are HURT.
total cycles in shared programs: 6797157 -> 6816686 (0.29%)
cycles in affected programs: 6365175 -> 6384704 (0.31%)
helped: 0
HURT: 74
HURT stats (abs) min: 16 max: 1690 x̄: 263.91 x̃: 181
HURT stats (rel) min: 0.14% max: 0.68% x̄: 0.32% x̃: 0.27%
95% mean confidence interval for cycles value: 199.74 328.07
95% mean confidence interval for cycles %-change: 0.29% 0.36%
Cycles are HURT.
total spills in shared programs: 703 -> 705 (0.28%)
spills in affected programs: 703 -> 705 (0.28%)
helped: 0
HURT: 1
total fills in shared programs: 1499 -> 1501 (0.13%)
fills in affected programs: 1499 -> 1501 (0.13%)
helped: 0
HURT: 1
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
The main purpose of this commit is to prepare for "soft-fp64/fadd: Move
common code out of both branches of an if-statement".
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 812590 -> 812094 (-0.06%)
instructions in affected programs: 672135 -> 671639 (-0.07%)
helped: 57
HURT: 0
helped stats (abs) min: 1 max: 32 x̄: 8.70 x̃: 7
helped stats (rel) min: <.01% max: 0.49% x̄: 0.12% x̃: 0.09%
95% mean confidence interval for instructions value: -10.46 -6.94
95% mean confidence interval for instructions %-change: -0.15% -0.09%
Instructions are helped.
total cycles in shared programs: 6798039 -> 6797157 (-0.01%)
cycles in affected programs: 5810059 -> 5809177 (-0.02%)
helped: 54
HURT: 2
helped stats (abs) min: 2 max: 68 x̄: 16.44 x̃: 12
helped stats (rel) min: <.01% max: 0.12% x̄: 0.03% x̃: 0.02%
HURT stats (abs) min: 2 max: 4 x̄: 3.00 x̃: 3
HURT stats (rel) min: <.01% max: <.01% x̄: <.01% x̃: <.01%
95% mean confidence interval for cycles value: -19.50 -12.00
95% mean confidence interval for cycles %-change: -0.03% -0.02%
Cycles are helped.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
Convert
} else if (...) {
...
} else {
...
}
to
} else {
if (...) {
...
} else {
...
}
}
Not doing this reformatting in the previous commit makes the previous
commit easier to review, and doing it before the next commit makes the
next commit easier to review.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
Previous condition checks already guaranteen that expDiff != 0 and
!(expDiff > 0), so expDiff < 0 is the only option left.
The main purpose of this commit is to prepare for "soft-fp64/fadd: Move
common code out of both branches of an if-statement".
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 815491 -> 812590 (-0.36%)
instructions in affected programs: 753668 -> 750767 (-0.38%)
helped: 74
HURT: 0
helped stats (abs) min: 3 max: 281 x̄: 39.20 x̃: 25
helped stats (rel) min: 0.29% max: 0.73% x̄: 0.42% x̃: 0.40%
95% mean confidence interval for instructions value: -48.50 -29.91
95% mean confidence interval for instructions %-change: -0.45% -0.40%
Instructions are helped.
total cycles in shared programs: 6813681 -> 6798039 (-0.23%)
cycles in affected programs: 6381699 -> 6366057 (-0.25%)
helped: 74
HURT: 0
helped stats (abs) min: 24 max: 1488 x̄: 211.38 x̃: 149
helped stats (rel) min: 0.20% max: 0.44% x̄: 0.26% x̃: 0.25%
95% mean confidence interval for cycles value: -261.68 -161.08
95% mean confidence interval for cycles %-change: -0.28% -0.25%
Cycles are helped.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
The main purpose of this commit is to prepare for "soft-fp64/fadd: Move
common code out of both branches of an if-statement".
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 822766 -> 817327 (-0.66%)
instructions in affected programs: 760943 -> 755504 (-0.71%)
helped: 74
HURT: 0
helped stats (abs) min: 8 max: 515 x̄: 73.50 x̃: 51
helped stats (rel) min: 0.58% max: 1.10% x̄: 0.77% x̃: 0.73%
95% mean confidence interval for instructions value: -91.17 -55.83
95% mean confidence interval for instructions %-change: -0.81% -0.74%
Instructions are helped.
total cycles in shared programs: 6816791 -> 6822826 (0.09%)
cycles in affected programs: 6384809 -> 6390844 (0.09%)
helped: 0
HURT: 74
HURT stats (abs) min: 6 max: 1179 x̄: 81.55 x̃: 50
HURT stats (rel) min: 0.02% max: 0.17% x̄: 0.09% x̃: 0.09%
95% mean confidence interval for cycles value: 48.99 114.12
95% mean confidence interval for cycles %-change: 0.09% 0.10%
Cycles are HURT.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
v2: Go to extra effort to avoid flow control inserted to implement
short-circuit evaluation rules.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 797779 -> 796849 (-0.12%)
instructions in affected programs: 3499 -> 2569 (-26.58%)
helped: 21
HURT: 0
helped stats (abs) min: 8 max: 112 x̄: 44.29 x̃: 44
helped stats (rel) min: 16.09% max: 33.15% x̄: 25.72% x̃: 24.62%
95% mean confidence interval for instructions value: -55.94 -32.63
95% mean confidence interval for instructions %-change: -28.14% -23.30%
Instructions are helped.
total cycles in shared programs: 6601355 -> 6588351 (-0.20%)
cycles in affected programs: 25376 -> 12372 (-51.25%)
helped: 21
HURT: 0
helped stats (abs) min: 156 max: 1410 x̄: 619.24 x̃: 526
helped stats (rel) min: 42.39% max: 53.98% x̄: 50.12% x̃: 50.75%
95% mean confidence interval for cycles value: -776.58 -461.89
95% mean confidence interval for cycles %-change: -51.57% -48.67%
Cycles are helped.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net> [v1]
Reviewed-by: Matt Turner <mattst88@gmail.com> [v1]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
fsat is defined as min(max(a, 0.0), 1.0), and IEEE defines both min and
max to return the non-NaN value when one value is NaN. Based on this,
fsat should definitely return 0.0 for NaN.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 841666 -> 841647 (<.01%)
instructions in affected programs: 122033 -> 122014 (-0.02%)
helped: 7
HURT: 0
helped stats (abs) min: 1 max: 4 x̄: 2.71 x̃: 3
helped stats (rel) min: 0.01% max: 0.02% x̄: 0.02% x̃: 0.01%
95% mean confidence interval for instructions value: -3.74 -1.69
95% mean confidence interval for instructions %-change: -0.02% -0.01%
Instructions are helped.
total cycles in shared programs: 6927246 -> 6926904 (<.01%)
cycles in affected programs: 1038987 -> 1038645 (-0.03%)
helped: 7
HURT: 0
helped stats (abs) min: 18 max: 72 x̄: 48.86 x̃: 54
helped stats (rel) min: 0.03% max: 0.05% x̄: 0.03% x̃: 0.03%
95% mean confidence interval for cycles value: -67.38 -30.33
95% mean confidence interval for cycles %-change: -0.05% -0.02%
Cycles are helped.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Fixes: a42163cbbc ("compiler: Add lowering support for 64-bit saturate operations to software")
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2585
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
__fabs64 doesn't do anything special, and the value is still NaN
regardless of the value of the MSB. In a strict sense, it's possible
that both functions should set the "signal" bit.
lts on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 844558 -> 843005 (-0.18%)
instructions in affected programs: 725975 -> 724422 (-0.21%)
helped: 53
HURT: 4
helped stats (abs) min: 1 max: 313 x̄: 29.87 x̃: 21
helped stats (rel) min: 0.01% max: 0.94% x̄: 0.30% x̃: 0.22%
HURT stats (abs) min: 4 max: 11 x̄: 7.50 x̃: 7
HURT stats (rel) min: 0.03% max: 0.09% x̄: 0.05% x̃: 0.04%
95% mean confidence interval for instructions value: -39.02 -15.47
95% mean confidence interval for instructions %-change: -0.34% -0.21%
Instructions are helped.
total cycles in shared programs: 6962024 -> 6944998 (-0.24%)
cycles in affected programs: 6185470 -> 6168444 (-0.28%)
helped: 59
HURT: 0
helped stats (abs) min: 64 max: 2863 x̄: 288.58 x̃: 208
helped stats (rel) min: 0.11% max: 0.87% x̄: 0.33% x̃: 0.27%
95% mean confidence interval for cycles value: -387.15 -190.00
95% mean confidence interval for cycles %-change: -0.38% -0.28%
Cycles are helped.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
Replace all of the bool(qSign) with qSign != 0u. Remove unnecessary
parenthesis from around most of the existing qSign != 0u.
This dramatically simplifies the next commit.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 848109 -> 848106 (<.01%)
instructions in affected programs: 53 -> 50 (-5.66%)
helped: 1
HURT: 0
total cycles in shared programs: 6969145 -> 6969125 (<.01%)
cycles in affected programs: 396 -> 376 (-5.05%)
helped: 1
HURT: 0
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
findMSB returns -1 for an input of zero, so 31 - findMSB(a) is
sufficient on its own.
There's only one user of findMSB in shader-db, and it does not match
this pattern.
TODO: Add a pattern in the backend code generator that emits 31 -
nir_op_ufind_msb(a) as if it were nir_op_uclz. That should save a couple
instructions.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 859509 -> 848109 (-1.33%)
instructions in affected programs: 841058 -> 829658 (-1.36%)
helped: 97
HURT: 0
helped stats (abs) min: 3 max: 1161 x̄: 117.53 x̃: 72
helped stats (rel) min: 0.98% max: 6.74% x̄: 1.70% x̃: 1.35%
95% mean confidence interval for instructions value: -147.21 -87.84
95% mean confidence interval for instructions %-change: -1.94% -1.46%
Instructions are helped.
total cycles in shared programs: 7072275 -> 6969145 (-1.46%)
cycles in affected programs: 6955767 -> 6852637 (-1.48%)
helped: 97
HURT: 0
helped stats (abs) min: 32 max: 10900 x̄: 1063.20 x̃: 560
helped stats (rel) min: 1.18% max: 7.58% x̄: 1.84% x̃: 1.45%
95% mean confidence interval for cycles value: -1339.43 -786.96
95% mean confidence interval for cycles %-change: -2.11% -1.57%
Cycles are helped.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
This doesn't help a lot of shaders, but it helps those few a LOT.
This could also be implemented using bcsel. That version is very
slightly worse because the generated SEL instruction wants to have two
immediate sources, so one of them usually needs an extra MOV instruction
to load.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 929619 -> 928859 (-0.08%)
instructions in affected programs: 1651 -> 891 (-46.03%)
helped: 8
HURT: 0
helped stats (abs) min: 38 max: 152 x̄: 95.00 x̃: 95
helped stats (rel) min: 42.70% max: 86.36% x̄: 49.88% x̃: 44.66%
95% mean confidence interval for instructions value: -132.97 -57.03
95% mean confidence interval for instructions %-change: -62.28% -37.49%
Instructions are helped.
total cycles in shared programs: 7280180 -> 7272912 (-0.10%)
cycles in affected programs: 12960 -> 5692 (-56.08%)
helped: 8
HURT: 0
helped stats (abs) min: 352 max: 1456 x̄: 908.50 x̃: 910
helped stats (rel) min: 52.45% max: 91.19% x̄: 59.24% x̃: 55.15%
95% mean confidence interval for cycles value: -1274.03 -542.97
95% mean confidence interval for cycles %-change: -70.06% -48.41%
Cycles are helped.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
The pattern is added to opt_algebraic because, for example, comparisons
with constant 0.0 will produce (a1 < 0).
Even with a pass that optimized Boolean expressions, I think this would
be very difficult to automatically recognize and optimize.
Results on the 308 shaders extracted from the fp64 portion of the OpenGL
CTS:
Tiger Lake and Ice Lake had similar results. (Tiger Lake shown)
total instructions in shared programs: 933054 -> 929619 (-0.37%)
instructions in affected programs: 784041 -> 780606 (-0.44%)
helped: 59
HURT: 0
helped stats (abs) min: 2 max: 213 x̄: 58.22 x̃: 44
helped stats (rel) min: 0.02% max: 2.51% x̄: 0.72% x̃: 0.46%
95% mean confidence interval for instructions value: -70.80 -45.64
95% mean confidence interval for instructions %-change: -0.92% -0.53%
Instructions are helped.
total cycles in shared programs: 7304712 -> 7280180 (-0.34%)
cycles in affected programs: 7176260 -> 7151728 (-0.34%)
helped: 92
HURT: 0
helped stats (abs) min: 8 max: 1414 x̄: 266.65 x̃: 166
helped stats (rel) min: 0.04% max: 2.34% x̄: 0.43% x̃: 0.22%
95% mean confidence interval for cycles value: -333.05 -200.26
95% mean confidence interval for cycles %-change: -0.54% -0.31%
Cycles are helped.
Regular shader-db changes:
No changes on any Intel platform.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4142>
Tapani Pälli