AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
9656dcedc2
skia2/resources/SkVMTest.expected

306 lines
5.8 KiB
Plaintext
Raw Normal View History

streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
A8 over A8
r0 = load8 arg(0)
r1 = splat 3B808081 (0.0039215689)
r0 = to_f32 r0
r0 = mul_f32 r1 r0
r2 = load8 arg(1)
r2 = to_f32 r2
r2 = mul_f32 r1 r2
r1 = splat 3F800000 (1)
r1 = sub_f32 r1 r0
r1 = mad_f32 r2 r1 r0
r2 = splat 437F0000 (255)
r0 = splat 3F000000 (0.5)
r0 = mad_f32 r1 r2 r0
r0 = to_i32 r0
store8 arg(1) r0
A8 over G8
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r0 = load8 arg(0)
r1 = splat 3B808081 (0.0039215689)
r0 = to_f32 r0
r0 = mul_f32 r1 r0
r2 = load8 arg(1)
r2 = to_f32 r2
r2 = mul_f32 r1 r2
r1 = splat 3F800000 (1)
r1 = sub_f32 r1 r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r1 = mul_f32 r2 r1
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r2 = splat 3E59B3D0 (0.21259999)
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = splat 3F371759 (0.71520001)
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r3 = splat 3D93DD98 (0.0722)
r3 = mul_f32 r1 r3
r3 = mad_f32 r1 r0 r3
r3 = mad_f32 r1 r2 r3
r2 = splat 437F0000 (255)
r1 = splat 3F000000 (0.5)
r1 = mad_f32 r3 r2 r1
r1 = to_i32 r1
store8 arg(1) r1
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
A8 over RGBA_8888
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r0 = load8 arg(0)
r1 = splat 3B808081 (0.0039215689)
r0 = to_f32 r0
r0 = mul_f32 r1 r0
r2 = load32 arg(1)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r3 = extract r2 FF
r3 = to_f32 r3
r3 = mul_f32 r1 r3
r4 = extract r2 FF00
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r4 = to_f32 r4
r4 = mul_f32 r1 r4
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r5 = extract r2 FF0000
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r5 = to_f32 r5
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r5 = mul_f32 r1 r5
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
r2 = shr r2 24
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r2 = to_f32 r2
r2 = mul_f32 r1 r2
r1 = splat 3F800000 (1)
r1 = sub_f32 r1 r0
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r3 = mul_f32 r3 r1
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r4 = mul_f32 r4 r1
r5 = mul_f32 r5 r1
r1 = mad_f32 r2 r1 r0
r2 = splat 437F0000 (255)
r0 = splat 3F000000 (0.5)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r3 = mad_f32 r3 r2 r0
r3 = to_i32 r3
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r4 = mad_f32 r4 r2 r0
r4 = to_i32 r4
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r5 = mad_f32 r5 r2 r0
r5 = to_i32 r5
add a x*y+0 -> x*y peephole This is mostly to test how easy rebaselining SkVMTest is. Change-Id: I27ab6f6bb8b7e126327374009783afd86d416f55 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218039 Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-03 21:32:49 +00:00
r0 = mad_f32 r1 r2 r0
r0 = to_i32 r0
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = pack r3 r4 8
r0 = pack r5 r0 8
r0 = pack r4 r0 16
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
store32 arg(1) r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
G8 over A8
r0 = splat 3B808081 (0.0039215689)
r1 = splat 3F800000 (1)
r2 = load8 arg(1)
r2 = to_f32 r2
r2 = mul_f32 r0 r2
r0 = sub_f32 r1 r1
r0 = mad_f32 r2 r0 r1
r2 = splat 437F0000 (255)
r1 = splat 3F000000 (0.5)
r1 = mad_f32 r0 r2 r1
r1 = to_i32 r1
store8 arg(1) r1
G8 over G8
r0 = load8 arg(0)
r1 = splat 3B808081 (0.0039215689)
r0 = to_f32 r0
r0 = mul_f32 r1 r0
r2 = splat 3F800000 (1)
r3 = load8 arg(1)
r3 = to_f32 r3
r3 = mul_f32 r1 r3
r2 = sub_f32 r2 r2
r2 = mad_f32 r3 r2 r0
r3 = splat 3E59B3D0 (0.21259999)
r0 = splat 3F371759 (0.71520001)
r1 = splat 3D93DD98 (0.0722)
r1 = mul_f32 r2 r1
r1 = mad_f32 r2 r0 r1
r1 = mad_f32 r2 r3 r1
r3 = splat 437F0000 (255)
r2 = splat 3F000000 (0.5)
r2 = mad_f32 r1 r3 r2
r2 = to_i32 r2
store8 arg(1) r2
G8 over RGBA_8888
r0 = load8 arg(0)
r1 = splat 3B808081 (0.0039215689)
r0 = to_f32 r0
r0 = mul_f32 r1 r0
r2 = splat 3F800000 (1)
r3 = load32 arg(1)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = extract r3 FF
r4 = to_f32 r4
r4 = mul_f32 r1 r4
r5 = extract r3 FF00
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r5 = to_f32 r5
r5 = mul_f32 r1 r5
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r6 = extract r3 FF0000
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r6 = to_f32 r6
r6 = mul_f32 r1 r6
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
r3 = shr r3 24
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r3 = to_f32 r3
r3 = mul_f32 r1 r3
r1 = sub_f32 r2 r2
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = mad_f32 r4 r1 r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r5 = mad_f32 r5 r1 r0
r6 = mad_f32 r6 r1 r0
r1 = mad_f32 r3 r1 r2
r3 = splat 437F0000 (255)
r2 = splat 3F000000 (0.5)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = mad_f32 r4 r3 r2
r4 = to_i32 r4
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r5 = mad_f32 r5 r3 r2
r5 = to_i32 r5
r6 = mad_f32 r6 r3 r2
r6 = to_i32 r6
r2 = mad_f32 r1 r3 r2
r2 = to_i32 r2
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r5 = pack r4 r5 8
r2 = pack r6 r2 8
r2 = pack r5 r2 16
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
store32 arg(1) r2
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
RGBA_8888 over A8
r0 = load32 arg(0)
r1 = splat 3B808081 (0.0039215689)
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
r0 = shr r0 24
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = to_f32 r0
r0 = mul_f32 r1 r0
r2 = load8 arg(1)
r2 = to_f32 r2
r2 = mul_f32 r1 r2
r1 = splat 3F800000 (1)
r1 = sub_f32 r1 r0
r1 = mad_f32 r2 r1 r0
r2 = splat 437F0000 (255)
r0 = splat 3F000000 (0.5)
r0 = mad_f32 r1 r2 r0
r0 = to_i32 r0
store8 arg(1) r0
RGBA_8888 over G8
r0 = load32 arg(0)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r1 = extract r0 FF
r2 = splat 3B808081 (0.0039215689)
r1 = to_f32 r1
r1 = mul_f32 r2 r1
r3 = extract r0 FF00
r3 = to_f32 r3
r3 = mul_f32 r2 r3
r4 = extract r0 FF0000
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r4 = to_f32 r4
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = mul_f32 r2 r4
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
r0 = shr r0 24
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = to_f32 r0
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r0 = mul_f32 r2 r0
r5 = load8 arg(1)
r5 = to_f32 r5
r5 = mul_f32 r2 r5
r2 = splat 3F800000 (1)
r2 = sub_f32 r2 r0
r1 = mad_f32 r5 r2 r1
r3 = mad_f32 r5 r2 r3
r2 = mad_f32 r5 r2 r4
r5 = splat 3E59B3D0 (0.21259999)
r4 = splat 3F371759 (0.71520001)
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = splat 3D93DD98 (0.0722)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r0 = mul_f32 r2 r0
r0 = mad_f32 r3 r4 r0
r0 = mad_f32 r1 r5 r0
r5 = splat 437F0000 (255)
r1 = splat 3F000000 (0.5)
r1 = mad_f32 r0 r5 r1
r1 = to_i32 r1
store8 arg(1) r1
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
RGBA_8888 over RGBA_8888
r0 = load32 arg(0)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r1 = extract r0 FF
r2 = splat 3B808081 (0.0039215689)
r1 = to_f32 r1
r1 = mul_f32 r2 r1
r3 = extract r0 FF00
r3 = to_f32 r3
r3 = mul_f32 r2 r3
r4 = extract r0 FF0000
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r4 = to_f32 r4
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = mul_f32 r2 r4
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
r0 = shr r0 24
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = to_f32 r0
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r0 = mul_f32 r2 r0
r5 = load32 arg(1)
r6 = extract r5 FF
r6 = to_f32 r6
r6 = mul_f32 r2 r6
r7 = extract r5 FF00
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r7 = to_f32 r7
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r7 = mul_f32 r2 r7
r8 = extract r5 FF0000
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r8 = to_f32 r8
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r8 = mul_f32 r2 r8
r5 = shr r5 24
r5 = to_f32 r5
r5 = mul_f32 r2 r5
r2 = splat 3F800000 (1)
r2 = sub_f32 r2 r0
r6 = mad_f32 r6 r2 r1
r7 = mad_f32 r7 r2 r3
r8 = mad_f32 r8 r2 r4
r2 = mad_f32 r5 r2 r0
r5 = splat 437F0000 (255)
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = splat 3F000000 (0.5)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r6 = mad_f32 r6 r5 r0
r6 = to_i32 r6
r7 = mad_f32 r7 r5 r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r7 = to_i32 r7
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r8 = mad_f32 r8 r5 r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r8 = to_i32 r8
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r0 = mad_f32 r2 r5 r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
r0 = to_i32 r0
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r7 = pack r6 r7 8
r0 = pack r8 r0 8
r0 = pack r7 r0 16
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
store32 arg(1) r0
streamline SkVM test rebaselining - keep expectations in resources/ - overwrite automatically if needed so we can see the diff in Git Change-Id: I2486b127ebcc7f40332fd0462e38b1af04d3e32b Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218038 Reviewed-by: Brian Osman <brianosman@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 21:27:46 +00:00
centralize test/bench SkVM builders Eliminate the duplicate functionality, and better testing for the bench builders. Change-Id: If20e52107738903f854aec431416e573d7a7d640 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218041 Reviewed-by: Mike Klein <mtklein@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 22:10:59 +00:00
I32 8888 over 8888
r0 = load32 arg(0)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r1 = extract r0 FF
r2 = extract r0 FF00
r3 = extract r0 FF0000
add pack instruction This instruction can lower to some useful SSE/NEON instructions, and even if not, is a handy way to express the frequent paring of << and |. I32_SWAR: 2.3 -> 1.9 I32: 2.6 -> 2.4 F32: 5.1 -> 4.7 Change-Id: Ia169ad40f0aaef32417e05d9bf91c2d2542e7b5f Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218238 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 15:41:49 +00:00
r0 = shr r0 24
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r4 = load32 arg(1)
r5 = extract r4 FF
r6 = extract r4 FF00
r7 = extract r4 FF0000
r4 = shr r4 24
r8 = splat FF (3.5733111e-43)
r8 = sub_i32 r8 r0
r5 = mul_unorm8 r5 r8
r5 = add_i32 r1 r5
r6 = mul_unorm8 r6 r8
centralize test/bench SkVM builders Eliminate the duplicate functionality, and better testing for the bench builders. Change-Id: If20e52107738903f854aec431416e573d7a7d640 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218041 Reviewed-by: Mike Klein <mtklein@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 22:10:59 +00:00
r6 = add_i32 r2 r6
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r7 = mul_unorm8 r7 r8
centralize test/bench SkVM builders Eliminate the duplicate functionality, and better testing for the bench builders. Change-Id: If20e52107738903f854aec431416e573d7a7d640 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218041 Reviewed-by: Mike Klein <mtklein@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 22:10:59 +00:00
r7 = add_i32 r3 r7
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r8 = mul_unorm8 r4 r8
r8 = add_i32 r0 r8
r6 = pack r5 r6 8
r8 = pack r7 r8 8
r8 = pack r6 r8 16
store32 arg(1) r8
centralize test/bench SkVM builders Eliminate the duplicate functionality, and better testing for the bench builders. Change-Id: If20e52107738903f854aec431416e573d7a7d640 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218041 Reviewed-by: Mike Klein <mtklein@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 22:10:59 +00:00
I32 (SWAR) 8888 over 8888
r0 = load32 arg(0)
add extract instruction Kind of the flip side of pack. Made slightly awkward by instructions having only one immediate... calling _BitScanForward / __builtin_ctz() at runtime seems to work fine, but it really could have been done at compile time. Change-Id: Ic83fe8e0a1603fb9189598dcc26c842cc797bf45 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218241 Commit-Queue: Mike Klein <mtklein@google.com> Reviewed-by: Brian Osman <brianosman@google.com>
2019-06-04 16:33:25 +00:00
r1 = extract r0 FF00FF
r0 = extract r0 FF00FF00
r2 = load32 arg(1)
r3 = extract r2 FF00FF
r2 = extract r2 FF00FF00
r4 = splat FF (3.5733111e-43)
r5 = shr r0 16
r5 = sub_i32 r4 r5
r4 = splat FF00FF (2.3418409e-38)
r3 = mul_i32 r3 r5
r3 = add_i32 r3 r4
r3 = extract r3 FF00FF00
r3 = add_i32 r1 r3
r5 = mul_i32 r2 r5
r5 = add_i32 r5 r4
r5 = extract r5 FF00FF00
r5 = add_i32 r0 r5
r5 = pack r3 r5 8
store32 arg(1) r5
centralize test/bench SkVM builders Eliminate the duplicate functionality, and better testing for the bench builders. Change-Id: If20e52107738903f854aec431416e573d7a7d640 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/218041 Reviewed-by: Mike Klein <mtklein@google.com> Commit-Queue: Mike Klein <mtklein@google.com>
2019-06-03 22:10:59 +00:00
Reference in New Issue Copy Permalink