6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-17 03:30:28 +00:00
Code Issues Projects Releases Wiki Activity
a6a0f12942
llvm-6502/test/CodeGen/X86/vector-shuffle-128-v16.ll

281 lines
14 KiB
LLVM
Raw Normal View History

[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+ssse3 -x86-experimental-vector-shuffle-lowering | FileCheck %s --check-prefix=ALL --check-prefix=SSSE3
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-unknown"
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
define <16 x i8> @shuffle_v16i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; FIXME-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00
; FIXME: # BB#0:
; FIXME-NEXT: punpcklbw %xmm0, %xmm0
; FIXME-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
; FIXME-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,1]
; FIXME-NEXT: retq
; FIXME-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00
;
; SSE2-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00
; SSE2: # BB#0:
; SSE2-NEXT: punpcklbw %xmm0, %xmm0
; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,3]
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00
; SSSE3: # BB#0:
; SSSE3-NEXT: pxor %xmm1, %xmm1
; SSSE3-NEXT: pshufb %xmm1, %xmm0
; SSSE3-NEXT: retq
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_00_00_00_00_00_00_00_00_01_01_01_01_01_01_01_01(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_01_01_01_01_01_01_01_01
; SSE2: # BB#0:
; SSE2-NEXT: punpcklbw %xmm0, %xmm0
; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,3]
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,5,5,5,5]
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_01_01_01_01_01_01_01_01
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1]
; SSSE3-NEXT: retq
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08
; SSE2: # BB#0:
; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,2,4,5,6,7]
; SSE2-NEXT: punpcklbw %xmm0, %xmm0
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,6,6,6,6]
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_00_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,0,0,0,0,0,0,0,8,8,8,8,8,8,8,8]
; SSSE3-NEXT: retq
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_00_00_00_00_01_01_01_01_02_02_02_02_03_03_03_03(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; ALL-LABEL: @shuffle_v16i8_00_00_00_00_01_01_01_01_02_02_02_02_03_03_03_03
; ALL: # BB#0:
; ALL-NEXT: punpcklbw %xmm0, %xmm0
; ALL-NEXT: punpcklwd %xmm0, %xmm0
; ALL-NEXT: retq
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1, i32 2, i32 2, i32 2, i32 2, i32 3, i32 3, i32 3, i32 3>
ret <16 x i8> %shuffle
}
[x86] Add another combine that is particularly useful for the new vector shuffle lowering: match shuffle patterns equivalent to an unpcklwd or unpckhwd instruction. This allows us to use generic lowering code for v8i16 shuffles and match the unpack pattern late. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212705 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-10 11:09:29 +00:00
define <16 x i8> @shuffle_v16i8_04_04_04_04_05_05_05_05_06_06_06_06_07_07_07_07(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; ALL-LABEL: @shuffle_v16i8_04_04_04_04_05_05_05_05_06_06_06_06_07_07_07_07
; ALL: # BB#0:
; ALL-NEXT: punpcklbw %xmm0, %xmm0
; ALL-NEXT: punpckhwd %xmm0, %xmm0
; ALL-NEXT: retq
[x86] Add another combine that is particularly useful for the new vector shuffle lowering: match shuffle patterns equivalent to an unpcklwd or unpckhwd instruction. This allows us to use generic lowering code for v8i16 shuffles and match the unpack pattern late. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212705 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-10 11:09:29 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 4, i32 4, i32 4, i32 4, i32 5, i32 5, i32 5, i32 5, i32 6, i32 6, i32 6, i32 6, i32 7, i32 7, i32 7, i32 7>
ret <16 x i8> %shuffle
}
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
define <16 x i8> @shuffle_v16i8_00_00_00_00_04_04_04_04_08_08_08_08_12_12_12_12(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_00_00_00_00_04_04_04_04_08_08_08_08_12_12_12_12
; SSE2: # BB#0:
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,3,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: punpcklbw %xmm0, %xmm0
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,2,2,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,6,6]
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_00_00_00_00_04_04_04_04_08_08_08_08_12_12_12_12
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,0,0,0,4,4,4,4,8,8,8,8,12,12,12,12]
; SSSE3-NEXT: retq
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4, i32 8, i32 8, i32 8, i32 8, i32 12, i32 12, i32 12, i32 12>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_00_00_01_01_02_02_03_03_04_04_05_05_06_06_07_07(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; ALL-LABEL: @shuffle_v16i8_00_00_01_01_02_02_03_03_04_04_05_05_06_06_07_07
; ALL: # BB#0:
; ALL-NEXT: punpcklbw %xmm0, %xmm0
; ALL-NEXT: retq
[x86] Add handling for splat-like widenings of v16i8 shuffles. These show up really frequently, not the least with actual splats. =] We lowered these quite badly before. The new code path tries to widen i8 shuffles to i16 shuffles in a splat-like way. There are still some inefficiencies in our i16 splat logic though, so we aren't really done here. Also, for certain patterns (bit of a gather-and-splat) we still generate pretty silly code, and I've left a fixme for addressing it. However, I'm not actually worried about this code pattern as much. The old shuffle lowering generates a 29 instruction monstrosity for it that should execute much more slowly. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211974 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-28 05:16:40 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 0, i32 1, i32 1, i32 2, i32 2, i32 3, i32 3, i32 4, i32 4, i32 5, i32 5, i32 6, i32 6, i32 7, i32 7>
ret <16 x i8> %shuffle
}
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
define <16 x i8> @shuffle_v16i8_0101010101010101(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; FIXME-LABEL: @shuffle_v16i8_0101010101010101
; FIXME: # BB#0:
; FIXME-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
; FIXME-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,1]
; FIXME-NEXT: retq
;
; SSE2-LABEL: @shuffle_v16i8_0101010101010101
; SSE2: # BB#0:
; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,1,0,3]
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,0,0,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,4,4,4]
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_0101010101010101
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
; SSSE3-NEXT: retq
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1, i32 0, i32 1>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_00_16_01_17_02_18_03_19_04_20_05_21_06_22_07_23(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; ALL-LABEL: @shuffle_v16i8_00_16_01_17_02_18_03_19_04_20_05_21_06_22_07_23
; ALL: punpcklbw %xmm1, %xmm0
; ALL-NEXT: retq
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 4, i32 20, i32 5, i32 21, i32 6, i32 22, i32 7, i32 23>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_16_00_16_01_16_02_16_03_16_04_16_05_16_06_16_07(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_16_00_16_01_16_02_16_03_16_04_16_05_16_06_16_07
; SSE2: # BB#0:
; SSE2-NEXT: punpcklbw %xmm1, %xmm1
; SSE2-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,0,0,0,4,5,6,7]
; SSE2-NEXT: punpcklbw %xmm0, %xmm1
; SSE2-NEXT: movdqa %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_16_00_16_01_16_02_16_03_16_04_16_05_16_06_16_07
; SSSE3: # BB#0:
; SSSE3-NEXT: punpcklbw %xmm1, %xmm1
; SSSE3-NEXT: pshuflw {{.*}} # xmm1 = xmm1[0,0,0,0,4,5,6,7]
; SSSE3-NEXT: punpcklbw %xmm0, %xmm1
; SSSE3-NEXT: movdqa %xmm1, %xmm0
; SSSE3-NEXT: retq
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 16, i32 0, i32 16, i32 1, i32 16, i32 2, i32 16, i32 3, i32 16, i32 4, i32 16, i32 5, i32 16, i32 6, i32 16, i32 7>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_03_02_01_00_07_06_05_04_11_10_09_08_15_14_13_12(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_03_02_01_00_07_06_05_04_11_10_09_08_15_14_13_12
; SSE2: # BB#0:
; SSE2-NEXT: pxor %xmm1, %xmm1
; SSE2-NEXT: movdqa %xmm0, %xmm2
; SSE2-NEXT: punpckhbw %xmm1, %xmm2
; SSE2-NEXT: pshuflw {{.*}} # xmm2 = xmm2[3,2,1,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm2 = xmm2[0,1,2,3,7,6,5,4]
; SSE2-NEXT: punpcklbw %xmm1, %xmm0
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,2,1,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,6,5,4]
; SSE2-NEXT: packuswb %xmm2, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_03_02_01_00_07_06_05_04_11_10_09_08_15_14_13_12
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[3,2,1,0,7,6,5,4,11,10,9,8,15,14,13,12]
; SSSE3-NEXT: retq
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 11, i32 10, i32 9, i32 8, i32 15, i32 14, i32 13, i32 12>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_03_02_01_00_07_06_05_04_19_18_17_16_23_22_21_20(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_03_02_01_00_07_06_05_04_19_18_17_16_23_22_21_20
; SSE2: # BB#0:
; SSE2-NEXT: pxor %xmm2, %xmm2
; SSE2-NEXT: punpcklbw %xmm2, %xmm1
; SSE2-NEXT: pshuflw {{.*}} # xmm1 = xmm1[3,2,1,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm1 = xmm1[0,1,2,3,7,6,5,4]
; SSE2-NEXT: punpcklbw %xmm2, %xmm0
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,2,1,0,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,7,6,5,4]
; SSE2-NEXT: packuswb %xmm1, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_03_02_01_00_07_06_05_04_19_18_17_16_23_22_21_20
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,xmm1[3,2,1,0,7,6,5,4]
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[3,2,1,0,7,6,5,4],zero,zero,zero,zero,zero,zero,zero,zero
; SSSE3-NEXT: por %xmm1, %xmm0
; SSSE3-NEXT: retq
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4, i32 19, i32 18, i32 17, i32 16, i32 23, i32 22, i32 21, i32 20>
ret <16 x i8> %shuffle
}
define <16 x i8> @shuffle_v16i8_03_02_01_00_31_30_29_28_11_10_09_08_23_22_21_20(<16 x i8> %a, <16 x i8> %b) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; SSE2-LABEL: @shuffle_v16i8_03_02_01_00_31_30_29_28_11_10_09_08_23_22_21_20
; SSE2: # BB#0:
; SSE2-NEXT: pxor %xmm2, %xmm2
; SSE2-NEXT: movdqa %xmm1, %xmm3
; SSE2-NEXT: punpcklbw %xmm2, %xmm3
; SSE2-NEXT: pshufhw {{.*}} # xmm3 = xmm3[0,1,2,3,7,6,5,4]
; SSE2-NEXT: movdqa %xmm0, %xmm4
; SSE2-NEXT: punpckhbw %xmm2, %xmm4
; SSE2-NEXT: pshuflw {{.*}} # xmm4 = xmm4[3,2,1,0,4,5,6,7]
; SSE2-NEXT: shufpd {{.*}} # xmm4 = xmm4[0],xmm3[1]
; SSE2-NEXT: punpckhbw %xmm2, %xmm1
; SSE2-NEXT: pshufhw {{.*}} # xmm1 = xmm1[0,1,2,3,7,6,5,4]
; SSE2-NEXT: punpcklbw %xmm2, %xmm0
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[3,2,1,0,4,5,6,7]
; SSE2-NEXT: shufpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
; SSE2-NEXT: packuswb %xmm4, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @shuffle_v16i8_03_02_01_00_31_30_29_28_11_10_09_08_23_22_21_20
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm1 = zero,zero,zero,zero,xmm1[15,14,13,12],zero,zero,zero,zero,xmm1[7,6,5,4]
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[3,2,1,0],zero,zero,zero,zero,xmm0[11,10,9,8],zero,zero,zero,zero
; SSSE3-NEXT: por %xmm1, %xmm0
; SSSE3-NEXT: retq
[x86] Begin a significant overhaul of how vector lowering is done in the x86 backend. This sketches out a new code path for vector lowering, hidden behind an off-by-default flag while it is under development. The fundamental idea behind the new code path is to aggressively break down the problem space in ways that ease selecting the odd set of instructions available on x86, and carefully avoid scalarizing code even when forced to use older ISAs. Notably, this starts off restricting itself to SSE2 and implements the complete vector shuffle and blend space for 128-bit vectors in SSE2 without scalarizing. The plan is to layer on top of this ISA extensions where we can bail out of the complex SSE2 lowering and opt for a cheaper, specialized instruction (or set of instructions). It also needs to be generalized to AVX and AVX512 vector widths. Currently, this does a decent but not perfect job for SSE2. There are some specific shortcomings that I plan to address: - We need a peephole combine to fold together shuffles where possible. There are cases where a previous shuffle could be modified slightly to arrange for elements to be in the correct position and a later shuffle eliminated. Doing this eagerly added quite a bit of complexity, and so my plan is to combine away these redundancies afterward. - There are a lot more clever ways to use unpck and pack that need to be added. This is essential for real world shuffles as it turns out... Once SSE2 is polished a bit I should be able to get interesting numbers on performance improvements on benchmarks conducive to vectorization. All of this will be off by default until it is functionally equivalent of course. Differential Revision: http://reviews.llvm.org/D4225 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@211888 91177308-0d34-0410-b5e6-96231b3b80d8
2014-06-27 11:23:44 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> %b, <16 x i32> <i32 3, i32 2, i32 1, i32 0, i32 31, i32 30, i32 29, i32 28, i32 11, i32 10, i32 9, i32 8, i32 23, i32 22, i32 21, i32 20>
ret <16 x i8> %shuffle
}
[x86] Teach the new vector shuffle lowering code to handle what is essentially a DAG combine that never gets a chance to run. We might typically expect DAG combining to remove shuffles-of-splats and other similar patterns, but we don't get a chance to run the DAG combiner when we recursively form sub-shuffles during the lowering of a shuffle. So instead hand-roll a really important combine directly into the lowering code to detect shuffles-of-splats, especially shuffles of an all-zero splat which needn't even have the same element width, etc. This lets the new vector shuffle lowering handle shuffles which implement things like zero-extension really nicely. This will become even more important when I wire the legalization of zero-extension to vector shuffles with the new widening legalization strategy. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212444 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-07 09:06:58 +00:00
define <16 x i8> @zext_to_v8i16_shuffle(<16 x i8> %a) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; ALL-LABEL: @zext_to_v8i16_shuffle
; ALL: pxor %xmm1, %xmm1
; ALL-NEXT: punpcklbw %xmm1, %xmm0
[x86] Teach the new vector shuffle lowering code to handle what is essentially a DAG combine that never gets a chance to run. We might typically expect DAG combining to remove shuffles-of-splats and other similar patterns, but we don't get a chance to run the DAG combiner when we recursively form sub-shuffles during the lowering of a shuffle. So instead hand-roll a really important combine directly into the lowering code to detect shuffles-of-splats, especially shuffles of an all-zero splat which needn't even have the same element width, etc. This lets the new vector shuffle lowering handle shuffles which implement things like zero-extension really nicely. This will become even more important when I wire the legalization of zero-extension to vector shuffles with the new widening legalization strategy. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212444 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-07 09:06:58 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> zeroinitializer, <16 x i32> <i32 0, i32 17, i32 1, i32 19, i32 2, i32 21, i32 3, i32 23, i32 4, i32 25, i32 5, i32 27, i32 6, i32 29, i32 7, i32 31>
ret <16 x i8> %shuffle
}
define <16 x i8> @zext_to_v4i32_shuffle(<16 x i8> %a) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; ALL-LABEL: @zext_to_v4i32_shuffle
; ALL: pxor %xmm1, %xmm1
; ALL-NEXT: punpcklbw %xmm1, %xmm0
; ALL-NEXT: punpcklbw %xmm1, %xmm0
[x86] Teach the new vector shuffle lowering code to handle what is essentially a DAG combine that never gets a chance to run. We might typically expect DAG combining to remove shuffles-of-splats and other similar patterns, but we don't get a chance to run the DAG combiner when we recursively form sub-shuffles during the lowering of a shuffle. So instead hand-roll a really important combine directly into the lowering code to detect shuffles-of-splats, especially shuffles of an all-zero splat which needn't even have the same element width, etc. This lets the new vector shuffle lowering handle shuffles which implement things like zero-extension really nicely. This will become even more important when I wire the legalization of zero-extension to vector shuffles with the new widening legalization strategy. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212444 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-07 09:06:58 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> zeroinitializer, <16 x i32> <i32 0, i32 17, i32 18, i32 19, i32 1, i32 21, i32 22, i32 23, i32 2, i32 25, i32 26, i32 27, i32 3, i32 29, i32 30, i32 31>
ret <16 x i8> %shuffle
}
[x86] Initial improvements to the new shuffle lowering for v16i8 shuffles specifically for cases where a small subset of the elements in the input vector are actually used. This is specifically targetted at improving the shuffles generated for trunc operations, but also helps out splat-like operations. There is still some really low-hanging fruit here that I want to address but this is a huge step in the right direction. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212680 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-10 04:34:06 +00:00
define <16 x i8> @trunc_v4i32_shuffle(<16 x i8> %a) {
[x86] Largely complete the use of PSHUFB in the new vector shuffle lowering with a small addition to it and adding PSHUFB combining. There is one obvious place in the new vector shuffle lowering where we should form PSHUFBs directly: when without them we will unpack a vector of i8s across two different registers and do a potentially 4-way blend as i16s only to re-pack them into i8s afterward. This is the crazy expensive fallback path for i8 shuffles and we can just directly use pshufb here as it will always be cheaper (the unpack and pack are two instructions so even a single shuffle between them hits our three instruction limit for forming PSHUFB). However, this doesn't generate very good code in many cases, and it leaves a bunch of common patterns not using PSHUFB. So this patch also adds support for extracting a shuffle mask from PSHUFB in the X86 lowering code, and uses it to handle PSHUFBs in the recursive shuffle combining. This allows us to combine through them, combine multiple ones together, and generally produce sufficiently high quality code. Extracting the PSHUFB mask is annoyingly complex because it could be either pre-legalization or post-legalization. At least this doesn't have to deal with re-materialized constants. =] I've added decode routines to handle the different patterns that show up at this level and we dispatch through them as appropriate. The two primary test cases are updated. For the v16 test case there is still a lot of room for improvement. Since I was going through it systematically I left behind a bunch of FIXME lines that I'm hoping to turn into ALL lines by the end of this. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@214628 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-02 10:39:15 +00:00
; FIXME-LABEL: @trunc_v4i32_shuffle
; FIXME: # BB#0:
; FIXME-NEXT: pand
; FIXME-NEXT: packuswb %xmm0, %xmm0
; FIXME-NEXT: packuswb %xmm0, %xmm0
; FIXME-NEXT: retq
;
; SSE2-LABEL: @trunc_v4i32_shuffle
; SSE2: # BB#0:
; SSE2-NEXT: pand
; SSE2-NEXT: pshuflw {{.*}} # xmm0 = xmm0[0,2,2,3,4,5,6,7]
; SSE2-NEXT: pshufhw {{.*}} # xmm0 = xmm0[0,1,2,3,4,6,6,7]
; SSE2-NEXT: pshufd {{.*}} # xmm0 = xmm0[0,2,2,3]
; SSE2-NEXT: packuswb %xmm0, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: @trunc_v4i32_shuffle
; SSSE3: # BB#0:
; SSSE3-NEXT: pshufb {{.*}} # xmm0 = xmm0[0,4,8,12],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
; SSSE3-NEXT: retq
[x86] Initial improvements to the new shuffle lowering for v16i8 shuffles specifically for cases where a small subset of the elements in the input vector are actually used. This is specifically targetted at improving the shuffles generated for trunc operations, but also helps out splat-like operations. There is still some really low-hanging fruit here that I want to address but this is a huge step in the right direction. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@212680 91177308-0d34-0410-b5e6-96231b3b80d8
2014-07-10 04:34:06 +00:00
%shuffle = shufflevector <16 x i8> %a, <16 x i8> undef, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
ret <16 x i8> %shuffle
}
Reference in New Issue Copy Permalink