runtime: add arm64 runtime.cmpstring and bytes.Compare

Add arm64 assembly implementation of runtime.cmpstring and bytes.Compare.

benchmark                                old ns/op     new ns/op     delta
BenchmarkCompareBytesEqual               98.0          27.5          -71.94%
BenchmarkCompareBytesToNil               9.38          10.0          +6.61%
BenchmarkCompareBytesEmpty               13.3          10.0          -24.81%
BenchmarkCompareBytesIdentical           98.0          27.5          -71.94%
BenchmarkCompareBytesSameLength          43.3          16.3          -62.36%
BenchmarkCompareBytesDifferentLength     43.4          16.3          -62.44%
BenchmarkCompareBytesBigUnaligned        6979680       1360979       -80.50%
BenchmarkCompareBytesBig                 6915995       1381979       -80.02%
BenchmarkCompareBytesBigIdentical        6781440       1327304       -80.43%

benchmark                             old MB/s     new MB/s     speedup
BenchmarkCompareBytesBigUnaligned     150.23       770.46       5.13x
BenchmarkCompareBytesBig              151.62       758.76       5.00x
BenchmarkCompareBytesBigIdentical     154.63       790.01       5.11x

* note, the machine we are benchmarking on has some issues. What is clear is
compared to a few days ago the old MB/s value has increased from ~115 to 150.
I'm less certain about the new MB/s number, which used to be close to 1Gb/s.

Change-Id: I4f31b2c7a06296e13912aacc958525632cb0450d
Reviewed-on: https://go-review.googlesource.com/8541
Reviewed-by: Aram Hăvărneanu <aram@mgk.ro>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

diff --git a/src/runtime/asm_arm64.s b/src/runtime/asm_arm64.s
index 68d0447f037ff1a073daa15b5556ef8520729519..a61d99c65be31c4252c3d8ae20199b08172e555f 100644 (file)
--- a/src/runtime/asm_arm64.s
+++ b/src/runtime/asm_arm64.s
@@ -801,6 +801,56 @@ eq:
        MOVB    R3, ret+16(FP)
        RET
 
+TEXT runtime·cmpstring(SB),NOSPLIT,$0-40
+       MOVD    s1_base+0(FP), R2
+       MOVD    s1_len+8(FP), R0
+       MOVD    s2_base+16(FP), R3
+       MOVD    s2_len+24(FP), R1
+       BL      runtime·cmpbody(SB)
+       MOVD    R8, ret+32(FP)
+       RET
+
+TEXT bytes·Compare(SB),NOSPLIT,$0-56
+       MOVD    s1+0(FP), R2
+       MOVD    s1+8(FP), R0
+       MOVD    s2+24(FP), R3
+       MOVD    s2+32(FP), R1
+       BL      runtime·cmpbody(SB)
+       MOVD    R8, ret+48(FP)
+       RET
+
+// On entry:
+// R0 is the length of s1
+// R1 is the length of s2
+// R2 points to the start of s1
+// R3 points to the start of s2
+//
+// On exit:
+// R8 is -1/0/+1
+// R5, R4, and R6 are clobbered
+TEXT runtime·cmpbody<>(SB),NOSPLIT,$-4-0
+       CMP     R0, R1
+       CSEL    LT, R1, R0, R6 // R6 is min(R0, R1)
+
+       ADD     R2, R6  // R2 is current byte in s1, R6 is last byte in s1 to compare
+loop:
+       CMP     R2, R6
+       BEQ     samebytes // all compared bytes were the same; compare lengths
+       MOVBU.P 1(R2), R4
+       MOVBU.P 1(R3), R5
+       CMP     R4, R5
+       BEQ     loop
+       // bytes differed
+       MOVD    $1, R8
+       CSNEG   LT, R8, R8, R8
+       RET
+samebytes:
+       MOVD    $1, R8
+       CMP     R0, R1
+       CSNEG   LT, R8, R8, R8
+       CSEL    EQ, ZR, R8, R8
+       RET
+
 // eqstring tests whether two strings are equal.
 // The compiler guarantees that strings passed
 // to eqstring have equal length.

diff --git a/src/runtime/noasm.go b/src/runtime/noasm.go
index ab9c7447891fc13668ea658e136af625a655b59e..9a6dbeec3df9b1135faca58dff40ab5def7db3fe 100644 (file)
--- a/src/runtime/noasm.go
+++ b/src/runtime/noasm.go
@@ -2,9 +2,9 @@
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.
 
-// Routines that are implemented in assembly in asm_{amd64,386,arm}.s
+// Routines that are implemented in assembly in asm_{amd64,386,arm,arm64}.s
 
-// +build arm64 ppc64 ppc64le
+// +build ppc64 ppc64le
 
 package runtime