This implementation utilizes the same registers found in the reference
implementation, aiming to produce a minimal semantic diff between the
Avo-generated output and the original hand-written assembly.
To verify the Avo implementation, the reference and Avo-generated
assembly files are fed to `go tool asm`, capturing the debug output into
corresponding temp files. The debug output contains supplementary
metadata (line numbers, instruction offsets, and source file references)
that must be removed in order to obtain a semantic diff of the two
files. This is accomplished via a small utility script written in awk.
Commands used to verify Avo output:
BASE="d66d9c31b4ae80d173d1187a9e40c188788dbdbc"
go tool asm -o /dev/null -debug \
<(git cat-file -p "$BASE":sha3/keccakf_amd64.s) \
> /tmp/reference.s
go tool asm -o /dev/null -debug \
sha3/keccakf_amd64.s \
> /tmp/avo.s
normalize(){
awk '{
$1=$2=$3="";
print substr($0,4)
}'
}
diff <(normalize < /tmp/reference.s) <(normalize < /tmp/avo.s)
Change-Id: I1c0ea516531355263b83d3b66a37df090e293cea
Reviewed-on: https://go-review.googlesource.com/c/crypto/+/594655
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Russell Webb <russell.webb@protonmail.com>
Reviewed-by: Roland Shoemaker <roland@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
keccakF1600 manually subtracts 200 from SP, but the generated prologue
already does this and the extra subtraction just means keccakF1600 is
defeating the stack growth check.
Remove the unnecessary SP adjustment.
Change-Id: I9450f6b12489bcd20e9ace30f9dd3066025d3a1a
Reviewed-on: https://go-review.googlesource.com/31653
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
NIST published a Keccak Code Package following the SHA-3 workshop
organized in 2014, containing optimized versions of various Keccak
functions for various architectures. This CL converts the GNU asm
code of the Keccak permutation for the x86_64 architecture into Go
assembly.
The code here is almost an identical copy of KeccakF1600_StatePermute,
with the only modification of converting the input state into the
implementation's internal representation and vice versa before return.
This keeps the algorithm an in-place version and avoids requiring
extra external state inits and data XORs before and after the permute.
The speed difference is:
benchmark old ns/op new ns/op delta
BenchmarkPermutationFunction-8 476 411 -13.66%
BenchmarkSha3_512_MTU-8 9910 8681 -12.40%
BenchmarkSha3_384_MTU-8 7124 6249 -12.28%
BenchmarkSha3_256_MTU-8 5666 4986 -12.00%
BenchmarkSha3_224_MTU-8 5401 4750 -12.05%
BenchmarkShake128_MTU-8 4614 3980 -13.74%
BenchmarkShake256_MTU-8 4935 4295 -12.97%
BenchmarkShake256_16x-8 71850 63798 -11.21%
BenchmarkShake256_1MiB-8 3784244 3285733 -13.17%
BenchmarkSha3_512_1MiB-8 7098875 6163359 -13.18%
benchmark old MB/s new MB/s speedup
BenchmarkPermutationFunction-8 420.11 486.35 1.16x
BenchmarkSha3_512_MTU-8 136.22 155.51 1.14x
BenchmarkSha3_384_MTU-8 189.49 216.03 1.14x
BenchmarkSha3_256_MTU-8 238.23 270.71 1.14x
BenchmarkSha3_224_MTU-8 249.91 284.19 1.14x
BenchmarkShake128_MTU-8 292.58 339.15 1.16x
BenchmarkShake256_MTU-8 273.53 314.28 1.15x
BenchmarkShake256_16x-8 228.03 256.81 1.13x
BenchmarkShake256_1MiB-8 277.09 319.13 1.15x
BenchmarkSha3_512_1MiB-8 147.71 170.13 1.15x
For further details, please see:
- http://csrc.nist.gov/groups/ST/hash/sha-3/Aug2014/documents/vanassche_keccak_code.pdf
- https://github.com/gvanas/KeccakCodePackage
Change-Id: I5b0b9395bba7d8c9acfe2b9c79f6e9c2cf858c7c
Reviewed-on: https://go-review.googlesource.com/17962
Reviewed-by: Adam Langley <agl@golang.org>
Garrett Bodley
Egon Elbre
Dmitri Shuralyov
Tobias Klauser
Tamir Duberstein
Austin Clements
Péter Szilágyi