On Thu, 9 Feb 2023 18:08:15 GMT, Scott Gibbons <sgibb...@openjdk.org> wrote:
>> Added code for Base64 acceleration (encode and decode) which will accelerate
>> ~4x for AVX2 platforms.
>>
>> Encode performance:
>> **Old:**
>>
>> Benchmark (maxNumBytes) Mode Cnt Score Error
>> Units
>> Base64Encode.testBase64Encode 1024 thrpt 3 4309.439 ± 2.632
>> ops/ms
>>
>>
>> **New:**
>>
>> Benchmark (maxNumBytes) Mode Cnt Score
>> Error Units
>> Base64Encode.testBase64Encode 1024 thrpt 3 24211.397 ±
>> 102.026 ops/ms
>>
>>
>> Decode performance:
>> **Old:**
>>
>> Benchmark (errorIndex) (lineSize) (maxNumBytes)
>> Mode Cnt Score Error Units
>> Base64Decode.testBase64Decode 144 4 1024
>> thrpt 3 3961.768 ± 93.409 ops/ms
>>
>> **New:**
>> Benchmark (errorIndex) (lineSize) (maxNumBytes)
>> Mode Cnt Score Error Units
>> Base64Decode.testBase64Decode 144 4 1024
>> thrpt 3 14738.051 ± 24.383 ops/ms
>
> Scott Gibbons has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Add URL to microbenchmark
src/hotspot/cpu/x86/stubGenerator_x86_64.cpp line 2399:
> 2397: VM_Version::supports_avx512bw()) {
> 2398: __ cmpl(length, 31); // 32-bytes is break-even for AVX-512
> 2399: __ jcc(Assembler::lessEqual, L_bruteForce);
The avx2 code needs the length to be atleast 0x2c (44) bytes. We could directly
go to non-avx code instead of L_bruteForce here. We will save one
subtract/branch.
src/hotspot/cpu/x86/stubGenerator_x86_64.cpp line 2658:
> 2656: // Check for buffer too small (for algorithm)
> 2657: __ subl(length, 0x2c);
> 2658: __ jcc(Assembler::lessEqual, L_tailProc);
This could be Assembler::less instead of Assembler::lessEqual.
src/hotspot/cpu/x86/stubGenerator_x86_64.cpp line 2699:
> 2697: __ addptr(dest, 0x18);
> 2698: __ subl(length, 0x20);
> 2699: __ jcc(Assembler::lessEqual, L_tailProc);
This could be Assembler::less instead of Assembler::lessEqual.
-------------
PR: https://git.openjdk.org/jdk/pull/12126
