On Wed, 13 Sep 2023 20:25:22 GMT, Smita Kamath <svkam...@openjdk.org> wrote:
>> Hi All,
>> I would like to submit AES-GCM optimization for x86_64 architectures using
>> AVX2 instructions. This optimization interleaves AES and GHASH operations.
>>
>> Below are the performance numbers on my desktop system with -XX:UseAVX=2
>> option:
>>
>> |Benchmark | Data Size | Base version (ops/s) | Patched version (ops/s) |
>> Speedup
>> |-------------|------------|---------------|------------------|-----------|
>> |full.AESGCMBench.decrypt | 8192 | 526274.678 | 670014.543 | 1.27
>> full.AESGCMBench.encrypt | 8192 | 538293.315 | 680716.207 | 1.26
>> small.AESGCMBench.decrypt | 8192 | 527854.353 |663131.48 | 1.25
>> small.AESGCMBench.encrypt | 8192 | 548193.804 | 683624.232 |1.24
>> full.AESGCMBench.decryptMultiPart | 8192 | 299865.766 | 299815.851 | 0.99
>> full.AESGCMBench.encryptMultiPart | 8192 | 534406.564 |539235.462 | 1.00
>> small.AESGCMBench.decryptMultiPart | 8192 | 299960.202 |298913.629 | 0.99
>> small.AESGCMBench.encryptMultiPart | 8192 | 542669.258 | 540552.293 | 0.99
>> | | | |
>> full.AESGCMBench.decrypt | 16384 | 307266.364 |390397.778 | 1.27
>> full.AESGCMBench.encrypt | 16384 | 311491.901 | 397279.681 | 1.27
>> small.AESGCMBench.decrypt | 16384 | 306257.801 | 389531.665 |1.27
>> small.AESGCMBench.encrypt | 16384 | 311468.972 | 397804.753 | 1.27
>> full.AESGCMBench.decryptMultiPart | 16384 | 159634.341 | 181271.487 | 1.13
>> full.AESGCMBench.encryptMultiPart | 16384 | 308980.992 | 385606.113 | 1.24
>> small.AESGCMBench.decryptMultiPart | 16384 | 160476.064 |181019.205 | 1.12
>> small.AESGCMBench.encryptMultiPart | 16384 | 308382.656 | 391126.417 | 1.26
>> | | | |
>> full.AESGCMBench.decrypt | 32768 | 162284.703 | 213257.481 |1.31
>> full.AESGCMBench.encrypt | 32768 | 164833.104 | 215568.639 | 1.30
>> small.AESGCMBench.decrypt | 32768 | 164416.491 | 213422.347 | 1.29
>> small.AESGCMBench.encrypt | 32768 | 166619.205 | 214584.208 |1.28
>> full.AESGCMBench.decryptMultiPart | 32768 | 83306.239 | 93762.988 |1.12
>> full.AESGCMBench.encryptMultiPart | 32768 | 166109.391 |211701.969 | 1.27
>> small.AESGCMBench.decryptMultiPart | 32768 | 83792.559 | 94530.786 | 1.12
>> small.AESGCMBench.encryptMultiPart | 32768 | 162975.904 |212085.047 | 1.30
>> | | | |
>> full.AESGCMBench.decrypt | 65536 | 85765.835 | 112244.611 | 1.30
>> full.AESGCMBench.encrypt | 65536 | 86471.805 | 113320.536 |1.31
>> small.AESGCMBench.decrypt | 65536 | 84490.816 | 112122.358 |1.32
>> small.AESGCMBench.encrypt | 65536 | 85403.025 | 112741.811 | 1.32
>> full.AES...
>
> Smita Kamath has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Removed isEncrypt boolean variable
src/hotspot/cpu/x86/stubGenerator_x86_64.hpp line 362:
> 360: // AVX2 AES-GCM related functions
> 361: void initial_blocks(XMMRegister ctr, Register rounds, Register key,
> Register len,
> 362: Register in, Register out, Register ct, Register
> subkeyHtbl, Register pos);
You could rename it to gcm_intial_blocks_avx2().
src/hotspot/cpu/x86/stubGenerator_x86_64.hpp line 365:
> 363: void gfmul_avx2(XMMRegister GH, XMMRegister HK);
> 364: void generateHtbl_8_block_avx2(Register htbl, Register rscratch);
> 365: void ghash8_encrypt8_parallel(Register key, Register subkeyHtbl,
> XMMRegister ctr_blockx, XMMRegister aad_hashx,
Rename to ghash8_encrypt8_parallel_avx2().
src/hotspot/cpu/x86/stubGenerator_x86_64.hpp line 367:
> 365: void ghash8_encrypt8_parallel(Register key, Register subkeyHtbl,
> XMMRegister ctr_blockx, XMMRegister aad_hashx,
> 366: Register in, Register out, Register ct,
> Register pos, bool out_order, Register rounds);
> 367: void ghash_last_8(Register subkeyHtbl);
Rename to ghash_last_8_avx2.
src/hotspot/cpu/x86/stubGenerator_x86_64_aes.cpp line 185:
> 183: if (VM_Version::supports_avx2()) {
> 184: StubRoutines::_galoisCounterMode_AESCrypt =
> generate_avx2_galoisCounterMode_AESCrypt();
> 185: }
This could be moved to line 192.
src/hotspot/cpu/x86/stubRoutines_x86.hpp line 40:
> 38: // AVX512 intrinsics add more code in 64-bit VM,
> 39: // Windows have more code to save/restore registers
> 40: _compiler_stubs_code_size = 30000 LP64_ONLY(+30000)
> WINDOWS_ONLY(+2000),
Since the stub is for 64 bit, the LP64_ONLY part needs to increase and not the
base.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/15410#discussion_r1325143256
PR Review Comment: https://git.openjdk.org/jdk/pull/15410#discussion_r1325143499
PR Review Comment: https://git.openjdk.org/jdk/pull/15410#discussion_r1325143801
PR Review Comment: https://git.openjdk.org/jdk/pull/15410#discussion_r1325145931
PR Review Comment: https://git.openjdk.org/jdk/pull/15410#discussion_r1325144430
