RFR: 8337632: AES-GCM Algorithm optimization for x86_64 [v2]

[Sandhya Viswanathan]

On Wed, 28 Aug 2024 22:42:33 GMT, Smita Kamath <svkamath at openjdk.org> wrote:

>> Hi, 
>> I want to submit an AES-GCM algorithm optimization. This implementation is using AVX512/VAES Instructions. Additionally, it reduces PARALLEL_LEN from 7680 to 512 bytes. The performance numbers are as below. Kindly review the code. Thank you.
>> 
>> Benchmark | Datasize | BaseJDK (ops/s) | Patch(ops/s) | %Gain
>> -- | -- | -- | -- | --
>> full.AESGCMBench.decrypt | 512 | 2928259.197 | 3269964.387 | 11.67
>> full.AESGCMBench.decrypt | 1024 | 2494254.611 | 3010987.731 | 20.72
>> full.AESGCMBench.decrypt | 1500 | 1883453.546 | 1934915.846 | 2.73
>> full.AESGCMBench.decrypt | 2048 | 1825780.711 | 2452861.368 | 34.34
>> full.AESGCMBench.decrypt | 4096 | 1275108.345 | 1806329.066 | 41.66
>> full.AESGCMBench.decrypt | 8192 | 1033936.634 | 1196836.052 | 15.75
>> full.AESGCMBench.decrypt | 16384 | 681494.768 | 711630.498 | 4.42
>> full.AESGCMBench.decrypt | 32768 | 385026.017 | 395043.193 | 2.6
>> full.AESGCMBench.decrypt | 65536 | 207373.924 | 214723.588 | 3.54
>>   |   |   |   |  
>> full.AESGCMBench.encrypt | 512 | 2658008.476 | 2882496.94 | 8.45
>> full.AESGCMBench.encrypt | 1024 | 2283709.63 | 2589534.403 | 13.39
>> full.AESGCMBench.encrypt | 1500 | 1794993.519 | 1817669.531 | 1.26
>> full.AESGCMBench.encrypt | 2048 | 1745532.435 | 2191097.29 | 25.52
>> full.AESGCMBench.encrypt | 4096 | 1203301.174 | 1649593.953 | 37.08
>> full.AESGCMBench.encrypt | 8192 | 985174.988 | 1132407.54 | 14.94
>> full.AESGCMBench.encrypt | 16384 | 658980.441 | 684765.771 | 3.91
>> full.AESGCMBench.encrypt | 32768 | 373543.798 | 391518.837 | 4.81
>> full.AESGCMBench.encrypt | 65536 | 202532.315 | 205084.833 | 1.260301597
>
> Smita Kamath has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Addressed review comments

src/hotspot/cpu/x86/assembler_x86.cpp line 8984:

> 8982: void Assembler::vinserti64x2(XMMRegister dst, XMMRegister nds, XMMRegister src, uint8_t imm8, int vector_len) {
> 8983:    assert(VM_Version::supports_avx512dq(), "");
> 8984:    assert(vector_len == AVX_256bit || VM_Version::supports_avx512vl(), "");

As this is an evex instruction, we could call it evinserti64x2. Also the assert should be: assert(vector_len == AVX_512bit || VM_Version::supports_avx512vl(), "");

src/hotspot/cpu/x86/assembler_x86.cpp line 11057:

> 11055: void Assembler::evbroadcastf64x2(XMMRegister dst, Address src, int vector_len) {
> 11056:   assert(VM_Version::supports_avx512dq(), "");
> 11057:   assert(vector_len == AVX_256bit || VM_Version::supports_avx512vl(), "");

The assert should be: assert(vector_len == AVX_512bit || VM_Version::supports_avx512vl(), "");

src/hotspot/cpu/x86/stubGenerator_x86_64_aes.cpp line 203:

> 201:     0x0000000000000003ULL, 0x0000000000000000ULL,
> 202:     0x0000000000000004ULL, 0x0000000000000000ULL,
> 203: };

This is same as COUNTER_MASK_LINC0[] in this file. Could we reuse that instead?

src/hotspot/cpu/x86/stubGenerator_x86_64_aes.cpp line 2813:

> 2811:   __ evshufi64x2(ZT5, ZT7, ZT7, 0x00, Assembler::AVX_512bit);//;; broadcast HashKey ^ 8 across all ZT5
> 2812: 
> 2813:   for (int i = 20, j = 52; i >= 0;) {

This should be i > 0.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/17515#discussion_r1735419450
PR Review Comment: https://git.openjdk.org/jdk/pull/17515#discussion_r1735419749
PR Review Comment: https://git.openjdk.org/jdk/pull/17515#discussion_r1735427066
PR Review Comment: https://git.openjdk.org/jdk/pull/17515#discussion_r1735420157