]>
Commit | Line | Data |
---|---|---|
2f98abbc AP |
1 | /* |
2 | * x86_64 BIGNUM accelerator version 0.1, December 2002. | |
3 | * | |
4 | * Implemented by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL | |
5 | * project. | |
6 | * | |
7 | * Rights for redistribution and usage in source and binary forms are | |
8 | * granted according to the OpenSSL license. Warranty of any kind is | |
9 | * disclaimed. | |
10 | * | |
11 | * Q. Version 0.1? It doesn't sound like Andy, he used to assign real | |
12 | * versions, like 1.0... | |
13 | * A. Well, that's because this code is basically a quick-n-dirty | |
14 | * proof-of-concept hack. As you can see it's implemented with | |
15 | * inline assembler, which means that you're bound to GCC and that | |
16 | * there must be a room for fine-tuning. | |
17 | * | |
18 | * Q. Why inline assembler? | |
19 | * A. x86_64 features own ABI I'm not familiar with. Which is why | |
20 | * I decided to let the compiler take care of subroutine | |
21 | * prologue/epilogue as well as register allocation. | |
22 | * | |
23 | * Q. How much faster does it get? | |
24 | * A. Unfortunately people sitting on x86_64 hardware are prohibited | |
25 | * to disclose the performance numbers, so they (SuSE labs to be | |
26 | * specific) wouldn't tell me. However! Very similar coding technique | |
27 | * (reaching out for 128-bit result from 64x64-bit multiplication) | |
28 | * results in >3 times performance improvement on MIPS and I see no | |
29 | * reason why gain on x86_64 would be so much different:-) | |
30 | */ | |
31 | ||
32 | #define BN_ULONG unsigned long | |
33 | ||
34 | /* | |
35 |