lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
Open Source and information security mailing list archives
| ||
|
Date: Tue, 31 Dec 2013 03:31:07 +0000 From: Samuel Neves <sneves@....uc.pt> To: discussions@...sword-hashing.net Subject: Re: [PHC] Best RNG for filling memory? On 31-12-2013 02:46, Solar Designer wrote: > Will ChaCha8 get below 0.85 cpb per core or 1.70 cpb per thread with 2 > threads/core on SB? This is unclear - would need to test. Alexander The speeds reported in [1] are much higher than the ones in eBACS; I don't know whether this is due to the measuring mechanism or the implementation. They seem to be more in line with your OpenSSL results, though: a good AES-CTR implementation should dispatch an AESENC at every cycle (by running 8 parallel instances in SB), which will bottleneck on Port5, shared competitively by both threads. That is, the effect of running a good AES-CTR implementation on two threads should be similar to that of running Chacha8. The best Chacha implementations already pretty much max out the available ILP. So AES-NI is likely, I suppose, to win on pre-Haswell processors. On the other hand, my experiments put Chacha8 on Haswell at 0.5-0.6cpb in a single thread. AES-NI also gets a little faster on Haswell, but by a small margin [1, slide 20]. [1] http://2013.diac.cr.yp.to/slides/gueron.pdf
Powered by blists - more mailing lists