linux-kernel - Re: [GIT PULL] x86/build changes for v4.17

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

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 5 Apr 2018 10:24:09 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Ingo Molnar <mingo@...nel.org>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Matthias Kaehlcke <mka@...omium.org>,
        Arnd Bergmann <arnd@...db.de>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Andrew Morton <akpm@...ux-foundation.org>,
        James Y Knight <jyknight@...gle.com>,
        Chandler Carruth <chandlerc@...gle.com>,
        Stephen Hines <srhines@...gle.com>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Kees Cook <keescook@...gle.com>,
        Guenter Roeck <groeck@...omium.org>,
        Greg Hackmann <ghackmann@...gle.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: Re: [GIT PULL] x86/build changes for v4.17

On Thu, Apr 05, 2018 at 10:04:46AM +0200, Ingo Molnar wrote:
>    http://www.agner.org/optimize/instruction_tables.pdf
> 
> The SkyLake costs for 'BT', 'AND' and 'TEST' variants are:
> 
>                   BT             m,i               2               2         p06 p23                   0.5

>                 TEST           m,r/i               1               2       p0156 p23          1        0.5

These two I would imagine (I tend to forget about the TEST instruction).
And while they're of equal speed, TEST has more ports available if I
read that right. But yes, on SKL it doesn't matter much.

But if you go back in history (a lot) then you'll find BT being far more
expensive than TEST.

On the original Pentium for example TEST-m,r/i is 2 cycles, but BT-m,i
is 4-9 cycles.

But yes, going by the tables that's all hysterical raisins, modern cores
don't much care.