[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 14 Sep 2015 22:39:40 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: SHA-1 H()
On 2015-09-14 19:40, Solar Designer wrote:
> On Mon, Sep 14, 2015 at 07:03:57PM +0200, magnum wrote:
>> That failure was unrelated. The PR is merged now, good stuff!
>
> Yeah, looks good to me.
>
> A minor detail, in this piece:
>
> #if __AVX512F__
> #undef vcmov
> #define vcmov(x, y, z) vternarylogic(x, y, z, 0xE4)
> #define vternarylogic _mm512_ternarylogic_epi32
> #endif
>
> I think we need to add:
>
> #undef VCMOV_EMULATED
>
> right after the "#undef vcmov". Since we have a one instruction
> implementation, it is as good as native, non-emulated vcmov.
Damn good catch. Fixed now.
BTW do you think we could use inline PTX to define a LOP3.LUT
instruction on nvidia, like you did with the funnel shifts? Or would it
possibly be worse than having the optimizer miss one or two, due to the
caveats of inline asm?
magnum
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
