[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <293683fc119625fdd90944257f6789ab@smtp.hushmail.com>
Date: Tue, 2 Jul 2013 20:32:56 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: bug: GPU use in CPU-only formats
On 2 Jul, 2013, at 18:43 , Solar Designer <solar@...nwall.com> wrote:
> Claudio, magnum -
>
> On Tue, Jul 02, 2013 at 01:20:56PM -0300, Claudio Andr? wrote:
>>> Yeah, but 62 MB? I'd love to be able to revert to the previous nvidia driver and compare.
>>
>> Too much, i agree.
>
> That's not the point. To me, the point is that _any_ GPU use when a
> CPU-only format is invoked is a bug. There should be zero GPU memory
> allocation, and zero load on the GPUs.
>
>> The easiest solution is to close these pointers and reopen them only
>> inside format init().
>
> Can't we do all initial "opening" of anything GPU-related in the
> format's init(), if the format is GPU-enabled, and not any sooner?
>
> Perhaps there's some implementation difficulty that I am missing.
In unstable, it works like you say. In bleeding, iirc the support for selecting multiple devices got us into a series of catch 22's depending on situation. I can't remember all details but I recall it was trickier than expected. I'll try to re-read the code and some old mail and see how we came to this situation.
We could start off with just doing what Claudio says and see if it solves the problem.
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.
