[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20130627122027.GB20753@openwall.com>
Date: Thu, 27 Jun 2013 16:20:27 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: Parallella: scrypt
Rafael,
On Thu, Jun 27, 2013 at 01:33:43AM +0100, Rafael Waldo Delgado Doblas wrote:
> It's true, but maybe in the Parallella's case is better use Pooler's
> Cpuminer* because it has ARM support and ckolivas/cgminer doesn't look like
> is going to improve the support any longer**. Also using Pooler's Cpuminer
> is suppose to mine ~2khash/s in ARM Cortex-A9 DualCore 1.2GHz***.
>
> BTW Pooler's Scrypt ARM implementation is a perfectly uncommented ARM-ASM
> code :-). There another Pooler's ASM versions for x86 and x64.
>
> What do you think?
We don't care about performance on ARM much. The only way ARM comes
into play for Litecoin mining is because the host CPU on Parallella
board is dual-core ARM. What we care about is getting Litecoin mining
to work on Epiphany - on all cores and using local memory (with TMTO) -
as a proof-of-concept.
We also care about having Litecoin mining as an end-user feature, on
usual x86 CPUs and on GPUs. Since this is not the primary focus of JtR,
and since this may become obsoleted soon (in fact, Litecoin mining on
CPUs is already mostly obsoleted, but we may support it for a while
anyway), I think we should just reuse and integrate cgminer code for this.
Alexander
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.
