[RC5] Amazon EC2 Configuration for GPU Instance

Pete Wilcock pete.wilcock at gmail.com
Thu Nov 13 07:53:42 EST 2014


Indeed, I eventually got this working using the CUDA 3.1 client and is
currently ticking along quite nicely at 324MKeys/sec which is around 8200
units per day.

The spot price instancing by Amazon is very appealing for these GPU boxes -
the cost of running this one is $0.08 per hour.

It makes me wonder how massively this could be scaled up with enough funds.
I wonder if the internet would be interest in crowdfunding an initiative to
finish off an old cryptography project ;)

I notice the keyrate history graphs are still broken, and the mailing list
archive stops at August. Does anyone know if these are likely to be fixed?

On 13 November 2014 08:37, Kees Nuyt <k.nuyt at zonnet.nl> wrote:

>
> Hi Mike,
>
> Ah, that's interesting, I wasn't aware of that.
>
> --
> Regards,
>
> Kees Nuyt
>
>
> On Thu, 13 Nov 2014 00:53:48 +0000, you wrote:
>
> >Hi Kees,
> >
> >Most of them are virtual machines on headless systems.. except the one
> >that Pete was using! The blurb on the AWS site says:-
> >
> >'G2
> >
> >This family includes G2 instances intended for graphics and general
> >purpose GPU compute applications.
> >
> >Features:
> >
> >High Frequency Intel Xeon E5-2670 (Sandy Bridge) Processors
> >High-performance NVIDIA GPU with 1,536 CUDA cores and 4GB of video memory
> >On-board hardware video encoder designed to support up to eight
> >real-time HD video streams (720p at 30fps) or up to four real-time FHD
> >video streams (1080p at 30 fps).
> >Support for low-latency frame capture and encoding for either the full
> >operating system or select render targets, enabling high-quality
> >interactive streaming experiences.'
> >
> >He said it gets about 420 Mk/sec. I am looking forward to when they
> >offer AMD based systems :)
> >
> >Kind regards,
> >Mike
> >
> >On 12 November 2014 21:21, Kees Nuyt <k.nuyt at zonnet.nl> wrote:
> >> On Wed, 12 Nov 2014 00:12:44 +0000, you wrote:
> >>
> >>>Hi folks,
> >>>
> >>>Had some trouble submitting to the list earlier so this is attempt #2!
> >>>
> >>>As a long-time user of the client faithfully running RC5-72, I thought
> it
> >>>would be 'fun' to see if I could make the client work on a GPU instance
> on
> >>>Amazon Web Services (g2.2xlarge).
> >>>
> >>>Sadly I have zero experience of the actual mechanics behind how the
> client
> >>>works, but have dutifully followed tutorials for installing CUDA 6.5,
> >>>OpenCL, and PyOpenCL on the Ubuntu instance.
> >>>
> >>>Despite my best efforts, when I run the client (v2.9111.520 PRE Linux) I
> >>>get the following:
> >>>
> >>>[Nov 11 18:05:30 UTC] No OpenCL platforms available!
> >>>[Nov 11 18:05:30 UTC] Unable to initialize OpenCL.
> >>>[Nov 11 18:05:30 UTC] *Break* Shutting down...
> >>>
> >>>I am possibly doing something rudimentarily stupid, or there may be some
> >>>fundamental incompatibility here that I'm not aware of, but if anyone
> can
> >>>point me in the right direction I'd be most grateful.
> >>
> >> I think (most) Amazon instances you can rent are virtual
> >> machines on headless systems.
> >>
> >> Servers typically don't need any human interface devices, like
> >> keyboard, mouse and screens. So, they don't contain any video
> >> hardware, thus no GPU.
> >>
> >> Even if the physical servers do have some video hardware, it's
> >> probably rudimentory (no fast GPU) and not exposed to the
> >> virtual machines.
> >>
> >> Hope this helps...
> >>
> >> --
> >> Regards,
> >>
> >> Kees Nuyt
> _______________________________________________
> rc5 mailing list
> rc5 at lists.distributed.net
> http://lists.distributed.net/mailman/listinfo/rc5
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.distributed.net/pipermail/rc5/attachments/20141113/4d92c1c9/attachment.html>


More information about the rc5 mailing list