Re: Intel Phi Coprocess

Bill Allombert on Tue, 28 Mar 2017 22:54:15 +0200

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: Intel Phi Coprocess

To: pari-dev@pari.math.u-bordeaux.fr
Subject: Re: Intel Phi Coprocess
From: Bill Allombert <Bill.Allombert@math.u-bordeaux.fr>
Date: Tue, 28 Mar 2017 22:54:13 +0200
Delivery-date: Tue, 28 Mar 2017 22:54:15 +0200
In-reply-to: <CAJ-x-Uigypr8_1bfAd5nOSOhSTmgRoMEKLX+BPRZg+jOagDSMQ@mail.gmail.com>
Mail-followup-to: pari-dev@pari.math.u-bordeaux.fr
References: <CAJ-x-Uigypr8_1bfAd5nOSOhSTmgRoMEKLX+BPRZg+jOagDSMQ@mail.gmail.com>
User-agent: Mutt/1.5.23 (2014-03-12)

On Tue, Mar 28, 2017 at 03:48:05PM -0400, Igor Schein wrote:
> Hi,
> 
> How feasible would it be to get pari/gp to take advantage of Intel Phi
> Coprocessor?

It is certainly possible. The coprocessors are standard ix86 processors
Even using MPI could work for some tasks that use little memory.

We have some Knight Landing Xeon Phi at Bordeaux, but this technology
moves too fast with respect to PARI/GP development, so we never had
time to try them.

Cheers,
Bill.

References:
- Intel Phi Coprocess
  - From: Igor Schein <igorschein@gmail.com>

Prev by Date: Intel Phi Coprocess
Previous by thread: Intel Phi Coprocess
Index(es):
- Date
- Thread