[Beowulf] China aims for 100 PF
cbergstrom at pathscale.com
Tue Jun 21 05:22:11 PDT 2016
Did you even other to read the article? It's IBM fab, but UC Davis
research.... unless I'm completely wrong.. Without real details like
the ISA (which I assume is some common RISC-like thing) who knows how
useful it could ever be outside of testing how many "cores" can be
stuffed on a single chip.. I'm curious how much the chip-to-chip
"networking" consumes compared to the cores themselves, as well as how
the hell you'd try to program that for any level of high utilization..
On Tue, Jun 21, 2016 at 8:14 PM, Remy Dernat <remy.dernat at univ-montp2.fr> wrote:
> 100 PF is really not far from reality right now:
> Is this why IBM just release its KiloCore processor ?
> Le 20/07/2015 06:43, Mark Hahn a écrit :
>>> well, this is always not bad for the world to have one more choice. I
>>> like to be more optimum in the future HPC market with cheaper product ...
>> except that many top500-leading systems are just national publicity
>> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
>> To change your subscription (digest mode or unsubscribe) visit
> Rémy Dernat
> Ingénieur d'Etudes
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit
More information about the Beowulf