[email protected] (Phil Smith III) writes:
> https://en.wikipedia.org/wiki/Xeon_Phi
>
> Up to 72 cores per chip, so up to 144 threads per socket. On an
> eight-socket motherboard, that's, um, a lot.

they announced they are discontinue Phi
https://www.extremetech.com/extreme/290963-intel-quietly-kills-off-xeon-phi

... but latest production server XEON announced last month have up to 56
cores per socket and up to eight sockets. from recent post
http://www.garlic.com/~lynn/2019c.html#9

Most recent announce (last month) 56-core (processors) Platinum 9200
https://www.anandtech.com/show/14182/hands-on-with-the-56core-xeon-platinum-9200-cpu-intels-biggest-cpu-package-ever</a>
https://www.servethehome.com/intel-xeon-platinum-9200-formerly-cascade-lake-ap-launched/</a>
https://www.storagereview.com/intel_releases_second_generation_intel_xeon_scalable_cpus</a>
https://www.hpcwire.com/2019/04/02/intel-launches-second-gen-scalable-xeons-with-up-to-56-cores/</a>

"We are delivering 8-core Xeons all the way up to 56-core, the highest
core count we've ever delivered on Xeon," said Shenoy. "We are
delivering support for 1- 2- 4- and 8-socket glueless support for Xeon."

... snip ...

aka eight socket, 56/socket, max 448 cores-processors sharing same
memory providing large number of TIPS (1000s BIPS) computation power in
single system.

IBM sold off its (intel) server business about the time the server chip
makers started saying that they are shipping over half their chips
directly to the big cloud megadatacenters ... for going on two decades,
the big cloud megadatacenters claim that they assemble their own servers
at 1/3rd the cost of brand name servers (aka cloud operators view
dataprocessing as a cost rather than profit).

big cloud megadatacenters have so radically reduced their server system
cost to a point that power & cooling have become major cost ... and they
are focusing on total costs, including electricity/cooling cost per
computation ... even getting special chip versions that improve
computation electricity/cooling costs. However, the highest performance
server chips can double the power reqirements for less than twice than
the computation throughput.

a big cloud megadatacenter will have over half million blade systems
with millions of processors, being operated by 80-120 people (enormous
automation) ... doubling the number of systems (for total computational
power), can easily be net financial win, for optimal computation per
power&cooling (and large cloud operations have several such
megadatacenters around the world).

-- 
virtualization experience starting Jan1968, online at home since Mar1970

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to