[Beowulf] Intel kills Knights Hill, Xeon Phi line "being revised"

[Christopher Samuel]

Interesting times (via a colleague on the Australian HPC Slack).

https://www.top500.org/news/intel-dumps-knights-hill-future-of-xeon-phi-product-line-uncertain/

Looks like fallout from the delayed Aurora system.

Rumours flying that the Xeon Phi family is in jeopardy, but the
article has an addendum to say:

# [Update: Intel denies they are dropping the Xeon Phi line,
# saying only that it has "been revised based on recent
# customer and overall market needs."]

cheers,
Chris

[Gerald Henriksen]

Post by Christopher Samuel
Interesting times (via a colleague on the Australian HPC Slack).
https://www.top500.org/news/intel-dumps-knights-hill-future-of-xeon-phi-product-line-uncertain/
Looks like fallout from the delayed Aurora system.
Rumours flying that the Xeon Phi family is in jeopardy, but the
# [Update: Intel denies they are dropping the Xeon Phi line,
# saying only that it has "been revised based on recent
# customer and overall market needs."]

Maybe worth pointing out that Intel has big changes in store, which
may or may not be a factor in the Xeon Phi future:

https://www.theverge.com/2017/11/9/16627470/intel-raja-koduri-graphics-amd-radeon-business-hire

_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo

[Christopher Samuel]

Post by Gerald Henriksen
Maybe worth pointing out that Intel has big changes in store, which

Might be a case of history repeating itself as Xeon Phi came out of the
work Intel did on the Larrabee discreet GPU (which they killed in 2010).

[Stu Midgley]

We have a few thousand KNL systems. They are amazingly fast when
programmed correctly and can be treated as a normal x86_64 system from a
cluster management point of view.

We love them.

Post by Christopher Samuel

Post by Gerald Henriksen
Maybe worth pointing out that Intel has big changes in store, which

Might be a case of history repeating itself as Xeon Phi came out of the
work Intel did on the Larrabee discreet GPU (which they killed in 2010).
--
Christopher Samuel Senior Systems Administrator
Melbourne Bioinformatics - The University of Melbourne
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[C Bergström]

Post by Stu Midgley
We have a few thousand KNL systems. They are amazingly fast when
programmed correctly and can be treated as a normal x86_64 system from a
cluster management point of view.

I'm torn between

The darkside runs deep in this one
or
That's some mightly strong coolaid you've been drinking..

[Christopher Samuel]

Post by C Bergström
I'm torn between

Knowing Stu and what he does I'll take the former over the latter.. :-)

[Stu Midgley]

we have over 4500 KNC's and love those as well :)

The KNL's are pretty bad at normal integer workloads, but for massively
threaded and vectorised codes, they are very very fast.

Post by Christopher Samuel

Post by C Bergström
I'm torn between

Knowing Stu and what he does I'll take the former over the latter.. :-)
--
Christopher Samuel Senior Systems Administrator
Melbourne Bioinformatics - The University of Melbourne
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[Jan Wender]

Hi,

Unfortunately I did not find the english version, but Andreas Stiller from heise has some news from SC:
https://m.heise.de/newsticker/meldung/Xeon-Phi-ist-tot-es-lebe-der-Xeon-H-3891026.html
Essentially yes Xeon Phi is not continued, but a new design called Xeon-H is coming.

Cheerio, Jan
--

Post by Stu Midgley
we have over 4500 KNC's and love those as well :)
The KNL's are pretty bad at normal integer workloads, but for massively threaded and vectorised codes, they are very very fast.

Post by Christopher Samuel

Post by C Bergström
I'm torn between

Knowing Stu and what he does I'll take the former over the latter.. :-)
--
Christopher Samuel Senior Systems Administrator
Melbourne Bioinformatics - The University of Melbourne
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

--
Dr Stuart Midgley
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

[Mikhail Kuzminsky]

Post by Jan Wender
Unfortunately I did not find the english version, but Andreas
Essentially yes Xeon Phi is not continued, but a new design called Xeon-H is coming.

Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some
important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.

Mikhail Kuzminsky

Zelinsky Institute
of Organic Chemistry
Moscow
_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowul

[Jonathan Engwall]

Maybe they felt married to government sponsorship while the competition has
found a way to compete with itself.
http://www.nag.co.za/2017/10/26/amd-launches-ryzen-processor-with-radeon-vega-graphics-for-notebooks/
Maybe such a huge contract even looks too good to be true.

Post by Jan Wender
Unfortunately I did not find the english version, but Andreas
Essentially yes Xeon Phi is not continued, but a new design called Xeon-H

Post by Jan Wender
is coming.

Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some
important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.
Mikhail Kuzminsky
Zelinsky Institute
of Organic Chemistry
Moscow
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[Scott Atchley]

Hmm, can you name a large processor vendor who has not accepted US
government research funding in the last five years? See DOE's FastForward,
FastForward2, DesignForward, DesignForward2, and now PathForward.

On Fri, Nov 17, 2017 at 9:18 PM, Jonathan Engwall <

Post by Jonathan Engwall
Maybe they felt married to government sponsorship while the competition
has found a way to compete with itself.
http://www.nag.co.za/2017/10/26/amd-launches-ryzen-
processor-with-radeon-vega-graphics-for-notebooks/
Maybe such a huge contract even looks too good to be true.

Post by Jan Wender
Unfortunately I did not find the english version, but Andreas
Essentially yes Xeon Phi is not continued, but a new design called Xeon-H

Post by Jan Wender
is coming.

Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some
important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.
Mikhail Kuzminsky
Zelinsky Institute
of Organic Chemistry
Moscow
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[C Bergström]

Actually, the better question is, which vendor received funds and actually
made a useful solution that can go production with the deliverables. From
my view it seems like history is repeating itself[1] and I wish more people
would wake up. The top down approach to funding scientific research and the
in-fighting between labs is just too much nonsense. If these research
projects were a start-up, it would have failed hard.

[1] https://en.wikipedia.org/wiki/X87

Post by Scott Atchley
Hmm, can you name a large processor vendor who has not accepted US
government research funding in the last five years? See DOE's FastForward,
FastForward2, DesignForward, DesignForward2, and now PathForward.
On Fri, Nov 17, 2017 at 9:18 PM, Jonathan Engwall <

Post by Jonathan Engwall
Maybe they felt married to government sponsorship while the competition
has found a way to compete with itself.
http://www.nag.co.za/2017/10/26/amd-launches-ryzen-processor
-with-radeon-vega-graphics-for-notebooks/
Maybe such a huge contract even looks too good to be true.

Post by Jan Wender
Unfortunately I did not find the english version, but Andreas
Essentially yes Xeon Phi is not continued, but a new design called

Post by Jan Wender
Xeon-H is coming.

Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some
important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.
Mikhail Kuzminsky
Zelinsky Institute
of Organic Chemistry
Moscow
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[Scott Atchley]

Some of the research has already made it into products and more is slated
for future products. As with all research, some did not pan out, but that
is to be expected.

Post by C Bergström
Actually, the better question is, which vendor received funds and actually
made a useful solution that can go production with the deliverables. From
my view it seems like history is repeating itself[1] and I wish more people
would wake up. The top down approach to funding scientific research and the
in-fighting between labs is just too much nonsense. If these research
projects were a start-up, it would have failed hard.
[1] https://en.wikipedia.org/wiki/X87

Post by Scott Atchley
Hmm, can you name a large processor vendor who has not accepted US
government research funding in the last five years? See DOE's FastForward,
FastForward2, DesignForward, DesignForward2, and now PathForward.
On Fri, Nov 17, 2017 at 9:18 PM, Jonathan Engwall <

Post by Jonathan Engwall
Maybe they felt married to government sponsorship while the competition
has found a way to compete with itself.
http://www.nag.co.za/2017/10/26/amd-launches-ryzen-processor
-with-radeon-vega-graphics-for-notebooks/
Maybe such a huge contract even looks too good to be true.

Post by Jan Wender
Unfortunately I did not find the english version, but Andreas
Essentially yes Xeon Phi is not continued, but a new design called

Post by Jan Wender
Xeon-H is coming.

Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some
important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.
Mikhail Kuzminsky
Zelinsky Institute
of Organic Chemistry
Moscow
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[Jonathan Engwall]

A consumer product has to provide more than performance. For example, just
last night I ordered a WYSE thin client to send to a friend for her kids to
use. It cost me $20.00. New, it is still available as a $400.00 machine.
The new/used price change (in four years with an unknown amount of use) is
-95%.

Does that make you think twice about putting your newest ideas on a
government shelf?
I like the link above:

[1] https://en.wikipedia.org/wiki/X87

<https://en.wikipedia.org/wiki/X87>
I had no idea x86 began its life as a co-processor chip, now it is not even
a product at all. Of course I could have missed it, in the eighties unless
it had to do with Pitfall or Zork I was oblivious.

Jonathan

Post by Scott Atchley
Some of the research has already made it into products and more is slated
for future products. As with all research, some did not pan out, but that
is to be expected.

Post by C Bergström
Actually, the better question is, which vendor received funds and
actually made a useful solution that can go production with the
deliverables. From my view it seems like history is repeating itself[1] and
I wish more people would wake up. The top down approach to funding
scientific research and the in-fighting between labs is just too much
nonsense. If these research projects were a start-up, it would have failed
hard.
[1] https://en.wikipedia.org/wiki/X87

Post by Scott Atchley
Hmm, can you name a large processor vendor who has not accepted US
government research funding in the last five years? See DOE's FastForward,
FastForward2, DesignForward, DesignForward2, and now PathForward.
On Fri, Nov 17, 2017 at 9:18 PM, Jonathan Engwall <

Post by Jonathan Engwall
Maybe they felt married to government sponsorship while the competition
has found a way to compete with itself.
http://www.nag.co.za/2017/10/26/amd-launches-ryzen-processor
-with-radeon-vega-graphics-for-notebooks/
Maybe such a huge contract even looks too good to be true.

Post by Jan Wender
Unfortunately I did not find the english version, but Andreas
Essentially yes Xeon Phi is not continued, but a new design called

Post by Jan Wender
Xeon-H is coming.

Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some
important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.
Mikhail Kuzminsky
Zelinsky Institute
of Organic Chemistry
Moscow
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[Christopher Samuel]

Post by Jonathan Engwall
I had no idea x86 began its life as a co-processor chip, now it is not
even a product at all.

Ah no, this was when floating point was done via a co-processor for the
Intel x86..

[Richard Walsh]

Well ...

KNL is (only?) superior for highly vectorizable codes that at scale can run out of MCDRAM (slow scaler performance). Multiple memory and interconnect modes (requiring a reboot to change) create a programming complexity (e.g managing affinity across 8-9-9-8 tiles in quad mode) that few outside the National Labs were able-interested in managing. Using 4 hyper threads not often useful. When used in cache mode, direct mapped L3 cache suffers gradual perform degradation from fragmentation. Delays in its release and in the tuning of the KNL BIOS for performance shrunk its window of advantage over Xeon line significantly, as well as then new GPUs (Pascal). Meeting performance programming challenges added to this shrink (lots of dungeon sessions), FLOPS per Watt good but not as good as GPU. Programming environment compatibility good, although there are those instruction subsets that are not portable ... got to build with

-xCOMMON-AVX512 ...

But as someone said “it is fast” ... I would say maybe now it “was fast” for a comparably short period of time. If you already have 10s of racks and have them figured out then you like the reduced operating cost and may just buy some more as the price drops, but if you did not buy in gen 1 then maybe you are not so disappointed at the change of plans ... and maybe it is time to merge many-core and multi-core anyway.

Richard Walsh
Thrashing River Computing

Sent from my iPhone

Post by Christopher Samuel

Post by Jonathan Engwall
I had no idea x86 began its life as a co-processor chip, now it is not
even a product at all.

Ah no, this was when floating point was done via a co-processor for the
Intel x86..
--
Christopher Samuel Senior Systems Administrator
Melbourne Bioinformatics - The University of Melbourne
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http:

[Stu Midgley]

We have found that running in cached/quadrant mode gives excellent
performance. With our codes, the optimal is 2threads per core. KNL broke
the model of KNC which did a full context change every clock cycle (so you
HAD to have multiple threads per core) which has had the roll-on effect of
reducing the number of threads required to run to get maximum performance.
However, if your code scales to 128 threads... it probably scales to 240.
So it probably doesn't matter.

The programming model is much easier than GPU's. We have codes running
(extremely fast) on KNL that no one has managed to get running on GPU's
(mostly due to the memory model of the GPU).

So you shouldn't write them off. No matter which way you turn, you will
most likely have x86 and lots of corse... and those cores will have AVX512
going forward (and probably later AVX1024 or what ever they'll call it).
So, make sure your code vectorises and has no thread-blocking points.

Post by Richard Walsh
Well ...
KNL is (only?) superior for highly vectorizable codes that at scale can
run out of MCDRAM (slow scaler performance). Multiple memory and
interconnect modes (requiring a reboot to change) create a programming
complexity (e.g managing affinity across 8-9-9-8 tiles in quad mode) that
few outside the National Labs were able-interested in managing. Using 4
hyper threads not often useful. When used in cache mode, direct mapped L3
cache suffers gradual perform degradation from fragmentation. Delays in
its release and in the tuning of the KNL BIOS for performance shrunk its
window of advantage over Xeon line significantly, as well as then new GPUs
(Pascal). Meeting performance programming challenges added to this shrink
(lots of dungeon sessions), FLOPS per Watt good but not as good as GPU.
Programming environment compatibility good, although there are those
instruction subsets that are not portable ... got to build with
-xCOMMON-AVX512 ...
But as someone said “it is fast” ... I would say maybe now it “was fast”
for a comparably short period of time. If you already have 10s of racks
and have them figured out then you like the reduced operating cost and may
just buy some more as the price drops, but if you did not buy in gen 1 then
maybe you are not so disappointed at the change of plans ... and maybe it
is time to merge many-core and multi-core anyway.
Richard Walsh
Thrashing River Computing

[Nathan Moore]

Post by Stu Midgley
So, make sure your code vectorises and has no thread-blocking points.

I remember hearing this when learning MPI as an undergrad back in the
90's... It's probably always been true!

Post by Stu Midgley
We have found that running in cached/quadrant mode gives excellent
performance. With our codes, the optimal is 2threads per core. KNL broke
the model of KNC which did a full context change every clock cycle (so you
HAD to have multiple threads per core) which has had the roll-on effect of
reducing the number of threads required to run to get maximum performance.
However, if your code scales to 128 threads... it probably scales to 240.
So it probably doesn't matter.
The programming model is much easier than GPU's. We have codes running
(extremely fast) on KNL that no one has managed to get running on GPU's
(mostly due to the memory model of the GPU).
So you shouldn't write them off. No matter which way you turn, you will
most likely have x86 and lots of corse... and those cores will have AVX512
going forward (and probably later AVX1024 or what ever they'll call it).
So, make sure your code vectorises and has no thread-blocking points.

Post by Richard Walsh
Well ...
KNL is (only?) superior for highly vectorizable codes that at scale can
run out of MCDRAM (slow scaler performance). Multiple memory and
interconnect modes (requiring a reboot to change) create a programming
complexity (e.g managing affinity across 8-9-9-8 tiles in quad mode) that
few outside the National Labs were able-interested in managing. Using 4
hyper threads not often useful. When used in cache mode, direct mapped L3
cache suffers gradual perform degradation from fragmentation. Delays in
its release and in the tuning of the KNL BIOS for performance shrunk its
window of advantage over Xeon line significantly, as well as then new GPUs
(Pascal). Meeting performance programming challenges added to this shrink
(lots of dungeon sessions), FLOPS per Watt good but not as good as GPU.
Programming environment compatibility good, although there are those
instruction subsets that are not portable ... got to build with
-xCOMMON-AVX512 ...
But as someone said “it is fast” ... I would say maybe now it “was fast”
for a comparably short period of time. If you already have 10s of racks
and have them figured out then you like the reduced operating cost and may
just buy some more as the price drops, but if you did not buy in gen 1 then
maybe you are not so disappointed at the change of plans ... and maybe it
is time to merge many-core and multi-core anyway.
Richard Walsh
Thrashing River Computing

--
Dr Stuart Midgley
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf

[Lux, Jim (337K)]

As someone who works in a government funded research lab (as do many other list members, I suspect), there’s a lot of other factors that go into the seemingly bizarre funding and development stuff. Here’s a partial list - not specific to this instance, but just in general.


1) The “name” sponsor isn’t always the one actually interested in the final product – money gets shuffled around via Interagency Authorizations and funding vehicles. Typically a given lab will have a “prime contract” or “master contract” with one government agency (NASA in the case of JPL, where I am). All the work you do is on task orders under that master contract. If someone else (e.g. DoD, DHS, DoE) wants something done by JPL, they IA or MIPR money to NASA, who sends it to JPL.

2) There is a desire to keep competencies at various centers – in order to be an intelligent consumer, it helps to have built one – I venture to say that folks who are on this list (almost all of whom have at least tried to build a cluster) are better at buying a cluster from someone else.

3) Political considerations – The senator from the great state of Ruritania takes pride in having the worlds fastest computer and jobs for all those STEM grad students connecting patch cords in his/her state.

4) Workforce retention – Building the worlds fastest computer is interesting work – for technical people, the single factor that keeps them in their job, and not jumping elsewhere (assuming they’re paid some reasonable amount) is “I get to do interesting work”. You may be HPC’ing for half your time, and doing lethally boring contract monitoring (watching Other People do interesting work) the other half of your time. You don’t leap out of bed in the morning saying, I hope that monthly financial statement has arrived so I can review it.



Jim Lux
[phone] (office)
[phone] (cell)

From: Beowulf [mailto:beowulf-***@beowulf.org] On Behalf Of C Bergström
Sent: Saturday, November 18, 2017 4:59 AM
To: Scott Atchley <***@gmail.com>
Cc: Beowulf Mailing List <***@beowulf.org>
Subject: Re: [Beowulf] Intel kills Knights Hill, Xeon Phi line "being revised"

Actually, the better question is, which vendor received funds and actually made a useful solution that can go production with the deliverables. From my view it seems like history is repeating itself[1] and I wish more people would wake up. The top down approach to funding scientific research and the in-fighting between labs is just too much nonsense. If these research projects were a start-up, it would have failed hard.

[1] https://en.wikipedia.org/wiki/X87


On Sat, Nov 18, 2017 at 8:50 PM, Scott Atchley <***@gmail.com<mailto:***@gmail.com>> wrote:
Hmm, can you name a large processor vendor who has not accepted US government research funding in the last five years? See DOE's FastForward, FastForward2, DesignForward, DesignForward2, and now PathForward.

On Fri, Nov 17, 2017 at 9:18 PM, Jonathan Engwall <***@gmail.com<mailto:***@gmail.com>> wrote:
Maybe they felt married to government sponsorship while the competition has found a way to compete with itself.
http://www.nag.co.za/2017/10/26/amd-launches-ryzen-processor-with-radeon-vega-graphics-for-notebooks/
Maybe such a huge contract even looks too good to be true.

On Thu, Nov 16, 2017 at 3:06 AM, Mikhail Kuzminsky <***@free.net<mailto:***@free.net>> wrote:

Unfortunately I did not find the english version, but Andreas

Essentially yes Xeon Phi is not continued, but a new design called Xeon-H is coming.
Yes, and Xeon-H has close to KNL codename - Knights Cove. May be some important (for HPC) microarchitecture features will remain.
But in any case stop of Xeon Phi give pluses for new NEC SX-Aurora.

Mikhail Kuzminsky

Zelinsky Institute
of Organic Chemistry
Moscow

_______________________________________________
Beowulf mailing list, ***@beowulf.org<mailto:***@beowulf.org> sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf


_______________________________________________
Beowulf mailing list, ***@beowulf.org<mailto:***@beowulf.org> sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf


_______________________________________________
Beowulf mailing list, ***@beowulf.org<mailto:***@beowulf.org> sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

[Gerald Henriksen]

Post by Christopher Samuel
Interesting times (via a colleague on the Australian HPC Slack).
https://www.top500.org/news/intel-dumps-knights-hill-future-of-xeon-phi-product-line-uncertain/
Looks like fallout from the delayed Aurora system.
Rumours flying that the Xeon Phi family is in jeopardy, but the
# [Update: Intel denies they are dropping the Xeon Phi line,
# saying only that it has "been revised based on recent
# customer and overall market needs."]

This should cause some confusion.

While Knights Hill was cancelled, Intel has quietly put information
about Knights Mill online as the next Phi product line:

https://www.anandtech.com/show/12172/intel-lists-knights-mill-xeon-phi-on-ark-up-to-72-cores-at-320w-with-qfma-and-vnni
_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/li

[Mikhail Kuzminsky]

Post by Gerald Henriksen

Post by Christopher Samuel
Rumours flying that the Xeon Phi family is in jeopardy, but the
# [Update: Intel denies they are dropping the Xeon Phi line,
# saying only that it has "been revised based on recent
# customer and overall market needs."]

This should cause some confusion.
While Knights Hill was cancelled, Intel has quietly put information
https://www.anandtech.com/show/12172/intel-lists-knights-mill-xeon-phi-on-ark-up-to-72-cores-at-320w-with-qfma-and-vnni

I partially disagree with "confusion". It's simple because KNM has
minimal microarchitecture changes vs KNL, and does not focus on normal
DP-precision. KNM focuses on SP etc, and is oriented to Deep Learning,
AI etc.

Mikhail Kuzminsky,
Zelinsky Institute of Organic Chemistry,
Moscow

_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit ht

[Gerald Henriksen]

I partially disagree with "confusion". It's simple because KNM has minimal
microarchitecture changes vs KNL, and does not focus on normal
DP-precision. KNM focuses on SP etc, and is oriented to Deep Learning, AI
etc.

By confusion I meant the product names - Hill and Mill will be easily
confused and mixed up in discussions and people's minds.

[Jonathan Engwall]

Hill and Mill...that is called weathering.
I love rumors.

Post by Gerald Henriksen

Post by Mikhail Kuzminsky
I partially disagree with "confusion". It's simple because KNM has
minimal microarchitecture changes vs KNL, and does not focus on normal
DP-precision. KNM focuses on SP etc, and is oriented to Deep Learning, AI
etc.

By confusion I meant the product names - Hill and Mill will be easily
confused and mixed up in discussions and people's minds.
_______________________________________________
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf