Nvm?? What's that?

gb

TCAE: Any insight into 3DXP claims by Charlie?

gb

IIRC, someone did an analysis of the profitability of AMD over it's entire life and found that it wasn't even break even. It had been kept afloat by "other peoples' money". ATI didn't help as it was similarly near break even before being bought.

And for those buying in now perhaps they should note the following:

http://www.gurufocus.com/term/Book+Value+Per+Share/AMD/Book-Value-per-Share/Advanced-Micro-Devices-Inc

TLDR: NAV -0.52/share...note the negative sign.

More of the same...

gb

It has the smell of more mortgaging of their future. Also that they're falling short of their original wafer use expectations with at least GF.

gb

AMD financial moves.

http://phx.corporate-ir.net/External.File?t=1&item=VHlwZT0yfFBhcmVudElEPTUyMzY0NTB8Q2hpbGRJRD02NDQ1NzM=

I can't figure out much about this beyond it is costing AMD.

gb

IDF sessions catalog now open.

Storage Snapshot and VTune now has Storage performance tools. Quite a few sessions on SSD, NVMe and presumably buried within: 3DXP.

Some interesting sessions on applications of iPhi.

www.intel.com/idf

gb

re: 3DXP competition.

If it was easy to do, they would have already done it.

I'm sure they'll acquire and do a tear down of 3DXP products but it's one thing to know what is there and quite another to replicate it.

gb

I think that this is a plant that GF bought from IBM.

fixed.

you're right.

Looks like VT is going after IBM however.

gb

Found on Facebook:

For all the ex-Intelers out there who may not know...

I just got word that Tom Piazza passed away yesterday (june 12). Pancreatic cancer. His family and friends were by his side.

Tom got me started in amateur astronomy. We observed up at the SVAS Blue Canyon site a couple of times.

RIP

gb

I'm not sure Tom has passed. Perhaps just left Intel. Nothing on his son's social media pages about his passing.

gb

IIRC, MSFT had some special hooks in the kernal that allowed virus products (theirs) to be more capable. The 3rd parties raised a stink about unfair advantage and MSFT gave them access which they promptly ignored and didn't use.

I don't remember the details but it was around Win7 timeframe.

gb

Since the next big HPC event is in November I'm guessing any announcement of 3DXP use in HPC will wait until then.

July 2016 will mark a year since Intel/Micron announced 3DXP. No products on the market yet that I'm aware of. This can't be going as planned...

gb

Since Intel has a top notch implementation of the Lustre file system I'm guessing that there will be some exciting implementation which exploits 3DXP in the not too distant future.

Intel's implementation is already widely used in HPC.

gb

I wonder if he has the chops to build and support modest sized HPC systems for his customers? I'm guessing the Intel software support he mentions would make it more feasible than GPU based approaches.

gb

True but I think there may be a large market that is yet untapped for HPC computing that isn't yet being served by anything. It's just too hard for most to adopt the types of systems and software so it has been limited to a small portion of the market.

I don't think HPC servers will replace Xeon traditional servers rather be additive.

And keeping Nvidia out of the driver's seat here means capturing more of the system $$. Along with Omnipath and probably specialized storage made possible by 3DXP.

gb

Looks like Intel is putting in place all the pieces for a much wider adoption of HPC. If they can build the parts this could be a nice upside.

gb

I wonder if the Win10 I/O system will be able to keep up with the performance capabilities of Optane. Particularly with NVMe implementations.

Not a gamer but daughter is a heavy duty user of Photoshop and Lightroom.
Both are I/O pigs at times even with an SSD.

I'd love to replace some more rotating rust.

gb

I'm personally less worried about the hardware and eventual availability of silicon than I am about software support.

"If you build it, they will come" may work for Kevin Costner but in computers it has proven to take the better part of a decade in some cases.

gb

AMD board shuffle. Rearranging the deck chairs...

http://fortune.com/2016/05/12/amd-chairman-board-john-caldwell/?xid=yahoo_fortune

gb

KNL is the first in the iPhi line to be a self hosting chip rather than an add-in card. That has the potential to be a game changer for the iPhi line.

Whether it has a broader uptake as a result is yet to be seen.

gb

There's a reason that Xeon Phi still has minority share in the HPC accelerator market compared to NVIDIA's Tesla GPUs.

Ya think that might have something to do with Phi availability?????

The versions preceding KNL were pretty much science experiments. KNL is a real product. I'm expecting a pretty broad deployment among those who need this kind of performance.

gb

And Intel is running significant training programs around the world to show programmers how to modernize their code so that it runs better on Xeon as well as iPhi.

https://software.intel.com/en-us/code-modernization-enablement

gb

Two quick points:

Intel's Lustre is built on the community version with value add in support, and additional interfaces, etc.

Getting 3DXP support in Lustre which is under Intel's control would certainly be easier than coaxing someone like MSFT to incorporate it.

gb

Still not seeing any mention of 3DXP in filesystem use. I would think that big one-off systems such as this would be perfect places to show the value of a new memory architecture.

Maybe something will be announced at the coming SC conference in June when the new Top500 list is announced.

Since Intel's Lustre filesystem seems to be pretty popular I would expect that to be a useful insertion point. But I'm not a software guy...

gb

Japanese HPC machine due end of this year. Lots of Intel content: iPhi, Omnipath, Intel's Lustre filesystem.

http://www.hpcwire.com/2016/05/10/japan-unveils-details-25-pflops-machine-operational-december-2016/

25PFLOPS. May end up as #2 on Top500. At any rate it should the first iPhi monster to power on that I'm aware of.

gb

Cray results:
http://www.hpcwire.com/off-the-wire/cray-reports-first-quarter-2016-financial-results/

Interesting quote:

While the Company continues to expect revenue to be in the range of $825 million for the year, there is an increased level of risk to achieving this target. This increased risk is driven by the Company’s reliance on key third-party components, some of which have already been delayed,...

I think all of Cray's recent intros and announcements have been Intel and iPhi based. I wonder if it is Intel or some other who is slipping.

gb

Of course there's more to sell! Intel's IP!

gb

I think they'll continue to do well in servers. I'm eagerly awaiting real results from Xeon Phi which will extend servers even further as HPC becomes a staple of business in big data and technical design and analysis.

I'm also very curious to see the uptake on 3DXP.

gb

As a former Intel employee (1982-2000) I can't remember many times Intel attacked a segment and failed. They had several products that failed of course (Itanium, i860, i432, etc.)

The closest I remember was printers where i960 was the vector. No where near the magnitude of phone volumes. There may have been automotive things that I wasn't aware of that failed. 8096 never really got traction.

The flop in cell phones takes the cake. Probably will be the subject of an MBA case study someday.

BTW, I remember Sean Maloney warning everyone long ago that "mobile" was a big issue. He made a big deal of it at the Sales and Application conference one year. Also at SLRP in the mid 90s. I don't think he was taken seriously enough. This was long before Apple and iPhone.

gb

AMD, out of its own body parts to sell, now decides to sell some of Intel's body parts.

What could possibly go wrong?

gb

Rometty also mentioned something about good progress on OpenPower but lacked the all important AMB: as measured by...

gb

IMO the potential for disruption is high and fairly quickly felt and any potential for improvement in products is years away.

"The market" is usually impatient so I don't see a stock price upside to this. Of course the yield will improve if the keep the dividend constant.

I'm glad I no longer own INTC but sad to see my former employer in such a mess.

gb

Just had a thought: in the old days Google would have thrown business to AMD to leverage Intel. Says volumes about how low AMD has sunk that Google would rather port all their software in hopes of a competition.

gb

Actually Intel's current pitch is about code modernization which benefits not only iPhi but pretty much any architecture except for the crazy stuff needed to make GPUs go fast.

https://software.intel.com/en-us/modern-code

https://software.intel.com/en-us/code-modernization-enablement

gb

That would make sense.

gb

IBM's Power fortunes if only for it's own system sales appears bleak.

IBM seems to be pushing it hard for Open Power sockets such as HPC and Google-like systems.

What I don't get is how they expect to benefit financially from this?

Is there a vast pot of gold in Power derived royalties ala ARM?

Seems like a pittance compared to the overall IBM financial picture.

gb

I wonder what impact this will have?

http://www.pcworld.com/article/3053092/ibms-power-chips-hit-the-big-time-at-google.html

gb

This is beginning to look like an "all in" strategy on solid state storage. In the past I viewed Intel products as mostly performance and reliability trendsetters leaving the lower price levels to others.

This is more like the i3, i5, i7, good better best approach.

Will be interesting to watch from a product launch and breadth standpoint and with the breakout of financial numbers we can see how successful they are. Note that unlike other parts of Intel business, Flash and storage really don't have anyone else paying the freight of process development.

gb

New financial reporting breakout. Looks like more transparency coming.

gb

One thing I wonder about with so much "house cleaning" is who is left from inside Intel to replace BK? ASG used to groom potential executives through the TA program. What remains of that? It would seem that such a program is only as good as the exec for whom the TA serves.

Or would we rather see someone like Fiorina start running the show?

And what could any replacement do to make any difference that would show tangible product results within the "patience" horizon of most investors or critics?

gb