Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Koog
I hope the board will put Krzanich in one of their autonomous cars and let it drive him to retirement
Who do you see replacing him?
The SI moderators will not allow this political discussion to continue. It's not me, it's them.
Time to drop it.
GB
Remember when we used to talk about exciting new product introductions and financial gains resulting from those?
Ya... And we could insult each other about AMD instead of politics.
The good old days...
willco
they will just make themselves some political enemies as Intel has now done
I think that's the most significant point. For the meaningless gesture of being politically correct, Intel has now made political enemies who are in power now.
chipguy
For a maker of widgets that go into boxes there is no upside in
getting publicly involved in political controversy on any side. Too
bad BK blew that right from the start
I agree but he needed to show how self righteous he was. I didn't expect any less from him.
borusa will never stop. He's been doing this for years and now it's going on decades. Your only option is to bang your head against a wall or ignore him. Just put him on ignore and save yourself the aggravation.
herb
Why not just ignore him?
all of this was brought out during the trial
Herb, there has never been a trial.
herb
Intel won backing from ECJ court adviser Nils Wahl last year who doubted if the company's actions had really harmed competition
An equally important question is, did Intel even do what the EU says they did?
simplytom
I wish I could understand what you just said.
And Intel hasn't lost it's lead in process technology. 14nm is still better than "10nm" on the foundry side
I have to wonder why it's taking Intel so long to come out with 10nm. There has never been such a long time lag before. I suspect there is some new technology there but I really don't know.
fpg
I think that unlike AMD's phony claims, Intel is doing it because there's an actual grievance here.
morrowinder
So how are those Epyc servers doing?
We all know that if AMD has a data sheet then Epcy should sell just fine because "Parts is Parts".
My point was that aside from a couple of blips, silver has gone nowhere in around 35 years. Don't get me completely wrong, I own physical but I'm not ignoring the historical record.
I was happy at SILVER $15.00
I remember in the early 1980s someone told me he couldn't believe people weren't loading up on silver at ~$15...
There always are Tim.
Infinity fabric is the server equivalent of duct tape because AMD CAN'T manage to produce monolithic die high core count processors
It's obvious that AMD's MCM approach is in response to anticipated low yield on large die. There's no other reason to do it.
morrowinder
In the FTC investigation of a few years ago, AMD fans charged Intel with degrading the Intel Compiler performance when running non Intel CPUs. They were certain this was criminal activity but in the settlement the FTC gave Intel their blessings recognizing that Intel had no obligation to empower their competition. Just document that non Intel hardware may not see optimization. I had it on good authority that the real problem was that AMD would not share their errata with Intel so that Intel could avoid code sequences that might crash some AMD chips and we all know how AMD fans would have a field day with that one claiming intentional dirty tricks!
morrowinder
Apparently AMD documents it's faking the numbers. Take a look at this:
At every targeted price point for two-socket processors, EPYC outperforms the competition, with up to 70% more performance in the eight hundred dollar price band and up to 47% more performance at the high-end of the market of four thousand dollars or more 5 Note footnote 5
5. Based on estimated SPECint®_rate_base2006 scores. 2P Intel Xeon E5 scores other than E5-2699Av4 were derived by AMD from the following ICC compiler-based test results published at www.spec.org, multiplied by 0.575 to convert from the ICC compiler to the GCC-02 v6.1 compiler used for EPYC testing:
http://www.globenewswire.com/news-release/2017/06/20/1026480/0/en/AMD-EPYC-Datacenter-Processor-Launches-with-Record-Setting-Performance-Optimized-Platforms-and-Global-Server-Ecosystem-Support.html
Got 15 grand waiting on the side to buy it, if it hits there
Why don't you write Puts?
morrowinder
AMD even penalized Intel for 20% because Intel has a vastly superior compiler on those ridiculous competitive slides.
Can you explain what you mean by this?
morrowinder
If I'm not mistaken, Intel has not explicitly said the 32 core device is monolithic, is that correct?
where is the "only the paranoid survive culture" that made me an Intel addict for over 2 decades?
It's been replaced by political correctness.
The worst case scenario for AMD would be an expensive chip respin
No, I think a recall would be called for and that would be worse.
It's the right thing to do. Lets see if AMD steps up and does the right thing.
morrowinder
Just put him on ignore. I have.
morrowinder
A premium brand NEVER drops prices by 23% with a successful product just months after launch
And certainly not one that is supply constrained. Only one that is demand constrained.
Of course.... What was I thinking.
You were Clarabell.
drjohn
According to this article Coffee Lake is not 10nm, it is 14nm. Cannon Lake will be 10nm. There is currently no release date.
https://www.neowin.net/news/intel-to-release-8th-generation-of-core-cpus-later-this-year
Didn't you used to play Dilly Dally on Howdy Doody?
https://www.slashgear.com/intel-core-i9-leads-new-x-series-cpus-for-the-most-demanding-users-30486837/
The company said that in its latest testing of Coffee Lake, it's "seeing a performance improvement of more than 30 percent" compared with the current generation, based on a SYSmark benchmark. It did add the caveat that "performance estimates are Pre-Silicon and are subject to change"
Why are they talking pre-silicon? Does that mean they don't have silicon or they haven't fixed the bugs well enough to run the benchmarks yet? Either way I don't like the wording.
morrowinder
Wait a minute... I thought Intel was caught completely flatfooted with no new products in the pipeline...
fpg
he knows they're bogus and opined anyways
I know nothing of the sort and I'd appreciate it if you wouldn't put words in my mouth. I expressed an opinion, and I stand by it. As Andy noted, at ~200mm2 it would take a defect density quite a bit under 0.2d/cm2. That would have been very difficult on a production basis in the past and even less likely today for reasons already discussed. Redundancy does help yields but only to a point. So I don't believe the rumor. That doesn't make me perfect but it's my opinion. You are entitled to yours. You can post here that you believe the rumor, along with an explanation as to why, without having your character assassinated, by me at least.
morrowinder
Yields are going to be a problem for the entire industry if you haven't figured that out yet
I don't think he's figured out much of anything. I believe yields are down across the board but I have no inside information. I don't know exactly what Intel's yields are anymore but I do know they aren't what they used to be either.
fpg
Don't you ever get tired of arguing?
fpg
You know "the number" isn't 44mm^2, and you know it isn't 200mm^2
I guess you don't know what it is either.
Bozo
You and El fudo mistake 44^2nm for ~ 200^2nm
I said if you had better numbers I'd gladly revise my figures and all you could do was sit there with your thumb up your butt, so don't blame me. And it's ^2mm not ^2nm.
morrowinder
Elmer: He looked no deeper...
That's why we need someone such as yourself to set the record straight.
Andy
Yup. You're right. 16MB cache on 44mm2 does not make sense. I should have looked deeper.
Andy
...of course it is not [line yield].............
You don't know that, you can only guess like the rest of us. Why should you be taken more seriously?
Andy
by saying that, there are only two possibilities..........
Chill out. I clearly said I did a google search and found the Zen die to be 44mm2. If someone had a better number then I'd rerun the numbers and stand corrected. Do your own search. Here's the first thing that comes up for me and that's what I went by:
http://wccftech.com/ryzen-smaller-die-intel-zen-architecture-not-good-hpc/
The Zen die is 44mm2 and the Intel die is 49mm2. By that measure 80%+ yield is not unreasonable but it has since been brought to my attention that that is not a proper comparison. A much larger die should be used and that would not allow an 80% yield on a modern process. So please unbunch your panties. I don't want to argue with you. You simply misunderstood my response.