Intel might have pulled one over on us, after all, assuming the last five years of effort designing a replacement for CMOS design bear fruit. Their new magneto-electric spin-orbit design not only uses significantly less power than traditional designs, but Intel also claims it offers five times better logic density. If they are able to bring this technology to fruition, their 10nm woes may not be as much of a setback as it currently seems. The Register has a link to the Nature article, if you would like to know more.
"Chipzilla claimed its magneto-electric spin-orbit (MESO) technology's important characteristics are low voltage (as much as five times below today's CMOS-based chips) and consequently lower power (between 10 and 30 times lower than CMOS)."
Here is some more Tech News from around the web:
- Epic Games to undercut Valve with own PC games store @ The Inquirer
- Samsung Caught (Again) Using DSLR Photo To Advertise Smartphone Camera @ Slashdot
- Windows 10 debacle is now affecting Cisco and Morphisec endpoints @ The Inquirer
Yet another lab revolution
Yet another lab revolution with no industrial application…
Good distraction from
Good distraction from everything else going on at Intel right now.
Yeah; this sounds like
Yeah; this sounds like marketing asked for something to tell shareholders about. The tech seems like something out of an advanced research project that may not be an actual product for many years, if ever. They can probably make all kinds of interesting things in the lab that will never be large scale manufacturable. They talked about phase change memory for about 20 or 30 years.
I don’t think intel really has an answer to AMD cpus for a couple of years. If the rumors are true about 16 core mainstream Ryzen 3000 processors and Intel only has a 10 core, then it isn’t going to look good. I wouldn’t take these rumors as truth yet though. The best way to make AMD look bad at this point might be to release over the top rumors such that the real product feels like a disappointment. Someone could also be trying to manipulate stock prices. The AdoredTV guy that reported them has a previous video saying that the IO die has a GDDR6 and an HBM memory interface on it. Or that there will be an IO die with those interfaces. That is almost certainly wrong. Using an interface chiplet for such fast interfaces would be incredibly wasteful, if it is even possible. Current HBM tech can reach 1 TB/s; how would you pass that through an interface chip? That doesn’t give me a whole lot of confidence in him.
We need to be free of these monopolies. I am hoping for the best for AMD. The PC market was stagnant for many years with very little reason to upgrade. With competition again, people have a lot more reason to upgrade now. I am trying to decide whether to wait for Ryzen 3000 or build a Ryzen 2000 system now. Perhaps I will wait until after CES to see what the schedule is.
Of course Intel’s Marketing
Of course Intel’s Marketing will try and oversell these performance metrics just like they did for 1st generation XPoint! And this product will take way longer than Intel’s 10nm time frame to get to market if it ever makes it that far.
Looking at the laptop market and the Mini-Desktop Market it look like Intel is up to their old market tactics once again and OEMs are being incentivised to use mostly Intel’s products with AMD getting almost no consumer Mini-desktop/NUC-Like devoce APU design wins. AMD has been getting more sucess in the industral/embedded market with some Epyc branded SKUs but not much love in the consumer facing Mini-Desktop PC market.