SI
SI
discoversearch

Technology Stocks
AMD, ARMH, INTC, NVDA
An SI Board Since December 2010
Posts SubjectMarks Bans Symbol
20986 219 0 AMD
Emcee:  neolib Type:  Moderated
Update on 8/16/16: At IDF, Intel announced they are broadly licensing and supporting ARM IP on their 10nm Foundry process, and presumably going forward. Having thrown in the towel on mobile SoC's and Contra Revenue earlier this year, this would appear to be RIP for x86 in both mobile and likely IoT. Instead, Intel appears to want to compete with the Foundries for the wafer volume those markets offer. Since it takes years to build IP and customer relationships, and Intel has a checkered history of being a reliable partner and silicon scaling is near the end, this will be a very interesting development to watch over the next 5-10 years.

Update on 1/28/16: TSMC & Samsung both exited 2015 in high volume FinFET production, while AMD is scheduled for mid-16 on GF's FinFETs. TSMC will start 10nm customer tapeouts in 1Q16 and production by 4Q16. The PC industry remains in contraction, now entering the 5'th year of decline, and mobile is showing hints of peaking as well. The big question for '16 is whether any new high volume applications for semis emerge. Automotive, wearables, VR, and drones are potentials that might develop in 2016.

Update on 3/6/15: With confirmation from WMC that Samsung will ship two devices with 14nm FinFET SoCs starting April 10, 2015, it is finely becoming clear how drastically the Foundries have closed the process gap with Intel. An interesting question now for 2015 is when the combination of Samsung/GF and TSMC will surpass Intel in wafer area on FinFET processes on a quarterly basis. My guess is that Q4'15 sees this milestone occur.

Update on 2/26/15: With WMC a few days away, it is widely anticipated that Samsung will debut 14nm SoCs into a very high volume product: The Galaxy S6. With an expected ramp for Apple A9 SoCs in Q2, and likely some additional SoC volume from Qualcomm, it would appear that Samsung will ship more CPUs at 14nm than Intel in 2015. Intel might possibly ship more wafer area at 14nm, but data on that is difficult to obtain. The mobile SoC's are as big or bigger than the lower end higher volume PC CPUs, and so far, Intel's slow ramp of 14nm product has not progressed well for the large die devices.

Update on 8/6/14: Since this board was started, the CPU space has changed considerably, and so has the semi Foundry space. The AMD/Intel competition in x86 processors has been replaced with x86 vs ARM as computing has increasingly shifted to mobile form factors. Intel has made a significant shift to use their leading edge process advantage as a Foundry. The huge volumes in mobile now dominate leading edge wafer production, and competition is fierce to land Apple's orders between TSMC and Samsung. Intel has struggled to launch 14nm, and the Foundries are making a determined effort to close the gap with Intel on process tech. What happens in late 2014 and through 2015 for 14/16nm FinFET processes now looks like the big story in semiconductors in the next year+.

While the x86 CPU contest between AMD & Intel has grown old, exciting things are happening in the CPU and graphics world. The age of APUs in the x86 world has begun, with both AMD and Intel rolling out new families in early 2011. Meanwhile, 2010 saw the explosive launch of tablet PC's, mostly powered by non-x86 designs, and running non-MSFT OSes and software. What will 2011 bring to the CPU/GPU world?

This is a moderated board. All views are welcomed, but please restrict posts to useful content about CPUs, GPUs, the companies involved, or related market conditions. Attacking/baiting other posters, and snark in general will lead to banning.

Additional Comment: The companies working in this space have talented engineers & managers, and while we all may have our views on them, please refrain from endless bashing of them, as it does not contribute much useful to the board.
Previous 25 | Next 25 | View Recent | Post Message
Go to reply# or date (mm/dd/yy):
ReplyMessage PreviewFromPosted
20986TSMC says HPC will become a major driver for them in 2020 and onward: digitimesneolibyesterday
20985As we approach earnings for AMD, this is worth looking at for the graphs: markeneoliblast Friday
20984Is ADBE the next NVDA? Yes. Adobe stock is growing far faster than the S&P,zzpatlast Friday
20983V* Re: Is that something being researched?> Endlessly by every programmer whTGPTNDRlast Friday
20982Yes, I assume, but $400m at the foundry level is a lot of silicon. That is cr wneoliblast Friday
20981> Can the software side keep up with 48 cores + 4 GCX? Do you have anything Vattilalast Friday
20980> If the CPU itself is smart enough to dispatch appropriate code to these [GPVattilalast Friday
20979Probably an ASIC for Bitcoin mining, where GPUs no longer play?Vattilalast Friday
20978Oops, I guess that comment from TSMC on crypto was not per se good news for eithneoliblast Friday
20977Further, TSMC referenced the cryptos in their Qtr report: digitimes.com <i&neoliblast Friday
20976Digitimes on the crypto outlook: digitimes.comneoliblast Thursday
20975Tightly-integrated GPU on-die = HSA accelerator. If the CPU itself is smart enoufastpathgurulast Thursday
20974Yeah, he has been negative on both Intel and AMD for quite some time. My big coneoliblast Thursday
20973Rasgon is at it again: Bernstein expects AMD also to report roughly in-line resPravin Kamdarlast Thursday
20972Tis a great pity that AMD can't tape out existing designs on this incrementaneoliblast Thursday
20971 Compared to its 14nm process, the 10nm node sports NMOS and PMOS current that iTHE WATSONYOUTHlast Thursday
20970Samsung Completes Qualification of 8nm LPP Process "8LPP provides up to 10Vattilalast Wednesday
20969HiSilicon looking at GF (and Samsung and Intel) for 7nm as a 2'nd source vs neoliblast Wednesday
20968...//... this scheme would up the ante considerably and Intel would have to copyrzborusalast Wednesday
20967From an EETimes article I linked earlier, it is 2.8x denser, so 12 cores + a GPUneoliblast Wednesday
20966I think AMD should hire your!neoliblast Wednesday
20965<i> Replace one of the CCXs by my hypothetical GCX and you may be able to THE WATSONYOUTHlast Wednesday
20964 > what do you estimate for a die size on 7nm? The EE Times article quotes Vattilalast Wednesday
20963<i>Is this too crazy an idea?</i> .......well...the (48) core leak THE WATSONYOUTHlast Wednesday
20962By the way, my crazy idea would mean that 7nm AMD processors would all be APUs. Vattilalast Wednesday
Previous 25 | Next 25 | View Recent | Post Message
Go to reply# or date (mm/dd/yy):
Copyright © 1995-2017 Knight Sac Media. All rights reserved.Stock quotes are delayed at least 15 minutes - See Terms of Use.