In Wake of Intel's 7nm Woes, AMD's Price per Stock Vaults Over the Blue Giant

[Ravenas]

Good news for AMD and Nvidia.

Intel to soon go semi fabless to address this.

 

[medi01]

how far ahead Nvidia is

With 5700XT AMD has narrowed the gap to "some perf/watt", which they actually aimed to address with RDNA2 (50% improvement)

 

[InVasMani]

Considering how fast stock price is changing, you would think it was AMD that made $18B this quarter. I think Intel is doing a good thing tapping into other foundries.
I hope AMD does good too. We need better AMD for more competition considering how far ahead Nvidia is.

Nvidia is ahead, but probably the most exposed of the three in terms of being susceptible to being a bit vulnerable however they are in arguably the most lucrative area's so financially sound right now. That said a big push from AMD and/or Intel could really squeeze Nvidia a fair bit and knock them down a peg or two because they aren't as diverse as perhaps they could be. AMD is arguably the most cohesively diverse of the three, but still has a bit of financial overhead that likely limits a bit of there R&D budget to leverage it's full potential on it's wealth of CPU/GPU ip that it's got available. Still AMD as a whole is being progressively more competitive I'd argue. Intel needs to get it's ducks in a row they have a wealth of IP and are a big company, but if they remain to stuck in their ways and can't innovate enough into other area's and maintain the advantages it had long held like it's fabrication advantages that it's struggled with as of late it could kind of doom them a bit.

 

[Toothless]

This is idiotism. You know productivity - gaming is for kids and people without jobs.

Huh, didn't know I'm a kid and jobless even though I can legally drink and have a full time job. Guess I'll go back and play my kiddie video games.

 

[Patriot]

People forget how wide Intel's portfolio is, Their failing X86 and Xe linup is by far not all they do, They created OneAPI for a reason, to prevent their Xeons from being useless.
They have a "strong" portfolio despite their failings with their FPGAs, VPU, Nirvana training and inference chips, Optane persistent dimms and nvme drive and Habana labs being their latest acquisition, they have a lot going on that is all capable of being accessed through a common API. So even when they release their slow low core count ice lake chips... they will still have a niche to fight in...

They know they fucked up, so they have hedge their bets wide.

Also note... 7nm via tsmc was in roadmaps like 1yr+ ago... so switching to TSMC causing a 1yr delay means they flipflopped internally several times.

ARM will hold the supercomputer crown for less than a year.
Summit arrives in 2021 at 1.5 Exaflops 3x the arm, and then in 2022 a 2 exaflop also all AMD supercomputer will be launched.
https://arstechnica.com/gadgets/201...1-5-exaflops-supercomputer-for-us-government/

Intel's 1 Exaflop appears to be delayed till 2022-23 due to their failings... and will be potentially half as powerful as the top AMD supercomputer when launched... Ouch.


Also...

Huh, didn't know I'm a kid and jobless even though I can legally drink and have a full time job. Guess I'll go back and play my kiddie video games.

Don't tell ARF about pro gamers, he will just spontaneously combust.

 

[medi01]

I'm rooting for AMD to get larger, but it would upset me greatly if Intel abandons own fabs and jumps to third party bandwagon.

Long term, especially with AMD getting to comparable size, it might not be sustainable for Intel to afford it, but nothing is stopping them from opening the gates for the third parties.

Their failing X86 and Xe linup is by far not all they do

Xe is not an actual product yet, and X86 just posted record profits.
Failing, my arse.

 

[Patriot]

I'm rooting for AMD to get larger, but it would upset me greatly if Intel abandons own fabs and jumps to third party bandwagon.

Long term, especially with AMD getting to comparable size, it might not be sustainable for Intel to afford it, but nothing is stopping them from opening the gates for the third parties.


Xe is not an actual product yet, and X86 just posted record profits.
Failing, my arse.

Xe will be 1.5 to 2 years late when launched powering a supercomputer half as powerful as the competition, that is pretty big fucking failure. 10nm was supposed to be out in 2015, Rome was meant to compete against icelake, but that has been delay after delay and cut down to same core counts as cascade lake... That is a failure. Multi year delays are failures. Especially when the competition is executing their plans. Spending 3B on marketing rather than innovating is a failure. Intels internal non aquires products are failing, and their mindshare is finally starting to slip... Once that happens there will be no more record profits.

 

[Mysteoa]

" A first in many years for the green company. "

What does NVIDIA have to do with this article? Seriously, AMD has been advertising themselves as " Team Red " for years!

The AMD logo used to be green, now it is black. They call them self team read when they talk about the GPUs since ATi was red.

 

[AsRock]

This is idiotism. You know productivity - gaming is for kids and people without jobs.

WTF, gaming is for those who all so work there asses of so they can too.

 

[watzupken]

Looking at AMD's success story after leaving GF (and of course also with a good CPU architecture), I feel Intel may need to reevaluate if it makes sense to keep their fab business. Over the last few years, we have seen quite a lot of issues with their fab, i.e. massive 10nm delays, shortage of 14nm parts, and now 7nm delays. In the past, it made sense because Intel is always ahead of their competitors when it comes to fab, which gave them a significant advantage in their chip business as well. That advantage is pretty much gone with a 4 to 5 years delay in their 10nm. Despite Intel's claim that 10nm have improved and we should see 10nm desktop chips soon, I feel we should wait and see because I don't think they are out of the woods yet. If a highly matured 14nm is not able to provide a stable supply, I wonder if Intel is able to avoid this issue with their 10nm which they are trying to aggressively release across all the market segments in starting this year with Tiger Lake U and 2021.