Intel CEO on 10nm delay: 'We prioritized performance at a time when predictability was...

[gamerk2]

Lastly, RTX is DOA. Developers are going to push AMD RT iteration because everybody will develop games on consoles and port them to PC. Proprietary RTX will be dead in waters making those Tensor and RT cores useless.

Unlikely. Even on consoles, you really don't code to the HW anymore; it's all higher level APIs now. For example. the XB1 is basically just a stripped down PC running a stripped down Win8 shell.

As for RTX cores, NVIDIA has absolutely the right idea. RT isn't something GPUs can brute force like they do for Rasterization; you absolutely need dedicated HW to do the calculations. The problem with that is that by adding RTX cores, you are not adding features that improve performance for non-RT calculations, which is why the 2000 series really wasn't a significant performance boost over the 1000 series. There's only so much die to go around.

Personally, if you want real-time RT, you probably need a dedicated GPU just for that. It's simply too much number crunching to be done alongside the Rasterization process.

 

[Markoni35]

Why are university professors allowed to have sex with their students? That's a lot bigger abuse of power than what the ex-Intel-CEO did.

 

[urbanman2004]

Boo hoo Intel, your market share's decreased ;-)

 

[urbanman2004]

Why are university professors allowed to have sex with their students? That's a lot bigger abuse of power than what the ex-Intel-CEO did.

A rule's a rule which Krzanich broke.

 

[jpuroila]

Why are university professors allowed to have sex with their students? That's a lot bigger abuse of power than what the ex-Intel-CEO did.

I'm fairly certain they're not, it's just that when someone has tenure, it's fairly difficult for the university to take a decisive action against them.

 

[RobJoy]

"The executive said they “prioritized performance at a time when predictability was really important.” In short, he added, “we learned from it, we’ll get our 10-nanometer node out this year.”"

We all know that Intel won't have a response for Zen 2 this year, performance-wise they are still competitive but price-wise there's a lot to improve and by the time they come with 7nm in "two years" AMD will have the upper hand with 5nm. At this moment just pure fanaticism will keep Intel from losing major share market this next 2-3 years.

Spec wise

Intel 10nm > 7nm TSMC
Intel 7nm > 5nm TSMC

Zen 2 on 7nm requires higher voltage @ 4.3 ghz than Intel's 14nm @ 5ghz, also temps are higher in Prime small ftt.

It all comes to Intel getting the 10nm node ready for 2020 and if they are successful, AMD is done. Ice Lake has been ready for years, it is already 15-18% faster clock to clock than the 14nm Skylake/Coffe Lake/etc so comfortably in front of Zen 2.

Yes, yes,.. but you pay a hefty premium for that.
When they will be able to compete in the price/performance bracket, call me.

 

[Markoni35]

I'm fairly certain they're not, it's just that when someone has tenure, it's fairly difficult for the university to take a decisive action against them.

We all know that university professors bang their students. It's a common topic in all media, even in movies. And it's never ever mentioned in relation to "abuse of power" even though it's evident it's one of the most common and biggest abuses of power. I find such double standards very annoying.