• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD THREADRIPPER VS INTEL SKYLAKE X

Is it true that Global Foundries 14Nm is designed for mobile device use rather than desktop use? AdoredTV mentioned it in his previous video but couldn't find anything online.
Also if it is true would that explain the problems that overclocking problem and the 4GHz wall that everyone is experiencing?

I'd almost certainly bet on GF 14nm being a factor here - some competing SoC designs get 20% higher clocks on Samsung 14nm or TSMC 16nm than the equivalent at GF - but also those in the industry still claim GF is running unstable yields so likely quite a variation on what individual batches of CPUs can clock to. The process I believe was designed with a focus around biasing for low voltage operation over clock speed.
 
So potentially Zen 2 next year if on 7nm; could potentially clock high. Depending how it is designed?

It should do yes.

As transistors get smaller, they are able to clock higher (physics reasons...). Also the specific 7nm process GloFo are going to use was designed by IBM, and the IP was bought by GloFo. There's a fair amount of talk about it being a very very good process.

Also it's going to be larger than a single node jump (before EUV lasers too), not as good as 2 jumps but better than 1. Talk of more than 60% size reduction, more than 50% power reduction, etc.

So, all else being equal, AMD should be able to fit around 10 Zen1 cores in the same space as 4 now, while also slightly lowering the power. That's if they used the node to it's maximum efficacy, and also if Zen2 had no changes to Zen1.

This is also why unless Intel pull something out of the bag, they're totally ****** in the server space for Zen2 and Starship. Intel currently have a process advantage (their 14nm is better than GloFo and TSMC and Samsung), but AMD's 48 core Starship CPU will be even lower power than their 32 core one, and the 32 core EPYC is already higher perf/w than Intel's offerings somehow. It'll be kind of hilarious if AMD manage to get Starship running on 150W or less.
 
Last edited:
I'd almost certainly bet on GF 14nm being a factor here - some competing SoC designs get 20% higher clocks on Samsung 14nm or TSMC 16nm than the equivalent at GF - but also those in the industry still claim GF is running unstable yields so likely quite a variation on what individual batches of CPUs can clock to. The process I believe was designed with a focus around biasing for low voltage operation over clock speed.

Is it true that Global Foundries 14Nm is designed for mobile device use rather than desktop use? AdoredTV mentioned it in his previous video but couldn't find anything online.
Also if it is true would that explain the problems that overclocking problem and the 4GHz wall that everyone is experiencing?

Zen is using the Samsung Zeppelin design process (Samsung 14nm LPP), which is designed for mobile devices with power efficiency in mind.
A good technical article here. Samsung Zeppelin+ which is used on Zen 2/Zen+ (same thing), is designed for +400Mhz higher speeds at same power envelop.

https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
 

Well, to be fair he actually said an average chip is in the 4.7 Ghz ballpark including delidding, since Intel decided to go for TIM...

So your average user, with average chip, not delidded, and also with mid-range AIO water, or high-end air; sounds like we should expect 4.5 really.

This is still great, but not as rosy as you've painted it.
 
Oh dear^^^ We delided like 50 chips no problems at all. Why are guys on here so obsessed with temps its stable around 88 on AIO I mean add custom loop its dropping another 10 or so C.......... These OC's are no problem at all.
 
Oh dear^^^ We delided like 50 chips no problems at all. Why are guys on here so obsessed with temps its stable around 88 on AIO I mean add custom loop its dropping another 10 or so C.......... These OC's are no problem at all.
I am not obsessed with speed, but the overall platform is sub par for the money.
 
It kind of is when AMD are offering more cores for less money. For the kind of work these CPU's will be doing pure IPC isn't going to cut it anymore.
Yep. I want it for rendering using Unity 5 and as gaming machine second though. 16cores is still 60% more cores compared to 10cores.
The Intel platform will have to be able to run at 6Ghz+ constantly to match the performance, and still with big IFs due to limitations on PCI-E lanes.
 
It kind of is when AMD are offering more cores for less money. For the kind of work these CPU's will be doing pure IPC isn't going to cut it anymore.
8 pack head is always in the clouds a dream world, don't worry.

yes Intel clocks well, cool the Overclockers out their will buy it, it doesn't mean its a good range of CPU's
De-lidding is not perfectly safe & the tool for x299 is not finalized, their was a video just a few days ago stating that.

8 pack cares about Performance and not about money, that is fine for the 2% of Benchers & media that get their chips for free. yes Intel will be faster, doesn't make it better.

In the real world people buy their Chips, Then will they void Warranty, by de-lidding i doubt it.

as for that video he posted, the guy said why all the negative press around x299? maybe because performance/core vs value?

Lot of these people who bench only care about performance that's fine, but the same people can not say what are the negatives about x299 or this not about the money, when they get handed chips to them by Sponsors.

real people pay for their products...
 
Many real people buy Intel though lol. Market share and all that!!!

I am not against AMD, Threadripper is shaping up nicely. Just saying RAW performance and OC is better on X299.
 
Last edited:
Back
Top Bottom