• 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 Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

https://www.reddit.com/r/Amd/comments/ahxxpg/der_8auer_thinks_5_ghz_on_ryzen_3000_is_very/
In his Q&A (ger) live stream 1:04:50 der8auer told that he thinks 5 ghz on ryzen 3000 is very very realistic. He claims that he got industry sources.

IIRC a 2700X can reach 5.5-6Ghz on LN2 so 5Ghz for R3 seems like an almost dead cert, i guess he could be talking about an OC on less exotic cooling but seeing as he's an 'extreme' overclocker that seems unlikely.
 

IIRC a 2700X can reach 5.5-6Ghz on LN2 so 5Ghz for R3 seems like an almost dead cert, i guess he could be talking about an OC on less exotic cooling but seeing as he's an 'extreme' overclocker that seems unlikely.

I keep changing my mind but I'm expecting 4.7-8 maybe 4.9 from the fastest choices at best, For me the big question is will we still have the same situation where there's no silicon lottery to speak of?
So far we can all pretty much get the same clocks with the right cooling which is good if you're not into overclocking like me but also a bit sad. :(
 
If they can bin chips that can reach 5 GHz, they will. Their marketing people will understand the allure of the "magic 5 GHz mark", even though in reality it means nothing.
 
IIRC a 2700X can reach 5.5-6Ghz on LN2 so 5Ghz for R3 seems like an almost dead cert, i guess he could be talking about an OC on less exotic cooling but seeing as he's an 'extreme' overclocker that seems unlikely.

If Ryzen 2000 can reach 5.5 - 6GHz using LN2, why would Ryzen 3000 be limited to only 5GHz using LN2? Surely he means 5GHz on air?

That's not to say I expect 5GHz on air myself, but that's what I think he's expecting.
 
I keep changing my mind but I'm expecting 4.7-8 maybe 4.9 from the fastest choices at best, For me the big question is will we still have the same situation where there's no silicon lottery to speak of?
So far we can all pretty much get the same clocks with the right cooling which is good if you're not into overclocking like me but also a bit sad. :(

I hate it when it's a silicon lottery. I never end up with the best and some people buy 5 or 10 processors, keep the best overclocker and send the rest back for a refund or sell them on eBay. It sucks when you end up with somebody's reject.

I like to overclock but I don't want to feel cheated.
 
Does the market department really have that much influence over the final clock speeds?
Many things are defined by marketing. They used to have more power than currently as specs have become more important again, but from Intel's marketing perspective they'll be screeching for more cores and the engineer's are likely scratching their heads saying, not possible with current monolithic tech...

Just look a the 9900K and it's power draw problems as well as the issues with heat evacuation in normal conditions (the current solder method and thicker silicon is an advantage in XOC conditions, just not normal conditions). Marketing would have been yelling for something to compete on cores with the 2700X. Even though it sacrificed power and needed solder to do...
 
With all the sensors on chip to verify operation these chips will be expertly binned, overclocking will only be for those willing to pump massive volts and extreme cooling etc, there will be next to nowt for free from a modern chip, previous two gens were the same.
 
With all the sensors on chip to verify operation these chips will be expertly binned, overclocking will only be for those willing to pump massive volts and extreme cooling etc, there will be next to nowt for free from a modern chip, previous two gens were the same.

The two previous gens were not on exactly stellar processes though, and it was at its limits in order to try and compete at the high end of the market. Depending on how good the 7nm process is and how good the yield is on each wafer, you could end up in a position where they need to bin down some of the silcon to fulfil demand at the low end of the market, making a part that can be overclocked not unlikely. Alas, as with everything we have both said, it is merely speculative and won;t be known until we have final silicon in end user hands.
 
If Ryzen 2000 can reach 5.5 - 6GHz using LN2, why would Ryzen 3000 be limited to only 5GHz using LN2? Surely he means 5GHz on air?

That's not to say I expect 5GHz on air myself, but that's what I think he's expecting.

I don't speak German but from what others have said he didn't say it would be limited to 5Ghz, he said 5Ghz is very very realistic.

There are reasons why a reduction in node size would reduce clockspeeds but from what i understand they don't really apply to R3 or TSMC's first foray into 7nm because AFAIK they're kicking of with a performance orientated design instead of a low power one like the did with 14nm.
 
There are reasons why a reduction in node size would reduce clockspeeds but from what i understand they don't really apply to R3 or TSMC's first foray into 7nm because AFAIK they're kicking of with a performance orientated design instead of a low power one like the did with 14nm.
Zen hasn't been any super low power design any time.
It's GloFo's manufacturing node which has been holding it down.
Because of their inability to develop working node themselves, GloFo bought Samsung's 14nm mobile-SOC node!
That's why many people were originally expecting some max ~3,5GHz clocks for Ryzen.
 
If i can get 5ghz out of the box on 2-3 cores on a decent AIO that'll do me, as that kinda speed should put it beyond the 9900k anyhow, i like AMD's XFR and PBO stuff, and the fact if the chip has the thermal headroom it will push itself. They done well putting the amount of sensors in for this type of thing, i understand people love manually overclocking, but im more than happy to just drop a chip in, enable a couple of settings and letting the chip find its sweetspot :)
 
Just tried getting into PS3 emulation (All legal) to run my PS3 games at higher resolutions.
Zen's nowhere near close to Intel in it still :(.

Hope Zen 2's much better.

At the moment I'm not even going to really try it, because the first game I wanted to try (God Of War 3)'s not really playable on my 2700 with it. Guess I'll have to play it on the PS3.
 
The two previous gens were not on exactly stellar processes though, and it was at its limits in order to try and compete at the high end of the market. Depending on how good the 7nm process is and how good the yield is on each wafer, you could end up in a position where they need to bin down some of the silcon to fulfil demand at the low end of the market, making a part that can be overclocked not unlikely. Alas, as with everything we have both said, it is merely speculative and won;t be known until we have final silicon in end user hands.

Good die will be TR/Epyc bound, the beauty of this single design throughout entire product family is you can use the best for the place that gets the best margins and sell the rest to the plebs. :)
 
Good die will be TR/Epyc bound, the beauty of this single design throughout entire product family is you can use the best for the place that gets the best margins and sell the rest to the plebs. :)

Or more specifically, the beauty of this design is there's no block and white answer to what is a "good die". A full 8 core chiplet that sips power will be a good die for the 64 core EPYC, but a full 8 core chiplet that clocks at 4.8GHz is too good for EPYC and is better suited for a 3600X or 3800X.

And this design actually leaves Threadripper potentially in a weird place in that it won't need the best dies this time around. For example, consider a theoretical 48 core 3990WX that boosts to 4.4GHz. That's a beastly package right there, but wouldn't actually need the best dies; a chiplet with only 6 working cores only capable of boosting reliably to 4.4GHz could be considered mediocre individually, but strap 8 of them together and you have a monster CPU.
 
Back
Top Bottom