Ryzen "2" ?

[kitfit1]

The fools are the ones that think this is real lol, £100 says the 2600x won't be doing 1823 multi and 250 single in cinebench R15 @ 4.2ghz as this video suggests.

For the 1st time in a long time, i agree with Gavin87. I agree the clock is wrong for a start, to get that result would need very good Phase or LN2. But i agree because i'm of the view the vid is fake anyway.

 

[gavinh87]

Good point. For me the multi thread score is believable it's the single thread score that's surprising people.

Except single threaded score has a direct impact on the multi thread.

This versus an 1800x at stock;
Multi - 33% increase (per core)
Single - 53% increase

They don't add up.

 

[Doobedoo]

Is that including smt and additional cores and threads?

 

[Vince]

Good point. For me the multi thread score is believable it's the single thread score that's surprising people.

It does look high doesn't it. High but perhaps not impossible?! I can't decide either way and can't be bothered to test the maths and see if it is possible. What we need to work out is the ratio on a ryzen single to multicore and see if that score is even possible.

We still have to guess around SMT cores vs physical core though.

Except single threaded score has a direct impact on the multi thread.

This versus an 1800x at stock;
Multi - 33% increase (per core)
Single - 53% increase

They don't add up.

I don't disagree but to add a little balance, is it not possible that a chip can boost a single core higher and then tank somewhat when all cores are loaded, threadripper at stock for example can run single thread at 4.2 but under full load will drop all core back to 3.4? You would expect some difference between a single core and all core on a stock chip, what I am not sure though is if the difference is 20%. I am almost at the point of getting the calculator out and working out the single core vs multi on my chip and working out if that score is feasible.

 

[gavinh87]

Is that including smt and additional cores and threads?

Excluding SMT, the percentage difference would have been the same anyway as I would have just halved the score per core.

 

[gavinh87]

It does look high doesn't it. High but perhaps not impossible?! I can't decide either way and can't be bothered to test the maths and see if it is possible. What we need to work out is the ratio on a ryzen single to multicore and see if that score is even possible.

We still have to guess around SMT cores vs physical core though.



I don't disagree but to add a little balance, is it not possible that a chip can boost a single core higher and then tank somewhat when all cores are loaded, threadripper at stock for example can run single thread at 4.2 but under full load will drop all core back to 3.4? You would expect some difference between a single core and all core on a stock chip, what I am not sure though is if the difference is 20%. I am almost at the point of getting the calculator out and working out the single core vs multi on my chip and working out if that score is feasible.

Do it, then tell me if you want to go ahead and part this fool from his money

 

[Doobedoo]

Excluding SMT, the percentage difference would have been the same anyway as I would have just halved the score per core.

So the maths are potentially wrong. Like I said though if that base frequency is in fact 4.2 then the boost frequency is possibly higher and may also be overclocked higher than that.

Is very hard to do theoretical maths when we have no real idea on the numbers. Yet alone taking the numbers of a different product without knowing potential IPC increases ect

 

[Vince]

Do it, then tell me if you want to go ahead and part this fool from his money

Id still take your money without a second thought Sometimes it's worth backing the outsider.

 

[muon]

Why are people saying there is a single core / multi core discreprancy?

Both CPU-Z and Cinebench show the multi core ratio and they are about right for a 6C/12T chip.

 

[sideways14a]

It would have to have a hefty increase in IPC and some great clocks to pull that kind of single thread score.
I could believe the multi thread score but not the single...

 

[Matkinson]

This looks insane if true.

But then this just makes me wonder (if true), what Zen 2 will be like?!

Not to fuel the hype train but.... AMD are selling off their current stock seemingly as quickly as possible....

(In reality I realise this probably isn't real, just let me enjoy my own hype)

 

[jigger]

TBH,if AMD managed to get near to 5GHZ on a new node it would be an achievement let alone a modified verison of an old one.

If they can another 10% extra clockspeed at stock and some slight IPC improvements that would still be a reasonable midlife update.

Compared to what we've seen in the past 6-7 years a 10-15% performance uplift in a single generation is unheard of. That's maybe 7 generations of Intel chips and half a dozen motherboards.

 

[Yaayuh!]

Expecting/hoping for anywhere near 5GHz is madness. Expect around 5% IPC and 5-10% clockspeed.

 

[jigger]

Sorry if this has been posted already but holy #### if this is true, 250 point's Cinebench single threaded, that's way higher than Coffeelake :O

Interesting. I can see a few members getting fully hyped dumping those Coffee Lake chips. Oh wait it's not an Intel chip.

Surprised this video hasn't got it's own thread though.

 

[sandys]

Looking at the Ryzen mobile stuff that has been on the wen they appear to have fixed the XFR precision boost, so it could perhaps clock much higher in single thread than multi

 

[jigger]

This looks insane if true.

But then this just makes me wonder (if true), what Zen 2 will be like?!

Not to fuel the hype train but.... AMD are selling off their current stock seemingly as quickly as possible....

(In reality I realise this probably isn't real, just let me enjoy my own hype)

Ryzen 2 on 7nm opens up a world of possibilty. Higher core counts, higher cluster counts, 10Tflop APU's, HBM memory and cache, integrated 2.5-100Gb/s networking, PCI-E 4.0 and 5.0, heterogeneous computing, improved coherent fabric, higher clock speeds lower power draw. Those are just the obvious things that spring to mind.

 

[Jono8]

That video only makes sense if AMD have cracked an easy 5.0ghz boost clock on their chips. I mean it is possible, but then if it was that easy, it is very surprising that Intel haven't been able to do it yet. Intel are going to be totally ***** if AMD suddenly bring out 6 and 8 core parts that can boost to 5.0ghz or more on all cores and have better ipc than coffeelake.

To put it into perspective, those scores are higher than a 5.3ghz on all core 8700k.

Although i want to believe, it seems too good to be true.

 

[DarrenM343]

This looks insane if true.

But then this just makes me wonder (if true), what Zen 2 will be like?!

Not to fuel the hype train but.... AMD are selling off their current stock seemingly as quickly as possible....

(In reality I realise this probably isn't real, just let me enjoy my own hype)

Not sure if selling off stock is what doing or just trying to generate more sales during a quarter where they're predicting lower sales and therefore profits.

 

[awaybreaktoday]

AMD have confirmed that the leaked video is true....

 

[SiDeards73]

AMD have confirmed that the leaked video is true....

Is it April 1st already??? Source of your statement?