More Ryzen 3000 info - 4.5GHZ boost & +10-15% IPC

[Loote]

So when are going to get official info?

[leadeater]

4 minutes ago, Derangel said:

but its a huge change from the current limits of any Ryzen or Threadripper chip.

That may or may not be due to fab limits because architecture limits could be equally as likely.

[Derangel]

Just now, Loote said:

So when are going to get official info?

Hopefully at the end of May during their Computex keynote.

[Derangel]

1 minute ago, leadeater said:

That may or may not be due to fab limits because architecture limits could be equally as likely.

True. Still, with the rumor stating that big of a change I'm going to remain doubtful until AMD says otherwise. There are far too many fake rumors shat out by people looking for their five minutes of fame these days.

[leadeater]

Just now, Derangel said:

True. Still, with the rumor stating that big of a change I'm going to remain doubtful until AMD says otherwise. There are far too many fake rumors shat out by people looking for their five minutes of fame these days.

Yep, just pointing out TSMC 7nm isn't a magic holy grail that will do everything.

[rickymohk]

23 minutes ago, Stefan Payne said:

The Translation looks like the "normal Frequency" is 4,5GHz.

That sentence is saying "The CPU samples are generally 4.5GHz". Nothing related to base clock.

[Stefan Payne]

Just now, rickymohk said:

That sentence is saying "The CPU samples are generally 4.5GHz". Nothing related to base clock.

That's what Google Translate says and "generally" can be interpreted as base clock or that the Samples they've seen are "normally" 4,5GHz.

 

It doesn't say that its maximum clock or boost clock.

[rickymohk]

28 minutes ago, Stefan Payne said:

That's what Google Translate says and "generally" can be interpreted as base clock or that the Samples they've seen are "normally" 4,5GHz.

 

It doesn't say that its maximum clock or boost clock.

I am a Hongkonger and I tell that by myself not by Google Translate. It really does Not mean base clock. Of course it also does not mean it is not base clock either. The article just does not mention which clock it is.

[mr moose]

1 hour ago, leadeater said:

Yep, just pointing out TSMC 7nm isn't a magic holy grail that will do everything.

But I heard all these rumours that 7nm was going to fix the ozone layer so I've already bought an AM4 mobo.

[Doobeedoo]

Well seems about expected more or less for what we can guess too. Potentially 5GHz will be a thing with this series. Really wish to see FX like clocks. 

[yolosnail]

7 hours ago, JoostinOnline said:

You're comparing a third gen to a ninth gen architecture.

That's very generous calling Intel's current lineup 9th generation, considering it's a refresh of 8th gen, which was a refresh of 7th gen, which was itself a refresh of 6th gen. 

Sure a few more cores have been tacked on here and there, but they're essentially the same with a few tweaks

[GoldenLag]

3 hours ago, leadeater said:

Yep, just pointing out TSMC 7nm isn't a magic holy grail that will do everything.

and we are currently so close to launch. speculating frequency to build hype is just dumb. 

 

why speculate when we will most likely have answers next month. 

[cj09beira]

3 hours ago, rickymohk said:

I am a Hongkonger and I tell that by myself not by Google Translate. It really does Not mean base clock. Of course it also does not mean it is not base clock either. The article just does not mention which clock it is.

would it be a fair assumption that it is all core boost? thats what i think it is, that or single core boost, either way we still don't have much to go on

[cj09beira]

5 hours ago, imreloadin said:

I like how everyone was shitting on me for saying I would be happy for a 10% IPC increase saying that was way too low because of the node shrink...everyone was like "Oh yeah 5GHz is totally possible and they'll do it because of AMD's 50th anniversary!"

 

Where you at now @Jack_of_all_Trades @MeatFeastMan and @cj09beira

 

Thread for reference:

 

how about we start that who was right talk when we have solid leaks at the very least 

4 hours ago, Derangel said:

 

Ignoring the incredibly dubious accuracy of Google translate, I have trouble believing that. That would put the base clock higher than the normal boost clock of any current Ryzen or Threadripper CPU and a full 800MHz higher than the highest base clock on current Ryzen processors (700MHz over the best base clock for Threadripper as well). 4.5GHz boost on some of the chips is easy to believe and might even mean 5GHz is possible on watercooling, on some chips. A 4.5GHz base clock would likely mean boost clocks of, at least, 4.8-5.0GHz. I would love that to be true, but its a huge change from the current limits of any Ryzen or Threadripper chip.

my guess is they are talking about the all core boost, which might be around that 4.5 ghz mark, i also expect lower end parts to overclock quite a bit less than higher end ones, 

and btw we cant really look behind to find what to expect expect by looking at Vega 10 vs Vega 20, because the nodes are really different, and 14nm was a really low clocking node, do you remember how people were surprised that zen 1 could not go past 4ghz

[GoldenLag]

3 minutes ago, cj09beira said:

ecause the nodes are really different, and 14nm was a really low clocking node

the only nice thing for AMD is that 14nm was a really high yielding node. like one of the best yielding nodes. 

 

the issue was that it was a LPP node. meaning it was never meant to clock very high. 

[GoldenLag]

4 hours ago, Derangel said:

Hopefully at the end of May during their Computex keynote.

we have confirmation from Gigabyte BIOstar that they will have X570 boards at display there, iirc. 

 

and having boards without CPUs would be kinda wierd. 

 

 

edit:

 

https://www.pcgamesn.com/amd/ryzen-3000-x570-motherboard-biostar-computex-2019?fbclid=IwAR1RnDL9tw-cRJ2g7PovkAiwNkiBmh1C0RzlCP0aEbJQ3SYrnrakGUfDi38

 

sorry it was BIOstar not Gigabyte. 

[Trixanity]

A 200 MHz higher clock speed and 15% increased IPC is still potentially a flat 20% performance increase from one generation to the next. That's ignoring core count increases and the massively increased FP performance (and any potential compounding of bits and pieces). That's not too bad all things considered.

[Dabombinable]

5 minutes ago, Trixanity said:

A 200 MHz higher clock speed and 15% increased IPC is still potentially a flat 20% performance increase from one generation to the next. That's ignoring core count increases and the massively increased FP performance (and any potential compounding of bits and pieces). That's not too bad all things considered.

Especially when you look at how huge an improvement Zen 1 itself was (Zen 2 was getting nearly all of its improvements from better cache and memory compatibility, so more a refresh).

[GoldenLag]

Just now, Dabombinable said:

Especially when you look at how huge an improvement Zen 1 itself was (Zen 2 was getting nearly all of its improvements from better cache and memory compatibility, so more a refresh).

we saw how Zen 1 scaled with better nodes while remaining unchanged with the 2000 series. we got a better memmory controller, but other than that it is the exact same CPU. 

 

looking at how Vega went from 14nm to 7nm gives us an idea of how much clockspeed we could get if they just dropped down the node. 

 

the issue is that Zen 2 isnt a copy of Zen 1. meaning all we can assure is that we will get bonuses from the node shrink, but everything else has to wait untill we get a launch. 

[leadeater]

2 minutes ago, Dabombinable said:

Zen 2 was getting nearly all of its improvements from better cache and memory compatibility, so more a refresh

Zen+?

 

FFS AMD this is all your fault, stupid ass naming conflicts.

[GoldenLag]

Just now, leadeater said:

FFS AMD this is all your fault, stupid ass naming conflicts.

wouldnt be nice of them to call it Zen 2 though. its a carbon copy with a slightly improved memmory controller. like there isnt even a dieshrink. 

 

they went into MS paint and copied the design over. 

 

at least they are skipping Zen 4 from what we know. so they should b back on track with AM5, using Zen 5 on a 5nm node while calling it the r5 5500x

[porina]

10 hours ago, ouroesa said:

Both talk about a 10-15% IPC improvement which I really hope is true, especially for gamers as this is where they have been lacking the most.

I've said it before, I'd be amazed if they did 10% IPC improvement across the board. What I suspect will turn out to be the case is that non-FPU intensive workloads will see moderate gains, of typically low single digit %. Where most of the gain comes from will be the updated FPU, offing twice the potential of current Zen cores and catching up with Intel's implementation. Of course, not everything will make use of that, nor will everything see the full gain. But I suspect, if taking a wide variety of applications, it is that which results in the bigger claimed IPC numbers.

 

If I'm wrong and we do see 10%+ across the board, I will be more than happy, but I'm leaving my hype dial set below 11.

[Dabombinable]

8 minutes ago, porina said:

I've said it before, I'd be amazed if they did 10% IPC improvement across the board. What I suspect will turn out to be the case is that non-FPU intensive workloads will see moderate gains, of typically low single digit %. Where most of the gain comes from will be the updated FPU, offing twice the potential of current Zen cores and catching up with Intel's implementation. Of course, not everything will make use of that, nor will everything see the full gain. But I suspect, if taking a wide variety of applications, it is that which results in the bigger claimed IPC numbers.

 

If I'm wrong and we do see 10%+ across the board, I will be more than happy, but I'm leaving my hype dial set below 11.

The FPU is heavily used in games, it's part of why the FX series did poorly as there were only 1-4 FPU in each CPU, which were further limited by the terrible cache design.

[Jito463]

11 hours ago, ouroesa said:

Both talk about a 10-15% IPC improvement which I really hope is true, especially for gamers as this is where they have been lacking the most.

Not really, IPC for Ryzen is already pretty close to Intel's.  Where AMD has been lacking the most is in clock speeds, though an IPC increase as well wouldn't hurt.

[Princess Luna]

4 minutes ago, Jito463 said:

Where AMD has been lacking the most is in clock speeds, though an IPC increase as well wouldn't hurt. 

Perfectly honest it's not about core frequency, it's about per core latency, Ring Bus still trashes Infinity Fabric.