Ryzen 2 slide in KitGuruTech video

[William Gaatjes]

I've heard this for ages and yet nothing still changed.

Have we not gotten a jump to 8cores /16 threads for a bargain price the last 12 months ?
It takes some time but we are getting there.

 

[Yotsugi]

Have we not gotten a jump to 8cores /16 threads for a bargain price the last 12 months ?
It takes some time but we are getting there.

That's very far and away from the actual manycore processing.

 

[William Gaatjes]

That's very far and away from the actual manycore processing.

It is a start.
a many core scalar cpu for consumer use is not going to happen just like that.
Companies need incentive.
Either government stimulating the development or waiting for market forces.

 

[Yotsugi]

It is a start.
a many core scalar cpu for consumer use is not going to happen just like that.
Companies need incentive.
Either government stimulating the development or waiting for market forces.

As of now most consumer tasks are not intended to scale to manycore numbers.
We'll see anyway.

 

[CatMerc]

Again, the "leaks" are coming whenever AMD wants them to.

At a certain point, controlling them is next to impossible. That point is usually when MB vendors have samples.

Early leaks are guerilla marketing, but there are certainly points when you can assume it's an actual leak.

 

[Mulrian]

Seems surprising that mobo manufacturers wouldn't have samples by now considering launch is only a couple months away. Pretty sure we had plenty more info at the same point last year with Ryzen.

 

[AtenRa]

Well to be honest, I find it extremely difficult for those specs to be true.

50% more cores + 22% higher Turbo clocks on the 12nm LP , unless those SKUs are at 125W TDP.

 

[Yotsugi]

but there are certainly points when you can assume it's an actual leak.

CPC Hardware getting an A0 Summit Ridge silicon was an actual leak.
The rest are guerilla marketing.

unless those SKUs are at 125W TDP.

Great, someone is finally starting to get it!

 

[CatMerc]

CPC Hardware getting an A0 Summit Ridge silicon was an actual leak.
The rest are guerilla marketing.

Great, someone is finally starting to get it!

You're assuming AM4's specifications are built to handle 125W TDP. While some of the higher end boards definitely could and then some, a lot of B350 boards or god forbid A320 boards are already stretching it with the 95W Summit Ridge.

 

[Yotsugi]

You're assuming AM4's specifications are built to handle 125W TDP. While some of the higher end boards definitely could and then some, a lot of B350 boards or god forbid A320 boards are already stretching it with the 95W Summit Ridge.

Even the shittiest AM3+ boards managed to handle 125W SKUs.
Though PR was 65-95W on the roadmaps.
We'll see how it goes.

 

[CatMerc]

Even the shittiest AM3+ boards managed to handle 125W SKUs.
Though PR was 65-95W on the roadmaps.
We'll see how it goes.

Being an older spec doesn't say much. Requirements for power delivery were stringent enough so that any AM3+ board could handle this much current. We can't say the same about AM4 unless someone here gets an internal document that details AM4's design requirements.

 

[Yotsugi]

We can't say the same about AM4 unless someone here gets an internal document that details AM4's design requirements.

Well we should end the discussion here.

 

[The Stilt]

You're assuming AM4's specifications are built to handle 125W TDP. While some of the higher end boards definitely could and then some, a lot of B350 boards or god forbid A320 boards are already stretching it with the 95W Summit Ridge.

The motherboards are built to support the infra variants available at the time of the design.
With Summit Ridge the available infra variants were 65W & 95W. Some motherboards will obviously have margins to support significantly higher power draws, but essentially nothing beyond 95W can be guaranteed on all boards.
If there will be a higher power (> 95W) infra variant for Pinnacle Ridge, then the official support for that variant will be introduced in 400-series PCH motherboards.

 

[DrMrLordX]

It would be fascinating to see what Intel / AMD could do with a high-performance RISC-V implementation.

There was that whole ARM thing they were looking to do. What was that, K12? Yeah, that was it. K12. Kind of MIA right now isn't it? Anyway they might have toyed with the idea of RISC-V but given what happened with K12, I doubt they'll put money down on RISC-V either.

Even the shittiest AM3+ boards managed to handle 125W SKUs.

Some of those 760G boards were pretty sketchy . . .

 

[Yotsugi]

There was that whole ARM thing they were looking to do. What was that, K12? Yeah, that was it. K12. Kind of MIA right now isn't it? Anyway they might have toyed with the idea of RISC-V but given what happened with K12, I doubt they'll put money down on RISC-V either.

K12 is a server first ARM core, just like Zen, and you know about the state of ARM servers right now.

Some of those 760G boards were pretty sketchy . . .

Some of the B350 boards are also pretty sketchy.

 

[f2bnp]

Some of those 760G boards were pretty sketchy . . .

You can say that again. Had a friend with a Phenom II x4 125W on a cheapo MSI 760G motherboard and a couple of VRMs (no cooling on them lol) just blew up one day. Where did this talk of 125W AM4 CPUs start from anyway?

 

[dark zero]

You're assuming AM4's specifications are built to handle 125W TDP. While some of the higher end boards definitely could and then some, a lot of B350 boards or god forbid A320 boards are already stretching it with the 95W Summit Ridge.

I feel that AM4 can support up to 225 watts, like the FX, but only on the Z tier.

 

[IRobot23]

I feel that AM4 can support up to 225 watts, like the FX, but only on the Z tier.

You mean X tier?

Not really, some X370 have poor power delivery and even 95W TDP is really a lot. Test are mostly done in open case etc. Real world scenario can be much worse.

Yet some X370 can do much more than 225W TDP CPU.

 

[Yotsugi]

Not really, some X370 have poor power delivery and even 95W TDP is really a lot.

Yeah, MSI boards are trash, we know it.

 

[DrMrLordX]

you know about the state of ARM servers right now.

Looking stronger every day, actually. At this point it's a matter of adoption to drive the software ecosystem. Though in terms of market share, they aren't there yet . . but ThunderX2 and um whatever is that Qualcomm chip, actually look fairly potent.

 

[Yotsugi]

Looking stronger every day, actually. At this point it's a matter of adoption to drive the software ecosystem. Though in terms of market share, they aren't there yet . . but ThunderX2 and um whatever is that Qualcomm chip, actually look fairly potent.

With x86 competition being back, it will be really, really hard for ARM to enter the market.

 

[DrMrLordX]

We will see. I would not count out ARM server cores just yet. Cavium is in their second product iteration phase - something Calxeda never managed. And they just got a $6 billion buyout offer so obviously there is market interest in the concept, if not the product.

Not sure how well-received Qualcomm's chip will be, but Qualcomm is big; powerful; and vicious.

 

[Chicken76]

@OP
So 5.1 GHz in a couple of months, eh?
If you're going to fake a slide like that, why won't you go straight to 10 GHz, for that "1st in the world to reach 10 GHz" wow factor?

 

[eek2121]

@OP
So 5.1 GHz in a couple of months, eh?
If you're going to fake a slide like that, why won't you go straight to 10 GHz, for that "1st in the world to reach 10 GHz" wow factor?

The OP did not fake the slide, he/she found it on the internet. I found that same slide myself a couple months ago and dismissed it as a fake. However, it's worth mentioning that 5 GHz is perfectly viable for a next-gen Ryzen chip. Overclockers have hit that now with the 7900x. If AMD moves to a performance optimized node for their refresh, I can definitely see them pushing out a 5 ghz chip.

 

[scannall]

The OP did not fake the slide, he/she found it on the internet. I found that same slide myself a couple months ago and dismissed it as a fake. However, it's worth mentioning that 5 GHz is perfectly viable for a next-gen Ryzen chip. Overclockers have hit that now with the 7900x. If AMD moves to a performance optimized node for their refresh, I can definitely see them pushing out a 5 ghz chip.

I don't see that happening on the 12nm node. On the 7? You bet.