Speculation: Ryzen 3000 series

[Markfw]

This is what they should do to the X570. I still have one of these operating, the best motherboard design EVER (well except maybe the last couple of years there are some, but certainly in its time)

https://www.gigabyte.com/us/Motherboard/GA-P35-DQ6-rev-10#ov

 

[Ajay]

A respin might not only be clock speeds, didn't the zepplin b0 fix the segfault that ryzen1000 had?
I agree about the clock speed optimism though. 'If' a part can hit a single core 5ghz it doesn't mean getting 16 or even 8 cores to 5ghz is easy. It's very likely to be hot and very power hungry at best. I don't think there's been a specific leak denying that the process can hit 5ghz, it's just that it might not be easy for everyone.

Yeah, there’s no way to know what as done in the respin - though they are more likely focused on fixing bugs (especially for Rome) - that doesn’t preclude AMD from resolving clock limiting circuitry.

I think a single core turbo of 5GHz is wildly optimistic for AMD's first 7nm chiplet. It took Intel years to hit 5 GHz on a highly refined process and a highly refined core. It’s not magic, it’s hard work, a lot of manpower and a lot of sim hours as well.

 

[Hans Gruber]

Intel excactly shows why this cadence planning and idea of execution is something from hell. You need flexibility so you can adapt quick. That's what gets you in control. Adaptability.
People say Intel execution was bad. Well the investor and marketing oriented yearly tick tock planning was the foundation for it.

What happens if you dont adapt and is 2 month late? Then people start to whine blame spread fear what not. All because of something that starts with some unrealistic investor circus show. Labelled shareholder value.

The consumer dictates the market based on price. This is why Intel is not well liked by consumers. They need not fear being the most hated of companies. Comcast has them beat in spades. The only thing the millennials are good for is using social networking to destroy brands.

This is what they should do to the X570. I still have one of these operating, the best motherboard design EVER (well except maybe the last couple of years there are some, but certainly in its time)

https://www.gigabyte.com/us/Motherboard/GA-P35-DQ6-rev-10#ov

I have that exact motherboard in my garage in a box still. When you and others mentioned heatpipes for the chipset. That is the board I thought of.

 

[B-Riz]

The consumer dictates the market based on price. This is why Intel is not well liked by consumers. They need not fear being the most hated of companies. Comcast has them beat in spades. The only thing the millennials are good for is using social networking to destroy brands.

I have that exact motherboard in my garage in a box still. When you and others mentioned heatpipes for the chipset. That is the board I thought of.

Nostalgia build time! Get that bad boy goin'!

 

[Fanatical Meat]

Nostalgia build time! Get that bad boy goin'!

Maybe it can be modded to fit a new X570?

 

[Ajay]

The only thing the millennials are good for is using social networking to destroy brands.

Geez dude, slow your roll!

 

[Asterox]

For memory bound workloads, 2950x will be faster due to quad channel RAM as you say but not all loads are memory bound. anything compute bound should be faster on ryzen 3000 16-core, if that part ever gets released. Rumors are only 12-core part at launch.

This is expected, for now or in 2019 there no big rush for 16/32 AM4 CPU.Regardless of expectations, AMD will probably launch green version until the end of year 2019.

 

[realibrad]

Yeah, I’m thinking maybe the chipset fan isn’t a big deal. Fans seem to work pretty good today, also could be a fun nerd project changing or modifying the fan.

**Ive never owned a PC with a chipset fan so my opinion may change if the fan sucks**

If its only needed for multiple m.2 drives, then I probably wont care about the fan. If it dies, then it wont matter in terms of perf. If someone makes a board without one, I would prefer that, but, its not going to stop me from getting the platform.

 

[DrMrLordX]

I don't know if you made a typo there, but intel certainly hasn't "launched 7nm in a timely fashion".

Sorry if I was not clear. I'm still holding out the possibility that they can do it. They have a lot of talent left, and a lot of money. Going to EUV should get them around most of the hurdles that they failed to clear on 10nm. So I give them credit for being able to launch it on time . . . not credit for actually having done it already.

We'll see if I'm being overly-optimistic. AMD can't afford to assume that they'll screw it up again.

Reports are meaningless. AMD said 2019 first, then mid-year, and now Q3.

Reports reflect what most tech writers see when AMD says, "Mid 2019". I interpreted AMD's statements the same way.

 

[Fanatical Meat]

Sorry if I was not clear. I'm still holding out the possibility that they can do it. They have a lot of talent left, and a lot of money. Going to EUV should get them around most of the hurdles that they failed to clear on 10nm. So I give them credit for being able to launch it on time . . . not credit for actually having done it already.

We'll see if I'm being overly-optimistic. AMD can't afford to assume that they'll screw it up again.

I don’t want to sound like an expert because I am not.
Isn’t intels 10nm more “dense” than AMD’s 7nm? Remember reading something like Intel 10nm had like 30% more circuitry stuff than AMD’s 7nm.
Strong possibility that AMD beats intel around for a year or a few quarters then intel magically pulls out a miracle. Intel has done this before.
I’m glad there is some competition and what appears to be forward progress again regarding CPUs

 

[dnavas]

Sorry if I was not clear. I'm still holding out the possibility that they can do it. They have a lot of talent left, and a lot of money. Going to EUV should get them around most of the hurdles that they failed to clear on 10nm. So I give them credit for being able to launch it on time

Yeah, I hear Intel saying 2021 for 7nm, I hear TSMC saying mass production of 5nm in 2021 ... and I'm thinking 2022 in both cases :>

AMD can't afford to assume that they'll screw it up again.

I don't get the sense that they are resting on their laurels, but I agree for sure. They're also going to want to manage product launches for maturity availability, because AMD can't afford to target a node that won't be there and spin up an emergency filler product like Intel has been doing for ... how long now? No money to burn!

 

[Shivansps]

This is what they should do to the X570. I still have one of these operating, the best motherboard design EVER (well except maybe the last couple of years there are some, but certainly in its time)

https://www.gigabyte.com/us/Motherboard/GA-P35-DQ6-rev-10#ov

That would not work, these heatpipes designs are for cooling the NB and some of the VRMs, not the SB, the SB there does not gain anything... the problem is simple, the distance that you can transfer heat passively is very limited.

It would be a far better option to give up one x16 slot and use big heat sink, multi gpu setup is not that common howday anyway.

They could also place the SB on top of the x16 slot... but i belive that would means they have to place the SATAs on the middle of the board.

 

[Markfw]

That would not work, these heatpipes designs are for cooling the NB and some of the VRMs, not the SB, the SB there does not gain anything... the problem is simple, the distance that you can transfer heat passively is very limited.

It would be a far better option to give up one x16 slot and use big heat sink, multi gpu setup is not that common howday anyway.

They could also place the SB on top of the x16 slot... but i belive that would means they have to place the SATAs on the middle of the board.

Just looking at that motherboard, and the X470 I have the NB/SB seems to be about the same distance away as the DQ6. The farthest away is between the 2 PCIE 16 slots.

What am I missing here ?

 

[Markfw]

That would not work, these heatpipes designs are for cooling the NB and some of the VRMs, not the SB, the SB there does not gain anything... the problem is simple, the distance that you can transfer heat passively is very limited.

It would be a far better option to give up one x16 slot and use big heat sink, multi gpu setup is not that common howday anyway.

They could also place the SB on top of the x16 slot... but i belive that would means they have to place the SATAs on the middle of the board.

Just looking at that motherboard, and the X470 I have the NB/SB seems to be about the same distance away as the DQ6. The farthest away is between the 2 PCIE 16 slots.

What am I missing here ?

 

[Veradun]

Reports reflect what most tech writers see when AMD says, "Mid 2019". I interpreted AMD's statements the same way.

Well, it was clear, it's not about someone's interpretation.

Mid 2019 means a neighborhood of '2019-06-30 23:59:59'.

The point is you usually use year, then refine to half then again to quarter. This is because the closer you get to release the clearer the picture gets on their side. Since there is no "half" that comprises Q2+Q3 they said mid 2019, then refining to Q3.

Pretty straightforward.

 

[Shivansps]

Just looking at that motherboard, and the X470 I have the NB/SB seems to be about the same distance away as the DQ6. The farthest away is between the 2 PCIE 16 slots.

What am I missing here ?

These older designs are intended to cool the NB and VRMs, remember that VRM/NB heat sinks are really close to each other. On X470/X570 the SB and VRMs are just too far away from each other, so you dont have were to connect the chipset heat pipes to. They would have to place a heat sink in the middle that may get in the way of CPU coolers, not to mention it is a area reserved for the M2 slot.

There are ways to fix this, for example a really big low profile heat sink like in the older Nvidia SLI chipsets for AM2 but it looks to me OEMs did not wanted to spend time and money OR they were caught whiout warning.

 

[beginner99]

OR they were caught whiout warning.

Entirely possibel since there were also rumors launch was delayed due to the chipset. Maybe heat was the issue and the delay was adding the fans to mobo designs.

 

[krumme]

Entirely possibel since there were also rumors launch was delayed due to the chipset. Maybe heat was the issue and the delay was adding the fans to mobo designs.

Much is possible but the likelyhood of an extra metal layer spin seems a bit more reasonably than eg a sudden need of active cooling of a chipset. Boring and not news worthy because well it's pretty standard.

No need to make drama here. Let's wait for the latency numbers to flow in before game dooming can start

 

[PotatoWithEarsOnSide]

Well, it was clear, it's not about someone's interpretation.

Mid 2019 means a neighborhood of '2019-06-30 23:59:59'.

The point is you usually use year, then refine to half then again to quarter. This is because the closer you get to release the clearer the picture gets on their side. Since there is no "half" that comprises Q2+Q3 they said mid 2019, then refining to Q3.

Pretty straightforward.

Too much logic and common sense for a online forum. Please re-evaluate your lifestyle choices.

 

[amd6502]

Ryzen 3200U/Athlon 300U and Ryzen Embedded R1606G/R1505G are indeed Raven2;

https://imgur.com/dQ8rEs9

Compared to Raven;

https://www.reddit.com/r/Amd/comments/927h5q

My opinion;
- Two years delayed?
- Expensive compared to what it was suppose to replace.
- Only a year away from Dali.

Sorry, this is a month old but I managed to miss this post. This is indeed a square die of half the size of RR/Picasso. Wow, that does mean goodbye Bristol Ridge. These should be ideal in mainstream sub $400 laptops. Would love to see more specs on these, including whether they are closer related to RR or Picasso.

 

[jpiniero]

https://ranker.sisoftware.co.uk/sho...d5e2daead2e6d1f785b888aecbae93a385f6cbf3&l=en

A Rome sample - 32 cores with a pretty low 1.7 Ghz base clock and 2.4 turbo. The Internets claim this is a Qualification sample so the clocks should be final but could be a lowerish TDP.

 

[IEC]

https://ranker.sisoftware.co.uk/sho...d5e2daead2e6d1f785b888aecbae93a385f6cbf3&l=en

A Rome sample - 32 cores with a pretty low 1.7 Ghz base clock and 2.4 turbo. The Internets claim this is a Qualification sample so the clocks should be final but could be a lowerish TDP.

For comparison, the 14nm Epyc 32c/64t parts:
Epyc 7501: 2.0GHz base/3.0GHz boost/2.6GHz ACT in a 155W/170W selectable TDP
Epyc 7551: 2.0GHz base/3.0GHz boost/2.55GHz ACT in 180W TDP
Epyc 7601: 2.2GHz base/3.2GHz boost/2.7GHz ACT in 180W TDP

Those clocks are significantly less than 1st generation Epyc parts. It must be a very low TDP part, or AMD is sandbagging hard.

 

[inf64]

Rome will not have lower clock speeds than 1st gen Epyc parts.

 

[realibrad]

Rome will not have lower clock speeds than 1st gen Epyc parts.

Why is that? I have seen discussions about clock regression on this new process.

 

[krumme]

Why is that? I have seen discussions about clock regression on this new process.

Vs 14lpp? I dont think so and havnt seen it - where?
That said there must be some limitations, one way or another, how fast they can drive a twice as fat fpu.