Rumor: Next gen AMD Epyc to get 64 cores?

[Taf the Ghost]

2 minutes ago, cj09beira said:

i think separating the server and consumer market would reduce profits a bit but might help them optimize the IF for one and the other, as long as they can lower latency and improve the clocks, we will see some good competition, 

7nm ryzen seems to me like a great potential perf increase, i am more worried about ryzen+ though with coffelake out and 8 cores next year, some ipc improvements would be very helpful 

Clocks alone will solve most of the problem. Gaming is about reaching saturation levels. This is why the Skylake-X CPUs can catch up in most games when you get them up to 4.4 Ghz or so. The Inclusive L3 cache + Ring Bus is still easier for game engines to leverage (because they've been leveraging it for years). Though, now we've got to ask if it's the DRM that is actually leveraging that better. (That >20% improvement in ROTR looks a lot more suspicious now.)

 

An 8% clock increase on the 1800X would take XFR from 4.1 Ghz to 4.4 Ghz. All-core at 4.3 Ghz would saturate a 1080 Ti in pretty much everything that the game engines could take advantage of the Zen uArch.

[Taf the Ghost]

2 minutes ago, leadeater said:

Unless the market and narrative shifts by the time these new CPUs launch the 12 core option does make the most sense. AMD finally releases a competitive product stack and within moments large waves of criticism comes in about how they are terrible and a failure because of slightly lower per thread performance. The amount of claims that Zen is useless for gaming is ridiculous.

 

So based on the above I'd say it's within AMD's best interest to focus on improving that single thread performance over adding more cores.

Intel, AMD & Nvidia all pay marketing agencies, and those agencies have been using Bots & paid shill posters a lot over the last few years. So they can push whatever narrative they want, but the Retail market pretty quickly figures out where the good value is.

[leadeater]

3 hours ago, Taf the Ghost said:

Intel, AMD & Nvidia all pay marketing agencies, and those agencies have been using Bots & paid shill posters a lot over the last few years. So they can push whatever narrative they want, but the Retail market pretty quickly figures out where the good value is.

Yep, this is why AMD needs to take on the mind-share battle more now. Ryzen and Threadripped reinforced this as being one of the most significant factors, it showed that again simply having a competitive product is not enough in the market as it stands now.

 

Logical analysis of a products performance and value is put aside and replaced with... I don't know what the hell to call it. Somehow magically the difference between 140 FPS and 125 FPS @1080p for 1440p gamers is a critical issue that warrants declaring all Zen based CPUs as failures and no body should buy them, the figures are made up before anyone brings that up.

[cj09beira]

1 minute ago, leadeater said:

Yep, this is why AMD needs to take on the mind-share battle more now. Ryzen and Threadripped reinforced this as being one of the most significant factors, it showed that again simply having a competitive product is not enough in the market as it stands now.

 

Logical analysis of a products performance and value is put aside and replaced with... I don't know what the hell to call it. Somehow magically the difference between 140 FPS and 125 FPS @1080p for 1440p gamers is a critical issue that warrants declaring all Zen based CPUs as failures an no body should by them, the figures are made up before anyone brings that up.

plus just saw a video on hardware canucks, almost 300fps on CS GO, i would say thats more than enough even for the more hardcore players

[cj09beira]

btw https://videocardz.com/73842/xfx-teases-custom-radeon-rx-vega

[leadeater]

Just now, cj09beira said:

btw https://videocardz.com/73842/xfx-teases-custom-radeon-rx-vega

Kinda off topic though 

[cj09beira]

Just now, leadeater said:

Kinda off topic though 

i will let a mod decide that  

[leadeater]

1 hour ago, wcreek said:

I hate how two things start sooner and sooner each year:

b.) Rumors for the next iteration of something that just came out

Was just thinking that too. EPYC is barely out, we have no idea how good it actually is in a wide variety of different server workloads but we're already talking about some distance future iteration of it.

[Taf the Ghost]

15 minutes ago, leadeater said:

Was just thinking that too. EPYC is barely out, we have no idea how good it actually is in a wide variety of different server workloads but we're already talking about some distance future iteration of it.

Epyc is only getting its first deliveries, haha. Same with Skylake-SP, actually. But AMD should also be taping out the designs right now, so that's probably where the news came from. It also means that AMD had worked up multiple designs and they approved going big. 4 CCXs with 16 mb of L3 cache each is going to be bigger than the Zen package, I'm pretty sure.

 

Part of me still hopes for a 12c mainstream design. I just really want to own a 9c CPU. Not sure why.

[Falconevo]

On 11/3/2017 at 11:58 AM, leadeater said:

Physical cores, however when virtualizing Windows they are shown as virtual cores but the limit is the same. The down side to VMs is that when you configure vCPUs that's 1:1 under Windows to this limit even though the hypervisor isn't necessarily giving the VM that number of real physical cores under the hood.

 

 

I'm under NDA, lets just say I have LOTS of information so far but it isn't worth my job to release it early.

 

What I can say is, per U space the cost vs density is far far greater than Intel currently have on offer.  Performance is right up there too but I can't provide any metrics just yet.  

Anything in particular you would like me to test?  Happy to disclose the findings minus all identifiable information when I can.

[leadeater]

On 11/3/2017 at 2:41 PM, Falconevo said:

I'm under NDA, lets just say I have LOTS of information so far but it isn't worth my job to release it early.

 

What I can say is, per U space the cost vs density is far far greater than Intel currently have on offer.  Performance is right up there too but I can't provide any metrics just yet.  

No problem, as I said anything you can share is great . Always good to get a slightly more unbiased view of things than I can get from say HPE, you don't want to sell me servers lol.

 

On 11/3/2017 at 2:41 PM, Falconevo said:

Anything in particular you would like me to test?  Happy to disclose the findings minus all identifiable information after the 12th of Dec.

Silly question, EVERYTHING of course. Actually seriously though the main things I'm interested in are VM hosting (ESXi), Storage Node performance and scaling and SQL Server performance.

[Humbug]

Is this based on the upcoming zen2 cores with improved ipc?

[Humbug]

5 hours ago, D Levy said:

Because with vulcan and if clock speeds can reach 4Ghz or higher....

it's the perfect CPU for both gaming and blender/editing.

Your performance will hit a wall even with the best optimized Vulkan renderer. It just won't scale to that many cores. There isn't that much work to do in terms of dispatching draw calls. Other aspects become the bottleneck such as physics, or the GPU etc. Maybe somebody can create a synthetic benchmark to show the theoretical possibilities but that's it....

[FloRolf]

Guess I'll not get a 1950x then?

 

Dunno, for me this 'rumor' looks more like a 'lel, we just doubled everything, that's Epyc 2 now!' 

[Trixanity]

4 minutes ago, FloRolf said:

Guess I'll not get a 1950x then?

 

Dunno, for me this 'rumor' looks more like a 'lel, we just doubled everything, that's Epyc 2 now!' 

Scheduled for 2019 so it'll be a long wait.

[ATFink]

I wonder what the TDPs of the next gen EPYC CPUs will be. Seriously, will they sacrifice clockspeed and voltage to reduce power consumption of the additional cores, increase TDP targets, or will the next generation of EPYC see a >50% (or some other significant percentage) improvement in power efficiency and negate much of the potential reduction in voltage/frequency that may occur? Maybe a combination of those possibilities?

[Taf the Ghost]

11 minutes ago, ATFink said:

I wonder what the TDPs of the next gen EPYC CPUs will be. Seriously, will they sacrifice clockspeed and voltage to reduce power consumption of the additional cores, increase TDP targets, or will the next generation of EPYC see a >50% (or some other significant percentage) improvement in power efficiency and negate much of the potential reduction in voltage/frequency that may occur? Maybe a combination of those possibilities?

7nm Zen 2 will run at the same clock speeds at somewhere around 40-50% less power draw. It's actually a node shrink from 14nm -> 10nm -> 7nm, it's just 10nm never happened. So the suggestion is probably a TDP increase, but frequencies will be around the same as current but at lower voltage. It's still going to be thirsty, but 64c per socket is still going to be nuts on "high performance" CPUs.

[Vode]

Damn this is starting to look like a GPU. Lol

[Taf the Ghost]

7 hours ago, Humbug said:

Is this based on the upcoming zen2 cores with improved ipc?

Yes. It'll be the first design-improved Zen cores. We'll probably know the details of the uplift somewhere in Q4 2018. If AMD is aggressive, and yields are good, they might be able to release the Mainstream Ryzen on Zen 2 in late 2018.

 

Oh, that's probably it. @cj09beira , @leadeater . The Big Zen is for the PCIe 4.0 boards. PCIe 4.0 won't come to Desktop until AM5 &/or Tigerlake, but there is already a use for it in the server space. This would let them make Eypc 2 & Ryzen 2 off the same die package, with "Big Eypc" on PCIe 4.0 with 32c, 48c or 64c. Premium products, but if they get a 10-15% IPC uplift, AVX2 and PCIe 4.0, what's Intel's advantage? That's some murderous amounts of performance in a 2U config.

 

If we think about the Roadmap, PCIe 4.0 was just finalized, though it was effectively done back in Spring, but DDR5 isn't until 2020 (if no delays). AMD, because of the new modular design approach, can just drop in whichever controller they need, however the rest of the tech needs to be there. This is why Zen 3 might follow Zen 2 after just one year. With IPC and design improvements. Zen 3 will have DDR5, which should help keep those CPUs really well fed. 

 

Introducing a "Big Zen" design branch let's AMD really get involved in the server space when Intel is going to have 32c designs as max, but the major key is it lets them adopt the Server space technologies faster. While at the same time letting them run Zen 2+ CPUs out if DDR5 gets delayed. Big Zen allows for more flexibility going forward, and it should thusly be assumed that's likely what is happening.

 

I still think the dies are going to be tiny compared to what we expect. The Big Zen is probably going to be a 200-240 mm² die that they can package in the 4-way cluster. Which is why it's going to be messy for Intel. Their next XCC core is probably around 750 mm² on 14nm++. 

[Urishima]

12 hours ago, BingoFishy said:

no

low clockspeed crap for gaming

also cost many monies

Also, lots of games will just shit themselves in panic when confronted with that many threads

[TidaLWaveZ]

AMD x64 x64

[System Error Message]

i think linus would sleep with the CPU if it came out soon

Still a 256 core 512 thread epyc server would be nice. Sparc had something similar before. If it comes out at the same price as the current epyc, it will disrupt the market even more.

[leadeater]

8 minutes ago, System Error Message said:

a 256 core 512 thread epyc server would be nice

Spoiler

 

[System Error Message]

1 minute ago, leadeater said:

Place 4 of those 54 cores into a single server and you'll have 256 cores and so many memory channels. You could have a lot of ram too. Imagine 4x16x64GB DDR4 ram.

[leadeater]

7 minutes ago, System Error Message said:

Place 4 of those 54 cores into a single server and you'll have 256 cores and so many memory channels. You could have a lot of ram too. Imagine 4x16x64GB DDR4 ram.

Isn't it currently 2TB ram per EPYC CPU? 4x2 = 8TB ram

 

Edit:

Updated to from 3TB to 2TB per CPU