AMD prepares 12 core 24 Thread 5.1Ghz Mainstream CPU

[DragonQ]

If it sounds too good to be true, it probably is. The huge increase in core counts and clock speeds, whilst retaining the same launch prices and launching in under 8 weeks...all I can say is I remain highly sceptical. Obviously amazing if true but I'll believe it when I see it.

If 12nm can do this, imagine what 7nm might do. I wonder if they built Ryzen to scale with die shrinks really well.

I don't think there's any evidence that GloFo's 7 nm will perform better than their 12 nm process right now. Considering Intel's 10 nm isn't expected to outperform their current 14 nm process until its THIRD iteration suggests it's best not to get your hopes up about 7 nm.

 

[Jono8]

A 5ghz 24 thread cpu for £300 also makes most of the TR line up totally redundant.

 

[AiiR]

8 cores struggling to do 4GHz up to 12 cores that breeze 5GHz in less than a year.

If true, then some folk at AMD are due a massive pay day.

 

[gavinh87]

Load of bs imo
Although might impressive if true.

 

[4K8KW10]

https://www.tweaktown.com/news/60100/amds-new-ryzen-7-2800x-teased-12c-24t-up-5-1ghz/index.html

I don't care about the insane amount of cores, although that is awesome its the clock speeds that i'm interested in, no doubt the IPC will also be a little higher so for the money something like the Ryzen 2600 will be an 8700K killer.

Nice

Excellent indeed.
With this process optimisation from 14nm to 12nm and really cheap real estate for more cores, I guess they went all in and easily gave the green light for this.

Very good that I missed the Ryzen 1700 upgrade time and will happily have a better opportunity coming February. Amazing !

 

[andy_mk3]

It's just not going to happen.

 

[Overlag]

Is it April fools? In December?

 

[muon]

I don't think there's any evidence that GloFo's 7 nm will perform better than their 12 nm process right now. Considering Intel's 10 nm isn't expected to outperform their current 14 nm process until its THIRD iteration suggests it's best not to get your hopes up about 7 nm.

I haven't seen evidence GloFo are struggling. Intel have been struggling with a super dense version of 10nm for ages. I don't see why those two have to correlate.

I don't think you'd be saying that if GloFo/TSMC were the ones struggling.

Intel were possibly trying something when the technology or knowledge wasn't there. Their 10nm++ will coincide with 7nm with everyone else timewise.

 

[DragonQ]

A 5ghz 24 thread cpu for £300 also makes most of the TR line up totally redundant.

Yep, it makes zero sense. 8c/16t at 4.4-4.5 GHz is my expectation.

 

[DragonQ]

I haven't seen evidence GloFo are struggling. Intel have been struggling with a super dense version of 10nm for ages. I don't see why those two have to correlate.

If I said I had a number between 1 and 10 in my mind, you wouldn't say "well there's no evidence it's a low number so it's probably a high number". There's no direct evidence either way, all we have is Intel's similar process which has run into huge delays and performance issues. They don't have to correlate but they obviously might. Maybe 7 nm will be great but we don't know yet.

I don't think you'd be saying that if GloFo/TSMC were the ones struggling.

Eh. All I know is Intel's first 14 nm iteration was better than GloFo's and CPUs based on it were released 18 months before Ryzen. For GloFo to leapfrog so far ahead requires something special. It's not impossible but certainly hard to believe right now.

 

[CAT-THE-FIFTH]

I really hope people have learnt better after the Polaris overhype. Everything pointed it to being more like R9 390 level and people believed WCCFTech with its "beast mode" rumours and then dissapointed themselves when the card ended up being R9 390 level.

GF 12NM is apparently a tweaked version of GF 14NM,so I would expect it more like a few percent increase in IPC and perhaps upto 10% more clockspeed - anymore and that would be quite large for a refresh. Personally I really hope,they can improve the bandwidth between the pair of CCX modules,which is probably the biggest issue IMHO.

 

[SiDeards73]

Highly doubtful they will increase core count AND frequency at the same price as previous tiers, especially as that 12/24 rumoured chip would beast their Threadripper.

I expect as many do 8/16 still on a refresh to hit a base clock of 4.1 or so and OC to around 4.6, and to be perfectly honest that's all they need to hit imho, that extra .5 or .6 GHz will make a world of difference.

 

[CAT-THE-FIFTH]

If it sounds too good to be true, it probably is. The huge increase in core counts and clock speeds, whilst retaining the same launch prices and launching in under 8 weeks...all I can say is I remain highly sceptical. Obviously amazing if true but I'll believe it when I see it.


I don't think there's any evidence that GloFo's 7 nm will perform better than their 12 nm process right now. Considering Intel's 10 nm isn't expected to outperform their current 14 nm process until its THIRD iteration suggests it's best not to get your hopes up about 7 nm.

IIRC,7NM actually might have some help from IBM and Samsung:

https://www.globalfoundries.com/new...ver-leading-performance-7nm-finfet-technology
https://www.hpcwire.com/2017/06/13/globalfoundries-7nm-chips-coming-2018-euv-2019/
https://www.anandtech.com/show/1155...nm-plans-three-generations-700-mm-hvm-in-2018
https://www-03.ibm.com/press/us/en/pressrelease/52531.wss

From the second link:

The 7nm process technology heads for prime time just two years after it was introduced by the IBM Research alliance which includes GlobalFoundries and Samsung. The original proof of concept chip was manufactured with extreme ultraviolet lithography (EUV), but initial products will go forward using optical lithography. This probably won’t be a surprise to those familiar with EUV’s uphill climb toward commercial viability.

From the 4th link:

IBM (NYSE: IBM), its Research Alliance partners GLOBALFOUNDRIES and Samsung, and equipment suppliers have developed an industry-first process to build silicon nanosheet transistors that will enable 5 nanometer (nm) chips. The details of the process will be presented at the 2017 Symposia on VLSI Technology and Circuits conference in Kyoto, Japan. In less than two years since developing a 7nm test node chip with 20 billion transistors, scientists have paved the way for 30 billion switches on a fingernail-sized chip.

So its an IBM,Samsung and Global Foundries collaboration.

 

[Doobedoo]

Yep, it makes zero sense. 8c/16t at 4.4-4.5 GHz is my expectation.

This is what I am expecting. The slide is what I would expect from 7nm Ryzen 2.

 

[4K8KW10]

This is what I am expecting. The slide is what I would expect from 7nm Ryzen 2.

This may turn out to an impossible shocker for quite many people. Especially, all the people who got used to the Intel domination and lack of innovation in all these recent years

 

[Doobedoo]

This may turn out to an impossible shocker for quite many people. Especially, all the people who got used to the Intel domination and lack of innovation in all these recent years

I'd rather wait to be honest. I'll likely be getting zen+ regardless so anything over my expectations will be a bonus, but I'd rather not give people a reason to bash.

 

[Martini1991]

I'd be very skeptical.
But if that's the successor to Ryzen then it's launch day buy for me.

Intel recovered from Ryzen quite well (IMO) with Coffeelake because of Intels IPC and clock speed domination. But AMD offering same clock speeds, slightly reduced IPC but double the cores? Then AMD will have as closed the gap and reversed it as much as possible.

I'd personally expect the same core count with higher clocks and IPC in my own conservative expectations. But even that can add about 25% performance.

 

[Martini1991]

Why is this only gaining traction now? That slide has been doing the rounds for months now.

Probably fake then.

 

[nzaptx]

I Don't expect much more than +300mhz on the clock of each model, at least for the refresh.

 

[4Qman]

This is most likely BS.

However, I had the same thoughts before Ryzen and TR was released and they turned out true.