Piledriver-based Vishera, Next AMD FX

[wuttz]

Originally Posted by kalelovil

It isn't helped by the fact AMD was claiming "up to 56% faster" graphics performance than Llano for Trinity, when the average has turned out to be around 20%.

so, you twist what was for a single benchmark into an "average" ???

-OMG- why?

The chipmaker also presented a few interesting if vague metrics for the mainstream and desktop versions. According to the slide we were shown, "productivity performance," i.e. anything to do with the CPU, is 29 percent higher on Trinity than on its predecessor, while "visual performance" is up by 56 percent. These results are based on Futuremark’s PCMark Vantage and 3DMark Vantage, respectively. For the notebook platform, AMD claims a "resting battery life" of more than 12 hours.

> toms @cebit2012

[wuttz]

Originally Posted by TESKATLIPOKA

kalelovil in desktop it will be much closer, the culprit is turbo because while it works nonstop in 3dmark where the 56% comes from the games are a different story.

Originally Posted by nvidiot

Since when did AMD's marketing department gives accurate estimates for future processors? AMD's marketing is all about hyping and very, very, very best case scenarios.

ribeyesteaknrice

[Opteron]

Ehh guys ... why are you discussing 3DMark and Trinity stuff in the Vishera / FX thread?

[Skywalker]

pardon me if this has been answered. Does anyone have any idea which BD iteration will finally have a uOP cache in the middle of the long pipeline? This, I believe is one big fix to make the CMT arch shine.

[Opteron]

Originally Posted by Skywalker

pardon me if this has been answered. Does anyone have any idea which BD iteration will finally have a uOP cache in the middle of the long pipeline? This, I believe is one big fix to make the CMT arch shine.

There is no official information, however AMD has some patents for it and Steamroller will be a bigger architecture update.
We can only hope that the µOp-cache will be part of it.

[esrever]

anyone know if steamroller will be AM3+?

[Seronx]

Originally Posted by esrever

anyone know if steamroller will be AM3+?

It won't it might have the same amount of pins but it won't be the same socket.

[kalelovil]

Originally Posted by esrever

anyone know if steamroller will be AM3+?

AMD have only announced APU versions of steamroller so far, so no.

Things still seem to be a bit up in the air though. It depends on when AMD manage to develop a PCI-E 3.0 controller, and whether they attempt to combat Intel again in the high-end desktop/workstation CPU market.

[265586888]

Up until some point, the new CPU will have to include PCI-E 3, and then this is when there is a complete change of socket pin-out for AMD enthusiast CPUs. And then Steamroller is 2013...

[Opteron]

Originally Posted by 265586888

Up until some point, the new CPU will have to include PCI-E 3, and then this is when there is a complete change of socket pin-out for AMD enthusiast CPUs. And then Steamroller is 2013...

Yes but there is a slight chance.. :

Our high-end 2012 desktop CPU will be “Vishera” which features up to 8 “Piledriver” cores and replaces “Komodo” from our previously communicated 2012 public roadmap. “Vishera” will also carry over into 2013 but with some enhancements over the 2012 version.

http://blogs.amd.com/work/2012/02/02...d-decoder-key/

... in the case that "some enhancements" = steamroller.

I would expect another codename in that case, but they recently did not deliver a codename for the announced 28nm Server CPUs, so it could just be that the new management is generally tight lipped with code names. Maybe a side effect of the execution, execution, execution plan.

Anyways, chances are slim, but one can hope (a bit) ^^