Phenom stock to 3.2 Ghz with new SB750 ?!

[Hekking]

Daily tech comes out with some interesting AMD news:

Taipei (Taiwan) - AMD has pulled a rabbit out of its hat to increase the performance of its existing Phenom triple-core (8000-series) and quad-core (9000-series) processors. Six “hidden pins” on the processors and chipsets are the secret, which, our sources told us, will enable simple overclocking through the southbridge – and accelerate the current processors on demand.

The name of the game is called “unlocking the multiplier”, which will be played with the SB700 and SB750 southbridge chips. If you are running an upcoming 2.8 GHz Black Edition CPU, a motherboard with the old SB600 model (RD690) will keep the processor cores operating at 2.8 GHz. However, if you have a motherboard with the SB700 chipset, you will receive a free upgrade to 3.0 GHz. And if you get a motherboard with a SB750 chip, your processor will run at 3.2 GHz, which matches the clock speed of the Athlon X2 6400+ - the highest clocked processor AMD ever offerred.

We have seen motherboards with 780G, 790FX and 790GX chipsets in ATX and mATX form factors, which all support this feature. Now it is up to AMD to execute and deliver a compelling user processor.

The actual overclocking is done either through the BIOS, a utility, or simply by pressing a physical button on certain motherboards. The release date of this technology is unknown, but we know that it will be available in time for the 790GX launch.

There you have it: An overclocking feature for AMD CPUs. It just does not get any more better than this. Now we just need to know how this overclocking feature will compare to Intel’s processors, how power consumption and heat dissipation will be affected and what that all means for AMD’s balanced platform marketing pitch.

Hardware zone already reported a little about it a few days ago: http://www.hardwarezone.com/articles/view.php?cid=2&id=2569&pg=4
This means Phenoms can't be higher clocked because SB600 is messing up I think.

 

[hoho]

I'm kind of skeptical considering that their current CPUs aren't exactly doing well in overclocking. Sure, it might be possible to OC some of them that far but I'm pretty sure most of the CPUs are not capable of it.

 

[Hekking]

I've forgotten something I see

If you are running an upcoming 2.8 GHz Black Edition CPU, a motherboard with the old SB600 model (RD690) will keep the processor cores operating at 2.8 GHz. However, if you have a motherboard with the SB700 chipset, you will receive a free upgrade to 3.0 GHz. And if you get a motherboard with a SB750 chip, your processor will run at 3.2 GHz

 

[ShaidarHaran]

This is *exactly* what AMD needed to do (given the existing process and micro-architectural limitations they face).

bravo!

 

[digitalwanderer]

Oh man, I wonder what my CPU would do with one of them new SB750s!

 

[Thorburn]

Sounds to me like boards with SB700/750 are just being designed with power circuitry that can handle the extra power. Can't see why else the south bridge would enable this, other than perhaps an incentive to buy more expensive SKUs.

 

[Davros]

Oh man, I wonder what my CPU would do with one of them new SB750s!

about 10% more than its doing now

 

[digitalwanderer]

Really, going from 2.6Ghz to 3.2Ghz?

I got a different %, I must still suck at the maths...

 

[Sc4freak]

2.6ghz to 3.2ghz is 23%, assuming linear scaling between clockspeed and performance.

 

[Davros]

oops thought you had a 2.8

 

[Thorburn]

2.6ghz to 3.2ghz is 23%, assuming linear scaling between clockspeed and performance.

In theory it won't scale linearly, as the IMC/L3 cache is clocked lower (2GHz?)

 

[digitalwanderer]

oops thought you had a 2.8

I had it clocked up there for a while, but I really wasn't noticing a lot of differences and had to put it at stock for a while for some benchies and never bothered to OC it again.

So you were sort of right in your assumptions, I was just being a prick.

 

[ShaidarHaran]

In theory it won't scale linearly, as the IMC/L3 cache is clocked lower (2GHz?)

They (IMC/L3) are clocked asynchronously (compared to core clock) but we don't know for sure what that clockspeed will be. After all, it is adjustable in 10h family products. We've seen 1.8-2.2GHz so far, IIRC. Hopefully the core clock bump will be accompanied by a bump in IMC/L3 clock as well. We all know they need it!

 

[Hekking]

Is L3 or 2 cache a tech they buy from Rambus or do they develop it themselves ?(AMD/Intel)

 

[ShaidarHaran]

Is L3 or 2 cache a tech they buy from Rambus or do they develop it themselves ?(AMD/Intel)

AMD's cache techology is nothing more than standard SRAM. They are not using any 3rd-party IP to develop products which utilize this SRAM for various levels of cache. I am sure the same goes for Intel. AMD has a bit tougher task though, since they produce MPUs on an SOI process, and Intel utilizes bulk silicon.

There has been a lot of speculation over the years about AMD's potential to use ZRAM for cache (due to its density and power advantages compared to SRAM), but it appears as though the window of opportunity here has all but disappeared.

 

[AlexV]

There has been a lot of speculation over the years about AMD's potential to use ZRAM for cache (due to its density and power advantages compared to SRAM), but it appears as though the window of opportunity here has all but disappeared.

You've spoken the magical word...Arun shall awaken and smite all those who do not believe in the magnificence of Z-RAM.

 

[ShaidarHaran]

LOL, I look forward to Arun's visit then

I'd love for ZRAM to be a viable alternative to SRAM in any product designed for a compute-intensive environment, but everything I've read about it says this is simply no longer possible given the intricacies of the currently-utilized manufacturing processes, and ZRAM itself.

 

[Mat3]

I haven't heard much about Z-RAM not being viable except some doubts from Toshiba engineers who question the validity of Innovative Silicon's Z-RAM. Since the Toshiba guys are also competing to come up with their own floating body memory, they're not exactly a neutral 3rd party.
Still, it doesn't look like it's going to be used in any real products any time soon. Is it just another Fast14?

Informative article on Z-RAM

 

[ShaidarHaran]

I haven't heard much about Z-RAM not being viable except some doubts from Toshiba engineers who question the validity of Innovative Silicon's Z-RAM. Since the Toshiba guys are also competing to come up with their own floating body memory, they're not exactly a neutral 3rd party.
Still, it doesn't look like it's going to be used in any real products any time soon. Is it just another Fast14?

Informative article on Z-RAM

I'd be happier to be wrong than right in this case. Thanks for the link!

 

[Bouncing Zabaglione Bros.]

Anyone know when SB750 motherboards are shipping? I would guess the new chips are a drop in replacement, but I might expect new model numbers for marketing reasons. I'm hoping it would make purchasing easier than waiting for the channel to work through old stock or try to ensure I get a newer board with the latest southbridge when there's lots of older Sb600 board on the shelves.