970FX Maxing out @ 2.4ghz?
Link to the 970FX evaluation board page on IBMs home page.
Click on the "more" link after "Solution description" and you'll see this:
As a side note from the www.970eval.com page
As it seems unlikely that it's the CPC925 northbridge that is holding the 970FXs back, the conclution that it's the 970FX that is maxing out @ 2.4Ghz is not far off. Then of course perhaps Apple can get access to the pool of few CPUs that can clock over the specified "public" max Ghz rate, like with the 1.42Ghz 7455's.
[ed] Ahrrrg.... the title should be 970Fx Maxing out @ 2.4 Ghz[/ed]
Click on the "more" link after "Solution description" and you'll see this:
Quote:
The Momentum Computer Maple Evaluation System and board is a reference design and software development platform for the IBM PowerPC 970FX processor and CPC925 Bridge. The baseboard is an ATX form factor and contains two 970FX processors that can operate up to 2.4GHz. The evaluation system is available in two configurations; a complete packaged system and the baseboard only. More information is available at www.970eval.com
The Momentum Computer Maple Evaluation System and board is a reference design and software development platform for the IBM PowerPC 970FX processor and CPC925 Bridge. The baseboard is an ATX form factor and contains two 970FX processors that can operate up to 2.4GHz. The evaluation system is available in two configurations; a complete packaged system and the baseboard only. More information is available at www.970eval.com
As a side note from the www.970eval.com page
Quote:
*PowerPC 970 NorthBridge.
Two point-to-point PowerPC interfaces, running at up to 1GHz.
*PowerPC 970 NorthBridge.
Two point-to-point PowerPC interfaces, running at up to 1GHz.
As it seems unlikely that it's the CPC925 northbridge that is holding the 970FXs back, the conclution that it's the 970FX that is maxing out @ 2.4Ghz is not far off. Then of course perhaps Apple can get access to the pool of few CPUs that can clock over the specified "public" max Ghz rate, like with the 1.42Ghz 7455's.
[ed] Ahrrrg.... the title should be 970Fx Maxing out @ 2.4 Ghz[/ed]
Comments
Originally posted by Eric_Z
Link to the 970FX evaluation board page on IBMs home page.
Click on the "more" link after "Solution description" and you'll see this:
As a side note from the www.970eval.com page
As it seems unlikely that it's the CPC925 northbridge that is holding the 970FXs back, the conclution that it's the 970FX that is maxing out @ 2.4Ghz is not far off. Then of course perhaps Apple can get access to the pool of few CPUs that can clock over the specified "public" max Ghz rate, like with the 1.42Ghz 7455's.
[ed] Ahrrrg.... the title should be 970Fx Maxing out @ 2.4 Ghz[/ed] [/B]
The point your missing is that this is a specialty system and with this system you can get up to dual 2.4Ghz performance. This is an old document. This has nothing at all to do with the potential clock speed of the 970FX. Your wild speculation isn't even well founded.
Originally posted by Programmer
And why is it unlikely that the northbridge is the problem? Northbridge chips have all sorts of timing related things in them, any of which might restrict the whole chip's clock speed.
Even with, say, a 1:6 or for that matter a 1:3 core clock:bus ratio set on the 970FX?
single 2.2GHz
dual 2.4 GHz
dual 2.6GHz
This is just wishful thinking but still quite likely.I hope .
Originally posted by Eric_Z
Even with, say, a 1:6 or for that matter a 1:3 core clock:bus ratio set on the 970FX?
That really messes with chip performance, however, and wouldn't reflect well on the design.
If there is a penalty in the 970 going for a 1:3 FSB instead of a 1:2 and Apple has to use a 1:3 bus then they simply have to wait until the 970 speed gain is at least offsetting the penalty. Preferably it has to make more than that.
A 3 GHz 970/ 1 GHz bus "must" be faster than a 2GHz / 1Hz bus. My guess is a 2.7GHz/900 MHz is that as well but hardly a 2.1 GHz/700 MHz. How about a 2.4 GHz/800 MHz
I think it better that Apple have the nerves to wait out the 970 systems that will be a real improvments compared to the current ones as opposed to do a MHz number similar to Intels PIII to P4 smoke and mirrors
Originally posted by Programmer
That really messes with chip performance, however, and wouldn't reflect well on the design.
So what's the upshot here, Programmer?
Originally posted by Big Mac
So what's the upshot here, Programmer?
i think the new northbridge-chipset will also be manufactured at 90nm .... so there shouldn't be a heat-problem. btw: do we have any info on the loction where the current apple-northbridge-chip is produced? however, the transition to 90nm could be a reason for possible delays but i can think of the Xserve to be one of the first products to use that chip. i for one would think about the possibility that apple will wait until they have enough 2.8GHz parts or something to get the wow-factor again when they come with new G5-systems...
[corrections by Brad]
Originally posted by Krassy
i think the new northbridge-chipset will also be manufactured at 90nm .... so there shouldn't be a heat-problem. btw: do we have any info on the loction where the current apple-northbridge-chip is produced? however, the transition to 90nm could be a reason for possible delays but i can think of the Xserve to be one of the first products to use that chip. i for one would think about the possibility that apple will wait until they have enough 2.8GHz parts or something to get the wow-factor again when they come with new G5-systems...
At least as of MWSF '04, the Xserve controller was based on the 130nm process and was verified by an Apple Xserve representative to be completely different than the PowerMac controller. Now it doesn't mean that the hold-up couldn't be because of transitioning to the 90nm controller . . . doing so could be blamed on "better" cooling capabilities. Not cooling problems, but we all know how the press mangles things, right.
edit (the controller is built at IBM's East Fishkill Foundry, according to that same Apple Rep.)
[corrections by Brad]
Originally posted by kim kap sol
How was it written? xServe?
or XServe, Brad is crack'n down on that lately, so hopeufully people won't keep screwing it up. Geez people these days