It could affect yield. Design affects manufacturing very much these days. They could rearrange a problematic area a little bit so it manufactures better. They could rebalance some pipeline stages a little so it can clock a bit higher. Some other minor tweaks here and there.

I wonder if anyone noticed this....

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3092&p=2

1.52V at 2.5GHz ?! Sounds like already (factory) OVERCLOCKED...

This post has been edited by lex: Sep 11 2007, 11:49 PM

Its an ES thats why..but still 2.5ghz at 1.5V?? Dissapointing.. Im going for Yorkfield next.. sorry AMD

Anyways heres a good read..

http://www.amdzone.com/modules.php?op=modl...rticle&sid=8343

This post has been edited by intune: Sep 12 2007, 12:33 AM

CPU-Z doesn't display voltages correctly on most of the newer chips.
But that "donut" in the logo...

Seems to be correct, looking at the power draw figures from here: http://www.techreport.com/articles.x/13176/10
The 2.5ghz chip draws 60W more than the 2ghz chip at load. The average power is supposed to be 75W for the 2ghz chip, so that's an 80% increase in power for a 25% increase in clock. This points to it having a substantially increased voltage over the 2.0ghz version.

Don't all mainboards have PCB routes for approximated equal delay?

Anyways, from the benches, you know what, i'm gonna continue my snooze fest, even if it couldn't beat an Intel, i was expecting it would bring in really fresh perspective, instead of just an FP boost. All this talk about Native, and it's still hobbled, caches are assumedly transparent, so any weakness is mostly likely a design flaw. Although i think they'll ramp up the clocks as the yields get better.

Interger performance is still Intel's domain from what i can see. But i'll have to admit, their FP performance for a 2ghz chip is seriously good, now the only question being is, can they use that little lead and sustain it by piling on the clockspeed? Is their process mature enough to propel them ahead, as Intel's 45 is going to give them a serious boost when it comes to clocks due to the lower leakage power. And yes, can IBM come up with something fast enough to save their asses?

It's not a design flaw as far as i can see. From the microarchitectural improvements they were talking about, it was clear that it was not going to be a core2 killer, despite what marketing would have everyone believe. So the performance we are seeing is really in line with reasonable expectations.

What IBM chip is this competing against?

Not IBM chip, but new joint researched process into newer and more efficient SOI production and strained silicon (yes, they're also replicating Intel's metal gate move). Just as like the boost the 90nm SOI imparted to the Venice, AMD damn well needs the same again for the Phenom if it's going to crank it's silicon up.

Ahhh, marketing, expect them to inject uncertainty and overexpectation into everything, yes i'm still waiting for em slides . When they said it would be 40 percent faster, i was personally expecting an average figure, or atleast in say 50 percent of benches, not only one arena. But i do see it doing well in HPC territory with it's FP

Err... the 2.5Ghz was B1.

B1 is the reason why they did NOT start selling it at 2.5ghz.

Just pointing out that the voltage really was that high.

I think it wasn't a real chip destined for production though.

Some vendors got 2.5 unreleased Opterons (99% B2), Anand o/ced their 2.0s.
DAAMIT most probably ended up with a rough spin but met urge to sell the chip immediately.

Some updates just to get this thread back to Pg 1 (what, the fanbois are tired already?).

1. Apparently FASN8 is dead
2. So is it B1, BA or B2?

so this new AMD proc is the replacement for the AMD AM2?

?.. this is the next generation AMD processor and yes for socket AM2..

and yes this this thread is getting absurd .. sounds more like Q&A.

Reopen Next November...until phenom launches



This post has been edited by intune: Sep 13 2007, 05:54 PM