Real and Ridiculous AMD News

There’s some real AMD news and some ridiculous news out there, too.

For the first, XBit Labs has told us when Denebs are supposed to show up and what the initial ones will be like.

They will:

• Show up January 8, 2009 at the CES show in Las Vegas
• Come in 2.8GHz and 3GHz
• Come in at the 125-watt power class 

All of this is quite believable based on what we already know about the chip.

In contrast, none of this is believable at all.  This purports to show, without any attribution at all, a CPU-Z of an Deneb popping along merrily at 4GHz while using a mere 1.168V. 

How believable is this?

Let’s look at a Deneb we know about and see what it did at what voltages.

• Engineering sample, Family F, Model 4, Stepping 0
• Apparent default voltage 1.224V, runs up to 2.8GHz at that voltage
• To hit 3GHz, 1.336V is needed
• To hit 3.2GHz, 1.424V is needed
• To hit 3.4GHz (and a bit beyond), 1.568V is required.

Let’s compare this to the alleged Deneb:

• Engineering sample, Family F, Model 4, Stepping 1, so this is just a single simple stepping change using the same process and process chemistry (no high-K) as the earlier Deneb.  Same process shrink, same processor chemistry.
• 4.0GHz, 1.168V

To believe the “new Deneb” numbers, you have to believe this new Deneb, with just a single stepping change, can:

• Run 43% faster using 5% less voltage than the old one
• Run at 4GHz while using 12% less voltage than the “old Deneb” did at 3GHz
• Run at 16% faster at 4GHz while using 27% less voltage than the “old Deneb” did at 3.441GHz. 

If you believe that, I have a bridge and a lot of other things I can sell you.  That’s not a stepping improvement, that’s a miracle.  Not even Barack Obama could get that level of improvement from a single stepping  .  Intel got far less from the 45nm process shrink and using high-k.  

And whoever came up with this fantasy didn’t stop there, I guess it wasn’t piled high or deep enough.  This was the little miraculous brother, there was supposed to be a 4.4GHz one around. 

Please understand, it’s not the speed that makes this BS; it’s the ludicrously low voltage being claimed that crapifies this claim.  If someone came up with a CPU-Z screen showing a Nehalem running at 5 or 5.5GHz at default voltage, I’d say exactly the same thing for exactly the same reasons.   

If AMD had a chip from the Deneb generation that could do this, even just one, they would be screaming so loudly and so publicly about this, you wouldn’t be able to hear anything else.  Almost overnight, the stock price would double, and probably go a lot higher than that afterwards.  It would save a company that desperately needs salvation.  If you think AMD would keep something like that secret, you don’t live in the real world; you live in Hollywood.

Ed