Thoroughbread 2800+ at CeBit

[rubank]

Here are a couple of <a href="http://www.clubic.com/n/n5377.html" target="_blank">pics</a> from a french site.

The new core seems to be little more than half the size of Palomino.

rubank

[Jammrock]

I wonder how well they overclock and how well they run in a dualie...

Jammrock

[KeiFront]

0,13 micron, less heat and higher overclock speeds

[Wombat]

Have to be careful with that Kei.

Everybody says "ooo pretty, smaller process = less heat." Well, the die is smaller too. So, less area, but less heat. With these newer processes, your wattage/cm tends to <I>increase</I>, and it's actually harder to keep the new die cool.

[Greebe]

which cause me to wonder why they don't integrate a heat spreader on them... doesn't cost much from the manufacturing standpoint and greatly improves the cooling factor.

[KeiFront]

Wombat you're partially right. Smaller die doesn't always mean higher overclock speeds, but they tend to overclock better.

I read somewhere that the voltage would be lowered to 1.65V instead of 1.75V so that also means that the wattage will decrease.

And offcourse that heat spreader would be very handy.

[GT98]

Are supposed to be compatable with the current chipsets that are out now?? I dont want to buy another mother board till Serial ATA and the like is widely avavble.

Scott

[CHHAS]

I saw that ECS released a BIOS upgrade for K7S5A with Thoroughbread support, this board just keeps on surpricing me

[Wombat]

Smaller die doesn't always mean higher overclock speeds, but they tend to overclock better.

That may not hold out for much longer. As we approach/pass the 100nm processes, the priorities reverse. You lose more power and time on your wiring than you do on your FETs, and leakage is bitch.

And it's design more than process that matters to overclock. People have been running their 25s at 33, 33's at 40, and 300a's at 450. It's just cyclical: sometimes the processor is really at the limits, sometimes it's got plenty of room.