What to expect from T-BRED and later variants?

Xtasy

Banned
Nov 23, 2001
568
0
0
I hope T-BRED will rock just as bad as the northwoord 1.6ghz! The specs seem to be officially 256kb, 266mhz fsb, same as palomino except the .13 micron, but will later variants carry 512kb or 333mhz fsb? Is barton still a go? Thanks.
 

DivideBYZero

Lifer
May 18, 2001
24,117
2
0
How exactly will the cache 'get in the way'?

Doesn't seem to hinder the 1.6A'S of this world too much
 

Mikewarrior2

Diamond Member
Oct 20, 1999
7,132
0
0
its very difficult (and wrong) to equate intel .13 process overclocking with amd .13 process overclocking.

THey're different architectures. Chances are that tbred will not overclock nearly as well as a its northwood counterparts. There are many factors, including the K7 Architecture, and the small contact surface with tbreds and a still high vcore/cpu wattage. This makes tbreds mcuh harder to cool than XP/MP chips.


Mike
 

Xtasy

Banned
Nov 23, 2001
568
0
0
It is a must that the t-bred must run cooler than palomino. Even though the northwoods look similiar to the skt 478 williamettes, under that heat spreader is a smaller die with smaller contact area, yet the northwoods still run cooler. Should be the same for t-bred as well.
 

Mikewarrior2

Diamond Member
Oct 20, 1999
7,132
0
0
I didn't say they didn't run cooler. I said they are harder to cool. Yes, the wording is similiar, but they are clearly different.

Tbreds are 140-150% hotter in terms of heat /sq mm versus the same speed Palomino counterpart.

Northwoods, are, compared to equivelant speed Williamettes, 115-120% hotter in terms of heat/sq mm.


The reason you see less of a heat/sq mm increase on Northwoods is because Northwood vcore is 1.5V, in addition to any other improvements. In the Tbred, the vcore is 1.60-1.65, which explains for the significantly higher heat/sq mm numbers.

Now, as to why heat/sq mm makes it harder to cool: Medium to lower quality heatsinks perform significantly worse iwth a smaller CPU die for heat transfer. Top tier heatsinks will likely have no adverse effect, but there are a good number that will.

Frosty Tech Heatsink Reviews
see that review: Notice several heatsinks where a heat drop will not result in a drop in temps.

For example, let's take several heatsinks from the list, apply 80W to the small contact surface, and 100W to the large contact surface. Temps are given as rise over ambient case temp.

Taisol CGK760: 27.9C with Large Block, ~43C with small block
Swiftech MCXC370: 32C wtih Large Block, ~35C with small block
Volcano 6CU+: 32.1C with large block, ~46C with small block

There are results that go both ways, but small contact surface will hurt cooling on many heatsinks.

No, the die sizes frostytech do not correspond exactly wtih tbred/palomino die sizes, but the significance of small contact surface when it comes to cooling cannot be overlooked. Lower overall heat wattage, in combination with some heatsinks, does not necessarily mean cooler cpu temps.


Mike