Trick that Duron into an Athlon XP

[Julio Franco]

Russian webmasters over at Overclockers.ru seem to have found a way to mod currently selling Duron processors into Athlon XPs by locking L2 bridges. If you can read russian, you can find the original article here, if you don't x-bit has an english translation here.

As with most hardware hacks, not all processors are guaranteed to work well with the mod, if you are lucky enough however Durons may function well with re-soldered L2 and functional 256KB L2 cache.

 

[Per Hansson]

This seems quite strange in my opinion...

Because a Duron CPU has a much physically smaller core (due to the fact that it does not have that extra 128kb cache onboard which the Athlon XP has)

However the old Celeron CPU's where infact Pentium3 CPU's on which not all of the cache worked after it was manufactured, thus Intel simply disabled half of the cache and sold them as Celerons instead... A program was available which you could manually force the CPU to use the remeining working cache (oftenly only a small part of the cache didn't work, so you could make use of almost as much cache as a P3 CPU had...)

 

[Didou]

I think the Durons are using the Thoroughbred leftovers, with half the cache disabled. It's thus smaller then the Barton core.

 

[Per Hansson]

The Duron used a smaller core than the Thoroughbred, I can't remember but think it was something like 100 vs 60 square millimeters...

I'm sure there must be a chart for this somewhere online... Someone with more time on their hands might wanna look further into this.

taken from this Cnet article: http://news.com.com/2100-1006_3-5072041.html?tag=fd_top

"Why the price difference? Making chips is like baking cookies. Although the recipe is identical, they don't always come out the same way. In the case of chips, not every one on a wafer will come out with a fully functional 6MB cache or perform at 1.5GHz.Those chips that don't meet the optimal requirements can be legitimately sold as 1.4GHz chips with a 1.5MB cache. These chips aren't defective (i.e., they work as described) and every chip maker engages in this practice as a way to cut costs and improve manufacturing yields."

 

[Didou]

They seem to be the same size to me.

The Duron spitfire core was smaller then the TBird & the Morgan core was smaller then the Palomino but at the time, the Duron was made from the Austin fab only. Both are being made in Dresden now, so maybe they just added the L2 cache modification to the TBred process that they've been using for a while now.

 

[Strakian]

That's a nice comparison of the cores... However, that's not a Barton. Besides starting at 2500+ a Barton core IS a bit longer (it has to be becuse it hold more on die Cache) so basically the 256k on die cache - twice.... the core has to be longer. I read about this in a maximum PC awhile back, they had a schematical drawing to show the difference between Barton and the previous cores. I wish I had a scanner to scan in the pic for you guys, but you'll just have to take my word for it

 

[Didou]

That's a nice comparison of the cores... However, that's not a Barton.

Exactly my point, I'm trying to show that the new Durons are probably defective or leftover ThoroughBreds. Thus the Duron's die on the left is the same size as the Thoroughbred 2400+ on the right.

 

[agissi]

I wonder if a Duron "1700+" would o/c then your normal TBredB AthlonXP