Some new Thoroughbred Info.

[Draco]

"XP 2500+ to have .13 micron core"



<< close to the AMD action tell the INQUIRER that the XP 2500+ will use a .13 micron core but that the 2200+ will use .18 micron technology. The 2200 is slated to appear this quarter, and it will support the usual cluster of multimedia instructions including MMX, 3DNow and SEE.

Other details are that it will run at 1800MHz, using a 133 x 13.5 multiplier, have a 64-bit dual pumped bus, operate at 1.75 volts, and have 64K data two way as well as 64K instructions, also two way.

There will be 256K on die unified level two cache, which will be 16 way exclusive, and the die size will be 128 square millimetres, with the thing having 37.5 million transistors.

There is no reason for AMD to move straight to 2GHz because the firm will be able to introduce intermediate speed steps, much like Intel has, and intro the 2100+, the 2200+, the 2300+ and the 2400+ before moving to the 2500+.

The XP-2500 using Thoroughbred will indeed launch in Q2 of this year but it seems the amount of on die level two cache is not decided - meaning it's likely to be higher.

It will have a 80 millimetre square die and use the .13 micron process that is currently sampling.

While there will be a slight decrease in raw megahurts when AMD moves from .18 micron to .13 micron, performance is expected to jump substantially, according to our reliable source, becauseof additional gates and a better design. µ
>>



Not sure how accurate it is, but it's interesting nevertheless. I guess this basically confirms what everyone has been speculating. Hopefully a ondie cache increase is in the list of improvments as well. Northwood probably won't seem all that competitive anymore if that's the case... but that's just what I think 🙂

Sorry if this is a repost.. didn't see any immediate threads on it.

 

[Adul]

I would like to see a wider datapath to the cache at least. But it is interesting to say the least.

 

[flexy]

>There is no reason for AMD to move straight to 2GHz because the firm will be able to introduce intermediate speed steps, much like Intel has, and intro the 2100+, >he 2200+, the 2300+ and the 2400+ before moving to the 2500+.

What means 'there is no reason' ? is the fact that people already are able to OC their northwoods to 2.5/2.7 Ghz NOT reason enough ? When i think about a 2.7 ghz running Intel cpu then i must admit i get green in the face because of envy !

I thought the ThBr will have at least real 2ghz now.....and honestly......XP2100 (ie. PR rating, not real MHZ) and XP2200, ehrm....does not sound too interesting to me as already (pseudo) XP2000 owner (xp1800@xp2000)


>The XP-2500 using Thoroughbred will indeed launch in Q2 of this year but it seems the amount of on die level two cache is not decided - meaning it's likely to be >higher.

Oh !!! They really SHOULD increase the l2 cache !!!! Just think about how fast you could calcualte a Seti workunit with a L2 a big bigger than 256kb... 🙂


>Not sure how accurate it is, but it's interesting nevertheless. I guess this basically confirms what everyone has been speculating. Hopefully a ondie cache increase is >in the list of improvments as well. Northwood probably won't seem all that competitive anymore if that's the case... but that's just what I think 🙂

If you can overclock them to REAL 2.7 ghz then they are in fact impressive......afaik they also have a bigger cache...not that i like intel....but that looks good to me...

I'd at least something similiar (if not FAR better) coming from AMD, too !

 

[Draco]

<< If you can overclock them to REAL 2.7 ghz then they are in fact impressive......afaik they also have a bigger cache...not that i like intel....but that looks good to me... >>



I agree. I plan to put together a new machine in the next few weeks. If Intel cuts their prices by a significant amount on the 27th, I'm definitely going to strongly consider a 2.0A P4, and an i845-D motherboard.

 

[Soccerman]

first of all, I don't even understand what they're trying to say in some place..

second of all, this is the FREAKIN INQUIRER. it so obvious just from the way they type things up.. sheesh..

Other details are that it will run at 1800MHz, using a 133 x 13.5 multiplier, have a 64-bit dual pumped bus, operate at 1.75 volts

wow who would have thought that. the logical steps up would REQUIRE the XP2200 to be running at 1.8 ghz.. as for the '64 bit dual pumped bus' are they just talking about the FSB?? I don't think they'd refer to the connection to the L2 cache as a bus would they?

64K data two way as well as 64k instructions, also two way

are they talking about the L1 cache? hello, could they be a bit more clear?

they need to use a spell checker.. While there will be a slight decrease in raw megahurts

the only significant peice of info that they put out there in this entire blurb, is that it (the 2200+) has 37.5 million transistors.

the rest either doesn't make sense, or we already knew..

 

[Viper GTS]

Anybody know how the multiplier situation will work? If they stick to a 133 FSB they're going to need some very high multipliers, will they require BIOS updates, or will they simply re-interpret lower multipliers (ala K6-2)?

I guess it all comes down to:

Will I be able to run dual .13 micron Thoroughbreds in my Tiger MP?

Viper GTS