AMD regressing back to 256k L2 cache?

[neptunefix]

Is it true that, while the Barton 3200 was boasted as an XP processor with twice the L2, future releases will no longer have the 512k L2 cache?

Why? As I understand, this will only make the cpu cooler and allow them to (over)clock higher without doing any real design improvement. Sounds like crap to me.

 

[GonzoDaGr8]

The thorton core is basically a Barton core with half the cache defective from manufacturing. They dissable the bad cache and then sell it as a lower-end proc. Kinda like the celeron is to the p4.

 

[neptunefix]

.. So they have no plans of making an XP cpu with 512k L2 faster than the Barton 3200? Not to start a flame war (I know people take this personally for some weird reason).. but the socket 478 seems to have a better upgrade path ahead of it.

 

[dexvx]

Its going to be the 3200+ Duron, where the PR rating is taken in perspective to a Duron instead of a TBird.



Ok in all seriousness, with a 166FSB and 256KB L2 cache, its going to have a pretty crappy PR rating. The 2800+/166 Tbred was clocked at 2.24Ghz, only a few Mhz off from the 3200+.

 

[FishTankX]

I believ eit's the end of the line for socket 462. It's at the end of it's life. Considering that it started out with teh T_bird 600, it's come quite a long ways going through a 4X speed increase. For socket 478 to do that, it would have to do reach 8GHZ. Which isn't going to happen. I'd say that this is most impressive. The only processor I can think of that has kept a socket this long is *maybe* the Pentium II with it's journey from 266MHZ to 1.13GHZ. YOu can still find slot II 1GHZ P3's around that will run on the BX boards.

 

[Need4Speed]

Originally posted by: FishTankX
I believ eit's the end of the line for socket 462. It's at the end of it's life. Considering that it started out with teh T_bird 600, it's come quite a long ways going through a 4X speed increase. For socket 478 to do that, it would have to do reach 8GHZ. Which isn't going to happen. I'd say that this is most impressive. The only processor I can think of that has kept a socket this long is *maybe* the Pentium II with it's journey from 266MHZ to 1.13GHZ. YOu can still find slot II 1GHZ P3's around that will run on the BX boards.

im not sure what you are talking about, but there is no slot II 1ghz P3....i assume you meant slot 1
the P2 went up to 450 in slot 1 form
then the p3 was offerend as fcpga and slot 1
here is a nice article on the history of cpus

 

[Jeff7]

AMD's 64-bit chips will probably be the next step - new socket design, new chip, new upgrade paths.
Now they'd better get them to the mass market soon, cause Intel's starting to enjoy being on top once again.

 

[sniperruff]

Originally posted by: neptunefix
.. So they have no plans of making an XP cpu with 512k L2 faster than the Barton 3200? Not to start a flame war (I know people take this personally for some weird reason).. but the socket 478 seems to have a better upgrade path ahead of it.

as everyone said AXP is pretty much dead already. if they release more chips with this LOOSE "performace rating" system the whole AMD reputation will be ruined.

good bye AXP. hello hammer

what kinda name will they tihnk of next?

 

[dexvx]

Originally posted by: FishTankX
I believ eit's the end of the line for socket 462. It's at the end of it's life. Considering that it started out with teh T_bird 600, it's come quite a long ways going through a 4X speed increase. For socket 478 to do that, it would have to do reach 8GHZ. Which isn't going to happen. I'd say that this is most impressive. The only processor I can think of that has kept a socket this long is *maybe* the Pentium II with it's journey from 266MHZ to 1.13GHZ. YOu can still find slot II 1GHZ P3's around that will run on the BX boards.

Thats an inherently false statement. No way can the same SocketA 750 Irongate run a 2.2Ghz (3200+) AXP Barton with remote efficiency, just like how a slot1 440LX cannot run a P3 Tualatin, or a socket 478 850 run a 3.2Ghz P4-c (BTW, it started a 1.6Ghz). The socket may be the same, but since you cannot run the latest chip of that socket, it doesnt make any difference anyways.

 

[neptunefix]

The fact that the XP 3200 is based on the same thing as the Athlon 600 is nothing to be proud of. AMD had my money when they started kicking some butt with the DDR XP solutions. They still provide good "bang for the buck", but now Intel has my money. Pretty soon, a lot of it. I'm told that the socket 478 format will go to and beyond 4ghz @ 800fsb. This means I don't have to change the motherboard to utilize this upgrade in the future. Maybe things will change when the A64 and an optimized WinXP64 comes out, but right now it just seems Intel has recovered a lot of lost ground and isnt really scrambling to please us. This also means higher prices for Intel.

None of us consumers win by one competitor getting ahead of another. You shouldn't root or be a fan for any of these corporations, unless you root for them both. One pushes the other. I'm hoping some other companies start competing in the cpu department. Maybe if Mac has good PR, the lower end dual G5's will push Intel and AMD. Then again, I've hoped the same thing about Linux vs. Windows for way too long.

 

[Gonad the Barbarian]

I'm told that the socket 478 format will go to and beyond 4ghz @ 800fsb. This means I don't have to change the motherboard to utilize this upgrade in the future.

I'm not trying to drag this off topic, but given Intel's track record I wouldn't count on that.

 

[Cosmic_Horror]

Personaly i do not think there is much life left in either the pentium 4 or Athlon lines, not with the athlon 64 just around the corner and intel new core, Prescott (?) soon afterward i believe... thankfully i can hold out for a while yet.. *grins*

 

[redhatlinux]

Hold on here..... AMD is NOT regressing with Thornton, Thorton replaces Tbreds in the mid speed range, thats why half the Cache is disabled. Makes alot of sense to just use 1 core not 2. Nothing happens to Barton. Not sure that I agree that these CPUs actually fail when testing the cache, People said that when Duron was announced, they were Wrong, the cache never was on on the die, the die size was physically smaller. Is it close to the end for XP, very likely. Is Athon64 the answer ?? Maybe for some but few desktop apps will benefit from 64bit. Still, as a 'value' processor, mid range Xp's are still good value. High end Bartons are just too expensive for their performance.

Would I bet that future P4's will still be able to use the same mobo, nope.

 

[FishTankX]

Originally posted by: Need4Speed

Originally posted by: FishTankX
I believ eit's the end of the line for socket 462. It's at the end of it's life. Considering that it started out with teh T_bird 600, it's come quite a long ways going through a 4X speed increase. For socket 478 to do that, it would have to do reach 8GHZ. Which isn't going to happen. I'd say that this is most impressive. The only processor I can think of that has kept a socket this long is *maybe* the Pentium II with it's journey from 266MHZ to 1.13GHZ. YOu can still find slot II 1GHZ P3's around that will run on the BX boards.

im not sure what you are talking about, but there is no slot II 1ghz P3....i assume you meant slot 1
the P2 went up to 450 in slot 1 form
then the p3 was offerend as fcpga and slot 1
here is a nice article on the history of cpus

Actually I was thinking wrong. I meant Xeon.

The Xeon has come a long way from slot II. Hasn't changed bus's much. You can find 900MHZ Xeons in SlotII format. Even with 2MB of cache although those are more expensive.

The P6 core went from PII-350 to P3 1GHZ without changing bus's or slot's.

By contrast the P4 has.. gone.. from 1.4GHZ to 2.6GHZ? *Snicker snicker* Either that or 2.26 to 3. *guffaw*

They're also changing sockets on the P4 to a BGA type array thing called SocketT. Sure, you might have 800MHZ bus. But they'll be as different as Socket478 is to Socket740. Completly different beasts.

 

[TerryMathews]

Originally posted by: dexvx
No way can the same SocketA 750 Irongate run a 2.2Ghz (3200+) AXP Barton with remote efficiency... The socket may be the same, but since you cannot run the latest chip of that socket, it doesnt make any difference anyways.

The original socket A chipsets can run the latest chips, as long as long as you're willing to toy with the multipliers. I can personally vouch for an A7V133 (KT133A) running a Tbred B at 2.26GHz. The original Socket A chipsets would top out at 2.0Ghz or so. Due to their limitation to the 100MHz FSB and the Athlon's maximum multiplier of 20.5.

Efficient? Depends on your definition of efficiency. There's no reason that a CPU intensive task can't take advantage of a super fast proc on a slow bus. RC5DES comes to mind. You won't gain any memory bandwidth, but you won't lose any either.

 

[imported_Phil]

I'm told that the socket 478 format will go to and beyond 4ghz @ 800fsb. This means I don't have to change the motherboard to utilize this upgrade in the future

Nope, after the P4 3.2 (or 3.4)Ghz, they're moving to Socket 775. The Prescott 3.2 & 3.4Ghz models will be the last Socket 478 Pentiums.

Source? Anandtech & THG.

Dopefiend

 

[bgeh]

Socket 775 will appear in Late Q1 2004/Early Q2 2004

 

[drag]

Originally posted by: neptunefix
The fact that the XP 3200 is based on the same thing as the Athlon 600 is nothing to be proud of. AMD had my money when they started kicking some butt with the DDR XP solutions. They still provide good "bang for the buck", but now Intel has my money. Pretty soon, a lot of it. I'm told that the socket 478 format will go to and beyond 4ghz @ 800fsb. This means I don't have to change the motherboard to utilize this upgrade in the future. Maybe things will change when the A64 and an optimized WinXP64 comes out, but right now it just seems Intel has recovered a lot of lost ground and isnt really scrambling to please us. This also means higher prices for Intel.

None of us consumers win by one competitor getting ahead of another. You shouldn't root or be a fan for any of these corporations, unless you root for them both. One pushes the other. I'm hoping some other companies start competing in the cpu department. Maybe if Mac has good PR, the lower end dual G5's will push Intel and AMD. Then again, I've hoped the same thing about Linux vs. Windows for way too long.

maybe you'd be suprised about linux then. It's not much in the desktop land, but server wise it has active support of companies like Novell, Oracle, IBM, Peoplesoft, Sun, etc etc whose ported their systems to run on linux. It's made it maindstream server-wise.

Anyways if youre a linux user you'd never have to wait on MS to get a cheap (non win2003 $999 server) OS for 64. Linux has x86-64 versions made already and are just waiting for AMD to release it the new platform so they can use it. (if your curious for proof, goto nvidia's website. They have drivers already built for x86-64 and ready to be downloaded)