Remarked/dud processor?

A friend of mine is upgrading a Socket A based machine, he decided to get an Athlon XP 3200+ on eBay. But he called me today to say that the board was detecting the processor as a 1000MHz Duron. I assumed an FSB issue so checked the jumper settings, they were correct for a 400MHz FSB.

I booted up XP and ran Everest, now this is the bit I don't get:

CPU Properties

[b]CPU Alias: Barton[/b]
CPU Stepping: A2
[b]CPU Type: Unknown, 1000MHz (5 x 200)[/b]
Instruction Set: x86, MMX, 3DNow!, SSE
L1 Code Cache: 64KB
L1 Data Cache: 64KB
[b]L2 Cache: 64KB (On-Die, Full Speed)[/b]

CPU Physical Info

Die Size: 101mm²
Package Size: 4.95 cm x 4.95 cm
Package Type: 453 pin PGA
Process Technology: 6Mi, 0.13 um, CMOS, Cu
Transistors: 54.3 million

The processor is a Barton core, which is correct, but it's only detecting 64kb of cache (hence the Duron detection) and a locked multiplier of 5x. I've tried the processor in two different boards so I know it's not that. Could this possibly be a remarked processor or just plain faulty? It states on the surface that it's a 3200+ but I don't think it looks right.

yoyo

Has he upgraded the motherboards BIOs? Could help

Nick

I checked the BIOS already, should have mentioned that. It's at version 1.3 which is the most recent available. The board is an AsRock K7-NF2 RAID (nForce2 Ultra 400) so it does support 400MHz FSB processors.

yoyo

Could it be the boards thats the issiue? No offence but ASrock mobos wouldn't be the greatest, looks like a weird issiue though, what does CPUz detect out of interest?

Nick

Anti

He needs to set the multi to what it should be, Its stuck on 5x200. It needs to be 10x200. If you even look at the everrest readout you can plainly see that it says 5x200

The board autodetects the multiplier so that shouldn't be an issue. As I said, I tried it on two different types of boards and both detected a 5x multiplier. If I put anything else into either board it detects it properly with the only change being the FSB jumpers (I tried a 2200+ and a 2400+). Even so, the L2 cache is being detected incorrectly, it should be 512kb for that processor.

I haven't had the chance to check CPUz but I think I'll do just that.

Ok I ran it, I've attached the result. It detects the processor correctly as an Athlon XP but only shows 64kb of cache again. Just now I read online that some Durons were Athlon XPs with some of their cache disabled. I wonder could that be what's up here? Still doesn't explain the pathetically low multiplier though.

Some time ago, AMD found that some of their Athlon XP CPU's had faulty 256kb L2 cache. At first these CPU's were thrown in the trash, until someone came up with the brain wave to cut the L2 cache back to 64kb, call them a Duron and resell them. The Durons were a hit, and to meet the demand, AMD started to laser cut the L2 cache on perfectly good Athlon processors to make more Durons. People discovered that you could re-enable the missing cache by joining two pins on the CPU, and if it not a "faulty" model, you were left with a perfectly good athlon XP.

Anti

set the multi manually. its a common problem.

TerrorFirmer

He said it's locked, so he can't go any higher..

Anti

Sorry, didnt read that part. Try reflashing the bios again.

duridian

Might not be much use to you, but I have a Barton 3200 in an oldish rig I've got here, so here is a capture of the info from CPU-Z so you can compare.


Note how the steppings appear to be the same, but mine has a multiplier of 11 and the full 512KB of L2 is being read. Also notice how the voltage reading on mine is 1.680V whereas your friend's is reading lower at 1.664V.
Is it possible that he has kept the power supply that was used with his older CPU and that it is not giving sufficient power to the Barton?
The main reason that springs to mind in this regard is because it was really about the same time that the Bartons/Nforce2-400 arrived on the scene that AMD motherboards first began to make use of the "Pentium 4" 4-pin auxiliary power connector, which previously hadn't been necessary for an Athlon based computer. Hence some power supplies used with Palomino or Thoroughbred systems still hadn't this auxilliary connector. If this connector was present on the motherboard and had nothing powering it, then possibly it could cause the processor/motherboard to step down to a lower performance level in an attempt to maintain stability.
Probably a long shot but that's about the only possibility I can think of off the top of my head.

He has a 550W PSU. Doesn't really matter though, the processor packed in on me last night. He's trying to source another one but accidentally got an Athlon 64 3200+ instead (Socket 754). So he'll have to change his board too.

I compared a photo of a new 3200+ to the one I have here and the L2 bridge is shorted on it whereas it isn't on the one in the photo. So I suspect someone was messing with it trying to unlock the multipliers and screwed it up.