Bad 2500 Barton chip?

[L8CS]

A few weeks ago, I replaced a 2200xp on my Abit NF-7mb with a 2500xp Barton CPU, Retail. I started having random reboots and lockups after a clean OS install. I ran a freeware version of Hot CPU Tester Pro 4 and found that it failed the 6 hour burn in test with Checksum and Exception errors. I RMA the Cpu to AMD and they replaced it with another Retail version and it did the same thing as the first 2500.

I ran the Hot CPU Tester program again and found the same Checksum and Exception errors at almost identical addresses. I ran Si Soft Sandras CPU benchmark and found it placed lower than a 2100xp. I replaced the 2500 with old 2200 and repeated the same test with no other hardware or software changes, so I'd have a point of reference. There were no errors on the 2200.

It appears that the AMD lot numbers are different so what gives? Has anyone else had this type of problem?

Thanks in advance

--------------------------------------------------------------------------------------------------------------------------
Test# 1, Chip# 1
000076l 9999 000004cc Nov, 28 - 14:37:12 | Error:CPU 0: MFC QuickSort: Checksums do not match; a[241530] = 19796, a[241530] = 19797

000077| 0000 000004cc Nov, 28 - 14:37:12 l Minidump file saved at C:\Program Files\Hot CPU Tester Pro 4 LE\\HCTNov28-143712.dmp

000078| 9999 00000258 Nov, 28 - 14:50:43 | Error:CPU 0: MFC QuickSort: Checksums do not match; Matrix1 = - 28115503930.188995, Matrix2 = 28115503930.188942

000079| 0000 00000258 Nov, 28 - 14:50:43 | Minidump file saved at C:\Program Files\Hot CPU Tester Pro 4 LE\\HCTNov28-145043.dmp

--------------------------------------------------------------------------------------------------------------------------
Test# 2, CPU# 2
000076| 0000 00000f0c Dec, 26 - 21:04:46 | Minidump file saved at C:\Program Files\Hot CPU Tester Pro 4 LE\\HCTEDec26-210446.dmp

000077| 0430 00000f0c Dec, 26 - 21:04:46 | Exception error:CPU 0: Exception ACCESS_VIOLATION (0xc0000005) at address 0x00407ba6.

000078| 9999 00000ccc Dec, 26 - 21:12:29 | Error:CPU 0: MFC QuickSort: Checksums do not match; a[58153] = 4768, a[58153] = 4769

000079| 0000 00000ccc Dec, 26 - 21:12:29 | Minidump file saved at C:\Program Files\Hot CPU Tester Pro 4 LE\\HCTDec26-211229.dmp

000080| 9999 000000ac Dec, 26 - 21:17:08 | Running...

000081| 0000 000000ac Dec, 26 - 21:17:08 | Minidump file saved at C:\Program Files\Hot CPU Tester Pro 4 LE\\HCTDec26-211708.dmp

 

[rogue1979]

The Barton probably runs your memory on at 333MHz, while the 2200 was probably 266MHz.

If this is the case, just run MemTest and see what happens.

Also, make sure your power supply is up to the task for the new cpu.

 

[L8CS]

I'm running two 512mb strips of Crucial PC2700 CL2.5 memory and the timings set to Auto in the BIOS. My PWS is an Enermax 350 so it should do the trick. The wattage on the 2500 is 53.7 and the 2200 is 57 so there's a decrease of approximately 3.5 watts on the 2500. I didn't change any BIOS settings between tests.

 

[L8CS]

If this is the case, just run MemTest and see what happens.

I finally ran MemTest-86 and found one of my two Crucial strips bad. After running it over an hour at the 166 mhz FSB for the 2500, it had over 20,000 errors and it was only through test # 7. The bad strip didn't show up with the slower FSB of the 2200, so I'll RMA it.


Thanks for the help rogue1979.