Upgraded CPU... Corsair Low-latency is unstable at stock 166MHz...

[brigden]

Before I begin, here are the system specs:

Athlon XP 2700+
EPoX 8RDA+ (Rev. 1.1 - Latest BIOS flashed)
512 MB of Corsair TWINX512-2700LLPT (2-2-2-5)
ATi Radeon 9700 Pro
Maxtor DiamondMax Plus 9 (6Y080P0) 80.0 GB @ 7200 RPMS
Sony CRX1611
Enermax EG365P-VE FM 350 Watt ATX PSU

Vdd: 1.58V
Vcore: 1.68V
Vdimm: 2.65V
+5V: 4.91V
+12V: 12.28 to 12.34
-12V: -12.52 to -12.61
-5V: -5.34 to -5.40
VBAT: 3.24V
5VSB: 5.04

CPU and DIMM voltages are set to auto.

CPU temperature is around 50-55 degrees celcius.

Now, the 2700+ is brand-new. The previous processor was an 1800+, and along with the RAM running underclocked, but synchronously at 133MHz with timings of 2-2-2-5, the system was rock-solid, passing memtest86 error-free and hours of 3DMark2001 SE and 3DMark2003.

I carefully popped in the new 2700+, reset CMOS, and re-entered the appropriate BIOS values. The motherboard reported no problems and posted. Windows quickly booted and I thought all was well. Within a few minutes I started getting errors all over the place. IE would fail, Outlook would fail, Explorer would crash, etc. I even got a few BSODs.

I got so fed up with all the errors I was receiving I simply decided to reformat and reinstall.

With the RAM running at 166MHz, no matter what the CPU's FSB was set at; 100, 133, or 166MHz, the Windows XP installation would fail. Yet with the RAM running at 133MHz, and the CPU's FSB at 166MHz the Windows installation was sucessful.

Upping the DIMM voltage from 2.65 to 2.77 allows me to install Windows error free at 166MHz, but once installed I experience errors. For instance, when I attempt to execute the 128MB SP1a file I received the following dialogue:

"File is corrupt."

I redownloaded the file, but receive the same error.

I've just finished running Memtest86 on the two Corsair DIMMs, with timings at 2-2-2-5, which is default. Each DIMM was tested seperately, in all three DIMM slots. At 133MHz, both DIMMs passed all tests in all DIMM slots. At 166MHz both DIMMs experienced thousands of errors in all DIMM slots.

Upping the voltage only seems to decrease the number of errors, but I still get errors every time. Could it be the PSU not getting enough power to either the CPU, or the system in general?

Now, I'm testing some Kingston ValueRAM. So far, the first DIMM passes all tests at 133MHz, but at 166MHz I get errors, fewer errors than the Corsair, but still hundreds of errors.

Keep in mind, when I first bought this memory I tested it along side my 1800+ with the FSB at 133MHz. I ran the RAM synchronously at 133MHz and passed all tests over a six hour period. Not one error. This leads me to believe the Corsair RAM is unstable and the cause of the errors. However, the RAM is allegedly tested at 166MHz with timings of 2-2-2-5. So, why is this RAM unstable at its default speeds? Could the CPU be at fault? Could it be the motherboard? Could it be the PSU? I need help isolating the offending hardware so I can RMA appropriately.

Please advise.

 

[Ionizer86]

RMA the CPU. I had a 1700+ that would make everything unstable. And all I switched in my setup was my Pally 1600+ with the Tbred 1700+. Raising the voltages a bit, increasing, or decreasing the clocks did nothing. Needless to say, my returned 1700+ worked flawlessly.

 

[xenos500]

turn the FSB on your AMD down to 100 and try again, if all your problems go away then your CPU is DOA. RMA it with your retailer or AMD direct. I have seen this 3 times with Athlon XPs once on a retail and twice on OEM.

 

[brigden]

Thanks alot, guys, but I'd some more advice before I RMA the CPU.

 

[MasterFlash]

Originally posted by: xenos500
turn the FSB on your AMD down to 100 and try again, if all your problems go away then your CPU is DOA. RMA it with your retailer or AMD direct. I have seen this 3 times with Athlon XPs once on a retail and twice on OEM.

That's good advice. Did you try changing the FSB in Bios down to 100? The other thing that could be possible is heat, perhaps, if you don't have good contact between your heatsink and CPU... Might want to try reseating the CPU and reattaching the HS, hopefully using some good thermal paste.

 

[Ionizer86]

If it's still lying around, try the old cpu again for some hours. If it's stable, you've pretty much confirmed the problem.

 

[MrDudeMan]

Originally posted by: Ionizer86
If it's still lying around, try the old cpu again for some hours. If it's stable, you've pretty much confirmed the problem.

exactly...i dont know what more advice you are seeking. everyone has told you the solution to your problem, or at least a way to figure it out. secondly, why do you need to remove the motherboard from the case to switch processors? that is extra effort and is more likely to cause a problem than help. you are more at risk to shock the board while it is out of the case.

 

[sandorski]

Do a combination of the previous posts:

set fsb to 100, if that works fine up the cpu voltage and set it back to 166. Try upping the voltage until it runs stable.

I suspect that either your power supply or motherboard is having problems supplying enough voltage to the new cpu.

It could be something as simple as your Ram settings though. Were you using CAS 2 at 133 because you had PC2700 ram? Try lowering your ram settings, you are now running at 166mhz fsb.

 

[HGC]

What is your power supply?

 

[brigden]

Originally posted by: HGC
What is your power supply?

The PSU is an Enermax EG365P-VE FM 350 Watt ATX.

 

[brigden]

The heatsink is stock AMD and I made sure that the thermal pad lined up with the processor when installing. The contact seemed to be good, but the processor was running at 55 degrees celcius once XP booted. A little hot, no?

The RAM timmings shouldn't be a problem. The RAM is brand-new and has never been a problem before, passing memtest86 with flying colours at the lower FSB. The modules are both low-latency and are designed to run stable with those timings at 333MHz.

I'm going to check the all the rails and the voltage, so you guys have more information to work with. If I can't get it going, I'll pop the old processor in...

 

[brigden]

OK, I checked the voltages:

Vdd: 1.58V
Vcore: 1.68V
Vdimm: 2.65V
+5V: 4.91V
+12V: 12.28 to 12.34
-12V: -12.52 to -12.61
-5V: -5.34 to -5.40
VBAT: 3.24V
5VSB: 5.04

CPU Voltage is set to auto.

 

[InlineFive]

It sounds to me like the cache on the processor is most likely bad. I would RMA the processor.

-Por

 

[HGC]

If it were me and I could get a new more powerful power supply at a local Best Buy, I would try that, then return it and RMA the CPU if it still doesn't work. These things can be real mysteries. I went through something similar once. I ended up replacing the power supply, CPU, and motherboard. It turned out that the motherboard, which had served me well for many months, would not post for the used computer shop I tried to sell it to.

 

[brigden]

Originally posted by: HGC
If it were me and I could get a new more powerful power supply at a local Best Buy, I would try that, then return it and RMA the CPU if it still doesn't work. These things can be real mysteries. I went through something similar once. I ended up replacing the power supply, CPU, and motherboard. It turned out that the motherboard, which had served me well for many months, would not post for the used computer shop I tried to sell it to.

Why would this setup require more than 350W? I don't believe it's a PSU-related problem.

 

[InlineFive]

Originally posted by: brigden

Originally posted by: HGC
If it were me and I could get a new more powerful power supply at a local Best Buy, I would try that, then return it and RMA the CPU if it still doesn't work. These things can be real mysteries. I went through something similar once. I ended up replacing the power supply, CPU, and motherboard. It turned out that the motherboard, which had served me well for many months, would not post for the used computer shop I tried to sell it to.

Why would this setup require more than 350W? I don't believe it's a PSU-related problem.

I agree 350w is more than enough for what he's running. Besides, BestBuy is a ripoff, go to NewEgg!

Barton 2500+ 1.8Ghz (before "optimizations" ) for $85. You can easily hit 11x200 with a good board at stock voltage.

-Por

 

[Maggotry]

I bought a 1GHz T-Bird a while back. As soon as I installed it everything started crashing. Almost every time I had a crash, it was related to a .dll file. I spent a lot of time on it. I finally RMA'ed it to AMD. They sent me a 1.1GHz T-Bird. All my problems went away. Your problem sounds very similar. Do as others have said and put your old CPU back in.

 

[HGC]

It might not, and I might just RMA the CPU straight away. It's a close call to me. I wasn't so much thinking of the 350 W as that the PS might not be reliable or might be older. My thought was with the ATI 9700 Pro and the XP 2700+, you definitely want plenty of reliable power.

 

[HGC]

Agree 100% Newegg is the best, but you can return it to Best Buy with no shipping or restocking fee if it doesn't solve your problem

 

[InlineFive]

Originally posted by: HGC
Agree 100% Newegg is the best, but you can return it to Best Buy with no shipping or restocking fee if it doesn't solve your problem

Plus NewEgg's retail 2500 Is $90 compared to outrageous prices in retail stores.

-Por

 

[RanDum72]

Another thing to consider is that the board may not be fully compatible with the 2700+ and needs a BIOS update. The 2700+ is a T-bred core so make sure the BIOS does support it besides the older Palomino core.

 

[sandorski]

Have you tried lowering the CAS latency on your ram?

 

[brigden]

Well, I've been playing with the CPU and RAM speeds. I left the timings at the default 2-2-2-5; keep in mind this is low-latency stuff.

With the RAM running at 166MHz, no matter what the CPU's FSB was set at; 100, 133, or 166MHz, the Windows XP installation would fail.

With the RAM running at 133MHz, and the CPU's FSB at 166MHz the Windows installation was sucessful.

This leads me to believe the RAM is unstable and the cause of the errors. However, the RAM is allegedly tested at 166MHz with timings of 2-2-2-5. So, why is this RAM unstable at its default speeds?

Next, I'll try to up the RAM to 166MHz, and drop the CAS latency to 2.5. We'll see how that goes...

 

[brigden]

Originally posted by: sandorski
Have you tried lowering the CAS latency on your ram?

That will be something I do, but why should the RAM not run at 166MHz with the timings of 2-2-2-5, that is what it is apparently advertised to run at.

 

[sandorski]

Originally posted by: brigden

Originally posted by: sandorski
Have you tried lowering the CAS latency on your ram?

That will be something I do, but why should the RAM not run at 166MHz with the timings of 2-2-2-5, that is what it is apparently advertised to run at.

Dunno, but, if your mobo allows you to up the voltage to the ram, that might help make it stable.