FX-53 fails Prime95 immediately!

[warbean]

Hey all,

I have been experiencing random crashing to BSOD lately while playing 3d games, and have been troubleshooting every part of my system. Last night it finally came time for my CPU to be tested.

PRIME 95 FAILED ALL 3 TESTS IMMEDIATELY - rounding errors stopped the test after 2-3 seconds on each one.

The computer is less than two years old, and has never been overclocked. THe average idle temperature is around 49 degrees celsius, and I am using an Antec 480 power supply.

SInce I bought the computer premade from ABS, AMD wont warranty the part itself, and my original warranty has expired from ABS. Am I completely screwed?

 

[keeleysam]

Open it up and check the fans and see if dust is clogging them.

Reinstall Windows.

 

[Duvie]

Check your ram as well....take out all of the sticks except one and try it....since the memory controller is ondie it is tough to say still whther it is a cpu issue alone, memory controller issue o the cpup, or a ram stick...

run memtest

 

[nccr]

Just in response to keeley and Duvie's posts: Warbean already emptied an entire can of compressed air into his system. It is quite dust free. He has reinstalled windows twice in the past week. The RAM was migrated over from a previous system and worked just fine. It also survived 7 hours of mem32 testing. Thanks for your suggestions, but is there anything else to check?

 

[Duvie]

POwer supply and cpu...probably at this point he should try some replacement parts...

This FX53 was a 130nm part? 49c idle seems high but would make sense with older 130nm part....49c idle and was using a stock cooler? I have a feeling that meant the chip loaded in the mid 60's....that is high for extended use....It may be damaged...


I would still try with less sticks...this can sometimes show issues with memory controller...testing single channle versus dual channel, but it also can test DIMM slots on the mobo.....

Power issues are the toughest to diagnos so definitely try a different power supply, or use a multimeter to test it...

 

[Noubourne]

Lower the multiplier and see if you can get the chip to pass at lower clockspeeds.

You might also try upping the voltage slightly to see if it will run stable at stock speed with a tad more voltage.

 

[Appledrop]

more likely mobo/psu/ram/*... cpus are pretty robust, but do as noubourne says to be sure i guess.

 

[warbean]

Well as nccr alluded, ventilation seems to be running well and memtest passed with no problems. Even if the memory controller on the CPU is to blame, doesn't that mean the CPU itself if still broken / must be replaced?

I agree the CPU runs a little hot at 50 degrees idle, but there is no way it should fail immediately like that, unless there is residual heat damage from the past.

I will try new PSU today and update.

 

[Noubourne]

Uhhh, my Venice fails Prime 95 almost immediately at any temp over 49C, at any clock speed.

Not sure if other A64 cores are the same, but 50C isn't exactly a comfortable idle temp. My 3000+ loads at about 47C running 1.55v and 2.6Ghz. It idles at 39C.

 

[Duvie]

Originally posted by: Noubourne
Uhhh, my Venice fails Prime 95 almost immediately at any temp over 49C, at any clock speed.

Not sure if other A64 cores are the same, but 50C isn't exactly a comfortable idle temp. My 3000+ loads at about 47C running 1.55v and 2.6Ghz. It idles at 39C.

The problem is his is a 130nm part but I think the operating temp was still in the 69-70c range...Venice chips are using 90nm and they were much cooler....

To OP, if the memory controller is at fault then yes it would still be the CPU...The question is did memtest pass running in this setp with this cpu??? Cause if it did then it would mean the cache and the memory controller is fine on the cpu....

 

[aka1nas]

Perhaps double check the chipset fan(if there is one) is still working and that it's heatsink is still making good contact.

Did the chip always run this hot? If not, and you have gotten all the dust out as you said, then perhaps the CPU heatsink could use a remount and new thermal compound. If the place that built your machine for you used a high performance thermal compound, than it may have lost it's potency a bit by now.

Can you give us readings for your CPU and PSU voltages? A flaky PSU could definately make that chip unstable once you start stressing the system.

 

[warbean]

Memtest was conducted on this setup and passed as stated before. CPU chipset and fan were remounted with arctic silver only about a week and a half ago. (I upgraded motherboards from AGP to PCI express)

I have no idea why it's so hot! Are all FX-53s this way? Does anyone think it would be worth it to invest in a brand new cpu cooler? I would only buy one if I thought there wasn't already permanent damage. If there was, I'd just say forget it and wait for Conroe.

Voltages:


Core: 1.52 (tested up to 1.6, still fails every time)
+3.3 : 3.29v
chipset: 1.52v
DIMM: 2.73v
+5v : 4.99v

 

[warbean]

Sorry to multipost, but after doing some research, it appears that my temperatures are quite normal for an FX-53. Anyone with a FX-53 at these temperatures or any Athlon 64 at these temperatures able to run Prime95 without any trouble?

 

[aka1nas]

Originally posted by: warbean
Memtest was conducted on this setup and passed as stated before. CPU chipset and fan were remounted with arctic silver only about a week and a half ago. (I upgraded motherboards from AGP to PCI express)

I have no idea why it's so hot! Are all FX-53s this way? Does anyone think it would be worth it to invest in a brand new cpu cooler? I would only buy one if I thought there wasn't already permanent damage. If there was, I'd just say forget it and wait for Conroe.

Voltages:


Core: 1.52 (tested up to 1.6, still fails every time)
+3.3 : 3.29v
chipset: 1.52v
DIMM: 2.73v
+5v : 4.99v

Ok, that gives us more to work with. Your previous posts suggested that the machine has been chugging along with the same hardware for 2 years, not that you have basically rebuilt it and are having troubles. Right off the bat I would suggest the new mobo or video card is going to be the problem.

What motherboard did you have before and what model is your new one? Did you keep the same power supply? If so, is it PCI-E capable? What video card do you have running in there? Have you guys tweaked the BIOS at all or is everything running on defaults? What memory configuration do you have?

 

[Kuroyama]

I've had similar issues in the past (with Athlon XP) with one problem caused by the motherboard, and a later one by the power supply. As everyone says, you need to start swapping parts and see if you can isolate the source of trouble.

 

[warbean]

Hello,

You can refer to my earlier thread for what else I've tried. New video card worked fine on friend's computer, so it's either bad mobo or PSU incompatible with PCI-E.

I am kind of new to hardware, so thanks all for being patient. This is painful yet addicting in its own way!

Finally....while it's true that a good strategy is to isolate components 1 by 1, I had downloaded Prime95 specifically as a way to "isolate" and test the CPU. Since it obviously fails that test (including the test that involves little or no RAM testing), and my temperatures seem normal, that leads me to conclude it can only be a faulty CPU or power supply.

How is my reasoning?

Originally posted by: aka1nas

Originally posted by: warbean
Memtest was conducted on this setup and passed as stated before. CPU chipset and fan were remounted with arctic silver only about a week and a half ago. (I upgraded motherboards from AGP to PCI express)

I have no idea why it's so hot! Are all FX-53s this way? Does anyone think it would be worth it to invest in a brand new cpu cooler? I would only buy one if I thought there wasn't already permanent damage. If there was, I'd just say forget it and wait for Conroe.

Voltages:


Core: 1.52 (tested up to 1.6, still fails every time)
+3.3 : 3.29v
chipset: 1.52v
DIMM: 2.73v
+5v : 4.99v

Ok, that gives us more to work with. Your previous posts suggested that the machine has been chugging along with the same hardware for 2 years, not that you have basically rebuilt it and are having troubles. Right off the bat I would suggest the new mobo or video card is going to be the problem.

What motherboard did you have before and what model is your new one? Did you keep the same power supply? If so, is it PCI-E capable? What video card do you have running in there? Have you guys tweaked the BIOS at all or is everything running on defaults? What memory configuration do you have?

 

[warbean]

Sorry, here's that earlier thread, in technical support:

http://forums.anandtech.com/messageview...atid=32&threadid=1887400&enterthread=y

Again, thank you all for your help on this!

 

[Duvie]

While the 130nm FX53's were hotter versus our current 90nm A64's, approximate 50c idle is waaayyyy to high even for stock.....Unless you have an 805D intel ofcourse....

1.5v is stock voltage so that looks good...

What I did find was there are 2 models of FX 53's...one is a 2000HTT version and one is 1600HTT.....what is you HTT multiplier in the bios 4x or 5x??? HTT link is part of the cpu so lets make sur eit is recognizing it properly...

 

[aka1nas]

I had that same Truepower PSU and it is not PCI-E compatable and I am pretty sure you might need it to be for a 7900GT. That may be a seperate issue though.

Just from what I am reading from a quick google search, that board is known to be somewhat hit or miss as far as stability. Are you running the latest BIOS for it? You can find them here:

http://www.epox.com.tw/eng/support.php?ps=335&type=1


I would suggest trying a different motherboard at this point. Could be something like that Epox board's VRMs can't handle the load required for the FX-53 as it is fairly power-intensive compared to newer .09 chips.

 

[Check]

My intel box did the same type of stuff. Crashed when it was in 3d games and failed prime95 every time within a few minutes. It was my mobo that was at fault (the northbridge was suggested to me by a geek friend) at any rate whenever I have had processor trouble in the past it would just completely die. Again I run intel boxes, so I dunno if my word helps that much. If you don't want to swap out boards swap a friends processor in, or yours into their box. If you have a local computer shop where the guys are friendly (like mine used to be) you could try just swapping processors down there.

 

[warbean]

Thanks for your replies. I am idling now at 46 degrees Celsius. According to Epox's own probe, I am runnin 200.93htt x 12 multiplier = 2411.11 mhz.

I am in fact running latest version of BIOS (April 2006) for this board.

I just installed a new Antec Truepower 2.0 (480 watt, pci-e compatible, 2 split +12 rails), and am about to run a battery of 3d tests on it. Unfortunately, Prime 95 continues to fail within seconds.

The two issues I see now are: (1) CPU failing Prime. (My next step is replace mobo, I am going with MSI K8nNeo4 Platinum next)

(2) Crashing in 3d games. Perhaps this may be fixed by the new PSU. Update soon...

Originally posted by: Duvie
While the 130nm FX53's were hotter versus our current 90nm A64's, approximate 50c idle is waaayyyy to high even for stock.....Unless you have an 805D intel ofcourse....

1.5v is stock voltage so that looks good...

What I did find was there are 2 models of FX 53's...one is a 2000HTT version and one is 1600HTT.....what is you HTT multiplier in the bios 4x or 5x??? HTT link is part of the cpu so lets make sur eit is recognizing it properly...

 

[warbean]

Update:

New 24 pin PSU installed with dedicated PCI-E support. Hasnot fixed game crash issue. Now resigned that mobo is broken.

Can anyone confirm that a faulty motherboard can cause Prime 95 to fail, even if the CPU is working fine? :::crosses fingers that my beloved FX-53 is still good:::

 

[aka1nas]

Originally posted by: warbean
Update:

New 24 pin PSU installed with dedicated PCI-E support. Hasnot fixed game crash issue. Now resigned that mobo is broken.

Can anyone confirm that a faulty motherboard can cause Prime 95 to fail, even if the CPU is working fine? :::crosses fingers that my beloved FX-53 is still good:::

A faulty motherboard can cause almost any failure symptoms as all your components have to communicate through it. It's much less likely, though not impossible, that you killed the CPU.

That MSI board is a much better model, we have about 4 machines running with them at work and they have been rock solid.

 

[blckgrffn]

Sorry to hear the news

Another model that I have had luck with is the EVGA/Jetway SLI board... Good stability, etc, for the price Not to mention great oc'ing! Well, for me at least, I haven't tested them all

 

[Noubourne]

Dude why don't you put back the old mobo and vid card and see if the chip passes Prime 95 that way?

That should tell you if it was the new mobo or not.