Dead CPU?

[Valantar]

So, I'm pretty sure my ancient Core 2 Quad has been paying attention to my drooling over Ryzen, as my PC BSOD'd yesterday while running Photoshop (not a very heavy load) , and now refuses to boot. The fun thing is that it returns a different error message with each BSOD, and it even BSOD's when trying to run startup repair from a W10 USB drive.

I'm running Memtest86+ now, which hasn't crashed, but so far has a 0% pass rate (~33% into test 3 at the time of writing). My take is that unless every single bit in each of my four sticks of memory has died at the same time, the CPU must be fried. Am I wrong?

For shits and giggles:
https://1drv.ms/i/s!Akhd5yRCmjurzjb-HzVZuyqYt1kW

If it had only held out a few weeks more!

I guess that's what happens when you run a 33% overclock on a Hyper 212 Evo for years and years.

 

[VirtualLarry]

Maybe it degraded slightly, and needs a "CLR CMOS" reset back to stock settings? Have you tried that?

One of my Q9300 rigs that was OCed pretty decently, had an issue hard-freezing when doing Distributed Computing loads, had to remove the OC, then it was still OK.

 

[Valantar]

Maybe it degraded slightly, and needs a "CLR CMOS" reset back to stock settings? Have you tried that?

One of my Q9300 rigs that was OCed pretty decently, had an issue hard-freezing when doing Distributed Computing loads, had to remove the OC, then it was still OK.

I haven't fully reset the CMOS, but I have removed the overclocks on both CPU and RAM (well, technically the RAM is rated for DDR2-1066, there just isn't any DDR2 equivalent of XMP). No dice. I'm pretty sure I've seen more blue screens now than I've had in total since I was first setting up the PC back in 2008.

Also, the memtest run with a seeming 100% error rate (at least the # of errors seems to match the current count #) is at stock clocks. I restarted it and reset the CMOS now just to be sure, and the result is the same. Funny thing is, only the first two bits in each test ever show any error - the remaining six are always fine ("000000" in the "error" column). Might the memory controller be broken? I have to admit I have no idea how to really read memtest's output.

If I'm right though, does that mean that the chipset is fried, rather than the CPU? The memory controller is AFAIK in the Northbridge, after all. Also, that might link this to the Southbridge overheat errors I got a few weeks ago. I added some fans to keep it from happening, and haven't seen those errors since, but I guess that might have been a symptom of chipset deterioration. The X48 chipset runs crazy hot after all, and having a heatpipe linking a crazy hot Northbridge to a not exactly cool Southbridge might... well, not be a good idea.

 

[UsandThem]

If I'm right though, does that mean that the chipset is fried, rather than the CPU? The memory controller is AFAIK in the Northbridge, after all. Also, that might link this to the Southbridge overheat errors I got a few weeks ago. I added some fans to keep it from happening, and haven't seen those errors since, but I guess that might have been a symptom of chipset deterioration. The X48 chipset runs crazy hot after all, and having a heatpipe linking a crazy hot Northbridge to a not exactly cool Southbridge might... well, not be a good idea.

Probably no way of knowing for sure if it is the CPU or the motherboard without throwing in a known working CPU. Generally speaking motherboards will die before a Intel CPU will, but having a 33% overclock all those years could have shortened the life of the CPU.

 

[Valantar]

Probably no way of knowing for sure if it is the CPU or the motherboard without throwing in a known working CPU. Generally speaking motherboards will die before a Intel CPU will, but having a 33% overclock all those years could have shortened the life of the CPU.

True. I could order a new Q9450 off Ebay, but it's not worth the outlay considering that I'm planning a Ryzen upgrade very soon anyway. I'd rather put those $30 into water cooling parts or a nicer motherboard.

I was wondering what I should use this computer for after the upgrade, but now I guess I won't have to worry about that

Also, the 33% OC is only about 2 years old, before that I used the motherboard's "CPU Level up" setting to match a QX9650 (3 or 3.2 GHz IIRC). Probably not the nicest voltages with the auto settings though. Also, I guess the memory controller running at 2V would take its toll over time.

 

[VirtualLarry]

I still had lingering throughts about donating my two Q9300 rigs, with R7 260X 2GB card, 160GB Intel SATAII SSDs, 8GB RAM, and Windows 10, to the local Boys and Girls Club in town. But I have no idea if they would even want PCs, especially ones that old. They both still work, but your thread gives me pause about that - what if they die suddenly a few months in? The PSUs are Antec VP450 supplies, which are 4+ years old, but still holding on. One of the rigs was overclocked, and started freezing up hard when doing DC, so I had to remove the overclock. I've since applied a more mild OC, because those Core2Quads at 2.5Ghz are actually a bit slow in ST performance these days.

 

[Valantar]

I still had lingering throughts about donating my two Q9300 rigs, with R7 260X 2GB card, 160GB Intel SATAII SSDs, 8GB RAM, and Windows 10, to the local Boys and Girls Club in town. But I have no idea if they would even want PCs, especially ones that old. They both still work, but your thread gives me pause about that - what if they die suddenly a few months in? The PSUs are Antec VP450 supplies, which are 4+ years old, but still holding on. One of the rigs was overclocked, and started freezing up hard when doing DC, so I had to remove the overclock. I've since applied a more mild OC, because those Core2Quads at 2.5Ghz are actually a bit slow in ST performance these days.

Yeah, I've been noticing some slow-downs even at 3,5GHz recently, so a Core 2 chip at 2,5GHz would probably feel rather slow even for everyday tasks. I have a feeling mine might have died of shame from the comparison to my new Oneplus 3T that just arrived. I don't have any direct comparisons , but I don't doubt that they would be comparable, at least at stock clocks.

What chipset are those hooked up to, though? Given that my motherboard is X48, which in turn is essentially a binned+overclocked X38, I'm not surprised if that was what failed. Even idling in the BIOS, the north- and southbridges run above 55C, and as noted above I started getting southbridge overheat errors (w/associated automatic shutdowns) recently - IIRC that means temps reaching 99C. Adding three fans to my case fixed that, but I can't help but wonder if years and years of running at 70-80C+ made one of those go kablooey.

I guess I'll try to give away the parts to someone who might have the parts lying around to find out what's broken. At least the RAM should be usable, and I bet 4x2GB kits of DDR2-1066 with heatspreaders are rather rare these days. Otherwise, it's off to the recycler.

 

[VirtualLarry]

Mine are plugged into P35-chipset mobos, with DDR2-800.

Yeah, my DFI X48 boards, the northbridge chipset heatsink, it came with a heatpipe tower for the chipset, and I think, a tiny fan to mount to it.

The P35 board just has a somewhat substantial block of Aluminum connected to it.

 

[Valantar]

Yeah, my Asus Rampage Formula has some rather beefy copper pin-fin-style heatsinks across the northbridge and VRMs (and a puny, sad little thing on the SB with a ROG sticker and a heatpipe running up to the others). These all hook into a rather small aluminium fin-style heatsink by the rear I/O where a tiny (30-40mm) radial fan can be attached, exhausting air through a tiny opening in the I/O shield. Yep, it's a hacked-together solution that makes the temperature issues of the X48 platform quite obvious. I ditched that tiny screamer of a fan pretty soon, but I've compensated by having some form of direct airflow there at all times. Perhaps I should be surprised it's worked for as long as it has. I was thinking of swapping out the thermal pads for some fresh ones, but never got around to it. ah heck.

I'd expect your P35 to last longer though, at least given that the components used are of decent quality.

 

[Valantar]

I finally had some time to actually troubleshoot a bit, and lo and behold:

Turns out one of my memory sticks has commited seppuku or some such. BSODs galore with it inserted (at any speed, in any slot, in any combination with other sticks or alone), while the system is as stable as before with the other three installed. Even got my OC back up and running. I guess it had some life left in it still.

As I placed my order for a Ryzen 7 1700X and a new motherboard yesterday, now I'm just benchmarking the living bejeezus out of this to have some fun comparing the two. Fun fact: the coil whine from my Fury X when producing 1000-1300fps in 3DMark Ice Storm Extreme sounds like someone letting the air out of a baloon. Sounds like it's going to explode. Oh, and turning off the global FPS cap in Radeon Settings is a good idea when benchmarking ...

Edit: so I figured I'd pull off the heatspreaders of the dead RAM stick and have a look. Turns out 9-year-old thermal adhesive is far stronger than 9-year-old BGA solder. Even with heat applied, it was no contest. I'm glad I didn't try this with working RAM.