Printable View
Increasing the RAM speed (note no FSB on this platform) shouldn't increase the power enough to blow the power circuitry. Unlocking cores, on the other hand...
What is the max TDP chip the board advertises as supporting? Granted, it would be a bit daft to allow core unlocking on a board that has little headroom but it does seem like the most likely scenario, not forgetting Prime95 will cause the CPU to draw more power than most 'normal' applications. Intel Burn Test does an even better job! ;)
The Jetway board is rated for 140w+ CPUs - one of my main criteria when purchasing it.
Been running an unlocked 5000+ @ Phenom FX 5000+ @ 3 to 3.3 GHz (that needed over 1.4v to get all cores stable) for the best part of a year. It's been through literally days of Prime95, IBT etc with that configuration with barely a hint of a problem.
The unlocked 5200+ was only on 1.25 vCore and similar on the CPU-NB voltage, so I'm a little baffled as to why the board died when it did. It had already done a 24 hour blend with DDR2 and 20x runs IBT on high a few times.
Pop in DDR3 on stock voltage and POP the first time Prime95 ran for about 10 minutes. Bizarre.
Edit: the temps in speedfan did rise alarmingly just before it went. I was just going to shut off Prime95 before it died. Maybe my fan controller decided not to work or something?
Hmm, maybe it was just on its way out or something? Like I said I can't think of a reason swapping the RAM out would burn up a motherboard like that, maybe the RAM is faulty? Other than that I can't really think of a valid reason.
I think the board manufacturers really underestimated the power draw on the vrms with an oc x6. All boards I run are sinked and a fan blowing on them directly. The jetway running a quad is a strange one, you would expect less stress, however the 5200 unlocked to quad that I ran showed 140w in cpuid hardware monitor, where as 2 x6 that I checked just now show 101.40w/105.30w.
Another point worth noting is that factory fitted sinks are making full contact, I have come across a few threads mention them not making contact due to various reason. It was one of the things I thought may have happened to my asrock 890gx as I had fitted a h50 and it did warp the board slightly after fitting, maybe some of the fets were not in full contact, thus the melting of the thermal tape.
Hopefully the fets issue will be better in the next generation
cheers
One other thing that springs to mind is that DDR2 and DDR3 run on with separate BIOS chips on the HA08 Combo. I had to flash the DDR3 BIOS to DDR3-A11 using a stick of DDR3 1333 so that the DDR3 1600 would be recognised.
There is always the possiblity that something in the BIOS was different and that me manually restoring the CPU settings that I'd used in BIOS DDR2-A09 actually changed something compared to before. I see that there one recorded change is something to do with correcting inaccurately set CPU voltage.
Having looked up the problem now, it seems like there's an epidemic of 790GX and 890GX boards popping mosfets with Phenom II X6s - especially on boards that use 4+1 phase power design. Far fewer cases of mosfets dying with X4s but it's certainly not rare.
These are very similar issues to what turned up on older AM2 boards when the original 140w Phenoms came along.
Still the ASROCK board that died on Max Tractor is an 8+2 phase power design with a mosfet heatsink, so there are definitely some exceptions.
Interesting that the unlocked 5200+ shows as 140w in CPU hardware monitor. I'll have to check that when I get the chance.
Just for comparison's sake, even my 95w TDP 1055T only draws 111.15W according to HWMonitor. I suppose it shows that there is a reason they are only binned as duals even though they work as quads, the VID must just be too high for the rated TDP.
Yes 8+2 on the asrock 890gx, and it ran a 125w x6 1055t around 3.8 for months. When it popped it was on a 95w x6. For reference I have had 2xAsus M4A785D-M PRO the first bought new and it ran the x6 125w@4.0ghz without sinked fets it would throttle, built into asus boards apparently, when sinked it ran the x6@4.0ghz 4gb ram 24/7 without any problems a great little board, my 2nd Asus M4A785D-M PRO is running the same x6 125w@3.8ghz 1.4v in my server 24/7 crunching wcg and a 4870 on collatz and does not miss a beat. This is a 4+1 design with my own sinks and a fan on it.
I also picked up the ASUS M4A88T-M recently, I may be wrong but the fets looks a lot beefier
cheers
To be honest I'll be very surprised if my 5200+ is pulling anything like 140w. It runs much cooler and on lower voltage than my 5000+. It usually doesn't even hit 50 degrees in Prime95 unlocked at 3.2 GHz with fans barely making a noise.
Great point. Although my 5200 did run cool, I had it in just last week as I was having bother with Asus M489GTD PRO USB3, as core temp is lost on unlocking, I use temperature/cpu(I use this at all times anyway) @3.3 1.36v it hit a max of about 43.Quote:
Originally Posted by watercooled
cheers
Hmm, you're not filling me with confidence - I have an X6 on a 4+1 770 Gigabyte board! :laugh:
Well at least it's a 95W TDP and it's not OC'd.
I ran a x6 95w in Gigabyte - GA-MA78GM-S2H(rev 1.0) maxed out @3.5 although hs on fets, the hs can be picked up very cheap, I think they are a must, although on a stock cpu not as much heat will be generated, but anything that gets rid of heat is always good
cheers
Further reading: http://hardforum.com//showthread.php?t=1524927
Yes I came across that thread previously, loads of msi boards dropping with major concerns of the power phase design/components
cheers
Hmm, had a quick skim over it, looks like Gigabyte 770 4+1 with a stock 95W CPU should be OK. Didn't read the whole thing though, anything I should see?
HWMonitor is showing 115v @ 3.24GHz 1.25 vCore on my unlocked x2 5200+
I'm not sure whether I trust HWMonitor, tbh. It shows my Opteron 1381 (essentially a 2.5GHz Phenom II) as drawing 120W, but that is at 1.3v, which seems pretty high: do the Opterons have higher stock vCores to ensure stability or something?