Question Overclocking X5650 past 3.9Ghz

Chunka666

Reputable
Aug 1, 2014
26
1
4,545
Hello you wonderful people!

I am trying my best to overclock an X5650 to 4.0 - 4.2Ghz on a GA-X58A-UD3R Rev 1.0 on BIOS F6
If I use any of the latest BIOS (Beta BIOS) I cannot get into the BIOS menu for some reason, with USB or PS/2 Keyboards, so that's why I'm on F6, it will POST but spamming DEL does nothing.

So the meat of it, my current overclock settings

20x CPU Multiplier - if I go any higher to the max of 22x, it resets to 20 when windows loads
195 BCLK
8.0x Memory multiplier (1520Mhz) Triple Channel - set to standard, not turbo
QPI and Uncore still on Auto - tried 36x and uncore x17 for 3.47Ghz but the system didn't benefit or wouldn't POST
LLC Level 2
Vcore 1.275 - tried all testing at 1.35 then lowered it to a still stable voltage with acceptable temps, the Vcore can still go lower but I stopped at 1.275.
Manually set RAM Timings, tried loosening them when the overclocks weren't sticking, but it didn't help
All C-states turned off along with the intel turbo boost

Everything else is Auto, I have tried upping the QPI/Vtt Voltage to 1.295 and the CPU PLL to 1.86 but it still didn't help the system POST.

with the listed above attempts I stopped increasing the BCLK at 195, it might go slightly higher, but 200 definitely does not allow the system to POST and I cannot figure out why, I see guides and videos with the same chip pushing 4.4 to 4.6, obviously they probably won the silicon lottery to some extent, but the basic "easy 4GHz on X5650" guides also yield no results for me.

I've got 3x2GB DDR3 1333Mhz Corsair Dominator RAM triple Channel
X5650 in a GA-X58A-UD3R Rev1.0 Mobo, I cleaned the north/southbridge chips and reapplied new Thermal Compound
Being Cooled by a Corsair H100i

so temps really aren't an issue, I just cannot seem to go any further than 3.9Ghz without accessing those last 2x multiplier available, anyone know why I cannot use the 22x?
any advice on pushing the BCLK past 195 to 200 or 205/210?
Seriously any help is appreciated at this point, I know the performance increase may seem negligible at this point, based on how old the hardware is, but I want to learn, and it's good fun right!

Thanks for reading and all of the wisdom you might impart!
 

compprob237

Distinguished
if I go any higher to the max of 22x, it resets to 20 when windows loads
That motherboard does that. It's the boost multiplier and some motherboards will only use it when there is 1 or 2 threads using 100% of the CPU. Otherwise, >2 threads results in the next step down: 20x.
QPI and Uncore still on Auto - tried 36x and uncore x17 for 3.47Ghz but the system didn't benefit or wouldn't POST
This is very likely the cause of your issues. Uncore == 2x DDR3 speed. At 195 BCLK and 6x (minimum) multiplier for DDR3 you're at ~DDR3-1170 so your Uncore should be 12x at ~2340MHz. Any higher and the QPI/Uncore get really unhappy and will crash or not even post. This is especially true with very high BCLK (>167 is "very high"). QPI link speed should be <7,500MT/s unless you have very good silicon for the QPI.
LLC Level 2
You'll have to do some research on this but the LLC setting goal should be one where the voltage stays the same from idle to load or stays very close to that.
Try the OC every X56xx chip can handle with auto everything else: 20x167 (DDR3 8x Multi is ~DDR3-1336).
Boot to windows, open CPU-Z, write down current VCore with low/idle load, load the system down with Prime95 or something and write down the VCore again. Restart to bios and set VCore to the highest voltage you wrote down. Boot with the different LLC settings until you find the one that doesn't change significantly, or at all, when under load vs idle and use that from here on.
Manually set RAM Timings, tried loosening them when the overclocks weren't sticking, but it didn't help
Set them to what they're rated at. If you go below their speed they wont care but leave the timings as is or on Auto. You want to isolate them from the equation. Uncore, VCore, QPI link speed, or QPI/VTT are usually what causes a failure to boot unless you're overclocking the RAM or their timings are wrong/not set to Auto.
All C-states turned off along with the intel turbo boost
Is VT-d/Virtualization support disabled as well? This tends to cause instability in some instances.

After doing a lot of research on the X56xx (Westmere-EP) chips I'll share a few things with you that are not covered by the overclocking guides here:
QPI/VTT: Never go above 1.35v for these 32nm Westmere-EP chips unless you're pushing for a very high overclock or using exotic cooling (Chilled water/DICE/LN2). The i7-900's 45nm can handle more but the 32nm X56xx chips cannot.
QPI/VTT additional: Sometimes the QPI gets unhappy with too MUCH voltage. So, if you're having trouble with booting you can always try lowering QPI to see if it solves it.
VCore: i7-900 specs claim 1.375v is the max but the real max on these chips should be <1.35v if you can. They're OK with up to 1.4v but they will get REAL hot very quickly.
Uncore: Unless you're running the DDR3 at 1066 or 800 then this should never, ever deviate from 2x the DDR3's speed. There's some instances where you can go higher by a step or two but they usually require a ridiculous amount of QPI voltage (We're talking +0.5v or more). Really never deviate from 2x DDR3 speed unless you're using the aforementioned exotic cooling.
DRAM voltage: Sometimes the QPI wants the RAM voltage stepped up a few notches. Seems to be board-specific. My Sabertooth X58 wants 1.58v for my DDR3-1600 kit to even boot whereas my P6T was fine with 1.52v.
QPI link speed: Never go above 7,500MT/s if you can avoid it. If you are above 7,500MT/s then it will need more and more QPI voltage to stay stable and I've had trouble going above 8,000MT/s without needing to blast QPI voltage (1.4v+) and these 32nm chips will die a very quick death using that voltage.
QPI voltage vs RAM voltage: Try to stay within 0.5v of each other. When you start to overclock this can cease to be a problem since you climb to 1.3v or more and I don't think you want to blast your RAM kit with 1.8v.
X58 RAM support: ALL X58 motherboards support LOW DENSITY 8GB RAM sticks. This effectively gives you a max of 48GB of RAM.

Here's my suggestion:
BCLK: 201
Multiplier: 20x
DDR3: 6x (DDR3-1206)
Uncore: 12x (2412MHz)
QPI link speed: Lowest you can set to (EXCEPT "Slow mode") if you have control over this. This is rated in MT/s (Mega-texels/second) and you want to aim for <7,500MT/s like I said earlier. I think you're stuck with 36x which lands you at ~7,500MT/s which should be OK. If this is expressed in MHz then double the MHz value to get the MT/s.
VCore: 1.35v (or a step down)
QPI Voltage: 1.3v (or a step down)
CPU PLL: Your setting is fine. I'd stop at 1.88v if you do increase it.
IOH Voltage: 1.12v, This sometimes needs a nudge to keep stability. Try it at this setting and if you're stable set it back to Auto and see if you're still stable.
RAM voltage: Bump this up a few steps if the system appears to be hanging during initial training or other voltage changes aren't working. That dominator kit can handle 1.6v without a problem but I wouldn't suggest going higher than that.
CPU Differential Amplitude: If you have this setting then set it to 800mV.

Increase either VCore or QPI one at a time. If it wont POST it's typically VCore that needs to be increased especially if you're already at 1.3v on QPI.

There's a list of BSOD codes and their cause that I could post if you want. It helps greatly with drilling down a BSOD cause and what it means.
 
Last edited:

Chunka666

Reputable
Aug 1, 2014
26
1
4,545
That motherboard does that. It's the boost multiplier and some motherboards will only use it when there is 1 or 2 threads using 100% of the CPU. Otherwise, >2 threads results in the next step down: 20x.
This is very likely the cause of your issues. Uncore == 2x DDR3 speed. At 195 BCLK and 6x (minimum) multiplier for DDR3 you're at ~DDR3-1170 so your Uncore should be 12x at ~2340MHz. Any higher and the QPI/Uncore get really unhappy and will crash or not even post. This is especially true with very high BCLK (>167 is "very high"). QPI link speed should be <7,500MT/s unless you have very good silicon for the QPI.
You'll have to do some research on this but the LLC setting goal should be one where the voltage stays the same from idle to load or stays very close to that.
Try the OC every X56xx chip can handle with auto everything else: 20x167 (DDR3 8x Multi is ~DDR3-1336).
Boot to windows, open CPU-Z, write down current VCore with low/idle load, load the system down with Prime95 or something and write down the VCore again. Restart to bios and set VCore to the highest voltage you wrote down. Boot with the different LLC settings until you find the one that doesn't change significantly, or at all, when under load vs idle and use that from here on.
Set them to what they're rated at. If you go below their speed they wont care but leave the timings as is or on Auto. You want to isolate them from the equation. Uncore, VCore, QPI link speed, or QPI/VTT are usually what causes a failure to boot unless you're overclocking the RAM or their timings are wrong/not set to Auto.
Is VT-d/Virtualization support disabled as well? This tends to cause instability in some instances.

After doing a lot of research on the X56xx (Westmere-EP) chips I'll share a few things with you that are not covered by the overclocking guides here:
QPI/VTT: Never go above 1.35v for these 32nm Westmere-EP chips unless you're pushing for a very high overclock or using exotic cooling (Chilled water/DICE/LN2). The i7-900's 45nm can handle more but the 32nm X56xx chips cannot.
QPI/VTT additional: Sometimes the QPI gets unhappy with too MUCH voltage. So, if you're having trouble with booting you can always try lowering QPI to see if it solves it.
VCore: i7-900 specs claim 1.375v is the max but the real max on these chips should be <1.35v if you can. They're OK with up to 1.4v but they will get REAL hot very quickly.
Uncore: Unless you're running the DDR3 at 1066 or 800 then this should never, ever deviate from 2x the DDR3's speed. There's some instances where you can go higher by a step or two but they usually require a ridiculous amount of QPI voltage (We're talking +0.5v or more). Really never deviate from 2x DDR3 speed unless you're using the aforementioned exotic cooling.
DRAM voltage: Sometimes the QPI wants the RAM voltage stepped up a few notches. Seems to be board-specific. My Sabertooth X58 wants 1.58v for my DDR3-1600 kit to even boot whereas my P6T was fine with 1.52v.
QPI link speed: Never go above 7,500MT/s if you can avoid it. If you are above 7,500MT/s then it will need more and more QPI voltage to stay stable and I've had trouble going above 8,000MT/s without needing to blast QPI voltage (1.4v+) and these 32nm chips will die a very quick death using that voltage.
QPI voltage vs RAM voltage: Try to stay within 0.5v of each other. When you start to overclock this can cease to be a problem since you climb to 1.3v or more and I don't think you want to blast your RAM kit with 1.8v.
X58 RAM support: ALL X58 motherboards support LOW DENSITY 8GB RAM sticks. This effectively gives you a max of 48GB of RAM.

Here's my suggestion:
BCLK: 201
Multiplier: 20x
DDR3: 6x (DDR3-1206)
Uncore: 12x (2412MHz)
QPI link speed: Lowest you can set to (EXCEPT "Slow mode") if you have control over this. This is rated in MT/s (Mega-texels/second) and you want to aim for <7,500MT/s like I said earlier. I think you're stuck with 36x which lands you at ~7,500MT/s which should be OK. If this is expressed in MHz then double the MHz value to get the MT/s.
VCore: 1.35v (or a step down)
QPI Voltage: 1.3v (or a step down)
CPU PLL: Your setting is fine. I'd stop at 1.88v if you do increase it.
IOH Voltage: 1.12v, This sometimes needs a nudge to keep stability. Try it at this setting and if you're stable set it back to Auto and see if you're still stable.
RAM voltage: Bump this up a few steps if the system appears to be hanging during initial training or other voltage changes aren't working. That dominator kit can handle 1.6v without a problem but I wouldn't suggest going higher than that.
CPU Differential Amplitude: If you have this setting then set it to 800mV.

Increase either VCore or QPI one at a time. If it wont POST it's typically VCore that needs to be increased especially if you're already at 1.3v on QPI.

There's a list of BSOD codes and their cause that I could post if you want. It helps greatly with drilling down a BSOD cause and what it means.

Wow ok that was a lot of information to digest.
Thank you very much for the in-depth explanations.
You indeed covered a few things I had researched into before beginning to consider an x58 project like the max (recommended) voltages for the QPI, VCore etc.
right now my Ram sits at 1520Mhz and the uncore was about 3400Mhz, so I can always try stepping that down so it matches the 2x ratio for a higher BCLK ratio, I will look over your recommended settings to see how they fare and if it will boot, when I said the system wouldn't POST I should have been specific in the fact it boot loops, the system at current is completely stable, it took me a while to dial the settings in and was worried that the fact there weren't even beep codes past 196BCLK was either bad luck on the CPU or Mobo.

So I will give this a go tomorrow and get back to you with any of my findings or results, I'm hoping that you are correct and that I was being a little lazy/unlucky with my research into the x58 platform for overclocking past 4.0 Ghz, even though the 4Ghz though totally unnecessary vs 3.9Ghz just upsets me somehow.

Have a pleasant day, and thanks again!
 
  • Like
Reactions: compprob237

Chunka666

Reputable
Aug 1, 2014
26
1
4,545
That motherboard does that. It's the boost multiplier and some motherboards will only use it when there is 1 or 2 threads using 100% of the CPU. Otherwise, >2 threads results in the next step down: 20x.
This is very likely the cause of your issues. Uncore == 2x DDR3 speed. At 195 BCLK and 6x (minimum) multiplier for DDR3 you're at ~DDR3-1170 so your Uncore should be 12x at ~2340MHz. Any higher and the QPI/Uncore get really unhappy and will crash or not even post. This is especially true with very high BCLK (>167 is "very high"). QPI link speed should be <7,500MT/s unless you have very good silicon for the QPI.
You'll have to do some research on this but the LLC setting goal should be one where the voltage stays the same from idle to load or stays very close to that.
Try the OC every X56xx chip can handle with auto everything else: 20x167 (DDR3 8x Multi is ~DDR3-1336).
Boot to windows, open CPU-Z, write down current VCore with low/idle load, load the system down with Prime95 or something and write down the VCore again. Restart to bios and set VCore to the highest voltage you wrote down. Boot with the different LLC settings until you find the one that doesn't change significantly, or at all, when under load vs idle and use that from here on.
Set them to what they're rated at. If you go below their speed they wont care but leave the timings as is or on Auto. You want to isolate them from the equation. Uncore, VCore, QPI link speed, or QPI/VTT are usually what causes a failure to boot unless you're overclocking the RAM or their timings are wrong/not set to Auto.
Is VT-d/Virtualization support disabled as well? This tends to cause instability in some instances.

After doing a lot of research on the X56xx (Westmere-EP) chips I'll share a few things with you that are not covered by the overclocking guides here:
QPI/VTT: Never go above 1.35v for these 32nm Westmere-EP chips unless you're pushing for a very high overclock or using exotic cooling (Chilled water/DICE/LN2). The i7-900's 45nm can handle more but the 32nm X56xx chips cannot.
QPI/VTT additional: Sometimes the QPI gets unhappy with too MUCH voltage. So, if you're having trouble with booting you can always try lowering QPI to see if it solves it.
VCore: i7-900 specs claim 1.375v is the max but the real max on these chips should be <1.35v if you can. They're OK with up to 1.4v but they will get REAL hot very quickly.
Uncore: Unless you're running the DDR3 at 1066 or 800 then this should never, ever deviate from 2x the DDR3's speed. There's some instances where you can go higher by a step or two but they usually require a ridiculous amount of QPI voltage (We're talking +0.5v or more). Really never deviate from 2x DDR3 speed unless you're using the aforementioned exotic cooling.
DRAM voltage: Sometimes the QPI wants the RAM voltage stepped up a few notches. Seems to be board-specific. My Sabertooth X58 wants 1.58v for my DDR3-1600 kit to even boot whereas my P6T was fine with 1.52v.
QPI link speed: Never go above 7,500MT/s if you can avoid it. If you are above 7,500MT/s then it will need more and more QPI voltage to stay stable and I've had trouble going above 8,000MT/s without needing to blast QPI voltage (1.4v+) and these 32nm chips will die a very quick death using that voltage.
QPI voltage vs RAM voltage: Try to stay within 0.5v of each other. When you start to overclock this can cease to be a problem since you climb to 1.3v or more and I don't think you want to blast your RAM kit with 1.8v.
X58 RAM support: ALL X58 motherboards support LOW DENSITY 8GB RAM sticks. This effectively gives you a max of 48GB of RAM.

Here's my suggestion:
BCLK: 201
Multiplier: 20x
DDR3: 6x (DDR3-1206)
Uncore: 12x (2412MHz)
QPI link speed: Lowest you can set to (EXCEPT "Slow mode") if you have control over this. This is rated in MT/s (Mega-texels/second) and you want to aim for <7,500MT/s like I said earlier. I think you're stuck with 36x which lands you at ~7,500MT/s which should be OK. If this is expressed in MHz then double the MHz value to get the MT/s.
VCore: 1.35v (or a step down)
QPI Voltage: 1.3v (or a step down)
CPU PLL: Your setting is fine. I'd stop at 1.88v if you do increase it.
IOH Voltage: 1.12v, This sometimes needs a nudge to keep stability. Try it at this setting and if you're stable set it back to Auto and see if you're still stable.
RAM voltage: Bump this up a few steps if the system appears to be hanging during initial training or other voltage changes aren't working. That dominator kit can handle 1.6v without a problem but I wouldn't suggest going higher than that.
CPU Differential Amplitude: If you have this setting then set it to 800mV.

Increase either VCore or QPI one at a time. If it wont POST it's typically VCore that needs to be increased especially if you're already at 1.3v on QPI.

There's a list of BSOD codes and their cause that I could post if you want. It helps greatly with drilling down a BSOD cause and what it means.

After a lot of trial and error, nothing seems to be working, the PC still just makes a sudden click, then reboots after about 2-3 seconds of power. i tried dialling in all your recommended settings, I also tried adjusting them slightly to see if the extra juice helped, still nothing. it just sits there rebooting.
Such a shame.
If you have any further ideas or advice hit me with anything you got.
Thanks again
 

compprob237

Distinguished
After a lot of trial and error, nothing seems to be working, the PC still just makes a sudden click, then reboots after about 2-3 seconds of power. i tried dialling in all your recommended settings, I also tried adjusting them slightly to see if the extra juice helped, still nothing. it just sits there rebooting.
Such a shame.
If you have any further ideas or advice hit me with anything you got.
Thanks again
VCore: 1.425v (Yes, I know this is really high but it's just a quick test)
Double-check that your QPI link speed is ~7,236MT/s (3,618MHz).

If this lets it boot then try 1.4v VCore.
Keep dropping the voltage by two steps until it wont boot.
When it wont boot, bump the voltage back up 3 steps.
At that point, see if you can get to Windows.

If it still wont boot, even with 1.425v VCore, then try lowering the Uncore to one step below 2x DDR3 speed. Yes, even though you're warned this is a bad idea it can still work and sometimes gives you some more stability. Although, the cost is an obvious reduction in RAM performance and increase in RAM latency.

If it STILL wont boot then set the QPI link speed to "SLOW MODE". This mode is painfully slow (We're talking like 1/10 the speed here) but sometimes it will let the system boot. If it does then that means you need to greatly increase the QPI voltage. If this is the case then you can try QPI at 1.4v, whilst setting link speed back to 7,236MT/s, but keep in mind that you shouldn't stay at that voltage for very long. You want to stay at, or below, 1.35v on QPI. You can run above 1.35v, all the way up to 1.375v, but be warned that the IMC will degrade at that voltage and eventually make it harder and harder to stay stable (more voltage).

If it wont boot fully into Windows: Increase VCore 2 steps.
If it boots to Windows but you get a BSOD or straight up restart: Depends on the BSOD/restart but very highly likely you need to increase VCore (sometimes QPI, only increase one at a time... I have a cheat-sheet for the BSODs - See spoiler)
After that you're looking at doing the stress tests to see if it's stable.

My guess at this point is that you have poor chip quality. Kind of like my second X5670 that needs a ridiculous amount of VCore to even run at 4.2GHz (175x24 or 191x22 -> 1.4v VCore). It's just silicon lottery. The X5650's are the lowest bin of the series so it's not surprising.

Last thing:
There is a slight possibility that your overclock is tripping your motherboard's over-current protection (OCP) or you're tripping OCP on the PSU. What's the make and model of PSU you have? Are you using a 2x4-pin CPU EPS power cable or are you using a 2x2-pin?

0x101 = CLOCK_WATCHDOG_TIMEOUT = increase VCore
0x124 = WHEA_UNCORRECTABLE_ERROR = increase/decrease VCore (Usually increase) or QPI/VTT (Depends, usually increase)... You have to test to see which one it is.
0x0A = IRQL_LESS_OR_NOT_EQUAL = unstable RAM/IMC, increase QPI first, if that doesn't work increase VCore
0x1A = MEMORY_MANAGEMENT = It usually means a bad stick of Ram. Test with Memtest or whatever you prefer. Also, try raising your RAM voltage.
0x1E = KMODE_EXCEPTION_NOT_HANDLED = increase VCore
0x3B = SYSTEM_SERVICE_EXCEPTION = increase VCore
0x3D = INTERRUPT_EXCEPTION_NOT_HANDLED = increase VCore
0xD1 = DRIVER_IRQL_NOT_LESS_OR_EQUAL = QPI/VTT, increase/decrease as necessary
0x9C = MACHINE_CHECK_EXCEPTION = QPI/VTT most likely, but increasing VCore has helped in some instances
0x50 = PAGE_FAULT_IN_NONPAGED_AREA = RAM timings/Frequency or Uncore multi unstable, increase RAM voltage or adjust QPI/VTT, or lower uncore if you're higher than 2x
0x109 = CRITICAL_STRUCTURE_CORRUPTION = Not enough or too Much memory voltage
0x116 = VIDEO_TDR_ERROR = Low IOH (NB) voltage, GPU issue (most common when running multi-GPU/overclocking GPU)
0x7E = SYSTEM_THREAD_EXCEPTION_NOT_HANDLED = Corrupted OS file, possibly from overclocking. Run sfc /scannow and chkdsk /r
0x139 = KERNEL_SECURITY_CHECK_FAILURE = Corrupt/out of date driver, possible corrupt OS file (I've had a bad drive), possibly bad/badly seated RAM, maybe increase VCore
0xF7 = DRIVER_OVERRAN_STACK_BUFFER = Bad system file read/corrupt. Issue with the OS or the storage. -> "HDD bad", or increase VCore
0x7F = UNEXPECTED_KERNEL_MODE_TRAP = QPI/VTT, Vcore, or RAM voltage. Have to test which.
0xC0000255 = Strange boot up = Corrupt BIOS config/behavior probably by RAM overclock. -> load defaults or reset CMOS to fix.
 
Last edited:

Chunka666

Reputable
Aug 1, 2014
26
1
4,545
VCore: 1.425v (Yes, I know this is really high but it's just a quick test)
Double-check that your QPI link speed is ~7,236MT/s (3,618MHz).

If this lets it boot then try 1.4v VCore.
Keep dropping the voltage by two steps until it wont boot.
When it wont boot, bump the voltage back up 3 steps.
At that point, see if you can get to Windows.

If it still wont boot, even with 1.425v VCore, then try lowering the Uncore to one step below 2x DDR3 speed. Yes, even though you're warned this is a bad idea it can still work and sometimes gives you some more stability. Although, the cost is an obvious reduction in RAM performance and increase in RAM latency.

If it STILL wont boot then set the QPI link speed to "SLOW MODE". This mode is painfully slow (We're talking like 1/10 the speed here) but sometimes it will let the system boot. If it does then that means you need to greatly increase the QPI voltage. If this is the case then you can try QPI at 1.4v, whilst setting link speed back to 7,236MT/s, but keep in mind that you shouldn't stay at that voltage for very long. You want to stay at, or below, 1.35v on QPI. You can run above 1.35v, all the way up to 1.375v, but be warned that the IMC will degrade at that voltage and eventually make it harder and harder to stay stable (more voltage).

If it wont boot fully into Windows: Increase VCore 2 steps.
If it boots to Windows but you get a BSOD or straight up restart: Depends on the BSOD/restart but very highly likely you need to increase VCore (sometimes QPI, only increase one at a time... I have a cheat-sheet for the BSODs - See spoiler)
After that you're looking at doing the stress tests to see if it's stable.

My guess at this point is that you have poor chip quality. Kind of like my second X5670 that needs a ridiculous amount of VCore to even run at 4.2GHz (175x24 or 191x22 -> 1.4v VCore). It's just silicon lottery. The X5650's are the lowest bin of the series so it's not surprising.

Last thing:
There is a slight possibility that your overclock is tripping your motherboard's over-current protection (OCP) or you're tripping OCP on the PSU. What's the make and model of PSU you have? Are you using a 2x4-pin CPU EPS power cable or are you using a 2x2-pin?

0x101 = CLOCK_WATCHDOG_TIMEOUT = increase VCore
0x124 = WHEA_UNCORRECTABLE_ERROR = increase/decrease VCore (Usually increase) or QPI/VTT (Depends, usually increase)... You have to test to see which one it is.
0x0A = IRQL_LESS_OR_NOT_EQUAL = unstable RAM/IMC, increase QPI first, if that doesn't work increase VCore
0x1A = MEMORY_MANAGEMENT = It usually means a bad stick of Ram. Test with Memtest or whatever you prefer. Also, try raising your RAM voltage.
0x1E = KMODE_EXCEPTION_NOT_HANDLED = increase VCore
0x3B = SYSTEM_SERVICE_EXCEPTION = increase VCore
0x3D = INTERRUPT_EXCEPTION_NOT_HANDLED = increase VCore
0xD1 = DRIVER_IRQL_NOT_LESS_OR_EQUAL = QPI/VTT, increase/decrease as necessary
0x9C = MACHINE_CHECK_EXCEPTION = QPI/VTT most likely, but increasing VCore has helped in some instances
0x50 = PAGE_FAULT_IN_NONPAGED_AREA = RAM timings/Frequency or Uncore multi unstable, increase RAM voltage or adjust QPI/VTT, or lower uncore if you're higher than 2x
0x109 = CRITICAL_STRUCTURE_CORRUPTION = Not enough or too Much memory voltage
0x116 = VIDEO_TDR_ERROR = Low IOH (NB) voltage, GPU issue (most common when running multi-GPU/overclocking GPU)
0x7E = SYSTEM_THREAD_EXCEPTION_NOT_HANDLED = Corrupted OS file, possibly from overclocking. Run sfc /scannow and chkdsk /r
0x139 = KERNEL_SECURITY_CHECK_FAILURE = Corrupt/out of date driver, possible corrupt OS file (I've had a bad drive), possibly bad/badly seated RAM, maybe increase VCore
0xF7 = DRIVER_OVERRAN_STACK_BUFFER = Bad system file read/corrupt. Issue with the OS or the storage. -> "HDD bad", or increase VCore
0x7F = UNEXPECTED_KERNEL_MODE_TRAP = QPI/VTT, Vcore, or RAM voltage. Have to test which.
0xC0000255 = Strange boot up = Corrupt BIOS config/behavior probably by RAM overclock. -> load defaults or reset CMOS to fix.

My PSU could be the Issue, its a fairly old Corsair CX750w, with 4+4 for the CPU, I can try swapping it for something newer with a tad more juice I have and see if it helps.
 

compprob237

Distinguished
My PSU could be the Issue, its a fairly old Corsair CX750w, with 4+4 for the CPU, I can try swapping it for something newer with a tad more juice I have and see if it helps.
Yes, if you have access to another PSU that's newer it couldn't hurt. It doesn't necessarily need to be >750w since I run "Gramps" with a 650w PSU. It definitely needs to be >550w I can tell you from experience. :p Power measurement equipment reads: 570w at the wall (513w output from the PSU assuming 90% efficiency). Tried SLI OC'd GTX 970s + OC'd CPU -> Trips OCP on the 650w PSU at startup. :ROFLMAO:
 

Chunka666

Reputable
Aug 1, 2014
26
1
4,545
Yes, if you have access to another PSU that's newer it couldn't hurt. It doesn't necessarily need to be >750w since I run "Gramps" with a 650w PSU. It definitely needs to be >550w I can tell you from experience. :p Power measurement equipment reads: 570w at the wall (513w output from the PSU assuming 90% efficiency). Tried SLI OC'd GTX 970s + OC'd CPU -> Trips OCP on the 650w PSU at startup. :ROFLMAO:

Does the overall Amperage and Wattage consumption add into this equation? or is it just the reliability of the PSU?
because I have a little GT210 I can drop in, lowest power consumption GPU I have instead of the RX580 I was planning to use and see if it still works, gonna try the different PSU anyways just because why not. but it's good to know for future reference.
Thanks
 

compprob237

Distinguished
Does the overall Amperage and Wattage consumption add into this equation? or is it just the reliability of the PSU?
because I have a little GT210 I can drop in, lowest power consumption GPU I have instead of the RX580 I was planning to use and see if it still works, gonna try the different PSU anyways just because why not. but it's good to know for future reference.
Thanks
Depends on the PSU. Some of them it is just the overall draw and others it depends on what the draw is on what line. Although, most systems will be drawing a lot of their power from the +12v line. The wattage output is the multiplication of the voltage and amperage it is outputting. If you're looking at buying a new PSU then I'd suggest a 750w or higher. I prefer to spend a bit more on a good quality unit (Seasonic, good Corsair, certain EVGA), lower output unit versus going for a high output, low quality unit.

If you suspect your PSU is the reason you're having issues then installing that low-power GT 210 will give you a healthy chunk of overhead. The only issue I have is that most GPUs at idle are using very little power already so it might not make a difference. Either way, it wont hurt to try with a higher output PSU. The old Corsair CX750 can handle up to 62A total draw on the +12v line so I doubt you're overdrawing unless there's an issue with the unit or you have 20+ HDDs and fans hooked up.
 

Chunka666

Reputable
Aug 1, 2014
26
1
4,545
Depends on the PSU. Some of them it is just the overall draw and others it depends on what the draw is on what line. Although, most systems will be drawing a lot of their power from the +12v line. The wattage output is the multiplication of the voltage and amperage it is outputting. If you're looking at buying a new PSU then I'd suggest a 750w or higher. I prefer to spend a bit more on a good quality unit (Seasonic, good Corsair, certain EVGA), lower output unit versus going for a high output, low quality unit.

If you suspect your PSU is the reason you're having issues then installing that low-power GT 210 will give you a healthy chunk of overhead. The only issue I have is that most GPUs at idle are using very little power already so it might not make a difference. Either way, it wont hurt to try with a higher output PSU. The old Corsair CX750 can handle up to 62A total draw on the +12v line so I doubt you're overdrawing unless there's an issue with the unit or you have 20+ HDDs and fans hooked up.

In the end no dice, it made no difference.
still get the weird clicking and no POST
 

compprob237

Distinguished
That clicking is usually a power protection feature.

Did you try a new PSU (What model?) and still no dice?
Try disconnecting everything unnecessary to just booting to BIOS. This includes HDDs/SSDs (including OS drive), CD drives, and external peripherals (like USB). Obviously keep the keyboard plugged in.

Beyond checking for failed caps on the motherboard I'm at a loss. I've read that those boards can handle clear past 4.4GHz on the X5660 or 4.2GHz for the X5650 (210x20).
 

TRENDING THREADS

Latest posts