Overclock.net › Forums › Intel › Intel CPUs › _₌= The Q9550 Club =₌_ [Please read first post]
New Posts  All Forums:Forum Nav:

_₌= The Q9550 Club =₌_ [Please read first post] - Page 228

post #2271 of 2408
Okay, I've run into my first issue and I haven't even really gotten started. I thought that before I started messing with anything I'd run some benchmarks to let me see my starting points. While running IntelBurnTest (v2.54) I keep getting the error message:

"WARNING! Your system was found to be unstable under IntelBurnTest! Please check your cooling system and/or lower your overclock! You may or may not encounter instability while working with other programs, but failing this test indicates that your CPU cannot maintain stabilty (produce correct results) in all situations!
Test executed for 62.06 seconds."

I've run it 5 times and it keeps stopping right around the 62 second mark. I have a temp gauge on my CPU and I've watched the temps in RealTemp and OCCT and they're not going over 60c (40 on the actual temp gauge on my case). What's up with this? I haven't overclocked anything yet. Could it be because all the setting are set to "auto" in the BIOS?

And just to add to the mystery... I decided to run Prime95 and for CPU's 1, 2, and 3 the torture tests ended in under one minute (during Test 1) with the lines:

Test 1, 7800 Lucas-Lehmer iterations of M9961473 using type-2 FFT length 512K, Pass1=512, Pass2=1K.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

CPU 4's worker ran for about 2 minutes, then stopped the same way. It feels like I should be concerned... any feedback will be greatly appreciated.
post #2272 of 2408
Quote:
Originally Posted by Slavenhook View Post

Okay, I've run into my first issue and I haven't even really gotten started. I thought that before I started messing with anything I'd run some benchmarks to let me see my starting points. While running IntelBurnTest (v2.54) I keep getting the error message:

"WARNING! Your system was found to be unstable under IntelBurnTest! Please check your cooling system and/or lower your overclock! You may or may not encounter instability while working with other programs, but failing this test indicates that your CPU cannot maintain stabilty (produce correct results) in all situations!
Test executed for 62.06 seconds."

I've run it 5 times and it keeps stopping right around the 62 second mark. I have a temp gauge on my CPU and I've watched the temps in RealTemp and OCCT and they're not going over 60c (40 on the actual temp gauge on my case). What's up with this? I haven't overclocked anything yet. Could it be because all the setting are set to "auto" in the BIOS?

And just to add to the mystery... I decided to run Prime95 and for CPU's 1, 2, and 3 the torture tests ended in under one minute (during Test 1) with the lines:

Test 1, 7800 Lucas-Lehmer iterations of M9961473 using type-2 FFT length 512K, Pass1=512, Pass2=1K.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

CPU 4's worker ran for about 2 minutes, then stopped the same way. It feels like I should be concerned... any feedback will be greatly appreciated.

are you overclocked? and if so at what, also dumb question I am sure you checked to see if the fan was working on the CPU and the fins are clean? Or if you are water cooled you might have algae in the block
post #2273 of 2408
Quote:
Originally Posted by Slavenhook View Post

Okay, I've run into my first issue and I haven't even really gotten started. I thought that before I started messing with anything I'd run some benchmarks to let me see my starting points. While running IntelBurnTest (v2.54) I keep getting the error message:
"WARNING! Your system was found to be unstable under IntelBurnTest! Please check your cooling system and/or lower your overclock! You may or may not encounter instability while working with other programs, but failing this test indicates that your CPU cannot maintain stabilty (produce correct results) in all situations!
Test executed for 62.06 seconds."
I've run it 5 times and it keeps stopping right around the 62 second mark. I have a temp gauge on my CPU and I've watched the temps in RealTemp and OCCT and they're not going over 60c (40 on the actual temp gauge on my case). What's up with this? I haven't overclocked anything yet. Could it be because all the setting are set to "auto" in the BIOS?
And just to add to the mystery... I decided to run Prime95 and for CPU's 1, 2, and 3 the torture tests ended in under one minute (during Test 1) with the lines:
Test 1, 7800 Lucas-Lehmer iterations of M9961473 using type-2 FFT length 512K, Pass1=512, Pass2=1K.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
CPU 4's worker ran for about 2 minutes, then stopped the same way. It feels like I should be concerned... any feedback will be greatly appreciated.

It could very well have something to do with Auto settings in the bios. This is why I always dial in my settings rather than rely on auto for anything. While it is a possibility that auto settings are what is causing this, you should verify a few things first.

First of all, you should be looking at your ram. In CPUZ, see what frequency and timings your ram is currently running at, and what the voltage is set at in a program such as this Rammon

If your voltage is too low, your timings are too tight, or your frequency is too high, it could cause instability.

Another thing to check is the reported Vcore in CPUZ. This voltage is different from chip to chip in terms of what is required to remain stable. However, if your voltage is set too low on idle, it can cause instability. Report back and let us know whats going on!

Edit:

If your bios hardware monitor displays memory voltage, that would be easier than downloading the program I linked. I also do not believe that the program I linked is an active monitor.
Edited by Mattb2e - 10/5/12 at 3:05pm
post #2274 of 2408
Quote:
Originally Posted by Mattb2e View Post

It could very well have something to do with Auto settings in the bios. This is why I always dial in my settings rather than rely on auto for anything. While it is a possibility that auto settings are what is causing this, you should verify a few things first.
First of all, you should be looking at your ram. In CPUZ, see what frequency and timings your ram is currently running at, and what the voltage is set at in a program such as this Rammon
If your voltage is too low, your timings are too tight, or your frequency is too high, it could cause instability.
Another thing to check is the reported Vcore in CPUZ. This voltage is different from chip to chip in terms of what is required to remain stable. However, if your voltage is set too low on idle, it can cause instability. Report back and let us know whats going on!
Edit:
If your bios hardware monitor displays memory voltage, that would be easier than downloading the program I linked. I also do not believe that the program I linked is an active monitor.

Sorry for the delay... busy...

So I ran CPUZ and RamMon and my RAMs frequency is 666.6MHz and my timing's 9-9-9-24 (my FSB:DRAM ratio is 1:2... so is that okay?). It's actually DDR3-1600 DRAM with voltage of 1.7 - 1.9v (from the specs) and after looking more online it looks like I need to change the timings and voltages in BIOS to get 1600. States that with setting set to "auto" they'll run at 1333... which is what RamMon shows. So first I'm going to try that (says that the module "has been tested to run at DDR3-1600 at low latency timing of 9-9-9-27 @ 1.65v").

What's confusing me more now is the Vcore CPUZ is showing me. It's showing the voltage at 1v or below (usually reading @ 0.976v or so). Isn't that too low? Remember, it's a Q9550, and I've been under the impression that I should be running it at 1.26 - 1.28v. So far anything I've tried to change in the BIOS has lead to either my machine totally locking up or the bsod... so feeling a little discouraged but not giving up. Thanks for any help. redface.gif
post #2275 of 2408
Quote:
Originally Posted by Slavenhook View Post

Sorry for the delay... busy...
So I ran CPUZ and RamMon and my RAMs frequency is 666.6MHz and my timing's 9-9-9-24 (my FSB:DRAM ratio is 1:2... so is that okay?). It's actually DDR3-1600 DRAM with voltage of 1.7 - 1.9v (from the specs) and after looking more online it looks like I need to change the timings and voltages in BIOS to get 1600. States that with setting set to "auto" they'll run at 1333... which is what RamMon shows. So first I'm going to try that (says that the module "has been tested to run at DDR3-1600 at low latency timing of 9-9-9-27 @ 1.65v").
What's confusing me more now is the Vcore CPUZ is showing me. It's showing the voltage at 1v or below (usually reading @ 0.976v or so). Isn't that too low? Remember, it's a Q9550, and I've been under the impression that I should be running it at 1.26 - 1.28v. So far anything I've tried to change in the BIOS has lead to either my machine totally locking up or the bsod... so feeling a little discouraged but not giving up. Thanks for any help. redface.gif

Haven't touched RAM setting yet, but I increased my Vcore to 1.2v (CPUZ shows it as 1.072v), then checked OCCT and it's reading my CPU as running @ 1.9GHz (-29.4% OC... ***?). This is with my BIOS reset to all original settings except the change to my core voltage. Sigh mad.gif
post #2276 of 2408
Quote:
Originally Posted by Slavenhook View Post

Haven't touched RAM setting yet, but I increased my Vcore to 1.2v (CPUZ shows it as 1.072v), then checked OCCT and it's reading my CPU as running @ 1.9GHz (-29.4% OC... ***?). This is with my BIOS reset to all original settings except the change to my core voltage. Sigh mad.gif

Try running your ram at 1333mhz until you have everything else running stable. That way, you can remove memory and memory controller instability as a possible cause of your issues.

Double check your bios settings, sometimes when an overclock fails, it will revert to a default which could be the reason of your really low clock speed.
post #2277 of 2408
Okay... well, I've got an overclock working (not too stable yet but it's a start). I don't know how to show a CPUZ validation, but I did get this:

http://valid.canardpc.com/show_oc.php?id=2538638

out of it. So I'm running at 3.8GHz or so... RAM's still higher than 1333 but i'll probably knock it back down to keep working with CPU/FSB stuff. My questions are...

I'm seeing a lot about having FSB:DRAM ratio of 1:1. If I have my FSB freq at say... 400, do I manually go in and change the RAM frequency to... 800... (is that right?)... or am I still mixed up on this?

Also, I tried to run OCCT just to see what would happen. It was running okay but when I saw my core temps go into the 80's I shut it down. What temps are danger zone for Q9550? Thanks.
post #2278 of 2408
Quote:
Originally Posted by Slavenhook View Post

Okay... well, I've got an overclock working (not too stable yet but it's a start). I don't know how to show a CPUZ validation, but I did get this:
http://valid.canardpc.com/show_oc.php?id=2538638
out of it. So I'm running at 3.8GHz or so... RAM's still higher than 1333 but i'll probably knock it back down to keep working with CPU/FSB stuff. My questions are...
I'm seeing a lot about having FSB:DRAM ratio of 1:1. If I have my FSB freq at say... 400, do I manually go in and change the RAM frequency to... 800... (is that right?)... or am I still mixed up on this?
Also, I tried to run OCCT just to see what would happen. It was running okay but when I saw my core temps go into the 80's I shut it down. What temps are danger zone for Q9550? Thanks.

The 1:1 FSB:Dram ratio is not preferred, generally speaking, frequency and memory subtimings have more of a bearing on performance than FSB:Dram ratios. The only thing that I have personally noticed with running lower FSB:Dram ratios/strap settings is that it puts less stress on the memory controller, however your results may vary. I was running on an X48 chipset, your running on a more efficient P45 chipset.

You definitely should not be so hot on your Q9550, I would double check that your cooler is seated properly, and that you have the proper amount of TIM applied. I was running mid 60's at full load at the same frequency with a Xigmatek DK in prime, and low 70's with IBT. You can go all the way to TJmax technically, but I wouldnt go over 75c for 24/7 usage. Thats just my personal preference, everyone has their own ideas on acceptable risks.

Either your running too much voltage, your cooler isnt seated properly, or your putting out more heat than your cooler can dissipate efficiently.

While IBT is good for finding gross instability, I find that Prime 95 is slightly more effective for determining overall system stability. I recommend 12 hours minimum of small FFT's to determine CPU stability, and 12 hours minimum of Blend for testing the memory controller/memory for instability. Most people do 12 hours of blend only, but I find that running just blend doesnt do an adequate job of determining overall stability.
post #2279 of 2408
Quote:
Originally Posted by Mattb2e View Post

The 1:1 FSB:Dram ratio is not preferred, generally speaking, frequency and memory subtimings have more of a bearing on performance than FSB:Dram ratios. The only thing that I have personally noticed with running lower FSB:Dram ratios/strap settings is that it puts less stress on the memory controller, however your results may vary. I was running on an X48 chipset, your running on a more efficient P45 chipset.
You definitely should not be so hot on your Q9550, I would double check that your cooler is seated properly, and that you have the proper amount of TIM applied. I was running mid 60's at full load at the same frequency with a Xigmatek DK in prime, and low 70's with IBT. You can go all the way to TJmax technically, but I wouldnt go over 75c for 24/7 usage. Thats just my personal preference, everyone has their own ideas on acceptable risks.
Either your running too much voltage, your cooler isnt seated properly, or your putting out more heat than your cooler can dissipate efficiently.
While IBT is good for finding gross instability, I find that Prime 95 is slightly more effective for determining overall system stability. I recommend 12 hours minimum of small FFT's to determine CPU stability, and 12 hours minimum of Blend for testing the memory controller/memory for instability. Most people do 12 hours of blend only, but I find that running just blend doesnt do an adequate job of determining overall stability.

So I think my current FSB:DRAM ratio is 3:5. Is that okay?

As for my overheating... I'm running my Vcore @ 1.26 (even though CPUZ shows lower) and it's rated for that I think, so I'm going to reseat my cooler and see what happens. My cooler's a ZALMAN CNPS9500 AT 2 Ball CPU Cooling Fan/Heatsink, which I though looked effective enough (all copper with the equiv. of 6 heat pipes), so I'll see what happens. If that doesn't do anything I guess I'll start pulling back on some voltage. thumb.gif
post #2280 of 2408
Quote:
Originally Posted by Slavenhook View Post

So I think my current FSB:DRAM ratio is 3:5. Is that okay?
As for my overheating... I'm running my Vcore @ 1.26 (even though CPUZ shows lower) and it's rated for that I think, so I'm going to reseat my cooler and see what happens. My cooler's a ZALMAN CNPS9500 AT 2 Ball CPU Cooling Fan/Heatsink, which I though looked effective enough (all copper with the equiv. of 6 heat pipes), so I'll see what happens. If that doesn't do anything I guess I'll start pulling back on some voltage. thumb.gif

Well, I am fairly certain that unless you have a golden chip, you are not going to pull off that voltage at that clock frequency. Additionally, to have 80c temps at such a Vcore voltage is insane, something is going on with your cooler.

You will need more than likely around 1.3v in order to achieve stability at 3.8ghz. For my chip, I was 24 hour prime stable at 1.3625v.

As far as you your CPUz reading. It will always be lower than what you set in bios. The difference between your bios setting and CPUz readout at idle is called Vdrop. Voltage drop between idle and load is called Vdroop. A feature called LLC (load line calibration) effectively removes Vdroop and Vdrop by circumventing the design characteristics of the CPU power system. With a board that has different levels of LLC, like mine for instance, with a max LLC setting my bios Vcore and idle CPUz vcore are almost identical. My load voltage is the same also. The only issue with LLC is that when your CPU goes from idle to load, and load to idle, the voltage will spike above the setting that you have dialed in. For this reason, LLC can be dangerous at higher vcore settings, as the voltage spike could lead to premature degradation or failure.

Your motherboard may only have a couple options for LLC, more than likely you will have the option to enable it, or disable it. If your board is anything like mine was, the LLC left something to be desired. In my case, even with it enabled, my Vdroop was still present, although not nearly as severe.

At any rate, the first thing you need to get situated is your cooling. 80c is no good, especially at that voltage. I suspect you have air bubbles in your TIM, or your cooler isn't secure. Let us know what you find!
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › _₌= The Q9550 Club =₌_ [Please read first post]