**Official** Asrock Z68 Extreme4 thread. Share your experiences, OCs & tech issues!

[jpm]

SB chips are odd things, you can pump more and more volts thrugh them and not gain any stability. I'd work on getting a slightly lower OC stable, then try simply turning up the multi. Every chip has its limits. Regards temps, simple logic. Go play any game or run any real world app thats considered demanding and I bet your temps will top out around 65c. Prime / IBT / whatever are designed to absolutely ravage the CPU, effectively showing instability that will only occur in situations you aren't ever going to run into.

not to stir up the pot, but i can understand the mentality of wanting to pass an extended torture test. systems that fail due to electrical instability tend to fail during peak bursts of activity, and this is generally the most inconvenient time for the system to fail. this is, for example, when someone in my house is streaming a transcoded video, and while downloading some torrents i start processing RAW photos, which momentarily tangle up my CPU with bursts of thumbnail generation, high memory usage, disk I/O, and JPG rendering. the only way to know that a system has the capacity to power through such peaks is to hit it with a maximum amount of randomized data processing at the maximum expected ambient temperature.

having a high-performance system that can run 24/7 without crashes or necessary reboots for days/weeks/months is a worthwhile pursuit, IMO.

[Ket]

I've found that to be absolutely untrue. Many times I used to run OCCT or something similar and OCCT would say all is well. I would then fire up a game or start BOINC and within minutes or even seconds the system would BSOD. The ONLY reliable way I've found to test any system is to run BOINC, it will load all the cores it detects to 100% and computes raw data, if a OC'd system survives BOINC, you ARE good. I've never had a single stability problem on my systems or other peoples when using BOINC as a test for stability. I run SETI with BOINC, and over the space of 2 years or so not once has it been inaccurate when I've been looking for the max stable OC.

[jpm]

I've found that to be absolutely untrue. Many times I used to run OCCT or something similar and OCCT would say all is well. I would then fire up a game or start BOINC and within minutes or even seconds the system would BSOD. The ONLY reliable way I've found to test any system is to run BOINC, it will load all the cores it detects to 100% and computes raw data, if a OC'd system survives BOINC, you ARE good. I've never had a single stability problem on my systems or other peoples when using BOINC as a test for stability. I run SETI with BOINC, and over the space of 2 years or so not once has it been inaccurate when I've been looking for the max stable OC.

it sounds like your objection isn't to the principle of stress testing, but rather the effectiveness of the specific synthetic tests that are often utilized?

[Ket]

Yep, precisely. IBT w\ AVX I've found to be the most accurate stability tester, but its still not as good as BOINC.

[jpm]

Yep, precisely. IBT w\ AVX I've found to be the most accurate stability tester, but its still not as good as BOINC.

fair enough. in the past i've combined prime calcs with looping 3DMark demos and video encoding. generally the more things the system is actively switching between, the more quickly it tends to trip up. will give IBT a try when i finally get to put together my asrock build next week.

for BOINC i presume the particular project choice is irrelevant? any particular settings involved that you have found best expose instability?

[Heavybomber]

SETI@Home(SSE3 + versions) & Einstein@Home BOINC application(I've ran only these from CPU bound apps) can ruin any "stable" system if it has weak points. The key is, they not only load the system like LinX(with no AVX) does, but also one will run them for 24/7 usually. If they don't make your machine to fail today, same cannot be expected for tommorow or day after tommorow etc.
I run SETI@Home on my 2500K(4.5 GHz @ 1.26 V, moderately clocked as it is not main crunching unit in the system, two GTX 460s are) and it is ultimate test. Of course, no one would install it just for testing stability.

[Ket]

fair enough. in the past i've combined prime calcs with looping 3DMark demos and video encoding. generally the more things the system is actively switching between, the more quickly it tends to trip up. will give IBT a try when i finally get to put together my asrock build next week.

for BOINC i presume the particular project choice is irrelevant? any particular settings involved that you have found best expose instability?

I just run SETI and make sure all CPU cores are @ 100% and let it run. Literally a 100% accurate test for me. When I first clocked my SB system I used SETI while playing a game. 8-10hrs passed (it was a sunday, gaming for long spurts is allowed :p ) and not a single glitch.

[noddi]

Oh i forgot to ask, My voltages across all my rails are lower than what other people get...
What does it mean?



Oh and my Core #1 temps off compared to my other cores... Been like that since I've got it... Anyone know why?
Edit: Added Load temps

[Anusha]

Man! 40C under load? *faints*

Ket, my man,

Today in the morning, I ran LinX with AVX with 6GB RAM and the temps were reaching almost 90C in a few minutes as per CoreTemp. But, when I touched the heatsink (all be it the top of the sink which may not be as accurate as the heatpipes or even fins in the middle), it was only slightly warm, say around 40-45C give or take. The air coming out of the exhaust was negligible too. For example, when my HD5870 is @70C, the air it puts out is very hot. Maybe it won't be same with the CPU, but I would expect the HS to be at about 60C, when the CPU core was at 90C, would you not?

So I think the problem could be with the contact. I'm certain it is not with the mounting, but could be with the thermal paste (it is one year old anyways), how much paste was used and how flat the HSF base is.

But my idle temps are very good, compared to ambient. Around 2-5C higher than ambient. One core is notably cooler.

So what I am going to do when I get home is this.

1. I would remove the left side panel and aim a fan towards the inside of the case.
2. If it doesn't bring down the temps, it means the case airflow is OK. So I would take out the HSF, and see if I have not applied thermal paste correctly, and I will use the same old thermal paste again.
3. If it doesn't work, I will use the hotplate and see if the heat gets transferred properly to the fins. Of course I cannot use the fans nor control the temp of the hotplate, but I will turn it ON for a few seconds and place the HS for a few seconds and see how it works. If it doesn't transfer heat quickly, something is wrong with the cooler itself. Heatpipe mechanism?
4. If there seems to be no problem with the cooler, I would lap it!
5. If all else fail, I will buy a new cooler because this has to stop! >_<

[noddi]

Man! 40C under load? *faints*

Ket, my man,

Today in the morning, I ran LinX with AVX with 6GB RAM and the temps were reaching almost 90C in a few minutes as per CoreTemp. But, when I touched the heatsink (all be it the top of the sink which may not be as accurate as the heatpipes or even fins in the middle), it was only slightly warm, say around 40-45C give or take. The air coming out of the exhaust was negligible too. For example, when my HD5870 is @70C, the air it puts out is very hot. Maybe it won't be same with the CPU, but I would expect the HS to be at about 60C, when the CPU core was at 90C, would you not?

So I think the problem could be with the contact. I'm certain it is not with the mounting, but could be with the thermal paste (it is one year old anyways), how much paste was used and how flat the HSF base is.

But my idle temps are very good, compared to ambient. Around 2-5C higher than ambient. One core is notably cooler.

So what I am going to do when I get home is this.

1. I would remove the left side panel and aim a fan towards the inside of the case.
2. If it doesn't bring down the temps, it means the case airflow is OK. So I would take out the HSF, and see if I have not applied thermal paste correctly, and I will use the same old thermal paste again.
3. If it doesn't work, I will use the hotplate and see if the heat gets transferred properly to the fins. Of course I cannot use the fans nor control the temp of the hotplate, but I will turn it ON for a few seconds and place the HS for a few seconds and see how it works. If it doesn't transfer heat quickly, something is wrong with the cooler itself. Heatpipe mechanism?
4. If there seems to be no problem with the cooler, I would lap it!
5. If all else fail, I will buy a new cooler because this has to stop! >_<

I presume you're OCed right? If not then there is def something wrong with the contact if its not able to lose the heat.
I remember mine hit about 90 Under Linx/IBT when i was at about 1.37v, only because my case had really bad cooling.

If you're using a stock cooler, i wouldn't hit over 1.3-1.35v with it.

[Anusha]

Overclocked to 4.7 with 1.37 but cannot test. Using TR MUX-120 with two silent fans in push/pull.

[noddi]

Hmm that's odd. Make sure when you reseat ur HSF dont put too much thermal paste, and not too little. I tend to use a small plastic card and spread the thinnest layer of thermal past i can over the cpu before i apply the HSF.
If your HSF is fastened on the mobo with screws, try screwing it on tighter, sometimes people don't put the HSF on with enough pressure applied to the CPU (be careful not to apply too much pressure). Too much thermal paste combined with little pressure will be bad when it comes to heat transfer.

If you can't test 4.7 just go for 4.5 which you can run at 1.29-1.32v. That way your temps will be a lot lower.

[Ket]

Oh i forgot to ask, My voltages across all my rails are lower than what other people get...
What does it mean?



Oh and my Core #1 temps are way off compared to my other cores... Been like that since I've got it... Anyone know why?
Edit: I have already reseated my cooler so i don't think that is the issue.
Attachment 116811
Edit: Added Load temps

One core will always be cooler than the others, I have the same thing and I'd expect 99.99% of other people do as well. The software isn't inaccurate, it'll tell you things as it is. You probably get lower readings on your rails than others simply because your system is pulling a bit more juice and / or your PSU isn't as powerful as what other people are using. Probably everybody would say my 850w Corsair unit is way overpowered for my system, but I look at things as whats the point in getting a PSU thats "enough" when for a extra £20 or so I can get something thats not only plenty powerful but more futureproof - saving me money in the long run because I know the PSU won't be a limiting factor for a long time, thus negating the need to replace it.

Man! 40C under load? *faints*

Ket, my man,

Today in the morning, I ran LinX with AVX with 6GB RAM and the temps were reaching almost 90C in a few minutes as per CoreTemp. But, when I touched the heatsink (all be it the top of the sink which may not be as accurate as the heatpipes or even fins in the middle), it was only slightly warm, say around 40-45C give or take. The air coming out of the exhaust was negligible too. For example, when my HD5870 is @70C, the air it puts out is very hot. Maybe it won't be same with the CPU, but I would expect the HS to be at about 60C, when the CPU core was at 90C, would you not?

So I think the problem could be with the contact. I'm certain it is not with the mounting, but could be with the thermal paste (it is one year old anyways), how much paste was used and how flat the HSF base is.

But my idle temps are very good, compared to ambient. Around 2-5C higher than ambient. One core is notably cooler.

So what I am going to do when I get home is this.

1. I would remove the left side panel and aim a fan towards the inside of the case.
2. If it doesn't bring down the temps, it means the case airflow is OK. So I would take out the HSF, and see if I have not applied thermal paste correctly, and I will use the same old thermal paste again.
3. If it doesn't work, I will use the hotplate and see if the heat gets transferred properly to the fins. Of course I cannot use the fans nor control the temp of the hotplate, but I will turn it ON for a few seconds and place the HS for a few seconds and see how it works. If it doesn't transfer heat quickly, something is wrong with the cooler itself. Heatpipe mechanism?
4. If there seems to be no problem with the cooler, I would lap it!
5. If all else fail, I will buy a new cooler because this has to stop! >_<

Seems like your problem is contact. Before going to all that hassle I would simply take the cooler off, clean it and re-apply TIM. Roughly the thickness of a expensive piece of paper. Alternatively, you could blob on about 1.5 grains worth of rice in the center of the CPU and let good old fashioned pressure do the rest, sometimes I get better results by doing this than spreading the TIM out.

[Anusha]

the thermal paste is a bit thick, so i cannot apply it as a layer. i tried, but the temps shot past 90C within a couple of minutes! and when i removed it, it wasn't looking nice on the bottom of the HS. so i applied a grain in the middle, and it brought down the temps to 80C. probably the thermal paste is no good anymore? should i buy some new thermal paste and see? but if it still is crap, i would still have to buy a new cooler and the money wasted on thermal paste for no reason. that always happens to me. >_<

for the moment, i dropped clocks down to 4.5GHz @ 1.32V in bios, but with level 2 LLC so only at 1.29x at max load. 50 passes of LinX with AVX went through with 2GB RAM work load.

[Ket]

80c with anything like IBT w\AVX utilising all the RAM possible is a very good temp. Short of going water temps won't get much lower than that.

[Anusha]

hehe. so you really want me to keep my current cooler ha? :P

anyways, from what i could see, i would need 1.37V+ for 4.7GHz, which i wouldn't try even if i had a better cooler. max i would feel OK is 1.35V. so, maybe you are right. i will try to get 4.6ghz stable on THIS cooler. maybe get ~85C for a few hours won't hurt the CPU much aye? (btw, just so that you know, my ambient is a whopping 31C right now!)

maybe some cheap MX-2 wouldn't hurt either.

[Ket]

If you get new TIM, make it good stuff like MX4, MX2 is just the stuff people should use when overhauling somebody elses system as its still a million times better than the crap TIM pre-builts use.

[Anusha]

of course MX4 it is. but they are not shipping it tomorrow! the bastards! i will have to get it on Thursday.

[Ket]

By Thursday I'll have my new games Been a while since I played any proper PC games, been hooked on football manager

[JoOs7eR]

Hi guys,

having some problems over here..

2600k with z68 extreme 4

on 4.7Ghz 1.3vcore / PLL voltage enable / LLC level 1 / pass IBT / Prime for 6 hrs with no errors

i will get BSOD when i surf the net randomly or during a game

any idea what is the problem ?

Cheers

[richierich]

Hi guys,

having some problems over here..

2600k with z68 extreme 4

on 4.7Ghz 1.3vcore / PLL voltage enable / LLC level 1 / pass IBT / Prime for 6 hrs with no errors

i will get BSOD when i surf the net randomly or during a game

any idea what is the problem ?

Cheers

Sounds like memory settings are off. Same thing happened to me before I tweaked my sub-timings.

[Ket]

Hi guys,

having some problems over here..

2600k with z68 extreme 4

on 4.7Ghz 1.3vcore / PLL voltage enable / LLC level 1 / pass IBT / Prime for 6 hrs with no errors

i will get BSOD when i surf the net randomly or during a game

any idea what is the problem ?

Cheers

Set your memory frequency, voltage and timings manually. You might need to pump up vcore and VTT a tad as well depending if you are OCing your memory or not.

[Anusha]

Well, if the error code in the BSOD is 124, you need more Vcore. I know, it is weird that LinX pass but still crashes when doing absolutely nothing. Happened to me. Had to increase Vcore. 4.7@1.3V might not work on almost all the chips.

[JoOs7eR]

Thanks for all the replies...

the rams i'm running on default

GSKILL 2x4GB 2133Mhz rated 1.65V

F3-17000CL9D-8GBXLD
System Type DDR3
Intel XMP certified, designed for Intel P67 & Z68
CAS Latency 9-11-9-28-2N
Capacity 8GB (4GB x2)
Speed DDR3-2133 (PC3 17000)
Test Voltage 1.65 Volts
Registered/Unbuffered Unbuffered
Error Checking Non-ECC
Type 240-pin DIMM

didnt change the timing for the rams , voltage was set to 1.65v manually..

[MacClipper]

on 4.7Ghz 1.3vcore / PLL voltage enable / LLC level 1 / pass IBT / Prime for 6 hrs with no errors

Post your LinX and HW Monitor screehshots showing the GFlops and Processor Powers reported.

Anyway most likely, inadequate vcore still.