OT...I'm back

[Von Paulus 8]

My Gigabyte EP45-DS3L board from 2008 came with an @BIOS utility that makes updating the BIOS a fairly painless one-click procedure. I don't know if that will apply to you Olham, but perhaps check your disks and manuals that came with your board and see what they have in the way of BIOS update utilites or other info.

 

Actually 77Scout, don't do it again with @BIOS. Never, and I say never use any program to flash the motherboard bios, running in Windows.

People, this is serious. I forgot to tell that in my last post. I've been tempted (OK I've done a couple of times but these doesn't count ) to do it, because it was really easy instead of using DOS. Never upgrade your BIOS with a Windows flash software.

 

Here is a easy guide, done by my friends at TweekTown. This forum is 1000 times better than the official one. In fact it should be the official forum of Gigabyte.

http://forums.tweakt...lash-guide.html

Edited October 28, 2010 by Von Paulus

[Olham 164]

Thank you for your detailed answer, von Paulus. I have meanwhile checked the updates again,

and as you say - they wouldn't improve anything to solve my problem.

As I have several jobs now, which will earn some money, I think I will upgrade my rig to a new

mobo with DDR3 and the ability to run Windows7 64bit.

Not sure if OFF or my graphic card (ATI HD4870) would benefit from that, but I would be in the

"next generation" for further upgrades with this, I think.

A first step is already done - I have bought a new power supply (Cooler Master 600W).

[Von Paulus 8]

Can't you return that PSU and exchange for a powerful one? 600 it's on the edge. You should have bought more 100 or 200w.

How many 12V rails and how many amp? If you don't know where to look, give me the PSU reference and I'll look it for you.

When did you bought? You've at least 15 to 30 days to return it. At least in Portugal works that way.

 

I wouldn't change your 4870. I'd wait for the ATI 6xxx series.

Edited October 28, 2010 by Von Paulus

[Olham 164]

The PSU is a "Cooler Master Silent Pro M600".

 

Model descriptions:

RS-600-AMBA-D3:600W Active PFC Power supply

 

Specifications:

Type: Intel ATX 12V V2.3 & SSI EPS 12V V2.91

Input voltage: 90~264V (Auto Range)

Input current: 10A@115Vac/5A@230Vac

Input freq. range: 47 Hz ~ 63 Hz

Fan: Super Silent 135 mm fan

Power Good Signal: 100 ~ 500 ms

Hold Up Time: > 17ms

Efficience: 85% Typically

MTFB: >100.000 hours

Protection: OVP / UVP / OCP / OPP / OTP / SCP / OLP

Dimension: ATX x 150 x 150 x 86 mm

Operation temperature: 0~40°C (Nominal Input Voltage)

Regulatory: NEMKO / TUV / GOST / CE / C-Tick / UL

 

Why would you prefer a PSU with less Watt ?

[almccoyjr 7]

Just to chime in....Also check the temperature of the Northbridge chip. From what I've read so far, your crashes are definitely symptomatic of memory and or a localized heat issue. You may also want to check your memory timings to make sure that they haven't drastically changed.

 

plug_nickel (Al)

[Von Paulus 8]

I want more power not less, Olham.

 

That PSU is a single rail. 40A might be enough. Don't know what kind of future upgrades you want to do it.

I know that some components consume less, nowadays, but... if you overclock one day your system, you might need more power.

[Olham 164]

Can you recommend two or three PSUs you know for good, von Paulus?

Cause as you said before - I still could bring it back.

 

almccoynjr, I know it can be several things - one friend even said, he had such problems

amd searched his a*s off, until he found, it was the on/off button of his rig.

What does plug_nickel (AI) mean???

 

This is now the second day with no crash so far. As soon as I have the next crash, I will

try running it with only two, like Parky had suggested, and when it crashes again, I'll take

the other - just to sort out, if it is the memory.

Edited October 28, 2010 by Olham

[Von Paulus 8]

I've a Corsair HX-750. So I think I can recommend it.

I also like a German brand of PSUs, BeQuiet. BeQuiet support is great. I've lost some of the cables. I've emailed them if they could sell me the missing cables.

Not only they freely sent a full new set of cables as they didn't even charge for the transport. Thats what I call a good customer service.

So I recommend BeQuiet DarkPowerPro p8 750W.

[almccoyjr 7]

almccoynjr, I know it can be several things - one friend even said, he had such problems

amd searched his a*s off, until he found, it was the on/off button of his rig.

What does plug_nickel (AI) mean???

When you're system starts to run under heavier loads so do the various chips on the mobo. The NB is in particular a "normal" hot spot. When running all dimms populated, you put a heavier load on the memory controller which is incorporated in the NB chipset, EVEN if you're running memory/voltage/timings at box stock values. I run 8GB of memory, 4 2GB modules. The temp difference on my EVGA 680i-SLI using 2GB, 4GB and 8GB is very noticeable. With 2GB, temp was around 42-45C. With 4GB, 4 1GB modules, temp was 50-55C. With 8GB, 4 2GB modules, the temp was approaching 70C under full load. Using a different NB cooler/fan reduced the temp from 70 to 47C under load and about 32C at idle.

 

When a heat load involving the NB and or dimms happens and lockups occur, the core memory voltage can droop first and then spike when the lockup occurs, rippling through the chipset, resetting the voltage value and or timing values memory is operating under.

 

When you're trying to troubleshoot possible heat/memory issues, you MUST revert to box stock settings on the cpu, memory speed, timings and core memory voltage. You should then make sure you have no ongoing heat issues with the NB chipset, THEN follow Parky and Von Paulus recommendations on troubleshooting memory modules. As you repopulate the dimms, you'll need to keep an eye out on the NB temp and note differences. This will aid you in determining if a better NB cooler might now be needed. IF you're going to populate all dimms, adjust memory voltage and try and tighten timings, you WILL need a better NB chipset cooler, not just on yours, but on anyone's motherboard.

 

For the record, I'm running 8GB, 2GBx4 modules of GSkill locked at 800 with bus at 1066. Stock voltage is 1.80v; I run 1.85v. Timings are 4-4-4-12 2T. I run at 4-3-4-9 2T.

 

plug_nickel (Al)

[Olham 164]

I am only yet learning about this all, so I don't know what NB stands for?

 

I have reset everything to default values, but haven't seen the timing yet - were do I check those?

I know that such numbers are printed on the labels of my memory, so I can check how they should be.

But were do I find them in BIOS?

[Herr Prop-Wasche 7]

NB stands for North Bridge controller. It's a chip that controls memory functions, etc. and can get almost as hot as the CPU, but frequently does not have a fan on it. You can usually find NB heatsinks and fans for sale at various computer stores.

 

Memory settings and timings can be found in the BIOS when you start the computer. Look for DRAM settings or something similar in the BIOS menus.

[Von Paulus 8]

That's a fine explanation, almccoyjr.

I never imagined that the difference of temperatures could be so huge between having 2 or 4 sticks.

[almccoyjr 7]

That's a fine explanation, almccoyjr.

I never imagined that the difference of temperatures could be so huge between having 2 or 4 sticks.

Not all motherboards will exhibit such temp differences. You must take into account that my 680i-SLI is cerca 2006! However, you WILL have temp increases based on number of ram modules, their speed, their core memory voltage requirements and their timings. As you populate the dimms, you add NB M.C. stress. As you populate the dimms AND increase the physical memory of the ram, you increase the stress even more. This stress is the efficiency and activity of of the memory controller and that simply relates to heat. It gets worse when you start to oc the ram or tighten the timings, even when you lock the speed of the ram. Add to this that just about ALL motherboard mfgs. NB coolers "suck", IMHO, just complicates the problem(s) even more when trying to diagnose mobo problems. You really need to know what the local temp is for the NB at at idle and at full memory load, then start your upgrades, mods or whatever.

 

And let's not get started on positive/negative case pressure and how that relates to case cooling when those tiny NB heatsinks are working (and that a stretch).

 

Olham, go into your mobo's manual and review the sections on ram, voltage, timing etc and also make sure that the Corsair modules you're using are listed by Gigabyte as compatible for your board. If they're not, go to the Gigabyte forum and see if others are using your type of ram in the mobo. It helps to eliminate another possible "oh man, I didn't know that..." scenario.

 

And as you see, your mileage (temp) does vary.

 

plug_nickel (Al)