I7 920 system acting strange

[Giorgio]

I'm currently building my new crunching rig. As most of you know it is not the first time that I assemble a PC so I was rather convinced that it would be a "walk in the park". hm, so it is not...

My config:
- CPU: Intel I7 920 D0 stepping
- CPU cooler: CoolerMaster V8
- Motherboard: Asus P6T
- RAM: OCZ 3 * 2GB PC3-16000 (DDR3-2000MHz)
- VGA: Gainward GTX260 Golden Sample, 896 MB DDR3 (216 cores)
- PSU: CoolerMaster M850
- Case: CoolerMaster HAF932
- Fan-controller: NXZT Sentry LX

As the part do not come from 1 shop and arrived over a couple of weeks timespan I started building up the rig as parts came in. 2 days ago the last parts arrived and so I thought I was ready to put on OS on the machine. euh not, the thing did not post or boot, no a single sound on the speaker either. => feeling the stress?

Checked everything several times but nothing helped (went through the normal process of disconnecting all peripherals etc, nothing helped). Then got the board out of the case and installed everything on my desk (with another PSU) went through hundreds of posts about problems with the P6T on multiple forums and after some messing around with the memory => bingo. The machine booted and complained immediately that I needed to upgrade the bios (still had V202 while V603 is available) to be able to support the CPU. Did everything, booted a couple of times, installed Windows and went to bed happy.

Today I installed everything back in the case and guess what... no go again. Got everything out again, had some problems getting it to boot again but once it did it continued to work correctly, so I tested it further on my desk and it is crunching for the moment.

Tomorrow I'll be testing the other PSU (with another motherboard) and I'll check if maybe something in the case is making a bad contact or short circuit.

Giorgio

[PowerToTheUsers]

Tomorrow I'll be testing the other PSU (with another motherboard) and I'll check if maybe something in the case is making a bad contact or short circuit.

My thoughts go that way too

[DevinK]

PSU should provide enough juice to power up the system, maybe a shortage like you said.
Googled for M850 + P6T issues? Or else, give the ATX connector a good check, could be a dodgy wire from any of the voltrails. Also possible: some pins of the mobos backsides touching the mbtray --> again you probably nailed it by saying shortage.

[Giorgio]

As I have 3 CoolerMaster M850 PSUs and the P6T is running now on one of them I assume it is not a compatibility problem between those. So today I tested the PSU which I had mounted in the HAF932 with another (heavy) config. The machine booted up correctly and ran fine. So I really have to look into the short circuit possibility. I'll put the PSU back in the case and test it with another (µATX) board.

There is nevertheless something strange with the config of the P6T rig. A I7 920 is a 2.66GHz CPU (multiplier 20). When looking at the detail with the newest version of CPU-Z it give me 2.8GHZ with multiplier 21? And I did not (yet) overclock anything...

Giorgio

[DevinK]

Turbo mode enabled? When using only 1 or 2 threads turbomode kicks in by upping the multiplier 1 step.

D0 should be a very good oc'er. Only problem is the heat.

What's your opinion about i7? Never had the chance to test one, is it as fast as they say?

[Raby]

Turbo mode enabled? When using only 1 or 2 threads turbomode kicks in by upping the multiplier 1 step.

Indeed it's called Dynamic Speed Technology. With the I920 the multiplier can vary between 12 and 22 !

D0 should be a very good oc'er. Only problem is the heat.

150W idle and over 500W with everything on (HT+DST)

OC : over 3.8Ghz on air cooling ...

[UnderGod]

Indeed it's called Dynamic Speed Technology. With the I920 the multiplier can vary between 12 and 22 !




150W idle and over 500W with everything on (HT+DST)

OC : over 3.8Ghz on air cooling ...

150W idle??? :o 500W with everything on????
that thing eats my octa for breakfast o_0

[Raby]

Here is a little graph from Tom's hardware. Of course the CPU is not responsible alone for the final result (TDP 130W though) but it's still impressive ...



Even bare bones systems suck between 120 and 220 W...

[Giorgio]

Turbo mode enabled? When using only 1 or 2 threads turbomode kicks in by upping the multiplier 1 step.

D0 should be a very good oc'er. Only problem is the heat.

What's your opinion about i7? Never had the chance to test one, is it as fast as they say?

Turbo mode was not enabled, I even do not see such an option in the bios. But the multiplier settings was on "auto" so indeed for some reason the bios decided itself to use 21.

As at this moment I'm still struggling to get the system stable with standard settings I'm not yet trying out the OC capabilities. I nevertheless already had a problem with the heat but it was probably a problem with the thermal compound which was not 100% OK. I deinstalled the cooler, re-applied the compound and reinstalled the cooler and the temp went down with about 8-10 degrees.

In the meantime I had the time to test the PSU with another (heavy loaded) machine and I also installed it with a MicroATX motherboard in the HAF932 case. Both tests were OK so the problem is the combination of the case and the P6T motherboard. I also think to have found the problem itself. The HAF932 has a square hole in the rear of the motherboard to allow you to easily mount/change the CPU-cooler. The edges of the hole are folded and I see that the 2 lower bolts of the cooler (which are in fact lower then the hole in the plate) are touching the plate. As this is currently the only thing which could be a reason to cause some sort of short circuit I think I'll have to get the Dremel out and extend the hole with about 2 cm in the direction of the bottom of the case.

Giorgio

[PowerToTheUsers]

Turbo mode was not enabled, I even do not see such an option in the bios. But the multiplier settings was on "auto" so indeed for some reason the bios decided itself to use 21.

There are two things: a CPU can automatically use a +1 multiplier to boost it's performance (I think it was when the temperature allowed it, or when not all cores were fully stressed: lower multi on unused cores, higher multi on used cores). but on the other hand: some mainboards can force the cpu to always use the higher multiplier: that can be set in the BIOS manually.

So I guess you experienced the first method: automatic boosting.