Project: Desk Rad

[JohnTedd]

Thanks for the reply, did you find it awkward getting under the piece of metal where the pilot holes for the fans are?

Thanks again, Joe

Luckily my hole was under one of the plates in the middle of the RAD rather than one of the corner ones. It was quite easy to lever the metal plate out of the way to get a good look at the hole. I was able to clean the area around the hole which is also a very important thing to do. I think I scratched it with something sharp. TBH I can't remember now. I also have a handy USB microscope which is great for getting a good look at the hole and later your repair. Failing that I bet a magnifying glass (or very good eyes) would suffice.

I do suggest building it up in layers rather than trying to fix it in one go with a large lump. I think I was lucky that my hole wasn't too big and so two layers were enough.

 

[JohnTedd]

More goodies have arrived.... In Piggin style, here's the unwrapping.

containing...

The main thing being the EK RAM WB:

Which contains:

and

The block itself is very heavy, so I am pretty confident it is going to do a good job. Hopefully I'll get to fit it tonight....

More photos once it is fitted.

 

[JohnTedd]

As promised here is it being fitted. After the heatsink fins are removed you put on MX3:

Then put the block on:

I think it's already looking better than the fans...

And with the pipes on:

and from another angle:

The result is much quieter. Later I will try overclocking the RAM much more than before...

 

[JohnTedd]

Although I have used heat pads on the GTX 480 I think they are most useful when the things you are cooling are at different heights. So I think thermal paste is fine for DIMMs as long as you believe they are all the same height. Otherwise pads would be better.

EK don't supply pads or TIM with the EK RAM Dominator, but they specify using TIM, so that's what I did (and, to be honest, I don't have any spare thermal pads lying around).

I've been a bit busy lately so I haven't had the chance to do any more overclocking yet. I can't see that situation changing in March... May be I'll have time in April.

 

[JohnTedd]

I think you will be fine with TIM and let's be honest it's not a huge deal to wipe off the TIM and re-apply. I had to do that a number of times with my i7 in order to get the CPU to see all 12 GB of RAM....

 

[JohnTedd]

That's pretty bad, CPU seating should not be that temperamental

I absolutely agree! I couldn't believe that the CPU would work but could not address all the memory - that just seemed weird to me. It's not as if it just happened to me, as this was the advice from corsair support and googling for the error showed others who had solved the issue by re-seating the CPU, including someone else on this forum.

 

[JohnTedd]

Sorry I missed this the first time m8.

Thanks for the mention in your build log.

Keep up the good work

No problem.

I certainly have plenty more to do but as the computer is working it is taking a back seat at the moment. Hopefully next month I can re-site the bigNG to keep it cool and re-route all the cables tidily.

 

[JohnTedd]

I guess if someone were really interested they could look at the pinout and work out where the high address lines are and see if they are somewhere peripheral. I also guess it points to the QA-ing of the sockets not extending to all the address lines or something like that.

Never mind, it is working well now. Nice and quiet with the RAM cooler

 

[JohnTedd]

Thank you for your comments.

I hadn't originally planned to water cool the RAM, but the sticks came with fans which were the noisiest thing in the system and drove me crazy. The good news is that with the watercooled RAM I can overclock more than before. I have been running my 1600 MHz RAM at 2000 MHz for ages now without any issues with the CPU running at 4.4 GHz.

 

[JohnTedd]

Having seen the GIGABYTE Masters of Overclocking comp I decided that since I had a Gigabyte board I would try some further overclocking....

In order to get some extra cooling I stuck my air conditioning unit behind the desk at full blast straight onto the rads. It was a bit awkward as the aircon unit is too high to hit the rads, so I raised the desk temporarily on two boxes to get the full effect of the aircon unit.

In this shot you can see the aircon by the rads:

This reduced my CPU temps by about 10-15 deg C so I was optimistic about getting some further overclocking, especially given that my CPU, NB, RAM and GFX cards are all watercooled.

Unfortuntely in the time I had I didn't manage ANY further overclocking

I will have another go, when I have some more time, but I am not that optimistic that the CPU can go higher. That is to say, without tweaking the voltages (which I have never done to date, so I'd need to learn a lot more about it).

 

[JohnTedd]

I enjoyed doing it, but I was disappointed with the results. Gotta have another go!

 

[JohnTedd]

I guess I will have to read up on what voltages to tweak then!

I also need to know how to underclock for the competition. I thought that would be a breeze, but even modest amounts of slowing down result in a non-booting PC and I have no idea why!

 

[JohnTedd]

During the week I did briefly up the Vcore quite significantly and got all the way to 5400 Mhz (200 MHz x 27 Mult x Vcore 1.8). It wasn't a stable configuration, but it kept going for long enough to qualify for the competition.

On Sunday I re-deployed the aircon unit, this time squarely facing one of the RADs. I got about 10 deg C cooling this time, perhaps not as good as the first time - no idea why. Idle temps came out at 22-25 for OC with aircon as opposed to 30-33 w/o aircon. I couldn't get a stable configuration above 5.1 GHz and a few times the temps shot up to 84 deg C!

My last test I went for a 3DMark Vantage score using 5 GHz and OC twin GTX480s and got a fully qualifying score of 40256 (using CPU Phyx) which I am pretty pleased with.

Afterwards I did some more tests w/o the aircon and soon found out that the aircon did make a +ve difference to the amount of OC possible. I don't know if it was these latter tests, but now my CPU is sick. I couldn't work out what was wrong, but I've disabled 4 of my 6 cores and now the PC boots. So it looks as though at least one core has gone kaput! I am going to try to RMA it, although I am not sure what I am supposed to say about OCing...

Anyone got any advice?

 

[JohnTedd]

Is it the CPU or motherboard.

It is definitely the CPU cos the whole computer runs fine when only 2 cores are enabled.

I doubt there will be any physical damage, so I will give the RMA a go and deny any OC. I will let you know how I get on.

 

[JohnTedd]

Well it took a month, but I finally got a new CPU from Intel and my computer is back working again with all 6 cores

The RMA was a little contorted as the original supplier of the CPU no longer stocked them so they suggested contacting Intel directly. Intel wanted me to test the CPU in another motherboard to confirm it was the CPU. As I didn't have another socket 1366 board to hand (nor any friends with them) I ended up sending the CPU back to the supplier who then tested it and worked with Intel to get me a replacement. It took a long time for Intel to cough up a new chip, but perhaps that was because it had been superseded by a newer version and they had none in stock either. Shame they didn't decide to give me the newer one, but I guess I can't really grumble as I now have a working computer again (albeit running with the stock settings ATM). Hooray!

 

[JohnTedd]

Thanks. I am not sure I am going to mess with the Vcore again. At least not with my primary CPU....