Alright, so now that the final push for the review is done, time to get back to work on this thread. I will start off with the insulation job for the LN2 session that was posted in the review. I will also be posting the benchmark results here sometime this week.
click for full size...
I start off with a 1/2" thick layer of armacell closed cell insulation cut from a large sheet. This is cut to fit perfectly inside the heat pipe cocoon and then a dremel with a dry wall carving tip is used to cut out the holes for the caps and inductors. Next is a double layer of the blue shop towels, thanks to Vince for the idea here. On top of that are two more layers of the 1/2" armacell with cutouts to accomodate the pot. Then the pot with a small donut makes an appearance on the scene. At this point the setup is ready for the inner armacell pipe insulation and a wrap of whop towels around the mounting hardware followed by another larger layer of armacell pulled over top the entire assembly. At this point we are snug as a bug and ready to rock.
click for full size...
There was one amendment that had to be made to the outer ring of armacell pipe insulation because of the Fusion block and it's flurry of worm clamps. When I find the thread that the stock heat sink uses I should be able to mount the MCW30 and will likely not have to have a chunk of the outer insulation missing. Benching was now in full swing and I finished a 9 hour LN2 session with the RE followed promptly by a nap. The board held up really well and the NB was great using just the Fusion block. It was suppose to be a feeling out session for the CPU as it hadn't been cold before but it turned into an all out benching romp when I easily found the CPUs limits. The RE benched at 600~615FSB on the 333Strap 5:8 ratio with a tRD of 9 the whole time. My E8600 had a crappy cold bug around -110C so 6100MHz was about the limit for all benches. Sorted a solid Pifast, 1M, and 32M for 2D which was followed by 01, 03, 05, & AM3 with CF HD4850s (volt modded but stock air) and single HD4850 01, 05, AM3. I have to go through all the results but here some photos...
click for full size...
From left to right, benching SPi early on, later I went back to it with a single HD4850 but I always start with my trusty 8600GTS with no drivers installed for 32M. My withdrawl device wouldn't build pressure so getting LN2 came down to manual labour. Setting the dewar on the milk crate at an angle let me rock it rather easily while using the second crate as a sort of safety stop in case I managed to lose the balance. It worked rather well but with a couple cracked ribs and recently separated left shoulder...it was a very tough 9 hours, considering I only had a 750ml thermos cup to fill
. The last photo was during tear down. There was the outer armaflex layer of insulation over what is shown here but there wasn't a drop of water in site. A good bit of frost under the shop towel but absolutely zero water anywhere...the setup could have easily gone a lot longer had more cold stuff been available. I then took a couple photos during tear down, not bad for 9 hours of being -105C at the CPU base.
click for full size...
As mentioned, the insulation held up remarkably well and will be my permanent insulation solution for this setup. Having a very alrge A/C unit with de-humidifier in the bench room really helps with condensation but even still, the setup stayed perfectly dry and was begging for more. Results from this session will go up before the weekend, followed by more benching, hopefully with some extra special GPU action
Bookmarks