DFI UT P35-T2R: Tweakers Rejoice!
- Posted in
- Motherboards
Going Sub-Zero With Our Cascades
Hardware used:
PCP&C; 1200W PSU
2GB OCZ Flex 9200 RAM
ASUS 8800 GTS 640MB
WD2500 KS 250GB Hard Drive
Intel X6800 Core 2 Duo CPU (Retail)
Dual Rotary Cascade (CPU cooling)
Dual Mini Cascade (GPU cooling)
Dimas Tech Benching Table
Moving over to sub-zero means that all voltage related conservatism was thrown out of the window, as evaporator loaded temps remained between -110 and -106C throughout testing. Due to the clean CPU socket area, neoprene insulation can be applied with relative ease, one issue with Digital PWM is that all the heat producing components are aligned on one side of the socket (to the left). Obviously due to a lack of heat producing circuitry around the top side of the socket, the associated area is very prone to condensation during extended benchmarking sessions, we recommend you take a little extra time checking insulation is sufficient enough to prevent any moisture penetration.
The initial plan was to scale up the settings which had worked well when benching the water-cooling setup with the QX6800. We had not accounted for the fact that our QX6800 was reluctant to run over 4.5GHz in both four and two core mode. It was taking copious amounts of time to try and hold the board steady with the QX6800 (both in two and four core mode) anywhere over 4.5GHz. We eliminated all possibilities of PSU based inadequacies by comparing clocking results on two separate models and also cross comparing motherboards by using the ASUS Striker Extreme.
Both boards exhibited the same CPU MHz wall, so it was time to move on to another processor. One other oversight is that we should have used an air-cooled graphics card while we were attempting to tweak GTL settings. The 8800 GTS was also being cascade cooled, and idle temps were hovering in the region of -81C. The side effect of leaving the card idling while trying to improve the CPU stability was that the whole graphics card froze over - lesson learned.
After a few hours of various attempts with several processors, our trusty Core 2 Duo X6800 was dusted off and given the thumbs up for a test run. After spending the better part of a day tweaking and twisting BIOS settings, we did manage to save a few screenshots with the X6800.
The 8800 GTS 640MB card clocked up really well at 940MHz on the core with around 1.55Vgpu while our X6800 ran without fault at these settings. It has to be said that the whole benchmarking experience, while tiring, was thoroughly enjoyable. Kudos to the folk who are able to manage LN2 on a couple of pots while trying to break records! Going sub-zero is not as easy as one would initially think. We will endeavor to provide more testing like this in the future on performance oriented motherboards with a standard set of peripherals; although the sacrifice in time is great, the end results are very rewarding and certainly a fun experience.
Hardware used:
PCP&C; 1200W PSU
2GB OCZ Flex 9200 RAM
ASUS 8800 GTS 640MB
WD2500 KS 250GB Hard Drive
Intel X6800 Core 2 Duo CPU (Retail)
Dual Rotary Cascade (CPU cooling)
Dual Mini Cascade (GPU cooling)
Dimas Tech Benching Table
Moving over to sub-zero means that all voltage related conservatism was thrown out of the window, as evaporator loaded temps remained between -110 and -106C throughout testing. Due to the clean CPU socket area, neoprene insulation can be applied with relative ease, one issue with Digital PWM is that all the heat producing components are aligned on one side of the socket (to the left). Obviously due to a lack of heat producing circuitry around the top side of the socket, the associated area is very prone to condensation during extended benchmarking sessions, we recommend you take a little extra time checking insulation is sufficient enough to prevent any moisture penetration.
The initial plan was to scale up the settings which had worked well when benching the water-cooling setup with the QX6800. We had not accounted for the fact that our QX6800 was reluctant to run over 4.5GHz in both four and two core mode. It was taking copious amounts of time to try and hold the board steady with the QX6800 (both in two and four core mode) anywhere over 4.5GHz. We eliminated all possibilities of PSU based inadequacies by comparing clocking results on two separate models and also cross comparing motherboards by using the ASUS Striker Extreme.
Both boards exhibited the same CPU MHz wall, so it was time to move on to another processor. One other oversight is that we should have used an air-cooled graphics card while we were attempting to tweak GTL settings. The 8800 GTS was also being cascade cooled, and idle temps were hovering in the region of -81C. The side effect of leaving the card idling while trying to improve the CPU stability was that the whole graphics card froze over - lesson learned.
After a few hours of various attempts with several processors, our trusty Core 2 Duo X6800 was dusted off and given the thumbs up for a test run. After spending the better part of a day tweaking and twisting BIOS settings, we did manage to save a few screenshots with the X6800.
![]() |
![]() |
![]() |
The 8800 GTS 640MB card clocked up really well at 940MHz on the core with around 1.55Vgpu while our X6800 ran without fault at these settings. It has to be said that the whole benchmarking experience, while tiring, was thoroughly enjoyable. Kudos to the folk who are able to manage LN2 on a couple of pots while trying to break records! Going sub-zero is not as easy as one would initially think. We will endeavor to provide more testing like this in the future on performance oriented motherboards with a standard set of peripherals; although the sacrifice in time is great, the end results are very rewarding and certainly a fun experience.
Originally published at https://www.anandtech.com/show/2354 (Wayback Machine).


