This is part of the applied piece for my senior project. Mostly self explanatory, but I did DICE cooling on a pc I had and tested gaming results. This is mostly to share my experiences with DICE and how all that turned out and such!
If you happen to be a gamer, like myself, then you undoubtedly want the best performance you can get. Also, if you're like me, you don't have an unlimited budget to spend on the highest end CPU's and GPU's. Believe me, if I could fork over $600 and not think twice for high-end components, I would do it. But alas, I lack that kind of disposable income, so I turn to alternative ways to increase performance in games. Most people that are in the 'inner circle' of gaming would tell you right off the bat that overclocking, increasing the clock speed of the computer through the bios or programs in windows, is the way to go if you want free performance. Throughout my years of computer usage, overclocking is never something I've been shy about doing. Ever since I got my first computer, overclocking was in my blood. For the lack of income I have, the thought of breaking parts has never been a concern. I never thought of the risks, only the rewards, which is probably why I am embarking on my current undertaking of dry ice cooling. Dry ice (commonly referred to as DICE) cooling is the next step for me as my lust for performance hasn't been quenched by high performance air, or water-cooling, and the next logical step, for me of course, was an alternative and slightly dangerous cooling method, reserved for the elites of the computer enthusiasts. I myself wish to join that higher echelon of overclockers, and experience it for myself. I will be taking my current rig, put it under DICE and see what kind of performance increases I can get out of it in games, and then judge it to see if it is worthwhile at all to do on a semi-regular basis. My choice of games came from a desire to pick three games that would test my system in different ways, and would give me a better idea of the effect of DICE cooling on gaming.
My dry ice setup will be similar to a few setups, but with a few differences not commonly seen. My setup will consist of the standard copper pot with 2 ˝” wide, 1/8” thick pipe. To seal the socket and eliminate condensation on the back of the board, plasticine based modeling clay to surround the socket and keep the small electronics around the board from corroding. Acetone will be used in the pot to draw away heat. The pot will be insulated with fiberglass pipe insulation. The mounting bracket will be made out of wood because it is easy to work with and is strong enough to hold the pot to the motherboard. Surprisingly enough dry ice cooling is quite cheap. My whole system ended up costing $77.61 including the dry ice, which was $30. A good dry ice setup can be cost efficient, if you make everything yourself, so it makes for a good option for quick power at a small cost if the user can find a way to keep the processor cool without dry ice on stock settings.
AMD Athlon 64 4000+x2 (2.1GHz stock) (2.9GHz Overclocked)
Zotac 8200 Motherboard
NVIDIA 8200 on board GPU (v182.5 Drivers)
4GB OCZ HTC Reaper 800mhz 4-4-4-12
Western Digital 250gb SATA 3.0 HDD
Resolutions of 1280x800 and 1360x768
Settings on lowest and highest.
2, 5 minute rounds on de_dust2 and ce_italy maps.
Crysis Single player Demo
Resolutions of 800x600 and 1176x664
Settings on lowest and highest
10 Minutes of play time.
F.E.A.R Single player Demo
Resolutions of and 640x480 and 1280x960
Settings on lowest and highest.
5 Minutes of play time.
Game Avg (Stock) Avg (Overclocked) Avg % INC
CS: S Low Res, Low Settings 14.629 14.327 -2.10%
CS: S High Res, Low Settings 16.449 11.825 -39.10%
CS: S High Res, High Settings 14.543 10.309 -41.10%
CS: S Low Res, High Settings 16.546 12.175 -35.90%
Crysis Low Res, Low Settings 11.102 17.342 64%
Crysis High Res, Low Settings 9.154 10.879 84.10%
Crysis High Res, High Settings 1.961 1.787 -9.70%
Crysis Low Res, High Settings 2.733 3.031 9.90%
F.E.A.R. Low Res, Low Settings 24.147 60.207 249.30%
F.E.A.R. High Res, Low Settings 7.159 17.323 241.90%
F.E.A.R. High Res, High Settings 3.972 3.649 -8.80%
F.E.A.R. Low Res, High Settings 18.059 24.556 35.90%
My baseline testing of these games confirmed the results I was expecting. My low powered system couldn't handle the intensity of either Crysis or F.E.A.R. Even on lower settings and lower resolutions they brought my system to its knee's. Counter-Strike: Source on the other hand faired moderately better. Average FPS weren't playable by most standards (averaging about 15), but in game there was little jitter or skipping, and it played relatively smoothly. During firefights, there were noticeable drops in FPS but overall it wasn't as unplayable as Crysis and F.E.A.R were. It was surprising how demanding even an older game such as F.E.A.R (released in 2005) was on even minimum settings and the lowest supported resolution. Undoubtedly part of that stems from the low-end nature of my components, but even so it was brutal trying to play. Even outside firefights, the game play was a grainy slide show. It was obvious that Sierra tried to make the game playable for older systems without needing to upgrade much, but even on my somewhat new system, F.E.A.R crippled it. Crysis put my system in a similar headlock, with little wiggle room. In certain open areas the game play was quite smooth and I was actually hopeful that I could get some enjoyment out of the game on my system. That desire was soon greeted with a full on body slam as firefights and dense jungle created the same slide show effect that F.E.A.R had on my rig. Overall I expected as much out of these games on my system. I expected Counter-Strike: Source to be the most playable of the three, with F.E.A.R being less playable than CS: S but more so than Crysis, and living up to it's hype as a system killer, Crysis made a mockery of my system
Overall the data from the second round of testing is a little surprising, specifically in the Counter Strike: Source tests. From my experiences, CS: S is a very processor oriented and directed game, and in 3 out of 4 tests the overclocked test rig FPS decreased, while in F.E.A.R. and Crysis experienced very high gains in low setting tests. An unstable overclock can be ruled out because the other two tests ran as expected and without issue. Overall it is very interesting to see that on all high resolution with high settings tests the average FPS went down be a decent amount, but there were gains in the other three tests for both F.E.A.R. and Crysis. I expected that there would be a gain in the low resolution and low setting tests, but not as much in the high resolution and high settings tests, which proved true. My testing shows that an overclocked processor increases FPS in most settings (CS: S excluded) but the highest increases came in low resolution and low settings. This would be because the resolution and settings are least dependant on the GPU which is the weak link in my testing setup, and when the CPU was overclocked, it drastically increased FPS in F.E.A.R. and there was a noticeable difference in Crysis.
My results are good, but I had hoped for better. I got some interesting test numbers. It was very surprising that the CounterStrike: Source tests didn’t show improvements, and some more extensive testing might offer different results. In respect to the overall testing experience, the quality of my motherboard was lacking, and there weren’t as many overclocking features as I would have liked to put the dry ice to the test. Coupled with the processor cold bugging (it would shut off below -13*C) my ability to test was limited and my overclocking headroom was limited. To fully test the effect that dry ice can have in games, a more high-end gaming motherboard and a different processor, either Intel or a new Deneb core AMD processor, would be able to get the full effect out of the dry ice. One thing that I did notice while testing was that while idling I only needed to put in a single pellet around every 5 to 10 minutes to keep the CPU around 13*C to 20*C. While running a stress test program, I had to add dry ice more frequently, but it was still in small dosages, generally one or two pellets at a time, and would need to add it every 3 to 5 minutes. While it would be slightly annoying to keep adding every so often, it is a feasible way to keep a processor cold for a period of time. With the remainder of the dry ice kept insulated and cool, it could last close to a day or two, and with it being used 50lbs could easily last 12 hours. With a more efficient way of keeping the dry ice cool, it is quite possible to have a dry ice setup in use for upwards of 4-6 hours of constant heavy use with no problems or issues. There is the potential for a dry ice cooling setup to be useful for long gaming sessions, that is if there is a game that could take advantage of an extremely high clocked processor.