REVIEW - der8auer's BEAST CPU LN2 POT
MaadDaawg's Maad Review of ...
THE der8auer BEAST LN2 CPU POT Well, it finally arrived from Germany one cold Saturday Chicago afternoon. BEAUTIFUL is perhaps the closest word to define this pot as it arrived with the entire outside of the pot polished to a mirror finish. The inside, as you saw when he first posted the BEAST on the forum, is a work of art in and of itself http://img808.imageshack.us/img808/5144/img1369ps.jpg http://img717.imageshack.us/img717/6773/img1372mc.jpg http://img3.imageshack.us/img3/7138/img1374de.jpg The mounting bracket and backplate are made of POM, and are slotted to accommodate virtually any socket type. The threaded posts are heavier than the #6 rod used by Kingpin but still fit through the motherboard holes nicely. Each rod is held in place by 2 nuts with nylon washers (back plate), with the top bracket using black flat washers, springs, and a nice large thumb screw to secure it all to the motherboard (see first pic above). Scope of the Review Compare the performance of this new BEAST against the performance of the venerable Kingpin F1EE throughout each phase of a sub zero over clock session. http://img263.imageshack.us/img263/5393/img1368n.jpg In one corner, the contender,the der8auer BEAST, weighing it at an incredible;
And in the other corner, the reigning champ, the Kingpin FIEE, weighing in at a respectible;
So as our two heavyweight contenders face off, predicting the outcome before getting into the trenches would be impossible as both pots are truly massive. Obviously, the inside of the BEAST appears to provide a significant increase in surface area over that of the F1EE, but it remains to be seen if that is advantageous or not. (You can see the F1 has seen its share of LN2 and map gas flame!) BEAST http://img717.imageshack.us/img717/6773/img1372mc.jpg F1 http://img88.imageshack.us/img88/6070/img1373h.jpg Testing Parameters Each pot will be tested from start (ambient temps) to finish (pot removal at -18C) of an extreme overclocking run, and parameters will be measured as best as possible.
Testing Environment Testing of the F1 took place in the dark dungeon of the MaadDaawg's subteranean laboratory, test of the BEAST took place in the dining hall (due to circumstances beyond my control). Both environments were artificially lowered to an ambient temp of +20C. As this required a good deal of A/C, both environments were cool and dry. I had planned on using my state of the art Seiko Chronograph for timing the various phases. Unfortunately, the rigors of monitoring and maintaining the test temps made that virtually impossible, except, for the first stage of ambient to +60C. Test Bed Both units will be tested on a single test bed to insure results are not distorted by the performance of differing components. Test bed is comprised of;
http://img810.imageshack.us/img810/4128/img1380b.jpg BEAST http://img221.imageshack.us/img221/4...1395nrurvm.jpg F1 http://img838.imageshack.us/img838/1364/img1392x.jpg Insullation The motherboard was water proofed using malleable eraser on the top covering the entire area of the socket between he VRs, the HS, and the DRAM slot 1. Three gaskets were cut to fit over the uprights of the pot supports, two of shop towels and one of ½ Armaflex. The holes in the towels were cut to sit relatively tightly around the CPU while the Armaflex was cut wider to allow the bottom of the pots to stick through and make contact with the CPU. The back of the motherboard was fitted with 4 layers of Frost King cut to cover the entire board. Each pot had it's own Armaflex gasket cut to fit its shape. Each of the pots were first wrapped in blue painters tape, then one layer of Armaflex tape, and then covered by ½ Armaflex pipe wrap. After mounting, the insulated pots were wrapped in shop towels held in place by rubber bands. Initial Thoughts BEAST Likes
Questionable
F1 Likes
General Notes
I learned after publishing the review, that there is a good reason that the hole for the digital thermometer is as large as it is in the BEAST. While we tend to use K-type probes (which are very tiny) the Germans use the solid bayonet probe, which is much thicker. http://img402.imageshack.us/img402/6...gaugeandpr.jpg Since the hole can be filled with TIM while using the K-type, the larger hole accommodates both types probe. PRICING The retail price of the BEAST is about $350 delivered from Germany. When it was in production, the F1 price was approximately $275 delivered locally. As the F1 is no longer being produced and sold retail an actual apples to apples comparison of pricing can only be conjectured.
PHASE 1 – The F1EE http://img220.imageshack.us/img220/2606/img1385d.jpg http://img20.imageshack.us/img20/3623/img1388qp.jpg I chose to run the testing on the F1 first only because I am familiar with it. I mounted the F1 with the aluminum extension sitting upside down so the aluminum was touching the copper of the F1 so they would freeze together and form a seal. I then started the rig without any LN2 in the pot and parked myself in the BIOS temperature monitoring screen to keep track of the CPU temps. Starting with an ambient temp of + 20C, it took 8 minutes and 7 seconds to reach +60C. A time I thought was very respectable. I then brought the temps back down by pouring the LN2 into the pot, slowly at first, and the a little more aggressively to bring it down to +18C where I exited turned off the rig and switched the jumper to enable the LN2 profile of the board. After enabling LN2 cooling, I turned on the rig and let it boot into Windows and used my K type thermocouple to monitor temps. My experience with the F1 has been that it takes a fair amount of both time and LN2 to bring it from +20 to -100C and this run was no different. One of the advantages of using the aluminum extension is that once the metals freeze together you can add a good deal more LN2 per pour without worrying about a boil over. My experience with the F1 is that temps tend to drop slowly, as in .1 or .2 C at a time until the pot is glazed with the torch. So I torched the pot at -60C and this increased the increments of temperature drops to 6C to 8C. As noted above, I hit the cold bug at -122C, which was very disappointing, but the good news was the cbb was -118C. My objective at this point was to keep the temp between -118C to -121C while running vantage at 5.5Ghz (I was able to reach 5.6Ghz but it wasn’t stable enough to run Vantage. http://img683.imageshack.us/img683/7325/img1389bc.jpg As always the F1 came through with flying colors and I was able to maintain the targeted temps. It did however, take a good deal of small pours but the F1 went up about as fast as it went down (temps) so it was relatively easy to control. http://img189.imageshack.us/img189/3265/kingpinpot.jpg Shortly after running Vantage, I experienced some strange board/software issues as all my ROG connect voltages were showing maxed out numbers (i.e. 10volts on the CPU) and the board locked up. After letting it warm up I tried rebooting at various temps but it wouldn’t come back to life so I torched it down to -18C and removed the F1. Fortunately it was apparently a software issue as everything tested fine the next morning when I threw a water cooling loop on the CPU and tested with my DMM. Still not sure what happened, but. PHASE 2 – The BEAST http://img801.imageshack.us/img801/8263/img1393tg.jpg http://img69.imageshack.us/img69/5295/img1399mq.jpg As per the test on the F1, I turned the rig on without LN2 in the pot, and then parked myself in the BIOS temp monitor and waited for the CPU to heat up to +60. 12 minutes and 37 seconds later I hit -60C. Assuming my mounts were of the same quality (I'm usually consistent) The mass of the der8auer BEAST appeared to offer better non LN2 cooling, at least for the period of time left running. Following the same temps used in the F1 test, I brought the temp down to +18C and shut down to change the jumper. The BEAST seems to respond faster to lesser amounts of LN2 than the F1, most likely because of it’s incredible surface area. It’s a shame I wasn’t able to quantify the time/LN2 differentials, but the F1’s strong point has always been temp control, not rapidly changing temps so this really wasn’t surprising. After booting back into Windows, I watched my temp meter drop as I added LN2 to the pot. The BEAST’s greater surface area again was evident in that the temperature dropped in 4 to 5C increments right off the bat. I glazed it with my torch at -60C and it increased to about .8 to 1C increments, so, after glazing the decrease in temperatures were about even with the F1 with the BEAST holding a slight advantage. The BEAST also warmed up incrementally slower than did the F1. I encountered the cb and the cbb at the same temps so again undertook to run 3DMark Vantage at approx 5.5Ghz between -118C and -121C. To my delight, the BEAST was just as good, if not a tad better, than the F1 at holding temps in this range. Whereas the BEAST tends to react faster to a pour, it gets warmer at a slower pace, so the need to pour to keep the temps in range was a little less daunting. http://img217.imageshack.us/img217/2998/3dmvbeast.jpg After running Vantage, I spent a little time trying to up the speed I could attain with this chip. The highest I was able to achieve while maintaining stability was 5.788Ghz. http://img17.imageshack.us/img17/6286/5788p.jpg Some more experimenting with this chip I suspect will get me to higher speeds, but, that is a story for another day. Again, unfortunately, I was hit with all my software tools telling me the motherboard should burst into flames in any second. Vcore was over 10 volts as was the Vdimm. This time, I checked with my DMM and all voltage were exactly where they was supposed to be, but, the board was locked up. I warmed it with my torch down to -60C then -40C but could only get an immediate FF whenever I tried to boot. Calling it quits for the day, I torched the pot until it hit around -18, then started to remove the hold down hardware and slipped the BEAST off the CPU. I tried booting with the pot off, and the board started right up, but, I wasn’t adventurous enough to let it run that way all the way to windows. CONCLUSIONS
NEXT STEPS Unfortunately, I didn’t have a Gemini pot to throw into the mix, but my doors are always open if Vince wants to jump from Detroit down to Chicago to repeat this exercise and compare notes. I'll rent a 230 liter dewar and buy him his favorite beer if he can ever make it!! (maybe he'll bring his new super secret pot too)?? I should be so lucky! In any event I will rerun these tests again in the near future and post the update. Perhaps it will change viewpoints, perhaps not. In any event, I'll try and recruit a daughter to keep track of temp change times to give a little more detail. I think it would be interesting to try both pots with DICE as well, so if time permits I'll run those as well and report the results. It was a gas being able to freeze both of these great pots and play in the lab for awhile. I am a little disappointed that my chip froze so dame warm, and that the board locked up when it got cold. Next time perhaps I’ll use another chip and my 760 Classified and see what the results are. Should be fun :D ~md PS. Those of you who noticed, yes, Phsyx was on during both runs of Vantage, what a great CPU score!! PSS. I made a video of both pots being cooled with 2.5 cups of LN2 but unfortunately my camera is so old the software won't work and my third party software won't download it either. As soon as I figure out to retrieve it I will post the video as well. |
Excellent write up. Don't see too many pot reviews or tests. Lovin it.
|
Awe Man, good stuff Maad!!!
|
Thx Dawg :good:
|
Thanks Guys, took a hellacious amount of time just to put it into [B] code much less pour and shoot pics.
A couple things I'm going to add to the review though that I'll start here with so those who have read the review won't miss them 1 - In regards to the F1, the aluminum extension is a $45 option while it's included in the cost of the BEAST. This raises the F1 cost to approximately $315 compared to the Beasts $350. This adjustment, and, equalizing the cost of the copper makes the der8aur pot the less expensive of the two. 2 - I've learned that in Germany, many over clockers do not use K-type probes, they use the solid probe with a handle, which accounts for the size of the hole in the BEAST being so large http://img402.imageshack.us/img402/6...gaugeandpr.jpg |
nice review :thumbsup:
|
Awsome work with the review!
Real Maad stuff in here :D |
Nice review MD :)
Sorry to hear about all the board troubles. |
Quote:
Hey - made the front page of the HWBOT.ORG news section :D |
sweet :)
|
All times are GMT -10. The time now is 08:05 AM. |
Copyright ©2009 Overclockaholics.com