Tek9 3.0 LN2/DI GPU unit

[rra]

Good heavens these are more pieces of art than anything I have seen before bravo Vince,and they work.

[blind_ripper]

you didn't stop, it's a temporary break.

for a good 2 years at least , but il still use the dragon now and then just to keep the feeling when pooring Ln²

[BeardyMan]

for a good 2 years at least , but il still use the dragon now and then just to keep the feeling when pooring Ln²

make it 5 months at it's best....
I'm at nearly two months now and it itches already...

I even drawed a new container yesterday FJ3.0 will not come close lol

anyways you'll be back soon

[k|ngp|n]

Something like PCMark?

3DMARK

vince u bearly gave the card more vgpu give it 1.75vgpu and u will see how hot they really are .
great pots to bad i stoped benching

I went up to 1.65v max and the heat output didnt seemt to effect the unit at all, but no gains in clocks just some instability. Card can't run colder than -84c or so. If it could do closer to -100c, I think I could benefit from additional vcore. -84c is better than the piece of x1900 I tried that couldnt even do below single stage/chiller temps.

[k|ngp|n]

FYI--1st run of these new Tek's will be done very soon and should start shipping out right when I get back from the greece event.
PM me for pricing and availablity.

[blind_ripper]

3DMARK




I went up to 1.65v max and the heat output didnt seemt to effect the unit at all, but no gains in clocks just some instability. Card can't run colder than -84c or so. If it could do closer to -100c, I think I could benefit from additional vcore. -84c is better than the piece of x1900 I tried that couldnt even do below single stage/chiller temps.

last one i had did -96°C evap on cascade befor coldbug hits .
btw there is a way to evade coldbug again , when getting my card's back il test some things and report them .

[eR1k]

Wow your work is amazing k|ngp|n !

[JacobF]

Nice pots

[ent]

nice containers kingpin! but why no backplates for GPUs? Arent the cards bend?

[MTP04]

WOW, nice work KP

The tube/oval shape really surprised me when I first saw the pictures, I just never thought of doing it that way before. Poor cnc must be working hard to get that shape. Looking forward to see more results.

[mXpS]

Man, I wish I had some money to spend on one of these bad boys. They are beautiful. Great work kp.

[k|ngp|n]

WOW, nice work KP

The tube/oval shape really surprised me when I first saw the pictures, I just never thought of doing it that way before. Poor cnc must be working hard to get that shape. Looking forward to see more results.

This wider design really rocks and is making benching my cards feel more like cpu units....massive ones . It's funny now when I go back to the old ones because they just dont have the internal volume that these have...not even close. These new units are the equivalent of two of the old ones side by side.
For the container to be really good to use with DI, this design works Xtremely well.

Speaking of DI and internal volume, my unit thrashed this 2900xt ES pretty good today on a nice slushy DI/acetone mix. I was able to get through Nature 01 solid at damn near 1238mhz on the core using 1.59v This card is an early ES sample that seems to run really hot.
Nature 01 is IMO the hardest test you can run on a gpu. You want to test your card hard, just loop nature 3-4 times and you will get your answer.
Unit always pulled warmer during 01' nature than at anytime during full 05 or 06. Full 05's were'nt too tough at 1211mhz core. Again, this isn't LN2 this is DI + acetone Unit could handle up to 1.6v and 1200mhz+ /-65c or so about 3/4 full though I only need to push near 1.6v for 1225mhz+
Difference in temps between idle at desktop with clocks/volts set and full load/volts that I saw was about 4c max at this OC. When pushing 1238mhz 1.6v through nature, I could make it about 3/4 way then it would lock so think that was the max for DI with this card.
Was really impressed with what it did and this is the reason behind the deisgn concept/base, and overall size. It was just really easy to use with DI and I could pack the unit with slush and just kick back and run full 05's...very nice.
Sitting at desktop taking screenshot unit maybe 3/4 full with 1.6v on the card, ATI tool temps were -42c to -44c:

Full 05' @1211/963 DI + acetone


Nature @ 1225/963 DI + acetone


Memory on this card is BRUTAL. Starts crapping out at 980mhz, so I bumped it down a little to focus on clocking the core with DI.

Here is a question I need some feedback on...
Would you guys rather have a gpu unit with a neutered footprint to allow for the ATI shim?? Or would you rather have a unit be the best it can be with ALL cards and pop the shim off to use with ATI cards??
Most that i have asked have said to leave the unit alone where it mates with the card, and that they will pop the shims off themsleves if the use them with the ATI's. They popped off so easy for me that I can't see why modifying the unit for the shim is necessary and I honestly never want to.
In fact on the run of these that we are doing now which should be done in about a week or so, I actually may increase the diameter of the button. It's really about what you guys who are getting the units want though ultimately.

nice containers kingpin! but why no backplates for GPUs? Arent the cards bend?

Nah...no problem with bending at all. I crank the container bolts so tight till the card bends so much, thats how I know I have good contact
Besides, backplate would cause headaches with xfire mounting with a unit this size.

[Gautam]

Would you guys rather have a gpu unit with a neutered footprint to allow for the ATI shim?? Or would you rather have a unit be the best it can be with ALL cards and pop the shim off to use with ATI cards??

Compromises are for suckers.

That's such crazy dry ice capability, being able to hold a load like that is amazing--couple hundred watts we're talking about e-z.

[BeardyMan]

lol vince, you could get the wr on di anyrime if you wanted

[hipro5]

Heh, heh, heh.......
I have the first Tek-9s and I'm happy....They suck a lot of LN2 during bench coz of the high clocked VGA but it's OK with me ......and my VGA's of caurse......They STILL CAN handle the VGAs and even drive them to cold bug if I pour more LN2 in them during a bench....

SO I believe that with those beauties, a user will not have to be "focused" of the VGAs that much, when driving them to the edge.....

[1Day]

Best performance for all cards - I would much rather have the best performance I can get and so what if I have to do some extra stuff before I bench with different cards. Performance is king in benching.

[T_M]

People who are using tubes, insulating, volt modifying should be capable of handling a shim
Go the full face, just like my kayl pots

[k|ngp|n]

People who are using tubes, insulating, volt modifying should be capable of handling a shim
Go the full face, just like my kayl pots

100% agree.

Heh, heh, heh.......
I have the first Tek-9s and I'm happy....They suck a lot of LN2 during bench coz of the high clocked VGA but it's OK with me ......and my VGA's of caurse......They STILL CAN handle the VGAs and even drive them to cold bug if I pour more LN2 in them during a bench....

SO I believe that with those beauties, a user will not have to be "focused" of the VGAs that much, when driving them to the edge.....

Couldn't have said it better than that with regards to LN2 performance. The old units can remove heat rapidly no matter the load. Never had any issues with those units ever in regards to getting proper cooling to the vga's with LN2.

These new units have a much different tune on them with LN2, they swing ALOT slower due to the new internal design. It was something to get used to at first because the old revs behaved quite differently.

Tuning on the new units really hold well. Not nearly the same ammount of pouring that the old ones need. MUCH easier to bench these units on LN2 making 3 unit benching by a single individual ALOT easier.
Can hold temp swings between pours less than one degree for full game tests pushing the cards very hard with 3.0 rev. The unit does all the work now and the user just have to make sure to have the right ammount of LN2 in the unit
for the load.

BTW--went back to LN2 last night on this 2900xt and was getting full 05's off at 1260mhz/1.63v, gonna push the cards voltage all the way today on LN2 to see how much it can take on the unit .

[dinos22]

BTW--went back to LN2 last night on this 2900xt and was getting full 05's off at 1260mhz/1.63v, gonna push the cards voltage all the way today on LN2 to see how much it can take on the unit .

[skylinepower]

thet was NICE
good joob

[BeardyMan]

indeed, that card at 1350/XXXX and some 5.4ghz C2D should give over 30K 05 single for sure

[fhpchris]

The containers look beautiful KP

Do not let the 45nm guys get you down just yet!

[k|ngp|n]

Getting closer to 1300mhz core 3D stable on the 2900xt using LN2 on the unit. So close to finishing full 05's in the 1290's at nearly 1.7v.
Still not bucking under load @1.66v/1270mhz 3D stable:

[dpa]

whats your secret..? you ALWAYS manage to get 50-60Mhz more out of gfx cards than others do..

anyways, why dont you give us a complete 05/06 run with high single gpu/cpu clocks?

32k withing reach in 05..?

[dinos22]

Vince those containers are excellent

holding that sort of load on ATI cards is unheard of
there is no other bencher i've seen do that yet >>> was gonna say something else but >>> errr no comment