SF3D OC Triple Point - New Ln2 pot for memory

[SF3D]

Here is the official launch text:

 

We are happy to announce new professional liquid nitrogen container for memory modules. This product is targeted to enthusiasts who like to push their hardware to absolute limits.

Container is designed by Finnish extreme overclocker Petri Korhonen (SF3D) and it is produced in Poland by another extreme overclocker Marcin Rywak (ryba).

The new liquid nitrogen cooler for memory modules is called SF3D OC Triple Point. This product will bring fresh innovations to memory cooling and take it to the level it belongs.

For example mounting mechanism is completely new, so you can use all types of memory modules since DDR1 era to latest DDR3 products.

Now you can connect the memory module directly to LN2 cooler, so you don't need to use stock heatsink with very bad thermal transfer abilities.

You can use up to 4* memory modules at once.

 

Triple Point cooling comes with the following bundle:

- Full solid main block (aluminium for easy temperature control)

- Cooling plates for memory modules (4 plates in each set fit for all memory types)

- Mounting screws

- Insulation (Armaflex mat)

 

*For quad channel operation you need to use 2 sets of Triple Point LN2 cooler.

 

Here is the product pictures:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

If you are interested: SF3DOCsales@sci.fi

Edited March 12, 2012 by SF3D

[Olesius]

Great! lovely block

[sergiorb]

Looks great...

Nice job

[TaPaKaH]

what is the price?

[pro]

not bad Petri, i like the mounting design

[bob80]

I live very much this ram pot beacause can be used not only with dominator .....

[ryba AICooling]

It can be used even with SDRAMs So Ney will love it for sure

 

Pro: thank you that you love what I made.

Edited March 16, 2012 by SF3D

[SF3D]

This pot definitely works!

 

Dual channel DDR3 2558MHz with 7-7-7 latencies.

 

I failed with CPU-Z, cause I did not remember to update it. My suicide dual channel 1300MHz 7-7-7 run is ok, but not able to publish, cause version was 1.59. I hate that thing in CPU-Z!!!

 

Temperature was -30C real and memory voltage 1.975V. There must be a lot left in this setup, cause IMC seems to be ok in this BD.

Max memory clocks are next on the line.

[TaPaKaH]

if it's "important" you can send your 1.59 dump to Franck or Sam from CPUz and ask them to validate

 

impressive Hyper clocks for AMD, I couldn't get Hypers to work above 1150 regardless of timings on mine

[SF3D]

if it's "important" you can send your 1.59 dump to Franck or Sam from CPUz and ask them to validate

 

impressive Hyper clocks for AMD, I couldn't get Hypers to work above 1150 regardless of timings on mine

 

Nah, it is not important. This was just the first quickie. Need to do insulation etc better and try more with correct CPU version.

 

Yeah, I also though my hypers did not like AMD, but I was wrong. These sticks have done 1145MHz max 7-7-7 before with Sandy Bridge, so there is quite a lot improvement in the results.

 

Some superpi 32M with 1200MHz+ 7-7-7 and 6.XX GHz Ivy Bridge will be interesting

[knopflerbruce]

What sort of numbers do you get using a high RPM fan on the sticks? I wonder what sort of gains you get more than the actualy numbers:)

[Hondacity]

impressive clocks sf3d! i want that pot

[SF3D]

What sort of numbers do you get using a high RPM fan on the sticks? I wonder what sort of gains you get more than the actualy numbers:)

 

It was on my last post on previous page: These sticks have done 1145MHz max 7-7-7 before with Sandy Bridge, so there is quite a lot improvement in the results.

 

So, quite a lot of improvement!

 

 

impressive clocks sf3d! i want that pot

 

Yes, everyone should want this pot!!!

 

I hope the final clocks will be really impressive This was just the start for my tests.

[I.nfraR.ed]

So you're comparing max clocks on SB with max clocks on Bulldozer. Nice

I think Knut asked about comparison on the same platform.

 

Anyway, great clocks on that screenshot.

Edited March 16, 2012 by I.nfraR.ed

[M.Beier]

Pretty product Petri, and very nice mounting.

Will Cu version be available at later point?

[Christian Ney]

Why do ya want Cu ?

 

I don't see the need of Cu over Al for a memory pot

[SF3D]

So you're comparing max clocks on SB with max clocks on Bulldozer. Nice

I think Knut asked about comparison on the same platform.

 

Anyway, great clocks on that screenshot.

 

Well, I have tested these sticks with some other intel CPU's as well and the one still under NDA did not change a thing, even though it is a lot better with some Hynix On X58 platform these did 1090MHz max, so I do know how these are clocking. Thanks for reminding about these things.

 

Pretty product Petri, and very nice mounting.

Will Cu version be available at later point?

 

Cu version would just add a lot of costs, but gain would be close to 0.

So, answer is no. There will not be any Cu version.

[Massman]

I agree with M.Beier, Cu version is needed for the massive heatload the memory is producing. Certainly close to 200W, which means that Cu version will give a crazy badass possibility to finetune the temperature to at least 0.1°C, making it perfect to work around coldbug and coldbootbug. Response is insanely important with memory, especially in quad channel configuration (image 200W x 4 !! wtf, crazy). The Alu version is nice, but I'm not sure if it will suffice to keep Hyper stable at DDR3-2400C6-6-6 at 2.3V, as M.Beier already indicated. So, yes, it would be nice to have a Cu version. Perhaps also with an updated internal rework since it's possible that the current design is not capable of handling fast response needed?

 

I agree with M.Beier.

[Hiwa]

CU is heavy some memories don't like high pressure, alu if mounted well enough for any load i think.

nice 7-7-7 run there petri

[SF3D]

PJ, dont feed the ....

 

CU is heavy some memories don't like high pressure, alu if mounted well enough for any load i think.

nice 7-7-7 run there petri

 

Thanks Hiwa!

 

Real attempts are close now. Did correct insulation etc.

 

I wish I would have some single stick for max mhz attempts. I have just random hynix here.

[der8auer]

Sweet looking setup, pete!

[M.Beier]

I agree with M.Beier, Cu version is needed for the massive heatload the memory is producing. Certainly close to 200W, which means that Cu version will give a crazy badass possibility to finetune the temperature to at least 0.1°C, making it perfect to work around coldbug and coldbootbug. Response is insanely important with memory, especially in quad channel configuration (image 200W x 4 !! wtf, crazy). The Alu version is nice, but I'm not sure if it will suffice to keep Hyper stable at DDR3-2400C6-6-6 at 2.3V, as M.Beier already indicated. So, yes, it would be nice to have a Cu version. Perhaps also with an updated internal rework since it's possible that the current design is not capable of handling fast response needed?

 

I agree with M.Beier.

 

Or for the lazy guy whom does not want to fill pot too often^^

[SF3D]

Target reached!

 

http://valid.canardpc.com/show_oc.php?id=2298571

 

ASUS Crosshair V Formula

AMD FX-8150 BD

Kingston HyperX 2133MHz CL8 Hypers

Nvidia GTX 550ti

Corsair AX1200W

SF3D OC Inflection point

SF3D OC Triple point

Win 7 64-bit

 

Memory V= 2.025V (not measured from pcb)

NBV (IMC)= 1.6V

Temperature of memory pot = -50C

Temperature of CPU pot = -100 - -180C (Cold does not help on these 7-7-7 runs)

 

Edited March 16, 2012 by SF3D
picture of setup

[Asmola]

Good Job Petri! Triple Point now officially released and i believe marketing works! Some kind of record here?

[Jimba]

nice result SF3D!

 

keep up the good work!

 

oh, and lovely pot too!

 

would pming/emailing you for a idea on cost/shipping to Australia for the pot be a good idea?