[Gautam]

This was my first experiment with subzero memory cooling, and I'd say it went all right. I wanted to get some SuperPi stability at speeds that are generally out of the norm.

I had to cripple the B2B pretty badly to make this happen, and run 2.14V through the ram, but nonetheless, here we are. I feel very dirty in doing so, but it was necessary for these speeds. (Hence the times are horrible) 1M at 2600 is no issue though at B2B 4, and with 1.8v vdimm FWIW. RTL remained at 60/62 throughout. The memory was at about -50. One of the sticks cold bugged fiercely past this, and due to the swing of my "container" (the heatspreader) it was a bit tricky to work around this.








Thanks Gigabyte for this silky smooth board.

[Reefa_Madness]

Very cool indeed (yes, of course the pun was intended)!

You are one xtreme fella when it comes to this kinda stuff...and I respect that in a guy.

I'm I understanding you correctly that your "container" is the ram's heatspreader? Looks like you've insulated around the sticks and dropped the DICE right on top/around the spreaders. Is that right? Don't miss the opportunity to share...teach us mere mortals how to do this ('cause I've got some BH-5 that might need to be chilled one of these days).

[sno.lcn]

Wow G, that's amazing!

When are you going to post a 3000mhz run?

[Gautam]

Maybe I'll try sometime, but the single channel suicides never appealed that much to me.

Reefa, it's liquid nitrogen, I poured it on the ram, using tape insulation as a "cup." Smeared the slots with vaseline and fitted thick neoprene around the ram. Dry ice should work well too.

[dyckah]

holy smokes dude, thats some crazy ****

[Reefa_Madness]

Originally Posted by Gautam

Maybe I'll try sometime, but the single channel suicides never appealed that much to me.

Reefa, it's liquid nitrogen, I poured it on the ram, using tape insulation as a "cup." Smeared the slots with vaseline and fitted thick neoprene around the ram. Dry ice should work well too.

Thanks for the info/clarification. Saw the ice crystals and just assumed it was dry ice.

I'm trying to figure out a way to mount some containers onto copper heatsinks and use DICE for cooling some Winbond...yes, I know that I'm still back in the dark ages, but then again, that's just about when I was born, too.

If anybody has ideas or links, they sure would be welcomed.

[Gautam]

Oh, don't overcomplicate things. You have a big advantage that'll make things significantly easier and safer; you can pull your heatspeaders.

So I'd remove the spreaders, then completely spray down the ram in conformal coating. Insulate the board properly with foam or clay around the DIMM slots, fill the DIMM slots in grease and then just place DI pellets on top of the bare sticks. Can't get much simpler and it'll be quite effective. Probably would pull the IC's around -50 without too much of a fuss. I think they scale decently, but I think DDR2 and DDR3 scales better cold.

Years ago a friend tried putting some DI around a stick of UTT BH5 and didn't get anything from it. He was just trying to get the ram below ambient and cold to the touch though, not extremely cold. Then again, he was stable at 279 as it was. I think it'll take a rare stick to scale properly with voltage and temperature in those conditions, but of course, you've got enough to find the pick of the litter.

[Reefa_Madness]

The use of spreaders is certainly optional. I just figured that I could use them for mounting the containers, however, you are absolutely right that it can't get much simpler than the approach you've just outlined...and that I've got a couple of sticks to choose from.

I'm trying to see if I can hit that magical 300MHz with tight timings and want to remove overheating from the equation.

Plus maybe my kids will like playing with the dry ice.


But enough on my future projects, let's get back on track...which is your fine results.