Cooling Memory

[Demon_Hunter]

Hey guys I want to know if it is better to let the heatspreads in the memory or to remove it and put this (http://store5.yimg.com/I/yhst-24067115789173_1797_4040794) instead, I know that it would remove the warranty but the temp wouldn't be a lot better ??

(the air flow in the computer is very good)

see ya

[Hell-Fire]

If the spreaders are warm to the touch then they are doign their job.

But many people would say remove them period.

Untimately its your choice.

Be careful if you remove them as you dont wanna remove anything else with them.

Those sinks you linked would work nicely if you use AS Epoxy.

[Karnivore]

Hellfire put it quite well, I'll also add that if you use copper sinks you need some decent air flow for them to be effective...

[Malves]

I made the transition. Didn't see any improvments in speed. Could be because I am using the thermal pad that came with it...or not!
At least they look nice!:D

[Demon_Hunter]

ok, i think I'm not going to remove them

thank you guys

[pucky124]

hey hunter y dont u try putting the heatsinks on the heat spreader that should give u the best performance because of the increased mass and surface area.

also where did u get those sinks and how much did they cost i mostly just use the cheap sinks out of old powersupplies and stuff.

DH is a great band 2

[slavik]

those sinks are really cheap, come in packs of eight ... like 7 or 8USD for pack, check SVC for coller master ramsinks ...

[enzoR]

you call that cheap? 8 tiny sinks for 8 dollars?

[Golias]

I used independent ramsinks on each chip (cooper) and also a 80mm fan on a Zalman fan bracket, i gained a lot more stability at high vdimm/fsb on my 3.3v BH5

[Jesus]

Well my opinion is that the heatspreaders are crap. The bst idea would be to put a linked HS onto both sides of the RAM. But you dfinetley need airflow form above to cool both sticks down.

[Jasonxxx]

question, does removing the heatspreaders improve anything? Has anyone done this and gained something?:D

[STEvil]

yeah, it does indeed help immensely if the heatspreaders arent touching the chips correctly.

Think about it (all situations assume fan placed over ram module(s)):

BH-6 chip @ 3.6v with heatspreader touching: its gonna be ok

BH-6 chip @ 3.6v with heatspreader not touching but blocking air flow: its gonna be HOT HOT HOT HOT

BH-6 chip @ 3.6v no heatspreaders installed: Nice and cool

[iddqd]

More often than not; heatspreaders are attached using cheap double-sided tape, which is full of air pores. It's like putting your ram in a warm, winter jacket.

[Reefa_Madness]

I had a friend send me a link to something that might just work for cooling ram. Check it out. It helped some with the overclock of this PC2700.

http://www.kaltmacher.de/viewtopic.php?t=32848&postdays=0&postorder=asc&start=0

[Kingsfan]

can you read that? it looks like something the guy made, pretty awesome work

[iddqd]

I had a friend send me a link to something that might just work for cooling ram. Check it out. It helped some with the overclock of this PC2700.

http://www.kaltmacher.de/viewtopic.php?t=32848&postdays=0&postorder=asc&start=0
Some good copper ramsinks, and a a fan would produce a similar effect :).

[FLYING FALCON]

lol

[justwOo]

Does the quality of the ram affect how well the heatspreaders are usually put on, and overall how well they work? Just curious :P
Like if there'd be a difference between how OCZ put heatspreaders on there high end ram vs how geil or somebody put it on their value series

[[XC] moddolicous]

Lets say I put OCZ ramsinks on each chip with a fan over then vs the ram with nothing but a fan over it. Would that increase my OC alot, or would it make like no difference.

[ZMarre]

Just that fan alone should be sufficient :)

[[XC] moddolicous]

So it wouldn't? I'm going to try for BH-5 @ 300 1:1 2-2-2-5 (bench able). I'm trying to think of a way to cool them that will help me reach this without resorting to water. We'll see.

[ZMarre]

So it wouldn't? I'm going to try for BH-5 @ 300 1:1 2-2-2-5 (bench able). I'm trying to think of a way to cool them that will help me reach this without resorting to water. We'll see.

Uh lol GL :stick:

Your best bet should be leave them without any HS, pump the juice through it, and if you can't touch them anymore, well then you could try to slap some damn HS's on it. But don't count on a big boost if you ask me...

Anyway, I'll thumb for you (if you say it like that in english :) ) :toast:

[Sup3rman]

http://xoxide.com/angeleye-technologies-akwaflo-ram-cooler-black.html

Found this the other day

[iddqd]

So it wouldn't? I'm going to try for BH-5 @ 300 1:1 2-2-2-5 (bench able). I'm trying to think of a way to cool them that will help me reach this without resorting to water. We'll see.
Watercooling memory is uselss... unless it's chilled of course. Hmm.. I suppose the only way to break 300MHz stable with BH-5 is use phase-change.

[Reefa_Madness]

Here is a picture of a nifty way to mount a fan to cool your ram. One of the guys over at OCF came up with it...it was inspired by a combination of the "rubber band" method and the Thermalright fan mounting clips. Here is a link to the thread should anyone be interested.

http://www.ocforums.com/showthread.php?t=350891

[iddqd]

Here is a picture of a nifty way to mount a fan to cool your ram. One of the guys over at OCF came up with it...it was inspired by a combination of the "rubber band" method and the Thermalright fan mounting clips. Here is a link to the thread should anyone be interested.

http://www.ocforums.com/showthread.php?t=350891
Interesting.
Here's mine:
http://img12.exs.cx/img12/2964/img00796wu.jpg
http://img12.exs.cx/img12/1250/img00617gm.jpg

[Reefa_Madness]

Nice "little" mod to that Super Talent ram cooler.

[iddqd]

Yep. Now with 50% less gayness! :rehab:

[amduser]

Some good copper ramsinks, and a a fan would produce a similar effect :).


sure, 30-50°C will have the same effect then -40° or less :stick: . just don'T suggest if you don't understand.

[matt9669]

sure, 30-50°C will have the same effect then -40° or less :stick: . just don'T suggest if you don't understand.Perhaps one should not be so quick to judge, especially when a mod is up at 6:45 and doesn't appreciate sarcasm against fellow XS members . . . :slapass:

Or perhaps it is YOU that does not understand. DRAM cores are primarily capacitor cells, connected to I/O buffers in rows and columns. While transitors have improved switching characteristics at lower temperatures, capacitors do not absorb or dissipate charge any faster at -40°C. Since the I/O buffers are rarely the bottleneck in modern DRAM, it is highly unlikely you would experience more than trivial gains at subzero temperatures.

[amduser]

Perhaps one should not be so quick to judge, especially when a mod is up at 6:45 and doesn't appreciate sarcasm against fellow XS members . . . :slapass:

Or perhaps it is YOU that does not understand. DRAM cores are primarily capacitor cells, connected to I/O buffers in rows and columns. While transitors have improved switching characteristics at lower temperatures, capacitors do not absorb or dissipate charge any faster at -40°C. Since the I/O buffers are rarely the bottleneck in modern DRAM, it is highly unlikely you would experience more than trivial gains at subzero temperatures.

and perhaps you should not judge about my knowledge :rolleyes: .

i thinkt you don't get it. just look how old this pics are, and 272 to 305 is quite a differnce and that with simple aircooling you would reach that, i don't suppose. by the way, this are not modern dimms, but old bh-6 kingston value stuff (*ok, it'S quite modern and atcually has the them prob*). by the way, you perhaps also suppose phasecooling doesn't bring higher overclocking in compartion to watercooling on a cpu :confused: .

[matt9669]

by the way, you perhaps also suppose phasecooling doesn't bring higher overclocking in compartion to watercooling on a cpu :confused: .A CPU is a different animal, tens of millions of transistors packed into an area smaller than your thumb. This is something I attempted to explain - not trying to belittle you, to help you understand.

By modern DRAM I mean post-vacuum tubes ;)

[amduser]

By modern DRAM I mean post-vacuum tubes ;)

:brick: , ok, my english ist not the best ...

[matt9669]

:brick: , ok, my english ist not the best ...NP :up: "modern" can mean a lot of things ;)

[iddqd]

sure, 30-50°C will have the same effect then -40° or less :stick: . just don'T suggest if you don't understand.
Hmm? The whole purpose is to remove just enough heat to prevent failure. And Watercooling can't do it any better than air.

[enzoR]

no, its used to reduce capacitance and resistance of the transistors. water can remove heat faster than air.

[iddqd]

no, its used to reduce capacitance and resistance of the transistors. water can remove heat faster than air.
You know, at 4.0vdimm, I'd be more worried about my chips not causing a household fire.

[enzoR]

You know, at 4.0vdimm, I'd be more worried about my chips not causing a household fire.


well anything under 70C is safe according to winbond but im sure they will overclock better the colder the chips are dont u think? ;)

[iddqd]

You could gain about 10-20Mhz by moving to phasechange. That's as far as that goes.

[matt9669]

You could gain about 10-20Mhz by moving to phasechange. That's as far as that goes.I'd doubt it, like I've said multiple times the limit does not lie with the transistor components - someone prove me wrong and build a phase change unit for DDR . . . :D

[shadowing]

lol. Who would? It'd be too big and expensive. Unless you use one of the coil lines for the condenser...

[iddqd]

I'd doubt it, like I've said multiple times the limit does not lie with the transistor components - someone prove me wrong and build a phase change unit for DDR . . . :D
Yes, it does rely on capacitors to store things, but first you have to understand how memory works.

[matt9669]

Yes, it does rely on capacitors to store things, but first you have to understand how memory works.You do realize you're talking to a third year computer engineering major . . .

[iddqd]

Good, then you know that there's a transistor for every capacitor on there. The capacitor holds its charge, but it is the transistor which switches, to control the capacitor. If you can increase its switching speed, you could for instance get rid of read delays. I'm not sure how much you could gain in terms of actual clock, but that's a boon within itself.

[matt9669]

I'll put it this way then - it is my opinion as an engineer that you would not see substantial gains from keeping the actual operating temperature of those transistors below freezing. Now it's possible this would allow you to run higher voltages safely and improve frequency/timings as a result, but the frequency overhead that i.e. cascade cooling creates on a CPU would not be nearly as prominent on DIMMs. Your average CPU has transistors switching ~10x as fast as your average DDR, thus any improvent to their switching characteristics is more readily apparent than in DRAM cells . . .

But, I am still stating an opinion. Overclocking could easily be termed a "lab science" and I would love to be proven wrong by an eager member with a DIMM phasechange system - though building such a system presents its own engineering problems ;)


there's a transistor for every capacitor on thereThis has not been my understanding of DRAM architecture BTW, SRAM operates in this fashion but in a column/row DRAM lattice fewer transistors are needed to switch to storage elements (than there are actual storage elements) . . . I may be explaining that poorly, I'm kind of a visual person myself, perhaps I should dig up some pics from digital logic :D

[iddqd]

I never said that you'd get the 50% overclock bonus you'd usually get on CPUs with phasechange. 10-20Mhz on say, BH-5 that's already doing ~280 is not that big a deal. Just my estimate...