Your thoughts on current RAM heatpreaders

[twwen2]

Personally i'm a fan of the naked look, where i can see the PCB and chips. However i'm also aware that memory can warm up considerably when overclocked, thus the need for heatspreaders or heatsinks.

What i'm interested in is this; how do you run your RAM?
Do you use a fan(s) over naked modules, a fan over modules with heatspreaders, heatspreaders with no fan, or nothing at all? What is your opinion on the traditional heatspreader? Does it actually help, or is it a hinderance? This (temperature testing) is an area that is often overlooked in memory reviews.

The only time i've seen RAM temp tested in a review was at bit-tech, where the corsair dominator sinks were proven to drop temps considerably (compared to some patriot sticks if i remember correctly). I've also read a review of some OCZ sticks, which found they ran cooler without the heatspreader.

I'd be interested to hears other people's thoughts on the matter:cool:

[nn_step]

honestly, I am attempting to get my rig to run under 3w total, so with heavy undervolting there is absolutely no need for any heatsinks or spreaders

[zanzabar]

naked + fan is better, but i cant get eh sinks off of mine without the risk of breaking them so i use fan + spreders


all that spreaders do (in most cases) is keep people from seeing the ics that they are buying, this was pioneered by ocz when they would ship a few shipments with good ocing ics then use cheap stuff affter the reviews and initial perches were done

[xlink]

honestly, I am attempting to get my rig to run under 3w total, so with heavy undervolting there is absolutely no need for any heatsinks or spreaders

your mouse and keyboard will use around 3W so it kinda kills the point
and your minitor(s) will use between 30-300W as well

look at the big picture


naked + fan is better, but i cant get eh sinks off of mine without the risk of breaking them so i use fan + spreders


all that spreaders do (in most cases) is keep people from seeing the ics that they are buying, this was pioneered by ocz when they would ship a few shipments with good ocing ics then use cheap stuff affter the reviews and initial perches were done
I think the goal was to sell more. Think racing stripes.

it's just that in this case the HS prevents air from getting to the chips and doesn't increase surface area enough to help so it hinders.

[M.Beier]

... I'd say.... Computex will be very interesting for this thread ;)

[Tony]

naked + fan is better, but i cant get eh sinks off of mine without the risk of breaking them so i use fan + spreders


all that spreaders do (in most cases) is keep people from seeing the ics that they are buying, this was pioneered by ocz when they would ship a few shipments with good ocing ics then use cheap stuff affter the reviews and initial perches were done

Our tests prove different, I know Corsair have tested also and heatspreaders do have a benefit over bare IC's

[massman]

Heatspreaders have to be mounted perfect to make a noticeable difference. I'd rather spend 25$ on a fan than 5$ on heatspreaders, tbh.

[Tony]

and you know this how?

[eva2000]

should be 2 options in that poll

bad contact HS + fan
or
good contact HS + fan = one I'd select as best option :)

good contact HS + fan > naked + fan > bad contact HS + fan

[M.Beier]

I agree with all of you... This is why I really really like the good thermal tape used on most ram...

[alexio]

It all depends on how much heat the chips are putting out and what the air temperature is. At recommended 24/7 settings of <2.25v for DDR2 and <1.9v for DDR and a cool case temperature (let's say around 25C) I'm pretty sure DDR2/3 D9 chips run cooler naked with a fan over them. When chips put out more heat and/or the case temp. is higher you're going to want to have a bigger surface area to cool down thus heat spreaders are the better option.

In any case heat spreaders with thick thermal pads and/or bad contact never do any good. For example the thick pads used with the ATI X19**XT(X) cards are so thick and mounting pressure is so low that the memory on these cards runs much cooler with just a fan over them. I could overclock the GDDR3 DDR-400! higher when the chips were naked and a fan blew cool air over them. GDDR3 is known to put out more heat than any DDR2 IC at a reasonable core voltage.

[saaya]

heatspreaders not only help with cooling, they also reduce the chance of accidently chopping of tiny smd parts, metal solder spits or small wire pieces falling on or sticking to the ram and shorting it, and last but not least ESD.

And depending on the heatspreader they add visual value as well :D

whether heatspreader actually improve the heat dissipation compared to no heatspreader... well, there is no easy answer for that... it depends.

watercooling isnt better than air cooling either, some watercooling kits are terrible :D

A good heatspreader and a good heatspreader with fan will definately keep memory chips cooler than no heatspreader at all and no heatspreader with a fan.

the rest... is up for opinions, and as we all know, "about opinions we can only argue, never agree" old german proverb :D

it would be interesting to see a review or article about this though, would be quite complex and not that easy to do though...

[saaya]

good contact HS + fan > naked + fan > bad contact HS + fan

> no contact HS ^^

[alexio]

it would be interesting to see a review or article about this though, would be quite complex and not that easy to do though...
Using single sided memory and readings taken on the other side of the PCB would be the solution. These readings won't reflect the actual core temperature of the ICs, but in any case a higher reading there means worse cooling preformance and visa versa.

Do you have a single sided kit to spare with a spreader that I can hack up? I have all the tools to do the damage needed :p:

[M.Beier]

... ICs isnt only thing that benefits from cold temps... I really wished somethin' like ASUS coolstack could be @ PCB... - This is why watercooling aint always beneficial...

[alexio]

... ICs isnt only thing that benefits from cold temps...
Theoretically colder traces could be beneficial. But don't forget that cooling the voltage traces gives you the same effect as increasing drive strength, yet only slighly. Furthermore cooler data traces are really only worth it if you choose high frequency (with slack timings) instead of a relatively low frequency with tight timings. In the latter case the difference would be neglectible.

[Neuuubeh]

wouldnt you need a very big drop in temperature on the traces to notice any difference? like 50K or so :D. IMO compared the benefits of cooling the ICs far outweigh those (just theoretically)

[Gogeta]

While we're on the topic of RAM heatspreaders...anyone have any thoughts about these?

http://newgskill.web-bi.net/bbs/view.php?id=g_ddr2&page=1&sn1=&divpage=1&category=55&sn=off&ss=on&sc=on&select_arrange=headnum&desc=asc&no=119

http://newgskill.web-bi.net/bbs/data/g_ddr2/DDR2_Pi_pack.jpg

[M.Beier]

alexio,
I disagree, I wont go into further details on this one...


While we're on the topic of RAM heatspreaders...anyone have any thoughts about these?

http://newgskill.web-bi.net/bbs/view.php?id=g_ddr2&page=1&sn1=&divpage=1&category=55&sn=off&ss=on&sc=on&select_arrange=headnum&desc=asc&no=119

http://newgskill.web-bi.net/bbs/data/g_ddr2/DDR2_Pi_pack.jpg

They look sexy, however, I would definantly wait for after Computex if you already have some good sticks... ;)

[Johnny Bravo]

They look sexy, however, I would definantly wait for after Computex if you already have some good sticks... ;)

Are you referring to Gigabytes concept piece from CeBIT?

[Hyperhorn]

When I push my RAM for benchmarks I use an additional fan + heatspreader everytime.
But for gaming etc. I leave out a fan, because my 24/7-settings are always unspectacular, for example 1.9V for 2.1V rated DDR2-sticks.

[M.Beier]

Are you referring to Gigabytes concept piece from CeBIT?

Nope... I didnt spend much time at Gigabyte this year, was only 1½ day at the exhibition, had to go to a European marketing competition (university) in France on 6th...
Only thing I did at Gigabyte was chatting with their marketing department.. :)
- And then try to hustle an engineer for info on upcoming products, especially if they had any plans of DDR2 I-RAM...

Computex is in a month, time will reveal.. :)

[twwen2]

Nope... I didnt spend much time at Gigabyte this year, was only 1½ day at the exhibition, had to go to a European marketing competition (university) in France on 6th...
Only thing I did at Gigabyte was chatting with their marketing department.. :)
- And then try to hustle an engineer for info on upcoming products, especially if they had any plans of DDR2 I-RAM...

Computex is in a month, time will reveal.. :)

Thanks for all the replies guys, it seems there is some ambiguity as to what is best. Each to their own i guess.

M.Beier it seems you know something that we don't! Would you care to share?:up:

[zanzabar]

who voted naked with no fan


and @ tony, how dose the mesh heat spreader help i have some ocz that has almost no contact and a loose spreaders and i was told that it was fine and for aesthetics, and i though that meant shiny + hidden ics, but i have some patriot were it has b-5 and spreaders and i had to take them off since they were hanging half off but they clocked better with out it

it seams to me that they just hide ics for the basic tape+ aluminum and no way to apply pressure, so it should work like a cpu, gpu, ext were u need pressure of u want get contact or heat transfer, the cellshock and thermal right have preasure and seam to work better, as well as your new flex. is this just an update in tech?

[nn_step]

who voted naked with no fan


and @ tony, how dose the mesh heat spreader help i have some ocz that has almost no contact and a loose spreaders and i was told that it was fine and for aesthetics, and i though that meant shiny + hidden ics, but i have some patriot were it has b-5 and spreaders and i had to take them off since they were hanging half off but they clocked better with out it

it seams to me that they just hide ics for the basic tape+ aluminum and no way to apply pressure, so it should work like a cpu, gpu, ext were u need pressure of u want get contact or heat transfer, the cellshock and thermal right have preasure and seam to work better, as well as your new flex. is this just an update in tech?

I, who am trying to set an undervolting record.

[Neuuubeh]

Some heatspreaders work very nice, other fail. Clamp-like headspreaders seem to suffer most, while carefully designed and machined ones seem to do very good - Corsairs Dominator DHX, Cellshock's milled headspreaders, Chaintech Apogee GTs, probably Flex by OCZ..

just my opinion :D.