RAM with heat spreaders vs. no heat spreaders

[zir_blazer]

I can touch the heat-spreader on my DDR3-2400 modules (G.Skill Trident X), and they're not even warm. Albeit, I might need to run a memory benchmark or two to really work them to properly check.

If I recall correctly, the vast majority of a DRAM chip power consumption comes from the refreshing cycle to maintain the data, more than the actual work of reading/writing.

 

[Tsavo]

I'd never heard the term "heat spreader" used in this context before Rambus used it. I always suspected that the term was used in preference to heatsink for marketing reasons, to make it sound less harsh. People are used to heatsinks for hot things, but if all you are doing is spreading the heat around, it might not sound quite as bad. Like, oh, there's no rush to get this heat away, let's just spread it around a bit, maybe place it over there for a while.

As I said before, RDRAM needed the heat spreader, because the active memory module became quite hot, and needed to spread the thermal load to less active modules.

The reasons why RAMBUS went for the heat spreader is packaging; they couldn't stick giant cooling fins out of the ram (cost) that needed another fan (cost), so they went with the lowest operative and effective device...sharing the thermal load across all the memory modules with a cheap stick of alu that didn't bulk out designs or make things expensive to cool.

Modern ram don't need this pocus-hokey. All the fancy stuff on ram does is add cost; make them look neat (not a small consideration), and get in the way of your 3rd party air cooler.

If anyone thinks otherwise, well, I'll gladly sell them SATA cable coolers and USB 2.0 heatsinks. :whiste:

 

[Emulex]

fb-dimm needed it too. terrible waste of energy. with no airflow (aka home built pc with no ducting) you might need it. but a well ducted 1U server can run 18 to 24 dimm at full power/speed without a problem.

with unknown air FLOW it doesn't hurt, with well designed air flow not really necessary for stable non-overclocking use at all (DDR3 RDIMM).

 

[Smoblikat]

Rambus started this because the active memory module got hot and needed a spreader.

Sinks on anything else are just there for looks. Like spinners, but for your ram.

Pure genius.........

 

[KingFatty]

I'd never heard the term "heat spreader" used in this context before Rambus used it. I always suspected that the term was used in preference to heatsink for marketing reasons, to make it sound less harsh. People are used to heatsinks for hot things, but if all you are doing is spreading the heat around, it might not sound quite as bad. Like, oh, there's no rush to get this heat away, let's just spread it around a bit, maybe place it over there for a while.

To me, the heatsink raises the notion of some fins sticking out to put the heat into the air, sinking the heat away from whatever it's touching.

Heat spreader is to eliminate hot spots, and provide a uniform temperature surface where heat is evenly spread across it.

So usually a heatsink will also function as a heat spreader, but if there are no fins sticking out, then you can't really call it a heatsink so instead you default back to the heat spreader concept.

 

[energee]

If anyone thinks otherwise, well, I'll gladly sell them SATA cable coolers and USB 2.0 heatsinks. :whiste:

Those interfere with the liquid cooling setup on my bios chip.

 

[Wall Street]

I would rather have 19 nm 1.35V sticks naked (for example, Samsung RAM) than sticks on a larger process that need 1.65V but have heatspreaders. The older stuff uses 150%-200% of the power of the new stuff, so a heat spreader one the latest and greatest RAM chips is now more for vanity than it used to be.