Why do RAM mfrs use heatsinks?

[mockingbird]

I just want a dual channel 8GB kit of simple ho-hum DDR3 1333 that is reliable and will work for the years to come. What is the purpose of putting those unsightly heatsinks on standard 1333mhz 9-9-9-24 DDR3?

I am loathe to buy Corsair, Kingston, or Patriot... I always prefer Crucial - in my opinion the only consistently reliable RAM, but the stock just isn't there. Last week I had to settle for Kingston "Hyper-X" with the blue heatsink...

Another reason I dislike those unecessary heatsinks so much is that removing can rip the BGA chip out right with the traces, and it happened to me a while back with Corsair DDR2 that I wanted to reflow. Someone had a great tip of letting them sit in hot water... I tried it with a hair dryer.

[mariushm]

Re: Why do RAM mfrs use heatsinks?

Marketing and less often heat dissipation.

The chips consume about 2-3 watts of power, maybe a bit more. But sometimes the 1333 Mhz chips are basically bottom binned chips, "leaky" chips so the manufacturer has to raise the voltage a bit just to keep them stable.

For example, the Hyper-X ones are running at 1.65v - the current generation ram chips will do 1333 Mhz at 1.35-1.4v with no problems, Kingston just buys in volume and then bins them in several categories and slaps heatsinks according to how good the chips are. Being leaky, they also dissipate more power so sometimes heatsinks are needed to keep them stable.

[mockingbird]

Re: Why do RAM mfrs use heatsinks?

This was Hyper-X Genesis "1600mhz". The SPD is actually programmed to 1333MHZ @ 1.5V and I left it at that because of the non-existent quality control at the manufacturers.

Yes, you're right about some really high-end ram needing sinks... But $20 4GB DDR3 sticks? Come on.

[larrymoencurly]

Re: Why do RAM mfrs use heatsinks?

I'm guessing they do it because heatsinks are cheaper than higher quality RAM chips but are more impressive to most customers. Here's a review of some G.Skill Ripjaws 2125 MHz modules, with a look beneath the heatsinks:

http://aphnetworks.com/reviews/g_ski...8gbxld_2x4gb/2

Notice the chips are Hynix rated H9 speed, or 1333 MHz. Apparently their slowest modules aren't made from overclocked branded chips but from no-name chips.

One company admitted that it tested with just a modified version of MemTest86+ and allowed up to 2 errors per module in final testing, except for their PC17000 modules, which had to pass with 0 errors. It seems that most companies test only with PCs, and few use Advantest machines costing over $1M each. Kingston has some of those machines, but whether they test with them depends on the quality grade of the product and the demands of the commercial customers. Also I don't know if Kingston still buys individual chips because I think they now buy only whole wafers and slice and package them in their own factory.

I used to believe that all Crucial/Micron memory was high quality if it didn't have heatsinks on it, but several months ago I saw a photo of some Crucials where the chips were marked "CRUCIAL" instead of with some Micron part number. So to play it safe I'm now buying only Samsung modules, which are always made from prime, non-overclocked chips. I don't overclock, but people who do say PC12800 Samsung modules can often run faster than other brands rated PC17000.

[Per Hansson]

Re: Why do RAM mfrs use heatsinks?

1: Marketing
2: Heat Dissipation
3: Cheap protection against bit flipping (What ECC memory does in hardware)

I know that "3" is kind of silly and it might all be in my (tinfoil protected) mind, but hey, every little bit helps

[smason]

Re: Why do RAM mfrs use heatsinks?

It's marketing. They'll sell more 'cause they're pretty.

BTW, Kingston RAM has always been reliable for me.

[mockingbird]

Re: Why do RAM mfrs use heatsinks?

Notice the chips are Hynix rated H9 speed, or 1333 MHz. Apparently their slowest modules aren't made from overclocked branded chips but from no-name chips.

This time I ordered this for a new build:

http://www.newegg.com/Product/Produc...82E16820139262

Yes, it's ECC, and it cost $10 more than Kingston HyperX Genesis or any other ordinary DDR3 8GB kit, but there are zero reports there of DOA or failure, so I'm confident it will last in the long run. And you don't have to use the ECC feature, it can be turned off in the BIOS. Hynix BFR pictured there. I know mine are Hynix too, I have them with me, but I didn't check yet exactly what.

Cheap protection against bit flipping (What ECC memory does in hardware)

You mean because of the heatsinks acting as shielding?

BTW, Kingston RAM has always been reliable for me.

Kingston ValueRAM used to be very reliable. Today it is very shoddy and I will not touch it. I'll even take bottom of the barrel Patriot DDR3 over ValueRAM.

[tom66]

Re: Why do RAM mfrs use heatsinks?

I have some RAM with LEDs on it (freebie.) And a heatsink. The heatsink stays cool, and the LEDs only show the address bus... it's supposed to give an idea of activity but it doesn't work that well.

[999999999]

Re: Why do RAM mfrs use heatsinks?

Another vote for marketing... and a bit of protection against failure from a tweaker who sets the voltage too high in the bios then tries to RMA it for failure pretending they did nothing wrong.

[Shocker]

Re: Why do RAM mfrs use heatsinks?

I think they could be useful in making the BGAs last longer, but by that logic ALL BGA RAM should have heatsinks, not just the high-speed stuff.

[larrymoencurly]

Re: Why do RAM mfrs use heatsinks?

BTW, Kingston RAM has always been reliable for me.

Many years ago, I was getting a 60% failure rate with their 512MB PC3200 DDR, right out of the box, and the number of errors was consistent with the markings on the chips, which differed by country of circuit board origin and their layout. OTOH Kingston apparently sends out replacements of higher quality than the originals if you get even semi-techincal with them and talk with higher level tech support.

The only modules that have been highly reliable for me have been those made of chips with the actual chip maker brand or part numbers clearly visible. Only one such module was bad, and that was because it had solder blobbed over its gold contacts.

[kc8adu]

Re: Why do RAM mfrs use heatsinks?

i would have sent that bad corsair in for replacement.
btw i have deployed thousands of sets of ram over the last few years and had the fewest failures which was none with g skill.
as for my own stuff i had a stick of corsair die for no apparent reason a while back.they replaced the whole kit of 4 2gb modules to ensure a match.
and the heatsinks are more bling than anything unless you overvolt and really push them.

[Per Hansson]

Re: Why do RAM mfrs use heatsinks?

You mean because of the heatsinks acting as shielding?

Yea, with the disclaimer that this might all be in my head too

[severach]

Re: Why do RAM mfrs use heatsinks?

I vote for marketing. Heat sinks look like a ploy to get not yet reliable new technology to higher speeds to sell to the chip kiddies that can't stand being second best. They upgrade every few months to stay ahead of the pack. I prefer last year's now reliable technology and RAM that works fine without heat sinks.