Low-Power DDR2 memory

I've seen two companies announcing 1.6V DDR2 SODIMMs this year. (Reminder, regular DDR2 is 1.8V.) These modules are marketed as Netbook memory, but should be usable in any notebook as well.

A-Data's is called XPG G series, DDR2-800G. It's not only lower voltage (1.65-1.75V) but only CL5, so it's also faster than garden variety DDR2-800 memory (typically CL6). While it was announced back in March/April, it hasn't been available anywhere for sale yet. I emailed A-Data to ask about it and today I got a reply saying it will be available on Amazon.com next week.

Here are some articles on this memory:

http://www.google.com/search?ie=UTF-8&oe=utf-8&q=a-data+xpg+g+ddr2

There's also ChainTech Apogee 1.6V memory

http://www.chaintech.com.tw/a511_newsrelease_detail.php?serno=62

which claims to use 12% less power than 1.8V SODIMMs.

I'm looking forward to trying these in my dv5z, it should lower the temperatures and increase the battery runtime. (By the way, a memory module's voltage is programmed into the module's SPD EEPROM so these things ought to work correctly in any DDR2 notebook. Will find out soon, I guess.)

 

I'd be careful with using different voltage RAM in the computer. If the computer is not able to adjust the DDR voltage levels down to 1.6V (many will not most likely) it will continue to run at 1.8V. Not only will there be no gain from a battery life or heat perspective, you may actually burn up the 1.6V chips.

To be honest, trying 1.6V RAM is not going to be worth the potential headaches. Also, the 12% reduction in power is not 12% reduction in total laptop power use...just 12% reduction in RAM power requirements.

 

The power difference between DDR2 and DDR3 is negligible in most systems too, so lowering DDR2 voltage won't give any noticeable benefit in terms of battery life and heat.

 

According to Samsung a 1GB DDR2 SODIMM consumes 2.25W of power.

http://www.samsung.com/global/business/semiconductor/products/dram/Products_DDR3SDRAM.html

So 4.5W total for a pair of sticks. (Dunno if a higher capacity module uses much more or not. I think since the memory controller will turn off banks that it's not accessing, the power of a 2GB module should not be 2X that of a 1GB module. But still, unused banks need to self-refresh, so at least the refresh power will double with doubled capacity.) A 12% reduction would amount to only .5W or so so yeah, maybe it's not worth it from that perspective.

On the other hand, the modules would still run cooler, which is still a good thing. Assuming the notebook reads the SPD and configures itself correctly.

Edit: this Intel presentation shows 2.7W per module. DIMM power use is dependent on the actual computer workload, of course. http://www.memforum.org/upload/MPGS002_100s.pdf

 

In ultraportables and netbooks, it could give up maybe 10 min more, but most mainstream notebooks the difference won't be noticeable.

 

For the most part, larger RAM sticks are not going to use more power (they will a little but most of the time not a lot). However, using the RAM certainly increases power needs .

0.5W is not that much for most systems, but yes it can mean a huge difference for ultraportables and netbooks.

 

Sure. I've got 4GB installed now, and usually 99% is in use. (Right now I have 200K free.) Switching between windows, editing files, compiling code - it all gets exercised pretty evenly.

This dv5z can use all the help it can get. Other folks have reported reduced temperatures from using lower power SODIMMs, but they never mentioned any product names. These two (Adata and Chaintech) were the only names I could find.

My temperatures (idle and loaded) have gone down ~5 degrees from replacing my G.Skill Titan SSD with the Samsung. Since this notebook is so poor at evacuating heat, any reduction will be noticeable.