Hardware Kingston LoVo: Power saving 1.25V DDR3 memory

In an extension of our recent investigation into energy efficiency, Kingston dropped in some of its latest LoVo DDR3 memory that can run at just 1.2-1.25V. How much power does it save and is it a significant difference? We check to find out.

http://www.bit-tech.net/hardware/memory/2010/03/18/kingston-lovo-power-saving-1-25v-ddr3-ram/1

 

Would you be able to run this RAM at 1.65V with a good overclock or will it just fry it?

 

:droool:

I'd love to have 8GBs running at 1.2(5) instead of 1.65. I wonder what GeIL's offerings are going to be like in comparison to these.

 

So are there any actual investigations in to how much of a difference low voltage RAM has on overclocking IMC CPUs?

 

What affect does this have on anything though? Overclocked ram (never mind lower timings for twice the price) always seemed like a scam to me (What effect if any does it have on fps?), and what sort of energy is saved in units/£'s by using this over the lifetime of a system even on 24/7 for say a generous 3 years, would be what?

If theres not much difference in price I got no complaints though, if its simply more marketing then I smell scam (not that i'm cynical or anything).

 

OK just worked out soem rough figures what it would be for the above on say a saving of 28W, at an average 8p per unit (a bit generous again) for 24/7 over 3 years would be the region of a not too insubstantial £44. Thats not too bad really, although energy is cheaper at night, most people here are unlikely to be using the ram for 3 years nor 24/7. For a NAS/networked RAID server though it could be a nice saving over the years.

 

Thats a good saving for folders and a saving is still a saving. As for your comment on faster ram, in games you mightn't see a huge performance difference but come memory intensive tasks such as some music production or video editing then thats where youll see your biggest benefit.

 

Is it possible to use these low voltage kits with Atom-ION boards that have DDR3 slots?

 

Probably, but you'll need to check compatibility (to see if the memory will work at all) and whether the board supports such low memory voltages (to see if you can run at it 1.25V). This is less than straightforward though, as Rich pointed out in the Final Thoughts section.

 

I see... I remembered this because it'd be interesting to see if they would cause any power saving on an already low powered system. I suspect that in this case we would notice any increase of efficiency though, but I'm just speculating.

EDIT: The lowest voltage in the Memory QVL of Asus AT3IONT-I is 1.35 volts.

 

If the cost is the same as a similarly performing kit then it would be something I would look into buying. Energy saving is something I like the idea of, but not at the cost of a huge performance drop and a price hike.

It may not have a huge effect on individuals in terms of energy/cost savings, but if many people adopted this a lot less energy could be used. /Hippy.

The main thing I like about this ram is its look, I am slightly obsessed with the colour green, I have been since I was a little'un.

 

So if the CPU will run cooler, with these modules, would that translate into higher overclocking potential?

 

Possibly

 

RAM is used for more than just gaming, you know. Apart from the plethora of video and image editing suites that benefit, also scientific simulation (what I do) aids from having fast memory - the faster the better.

 

I ran a review for some low volt stuff on another site... ll look for some screenies of an OC.

Andy

EDIT





1.35v 1600mhz (8-8-8-24) clocked in a 1.6v 2006mhz (10-9-9-27)

rather good really

 

Not really. Faster RAM (1066 vs 1600MHz) will only give you about 3% performance increase in case of cryptography (what I do), with ~10GB file encryption, multithreaded and RAM to RAM so no HDD slowing things down. The most limiting factor is the CPU, x86 is a weak architecture performance-wise, so there is really no need for 16GB/s when the CPU can process 10% of it. Even 667MHz DDR2 could easily cope with fastest quad- or hexa-core x86 CPUs.

And besides, DDR3 was introduced because motherboard market was getting a bit slow and manufacturers needed something to make people buy new things.

As for the "green" stuff, I don't really care. I like my car drinking an ocean of fuel per kilometer and my computers eating as many kilowatts as they wish. With everything watercooled heat isn't an issue. Instead of stupid whining, EU should just build more nuclear power plants, and everyone would be happy - more workplaces for engineers and cheaper power for all.

 

I'm sorry, but that's just plain daft.

The fuel your car drinks an ocean of - it's electricity is it? Of course it's not, it's a fossil fuel, fuel which is polluting the local air you breath with far nastier gunk than the vastly over-hyped CO2. I agree with your point re: nuclear power plants, but what has that to do with this?

If sports car 'A' can do 0-60 in 4.3 seconds with a top speed of 170mph on 12mpg, but identical sports car 'B' with an improved engine does 0-60 in 4.3 with a top speed of 170 - but on 20mpg - then sports car 'B' is simply a better car. If you can do the same job with less expenditure, (energy, cost, or both) then it's an improvement. It's not a question of doing less, it's just finding ways to do the same, (or better) more effectively.

Greater efficiency is always better.

Saving between 5 watts idle to 32 watts load in one machine is almost negligible, but, multiply it by 10-20 million machines nationally, by hundreds of millions of machines globally - it becomes significant.

You shouldn't have to be a lentil eating tree-hugger to see that...

 

Thanking you! Mmm, cheap 2000MHz.

 

Can you repeat the tests with LinX as it will load up pretty much all your memory rather than the paltry amount Prime Blend uses. It's far better for testing memory overclocks.

 

That at me mate??

If so thats 50 instances of IBT max mem stable. didnt try any more as I figured that would about do it..

Andy