Memory Scaling on Core i7 - Is DDR3-1066 Really the Best Choice?
by Gary Key on June 24, 2009 9:00 AM EST- Posted in
- Memory
Memory Prelude
Intel designed Core i7’s performance around triple-channel mode and low latency DDR3-1066. In fact, with three channels of DDR3-1066 memory, the platform will provide up to 25.6GB/s of memory bandwidth and nearly 38.4GB/s at JDEC approved DDR3-1600 memory speeds. Intel figures the bandwidth numbers available at DDR3-1066 coupled with the on-die memory controller and relatively fast secondary cache speeds should provide enough performance to satisfy anyone. For the most part, they are right.
However, we are never satisfied so we set out to determine if applications we utilize everyday would benefit from increased bandwidth, improved latencies, or a combination of both. Obviously synthetic memory benchmarks improve greatly as you increase bandwidth and decrease latencies. The memory companies (with help from us at times) tout their latest and greatest memory products with these benchmarks while they are staples on the benchmarking circuit.
We were discussing our initial overclocking results with a couple of BIOS engineers and came away surprised. We learned they typically tune memory performance utilizing a combination of Sandra/Everest and Super Pi. That did not shock us but it explains why time after time our various test systems would perform exceptionally well in the synthetic benchmarks and then fall flat on their collective faces when faced with heavy multitasking situations. We believe too much emphasis is placed on trying to win a benchmark that in no way reflects the true performance of applications or enhances the stability of the system.
That brings us to another story. We had planned to incorporate a full overclocking section in this article but our DDR3-1866 and DDR3-2000 kits based on the Elpida DJ1108BASE, err Hyper ICs, have been experiencing technical difficulties as of late. As in, the damn things are failing faster than we can keep replacements in the labs. This is not a vendor specific problem, nor does it seem to be a voltage problem either, as we have had modules from various suppliers fail at stock VDimm or at 1.8V when trying to push through to the DDR3-2300 level. We have asked the memory suppliers a few pointed questions and are waiting for answers. In the meantime, replacements are arriving shortly. We will follow up with additional DDR3-2000 overclocking scenarios in the near future.
Facebook
Twitter
RSS
47 Comments
View All Comments
ilkhan - Wednesday, June 24, 2009 - link
running a small cross section of the tests in dual channel mode would be the improvement I can see. Awesome article.Gary Key - Wednesday, June 24, 2009 - link
We will have dual channel results in the 3GB, 4GB, 6GB, 12GB article in a couple of weeks. Right now, you are not giving up that much if any at all in most of these apps with a dual channel 4GB/8GB setup.The0ne - Wednesday, June 24, 2009 - link
The use of percentages and the comments made for them is vastly different than comments made for video cards. A 14% gain in minimal FPS isn’t much, especially for Dawn of War II. To state the game is a “stutter fest” from a low of 12FPS to “smooth” of a high 17FPS is really exaggerating the picture. 17FPS is still a “stutter fest.”From the data collected it really can be said, much like video card reviews is, that if you have the money and want the best then buy the faster memory, otherwise it is a waste of your hard earn money. My point of posting this comment is that the objectivity should not be any different when talking about FPS gains. Here it appears to sound more pleasing even though the numbers don’t show much gain at all.
GourdFreeMan - Thursday, June 25, 2009 - link
For nearly all human beings the perception of motion as opposed to a progression of still frames lies in the 8-20 fps range. It is not beyond the realm of possibility that Gary's perception of stutter is from crossing this threshold at least momentarily while playing Dawn of War II. Of course, you could probably more cheaply improve your minimum frame rate by buying a better video card than faster RAM unless the game really is (CPU) memory bound.SiliconDoc - Sunday, July 5, 2009 - link
I kind of thought the opposite of the two prior comment (except I agree it may have been exagerrated to go with smooth as silk)- it seems to me that 2%-5%-7%-14% framerate gains are usually considered quite impressive and quite a win in videocard comparisons, and especially in minimum framerate areas, that would be quite nice.I understand it's a different review person, hence perspective and emphasis to a large degree, but it impressed me in the sense that those sized percentages are the end all and be all in video card comparisons - oh golly the declatory winners with that kind of spread based on just videocard performance... so discounting it here - no way.
So, except for the statement that overclocking the cpu is as much or more a gain and overpowers and negates ram timings to a degree (if I caught that latter part intent correctly in the article), I'd have to say the ram advantage is very important to the hardcore videocard shoppers - it can really add quite an edge - as much as a videocard / head to head choice based on benches. Maybe enough to wait for higher clocked ram prices to drop, or score that great deal on overclockable ram.
I enjoyed the article mainly because of those FPS benches shown.
fishbits - Wednesday, June 24, 2009 - link
"Of course, those that are doing all of these activities and more will benefit from purchasing fast low-latency memory and we even suggest getting 12GB while you are at it."How much of a performance hit (if any) is there typically in populating 6 banks on an i7 system versus 3?
bh192012 - Wednesday, June 24, 2009 - link
I'm not sure I understand the conclusion. You recommend DDR3-1333 C6 for people who want a little more speed, but it seems to me that your data shows that DDR3-1600 C9 is faster and cheaper?Example:
1066 c5 / min fps H.A.W.X. 80/50 = 1.60$ per frame
1333 c6 / min fps H.A.W.X. 125/52 = 2.40$ per frame
1600 c9 / min fps H.A.W.X. 85/54 = 1.58$ per frame (winner)
1600 c6 / min fps H.A.W.X. 175/56 = 3.13$ per frame
QChronoD - Wednesday, June 24, 2009 - link
I would have to assume that you are doing your calculations on the Min frame rate?Personally, I would look at that and see that they are about equal at the minimum, but 1333c6 is almost 50% faster on average!
I'd suggest redoing your $/fps with the average rates.
bh192012 - Wednesday, June 24, 2009 - link
Where is 1333 c6 50% faster than 1600 c9? I think you have confused the price chart with a benchmark or something. Also, min FPS are more important.Affectionate-Bed-980 - Wednesday, June 24, 2009 - link
Uhh obviously no one read this article really because Page 11 is supposed to be about choosing a kit, yet it has 3D rendering benchmarks which should be on page 12.... Yeah..