Memory Performance: 16GB DDR3-1333 to DDR3-2400 on Ivy Bridge IGP with G.Skill
by Ian Cutress on October 18, 2012 12:00 PM EST- Posted in
- Memory
- G.Skill
- Ivy Bridge
- DDR3
Conversion – Xilisoft 7
Another classic example of memory bandwidth and speed is during video conversion. Data is passed between the memory and the CPU for processing – ideally faster memory here helps as well as memory that can deal with consecutive reads. Multiple threads on the CPU will also provide an additional stress, as each will ask for different data from the system. Our test uses two sets of conversions: first, a series of 160 videos have the first three minutes of each converted from various formats (480p to 1080p mkv) to AAC audio; and second is the motherboard testing, converting 32 1080p videos to iPod format. Each test is measured by the time taken to complete.
Converting to AAC seems to depend not on the memory – the movement of data from storage to memory to CPU is faster than the CPU can compute.
Video conversion is often one area quoted as being beneficial for memory speed, however these does not seem strictly true. As data is moved from storage to memory to the CPU, only if that memory transfer is the limiting factor does having faster memory help. In conversion to an iPod video format, that seems true moving from DDR3-1333 to DDR3-2133 just about, however it seems the limiting factor is still the CPU speed or the algorithm doing the conversion.
Folding on GPU
Memory usage is all algorithm dependent – if the calculation has a lot of small loops that do not require additional reads memory, then memory is unimportant. If the calculation requires data from other sources in those calculations, then memory can either be stressed randomly or sequentially. Using Ryan’s Folding benchmark as a platform, we are testing how much memory affects the serial calculation part of a standard F@H work unit.
Unfortunately this test is only available to the nearest second, thus the benchmark finishes in either 89 or 90 seconds, giving appropriate ns/day. From the results, folding on GPUs is not affected by memory speed.
WinRAR x64 4.20
When compressing or converting files from one format to another, the file itself is often held in memory then passed through the CPU to be processed, then written back. If the file is larger than the available memory, then there is also loading time between the storage and the memory to consider. WinRAR is a variable multi-threaded benchmark, whereby the files it converts and compresses determines how much multi-threading takes place. When in multithreaded mode, the rate of cache misses can increase, leading to a less-than optimal scaling. Having fast memory can help with this.
WinRAR is one of the benchmarks in our testing suite that benefits immensely from having faster memory. Moving from DDR3-1333 to DDR3-2400 speeds the process up by 20%, with the biggest gain moving from 1333 to 1600, and noticeable gains all the way up to 2133 C9.
Greysky's x264 HD 5.0.1
The x264 HD test, now version 5.0.1, tests the time to encode a 1080p video file into a high quality x264 video file. This test is standard across a range of reviews from AnandTech and other websites allowing for easy comparison. The benchmark is capable of running all cores to the maximum. Results here are reported as the average across four attempts for both the first and second passes.
In another conversion test, we see that this benchmark gets a ~5% boost with faster memory, although Pass 1 sees a bigger boost than Pass 2. If conversion into x264 is the main purpose of the system, then the price premium of the faster memory could easily be justifiable.
Facebook
Twitter
RSS
114 Comments
View All Comments
Peanutsrevenge - Friday, October 19, 2012 - link
Thanks Ian.Well, except for making me feel ludicrously old, first memory kit of 4GB DDR2?
Mine was back in SIMM days, when I think I added an 8MB 72pin stick to my existing 4MB stick.
Although the external math co-processor might have come first.
And I'm only 31.
You shall now always be Dr Evil Cutress to me.
IanCutress - Friday, October 19, 2012 - link
First *purchased* memory kit. I dealt with plenty of older memory thanks to hand me downs or prebuilt systems from my family at the time. I still have some SDRAM around somewhere, or some 8MB sticks of something or other. It's in a box under the desk ;)Haha, I've been called worse :D
Ian
alpha754293 - Friday, October 19, 2012 - link
I would have figured that with a memory test/benchmarking that you would be running Stream test.And with all this talk about the various latencies (measured in clock cycles) - a) a comparison should be given between the theorectical calculations and the actual performance and b) that you would think that you'd use something like lmbench in order to try to better quantify/test that (in addition to the actual games, tools, and applications).
Most of the results are pretty much inconclusive since the standard deviation is within the margin of error.
IanCutress - Friday, October 19, 2012 - link
Main reason is to steer away from synthetics. Synthetics frustrate me so - they will easily show the difference between a 1600 C9 and 2400 C10 kit, but what is that difference in real life? If latencies and burst speeds are x% difference in the synthetic, does that actually make a difference when playing Portal 2? Hence the requirement of this review to focus on the practical rather than the synthetic.Regarding being within standard deviations, the results you see are the culmination of multiple tests. The standard deviations are actually quite low as the results are enormously repeatable. I did a science doctorate, I make sure my numbers are valid.
Ian
Tchamber - Friday, October 19, 2012 - link
Back in 2009 I picked up a 3x2GB kit of Mushkin DDR3 1600 with timings of 6-7-6-18. Why don't we see low latency like that any more?IanCutress - Friday, October 19, 2012 - link
Those were linked to different types of memory chips at the time - the Elpida 'Hyper' ICs (http://www.anandtech.com/show/2799). Nice speeds, but high fail rates and low yields. They have been replaced by chips that are slightly slower, but a lot more reliable. Also to note that those Elpida Hyper kits worked great with Clarkdale and Nehalem, but are poor with Sandy Bridge and Ivy Bridge.Ian
CherryBOMB - Friday, October 19, 2012 - link
Can you explain why you say Hyper' IC's are " are poor with Sandy Bridge and Ivy Bridge."As I stated "I have 16gb of the fastest money could buy around that era running on x79 @ 1666 6-6-6-18-1t right now."
This was a tri channel run >
http://www.overclock.net/t/872945/top-30-3d-mark-1...
post #1054
IanCutress - Saturday, October 20, 2012 - link
Because Hyper ICs fell out of favor, motherboard manufacturers are now reluctant to spend time in optimizing the Hyper IC kits to work with their systems. Thus the kits often have to fall back onto default settings, and they sometimes do not work. As one set of ICs is phased out, and new ICs come in, the newer ICs get priority.Ian
PS. You'll find me on the overclock.net HWBot team :)
CherryBOMB - Friday, October 19, 2012 - link
I have 16gb of the fastest money could buy around that era running on x79 @ 1666 6-6-6-18-1t right now.well over $1000 invested. Each 6gb kit was over $450 - bought the extra to future proof to quad lanes today.
2x CMT6GX3M3A1600C6
1x CMT4GX3M2A1600C6
http://www.newegg.com/Product/Product.aspx?Item=N8...
http://www.newegg.com/Product/Product.aspx?Item=N8...
saturn85 - Friday, October 19, 2012 - link
how about adding a folding on cpu benchmark with different memory speed?