DDR4 Haswell-E Scaling Review: 2133 to 3200 with G.Skill, Corsair, ADATA and Crucial
by Ian Cutress on February 5, 2015 10:10 AM ESTProfessional Performance: Windows
Agisoft Photoscan – 2D to 3D Image Manipulation: link
Agisoft Photoscan creates 3D models from 2D images, a process which is very computationally expensive. The algorithm is split into four distinct phases, and different phases of the model reconstruction require either fast memory, fast IPC, more cores, or even OpenCL compute devices to hand. Agisoft supplied us with a special version of the software to script the process, where we take 50 images of a stately home and convert it into a medium quality model. This benchmark typically takes around 15-20 minutes on a high end PC on the CPU alone, with GPUs reducing the time.
Photoscan, on paper, would offer more possibilities for faster memory to make a difference. However it would seem that the most memory dependent stage (stage 3) is actually a small part of the overall calculation and was absorbed by the natural variation in the larger stages, giving at most a 1.1% difference between times.
Cinebench R15
Cinebench is historically CPU dependent, giving a 2% difference from JEDEC to peak results.
3D Particle Movement
3DPM is a self-penned benchmark, taking basic 3D movement algorithms used in Brownian Motion simulations and testing them for speed. High floating point performance, MHz and IPC wins in the single thread version, whereas the multithread version has to handle the threads and loves more cores.
3DPM is also relatively memory agnostic for DDR4 on Haswell-E, showing that DDR4-2133 is good enough.
Professional Performance: Linux
Built around several freely available benchmarks for Linux, Linux-Bench is a project spearheaded by Patrick at ServeTheHome to streamline about a dozen of these tests in a single neat package run via a set of three commands using an Ubuntu 14.04 LiveCD. These tests include fluid dynamics used by NASA, ray-tracing, molecular modeling, and a scalable data structure server for web deployments. We run Linux-Bench and have chosen to report a select few of the tests that rely on CPU and DRAM speed.
C-Ray: link
C-Ray is a simple ray-tracing program that focuses almost exclusively on processor performance rather than DRAM access. The test in Linux-Bench renders a heavy complex scene offering a large scalable scenario.
Natural variation gives a 4% difference, although the faster and more dense memory gave slower times.
NAMD, Scalable Molecular Dynamics: link
Developed by the Theoretical and Computational Biophysics Group at the University of Illinois at Urbana-Champaign, NAMD is a set of parallel molecular dynamics codes for extreme parallelization up to and beyond 200,000 cores. The reference paper detailing NAMD has over 4000 citations, and our testing runs a small simulation where the calculation steps per unit time is the output vector.
NAMD showed little difference between our memory kits, peaking at 0.7% above JEDEC.
NPB, Fluid Dynamics: link
Aside from LINPACK, there are many other ways to benchmark supercomputers in terms of how effective they are for various types of mathematical processes. The NAS Parallel Benchmarks (NPB) are a set of small programs originally designed for NASA to test their supercomputers in terms of fluid dynamics simulations, useful for airflow reactions and design.
Despite the 4x8 GB results going south of the border, the faster memory does give a slight difference in NPB, peaking at 4.3% increased performance for the 3000+ memory kits.
Redis: link
Many of the online applications rely on key-value caches and data structure servers to operate. Redis is an open-source, scalable web technology with a b developer base, but also relies heavily on memory bandwidth as well as CPU performance.
When tackling a high number of users, Redis performs up to 17% better using 2800+ memory, indicating our best benchmark result.
Facebook
Twitter
RSS
120 Comments
View All Comments
JlHADJOE - Thursday, February 5, 2015 - link
Will be interesting to see another article like this when we have CPUs with integrated graphics and DDR4.OrphanageExplosion - Thursday, February 5, 2015 - link
"For any user interested in performance, memory speed is an important part of the equation when it comes to building your next system."Doesn't your article actually disprove your initial statement?
And surely your gaming benchmarks might make more sense if - once again - you actually tested CPU intensive titles as opposed to the titles you've tested? The GPU will barely touch your expensive DDR4, if at all.
The only scenario I can see DDR4 making a real difference will be in graphics work with AMD APUs, and even then we'll need to see really high-end, fast kits that should just about offer comparable bandwidth with the slowest GDDR5 to offer a literally game-changing improvement.
Sushisamurai - Thursday, February 5, 2015 - link
Errr... Memory speed did make a difference (small IMO) when it came to DDR3. This article tests if it holds true to DDR4 - however, without an iGPU the other tests don't really show a significant difference when price is factored in. I mean, sure, there's a difference, but not worth the price premium IMO.A future AMD comparison would be nice, when AMD decides to support DDR4... Otherwise, it was a nice article.
FlushedBubblyJock - Sunday, February 15, 2015 - link
That's called the "justify wasting my life to write this article, tag and hook and sinker line, plus the required tokus kissing to the kind manu's that handed over their top tier for some "free" advertising and getting out the word.It's not like the poor bleary eyed tester can say: " I didn't want to do this because one percent difference is just not worth it, my name is not K1ngP1n and I'm not getting 77 free personal jet flights this year to go screw around in nations all over the world.
vgobbo - Thursday, February 5, 2015 - link
I really enjoyed this review!But... Intel processors are massive cache beasts, which reduces a lot the pressure put on memory (except for games, which I believe was the most interesting part of this review). Said that, I wish to see a review on an AMD system, which have a lot weaker cache structure and memory buses.
Is this possible to happen, or I'm just a dreamer? ;D
Anyway, this was another outstanding review of Anandtech! Loved it! Thank u guys!
dazelord - Thursday, February 5, 2015 - link
Interesting, but isn't Haswell-E/X99 accessing the memory in 256bit mode using 4 dimms? I suspect the gains would be much more substantial in 128bit/ 2 dimm systems.willis936 - Thursday, February 5, 2015 - link
Good stuff but after seeing a fair bit of memory roundups in my time I think this mostly confirms what everyone has been thinking: DDR4 is incredibly underwhelming in the performance space. You not only get better bang for buck with DDR3 right now but comparable, if not better, performance in the high end kits.galta - Thursday, February 5, 2015 - link
You've got it wrong. Nobody goes for DDR4 because of the memory, it's because of the new CPU and chipset.Ask yourself: do you really need extra cores and/or pci lanes? Or, do you want them and have the money to pay for it? If the answer is "yes" than you'll go for 5xxx and DDR4 is incidental.
Otherwise, go 4xxx and DDR3 will also be incidental.
It makes no sense to talk about memory as if it could be chosen independently from CPU/chipset.
rmh26 - Thursday, February 5, 2015 - link
Ian could you post more information about the NPB fluid dynamics benchmark. Specifically which benchmark CG, EP, FT ... and which class problem S, W, A, ...etc. In my own research I have found the simulation time to scale nearly linearly with the memory frequency for large enough problems. I am wondering how much the cache has to do with masking the effects of memory frequency on performance. As a the size of the problem gets larger the cache will no longer be able to mask the slowness of the memory. In general memory, and moreover interconnects between computers play a very important role in some HPC applications the rely on solving partial differential equations. In fact there have been suggestions to move away from the standard HPC Linpack benchmark used to create the top 500 lists as this compute intensive benchmark does not accurately reflect the load placed on supercomputers.http://insidehpc.com/2013/07/replacing-linpack-jac...
Dasa2 - Thursday, February 5, 2015 - link
Congrats anandtech you screwed up another ram review further misleading peopleThe games you chose to review are so badly GPU bottlenecked its sad. Do you not know that ram performance affects cpu performance?
You could run Dirt 3 with a i3 2100 vs a 5ghz 5960x and get the same score
How about putting some different CPU in amongst your ram benchmarks like 4460-4690 5820-5960x so people can see how faster ram compares to spending more on the CPU...
A 4690k with 1600c11 ram can perform slower in games than a 2500k with 2133c9 ram