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.
120 Comments
View All Comments
galta - Thursday, February 5, 2015 - link
Yes, yes, it is wrong: whoever spends money on "enthusiast" RAM has more money than brains, except for some very specific situations.The golden rule is to buy a nice standard RAM from a reputable brand and use the savings to beef-up your CPU/GPU or whatever.
Murloc - Thursday, February 5, 2015 - link
yeah but e.g. with corsair ram I always bought the mainstream XMS one instead of the Value Select sticks, but given that I haven't done any tweaking in my last rig, I might just as well have bought the cheaper one without the heatsinks.Maybe in my next build I will do that if there is a significant price difference.
galta - Thursday, February 5, 2015 - link
You just proved my point: crucial is pretty reputable and they have no thrills RAM that are generally the cheapest on the market.Corsair is always fancy ;)
Kidster3001 - Friday, February 6, 2015 - link
The word "Enthusiast" with respect to computers is synonymous with "Spends more than they need to because they want to." If you're making the Price/Performance/Cost purchase then you are not an Enthusiast. Every year I spend money on computer stuff that I do not need. Why? Because I am an Enthusiast. You may consider this "wasting money", perhaps it is. I don't "need" my 30" monitor or my three SSD's or my fancy gaming keyboard and mouse. I did spend money on them though. It's my hobby and that's what hobbies are for.... spending money you don't need to spend.Stick with your cost conscience, consumer friendly computer parts. They are good and will do what you need them to do. Just don't ever try to call yourself an Enthusiast. You'll never have the tingly feeling of powering up something that is really cool, expensive and just plain fun. Yeah, it costs more money but in reality, that's half the fun. The tingly feeling goes away in a month or so. That's when you get to go "waste" more money on something else. :-)
sadsteve - Friday, February 6, 2015 - link
Hm, I don't necessarily agree with you on size. With the size of digital photos today, a large amount of RAM gives you a lot more editing cache when Photoshopping. I would also imagine it's useful for video editing (witch I don't do). For all my regular computer use, yeah 16GB of RAM is not too useful.Gunbuster - Thursday, February 5, 2015 - link
So a 4x4 2133 kit for $200 or a 3333 kit for $800 and 2% more speed in only certain scenarios. Yeah seems totally worth $600 extra.You could buy an extra Nvidia or two AMD cards for that and damn sure get more than 2-10% speed boost.
FlushedBubblyJock - Sunday, February 15, 2015 - link
Shhh ! We all have to pretend 5 or 10 dollars or maybe 25 or 50 is very, very ,very very important when it comes to grading the two warring red and green video cards against each other !just4U - Thursday, February 5, 2015 - link
Is there no way for memory makers to come up with solutions where they improve the latencies rather than the frequencies? The big numbers are all well and good at the one end but the higher you go at the other end offsets the gains.. at least that's the way it appears to me.menting - Thursday, February 5, 2015 - link
there is. The latency is due to physical contraints, so you can improve it by stacking (technology is just starting to slowly become mature for this), or by reducing the distance a signal needs to travel, which is done by smaller process size as well as shortening the signal distance (smaller array, smaller digit lines, etc). But shortening the signal distance comes at a cost of either|or|and smaller DRAM density, more power, etc, so companies don't really do it since it's more profitable to make larger density DRAM and/or lower power DRAM. The only low latency DRAM I know of is the RLDRAM, which has pretty high power and is fairly expensive.ZeDestructor - Thursday, February 5, 2015 - link
That, and with increasingly larger CPU caches, less and less of an issue as well.