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
menting - Friday, February 6, 2015 - link
simulation software, pattern recognition, anything that does a lot of data analysis and/or data transformation. Heck, whatever an SSD's BW is good for, high BW memory can also be good forretrospooty - Sunday, February 8, 2015 - link
Simulation and pat rec maybe, not that anyone uses it other than rare outliers... But Anything an SSD is good for? No, not at all. An SSD improves launch times for anything and everything from browsers to office apps, to graphic suite apps like Adobe CS to games. Everything that normal people do. HB ram improves almost zero for the vast majority of what people actually do with computers. Even what enthusiasts do.menting - Monday, February 9, 2015 - link
"SSD improves launch times for anything and everything from browsers to office apps, to graphic suite apps like Adobe CS to games. Everything that normal people do". Exactly my point. Everything that normal people do. All that an SSD does is to provide faster storage (granted, it's non-dynamic storage, unlike DRAM) such that when a CPU can't find the data in the cache nor the DRAM, it can go to it for data. If you have enough DRAM, all these software can reside in DRAM (provided that power does not go out). Also, a smart algorithm will know what software you used before so it will keep some parts in memory so the next time you launch it it will be faster. And the speed that it can relaunch will heavily depend on DRAM BW.If you want to talk about rare outliers, people who are serious enough Adobe users or gamers who really get affected in a major way by using a SSD vs a traditional HD are rare outliers. I won't be surprised if the number of those people are on par or lower than people that use simulation software or pattern recognition software.
BTW, if memory BW doesn't make much of a difference, why do graphic cards go for GDDRx instead of plain DDRx?
retrospooty - Monday, February 9, 2015 - link
"BTW, if memory BW doesn't make much of a difference, why do graphic cards go for GDDRx instead of plain DDRx?"You are seriously clouding the issue here. With a graphics card assuming you are utilizing it with a modern 3d game or such, would benefit immensely from memory bandwidth on the card... It hardly benefits at all from system RAM bandwidth. Like everything, memory bandwidth doubling while timing/latency more than doubles doesn't improve system performance much at all. Some cases ahve it diminishing performance. This is NOT about Video RAM, its system RAM and based on the Intel CPU architecture for the past 15+ years, improving memory bandwidth at the cost of latency doesn't help much...
AGAIN, as I said in my original post - "We saw almost no improvement going from DDR400 cas2 to DDR3-1600 CAS10 now the same to DDR4 3000+ CAS [rediculous]
menting - Tuesday, February 10, 2015 - link
if you just want to talk about system RAM, the biggest blame is on software and CPU architecture, since with the exception of a few, are not optimized to take advantage of BW, only latency, even when the usage condition is ripe for doing so.retrospooty - Wednesday, February 11, 2015 - link
This is an article about system RAM. Why would anyone be talking about video RAM? Agreed, it is a CPU architecture issue, however this is the world we live in and this is the CPU architecture we have... Intel is pretty much the top of the heap.FlushedBubblyJock - Sunday, February 15, 2015 - link
Don't worry there are endless thousands of bonerheads who cannot wait to be "sporting" DDR4.For most, if they are made to think it should be faster, it is, no matter what occurs in reality.
I'd say 75% of it is how happy hyped their mental attitude is about how awesome their new bonerhead equipment is marketed to be, including any errors and rumors about what it is they actually purchased and installed, which they in many cases are not clear on.
xTRICKYxx - Thursday, February 5, 2015 - link
This would be an interesting topic to return to when DDR4 becomes mainstream with higher speeds.WaitingForNehalem - Thursday, February 5, 2015 - link
Now this is an excellent article. Thank you!ExarKun333 - Thursday, February 5, 2015 - link
AMAZING article! Been waiting for this! :)