Linux Performance

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 11.04 LiveCD. These tests include fluid dynamics used by NASA, ray-tracing, OpenSSL, 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.

All of our benchmark results can also be found in our benchmark engine, Bench.

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.

Linux-Bench c-ray 1.1 (Hard)

C-Ray is another bench that likes threads and high IPC cores, hence the i5-6600 sits out ahead followed by the FX-8350. At a lower price bracket, the FX-6350 with six threads wins out against the set of Core i3s, which follow their regular staircase pattern. The difference in frequency accounts for 4 seconds difference, and it seems the L3 cache difference accounts for another 4, or about 2%.

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.

Linux-Bench NAMD Molecular Dynamics

Similarly with NAMD, cores and IPC matter most. The Athlon X4 845 gets a good showing here for a $70 chip, beating the price/performance ratio of the Core i3s. Again, the AMD FX-6350 is ahead of the Core i3s, and the lopsided staircase means that the L3 cache difference accounts for another 2% loss in performance.

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.

Linux-Bench NPB Fluid Dynamics

NPB clearly favors Intel CPUs, and the name of the game is IPC and frequency due to the i5 being behind the Core i3 parts. On the AMD side, the highest frequency part takes the lead and the thread count doesn't seem to matter much; neither does the microarchitecture, indicating that NPB is likely tuned for Intel.

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 strong developer base, but also relies heavily on memory bandwidth as well as CPU performance.

Linux-Bench Redis Memory-Key Store, 1x

Linux-Bench Redis Memory-Key Store, 10x

Linux-Bench Redis Memory-Key Store, 100x

Redis is similar - high IPC and lots of fast cache are needed, and again it seems to be Intel tuned. The i5 sits below the set of Core i3s until the 100x loading comes along, where having physical cores and more cache seems to help. Despite saying more cache seems to help, the effect of the lower L3 cache on the i3-6100 seems to matter more in the 1x and 10x tests, but less in the 100x test, which is somewhat odd.

Performance Comparison: Office Performance Comparison: Legacy
Comments Locked


View All Comments

  • Morawka - Monday, August 8, 2016 - link

    It's still suprising to see it on a low end CPU. i didnt know the i3's had ECC, now i'm thinking of building a FREENAS box off of one of these
  • Samus - Tuesday, August 9, 2016 - link

    You'd be surprised how many servers actually ship with i3's...those HP ML10's are incredibly common and I've seen a number of Lenovo SMB servers run i3's.

    The i3 is more than adequate for most small business servers unless they plan to run Hyper-V. Most other CPU intense services are cloud-based now (I personally think a business is crazy to maintain their own exchange server these days when Microsoft offers a $4/month/mailbox turn-key solution)

    Basically all servers do now is run the domain and a few basic services like file sharing and routing. SQL, Exchange, even Hyper-V are all inexpensive Azure\Office365 services. It really comes down to who costs more, your IT guy, or Microsoft. Odds are, the IT guy. Unfortunate because I am one.
  • jardows2 - Monday, August 8, 2016 - link

    Low end consumer CPU's. Intel likes the product segmentation between Xeon's and i5/i7. This is also why Intel forced the use of the "c" series chipsets for the Xeon processors. a Xeon E3-1240 v5 is about $30.00 cheaper than an i7 6700, with a higher base frequency, but slightly lower turbo frequency.

    Of course, this helps people who are wanting server grade, but only need low end processing power. A Pentium or an i3 would be a great home server chip, but i5 or i7 overkill. If you are wanting to use a higher-end production computer with ECC, you probably are looking at higher-end processors than i5 or i7 anyway.

    However, AMD includes the support in all their AM3+ processors, and I believe in all their FM2 processors as well. Not every motherboard supports it.
  • sheh - Monday, August 8, 2016 - link

    Yes, I was referring to non-server CPUs.

    But why is it on the i3? Not that I mind, but the surprising aspect and the problem is that it's not on i5 and i7.
  • Black Obsidian - Monday, August 8, 2016 - link

    jardows2 covers that in his/her product segmentation reference.

    If you want ECC on the low end, Intel is happy to sell you an i3. If your needs are any higher, Intel wants to push you towards a Xeon (and C-series chipset), which IIRC are higher-margin parts than the i5 and i7, and happen to have gone through additional server-related validation.
  • satai - Monday, August 8, 2016 - link

    I understand this concept but I still don't get why there are cheap ECC enabled i3s instead of more expensive (but still cheaper than 4C Xeons) dual-core Xeons...
  • extide - Monday, August 8, 2016 - link

    Yeah, that is kinda weird, you would think Intel would do that, and create even more segmentation, which is something they definitely tend to like to do.
  • rhysiam - Monday, August 8, 2016 - link

    My guess would be (and full disclosure... this an educated guess): at some point the market becomes too niche and the higher profit margins get lost to the additional costs of segmenting product lines, keeping different lines in stock, etc. The cheapest Xeon quad core on Newegg is already only $90 more than the cheapest i3. How much more could Intel actually charge for a dual core Xeon over the i3, the only benefit of which is ECC? Then they'd have to maintain a whole new product line, manage stock levels, etc. My guess is that for the relatively small number of customers pairing a dual core with ECC memory, it's just more trouble than it's worth.
  • satai - Tuesday, August 9, 2016 - link

    Thet sounds like a believable explanation.
  • DanNeely - Monday, August 8, 2016 - link

    What exactly is the point of the Core i3-6098P supposed to be? Compared to the equally priced I3-6100, it's slower, has a weaker GPU, and a higher TDP. On paper I can't see any reason to buy the former instead of the latter?

Log in

Don't have an account? Sign up now