CPU Tests: Office and Science

Our previous set of ‘office’ benchmarks have often been a mix of science and synthetics, so this time we wanted to keep our office section purely on real world performance.

Agisoft Photoscan 1.3.3: link

The concept of Photoscan is about translating many 2D images into a 3D model - so the more detailed the images, and the more you have, the better the final 3D model in both spatial accuracy and texturing accuracy. The algorithm has four stages, with some parts of the stages being single-threaded and others multi-threaded, along with some cache/memory dependency in there as well. For some of the more variable threaded workload, features such as Speed Shift and XFR will be able to take advantage of CPU stalls or downtime, giving sizeable speedups on newer microarchitectures.

For the update to version 1.3.3, the Agisoft software now supports command line operation. Agisoft provided us with a set of new images for this version of the test, and a python script to run it. We’ve modified the script slightly by changing some quality settings for the sake of the benchmark suite length, as well as adjusting how the final timing data is recorded. The python script dumps the results file in the format of our choosing. For our test we obtain the time for each stage of the benchmark, as well as the overall time.

(1-1) Agisoft Photoscan 1.3, Complex Test

For a variable threaded load, the i9-10900K sits above the Rocket Lake parts.

RISC-V Toolchain Compile

Our latest test in our suite is the RISCV Toolchain compile from the Github source. This set of tools enables users to build software for a RISCV platform, however the tools themselves have to be built. For our test, we're running a complete fresh build of the toolchain, including from-scratch linking. This makes the test not a straightforward test of an updated compile on its own, but does form the basis of an ab initio analysis of system performance given its range of single-thread and multi-threaded workload sections. More details can be found here.

(1-4) Compile RISCV Toolchain

One place where Intel is winning in absolute terms in our compile-from-scratch test. We re-ran the numbers on Intel with the latest microcode due to a critical issue, but we can see here that AMD's best are single chiplet designs but Intel ekes out a small lead.

 

Science

In this version of our test suite, all the science focused tests that aren’t ‘simulation’ work are now in our science section. This includes Brownian Motion, calculating digits of Pi, molecular dynamics, and for the first time, we’re trialing an artificial intelligence benchmark, both inference and training, that works under Windows using python and TensorFlow.  Where possible these benchmarks have been optimized with the latest in vector instructions, except for the AI test – we were told that while it uses Intel’s Math Kernel Libraries, they’re optimized more for Linux than for Windows, and so it gives an interesting result when unoptimized software is used.

3D Particle Movement v2.1: Non-AVX and AVX2/AVX512

This is the latest version of this benchmark designed to simulate semi-optimized scientific algorithms taken directly from my doctorate thesis. This involves randomly moving particles in a 3D space using a set of algorithms that define random movement. Version 2.1 improves over 2.0 by passing the main particle structs by reference rather than by value, and decreasing the amount of double->float->double recasts the compiler was adding in.

The initial version of v2.1 is a custom C++ binary of my own code, and flags are in place to allow for multiple loops of the code with a custom benchmark length. By default this version runs six times and outputs the average score to the console, which we capture with a redirection operator that writes to file.

For v2.1, we also have a fully optimized AVX2/AVX512 version, which uses intrinsics to get the best performance out of the software. This was done by a former Intel AVX-512 engineer who now works elsewhere. According to Jim Keller, there are only a couple dozen or so people who understand how to extract the best performance out of a CPU, and this guy is one of them. To keep things honest, AMD also has a copy of the code, but has not proposed any changes.

The 3DPM test is set to output millions of movements per second, rather than time to complete a fixed number of movements.

(2-1) 3D Particle Movement v2.1 (non-AVX)(2-2) 3D Particle Movement v2.1 (Peak AVX)

When AVX-512 comes to play, every-one else goes home. Easiest and clearest win for Intel.

y-Cruncher 0.78.9506: www.numberworld.org/y-cruncher

If you ask anyone what sort of computer holds the world record for calculating the most digits of pi, I can guarantee that a good portion of those answers might point to some colossus super computer built into a mountain by a super-villain. Fortunately nothing could be further from the truth – the computer with the record is a quad socket Ivy Bridge server with 300 TB of storage. The software that was run to get that was y-cruncher.

Built by Alex Yee over the last part of a decade and some more, y-Cruncher is the software of choice for calculating billions and trillions of digits of the most popular mathematical constants. The software has held the world record for Pi since August 2010, and has broken the record a total of 7 times since. It also holds records for e, the Golden Ratio, and others. According to Alex, the program runs around 500,000 lines of code, and he has multiple binaries each optimized for different families of processors, such as Zen, Ice Lake, Sky Lake, all the way back to Nehalem, using the latest SSE/AVX2/AVX512 instructions where they fit in, and then further optimized for how each core is built.

For our purposes, we’re calculating Pi, as it is more compute bound than memory bound. In ST and MT mode we calculate 250 million digits.

(2-3) yCruncher 0.78.9506 ST (250m Pi)(2-4) yCruncher 0.78.9506 MT (2.5b Pi)

In ST mode, we are more dominated by the AVX-512 instructions, whereas in MT it becomes a mix of memory as well.

NAMD 2.13 (ApoA1): Molecular Dynamics

One of the popular science fields is modeling the dynamics of proteins. By looking at how the energy of active sites within a large protein structure over time, scientists behind the research can calculate required activation energies for potential interactions. This becomes very important in drug discovery. Molecular dynamics also plays a large role in protein folding, and in understanding what happens when proteins misfold, and what can be done to prevent it. Two of the most popular molecular dynamics packages in use today are NAMD and GROMACS.

NAMD, or Nanoscale Molecular Dynamics, has already been used in extensive Coronavirus research on the Frontier supercomputer. Typical simulations using the package are measured in how many nanoseconds per day can be calculated with the given hardware, and the ApoA1 protein (92,224 atoms) has been the standard model for molecular dynamics simulation.

Luckily the compute can home in on a typical ‘nanoseconds-per-day’ rate after only 60 seconds of simulation, however we stretch that out to 10 minutes to take a more sustained value, as by that time most turbo limits should be surpassed. The simulation itself works with 2 femtosecond timesteps. We use version 2.13 as this was the recommended version at the time of integrating this benchmark into our suite. The latest nightly builds we’re aware have started to enable support for AVX-512, however due to consistency in our benchmark suite, we are retaining with 2.13. Other software that we test with has AVX-512 acceleration.

(2-5) NAMD ApoA1 Simulation

The Intel parts shows some improvement over the previous generations of Intel, however the 10-core Comet Lake still wins ahead of Rocket Lake.

AI Benchmark 0.1.2 using TensorFlow: Link

Finding an appropriate artificial intelligence benchmark for Windows has been a holy grail of mine for quite a while. The problem is that AI is such a fast moving, fast paced word that whatever I compute this quarter will no longer be relevant in the next, and one of the key metrics in this benchmarking suite is being able to keep data over a long period of time. We’ve had AI benchmarks on smartphones for a while, given that smartphones are a better target for AI workloads, but it also makes some sense that everything on PC is geared towards Linux as well.

Thankfully however, the good folks over at ETH Zurich in Switzerland have converted their smartphone AI benchmark into something that’s useable in Windows. It uses TensorFlow, and for our benchmark purposes we’ve locked our testing down to TensorFlow 2.10, AI Benchmark 0.1.2, while using Python 3.7.6.

The benchmark runs through 19 different networks including MobileNet-V2, ResNet-V2, VGG-19 Super-Res, NVIDIA-SPADE, PSPNet, DeepLab, Pixel-RNN, and GNMT-Translation. All the tests probe both the inference and the training at various input sizes and batch sizes, except the translation that only does inference. It measures the time taken to do a given amount of work, and spits out a value at the end.

There is one big caveat for all of this, however. Speaking with the folks over at ETH, they use Intel’s Math Kernel Libraries (MKL) for Windows, and they’re seeing some incredible drawbacks. I was told that MKL for Windows doesn’t play well with multiple threads, and as a result any Windows results are going to perform a lot worse than Linux results. On top of that, after a given number of threads (~16), MKL kind of gives up and performance drops of quite substantially.

So why test it at all? Firstly, because we need an AI benchmark, and a bad one is still better than not having one at all. Secondly, if MKL on Windows is the problem, then by publicizing the test, it might just put a boot somewhere for MKL to get fixed. To that end, we’ll stay with the benchmark as long as it remains feasible.

(2-6) AI Benchmark 0.1.2 Total

Every generation of Intel seems to regress with AI Benchmark, most likely due to MKL issues. I have previously identified the issue for Intel, however I have not heard of any progress to date.

CPU Tests: Microbenchmarks CPU Tests: Simulation and Rendering
Comments Locked

279 Comments

View All Comments

  • GeoffreyA - Tuesday, March 30, 2021 - link

    It could be due to x264 limiting the number of threads because when vertical resolution divided by threads drops below a certain threshold---I think round about 30 or 40---quality begins to suffer.
  • GeoffreyA - Wednesday, March 31, 2021 - link

    I tested this now on FFmpeg but it should be the same on Handbrake because the x264/5 libraries are doing the actual encoding.

    I only have a 4C/4T CPU but used the "-threads" switch to request more. On x264, regardless of resolution, once more than 16 threads are asked for, it logs a warning that it's not recommended but goes ahead and uses the requested count, up to 128. I assume that running at default settings, like AT is probably doing with Handbrake, will let x264 cut off at 16 by itself. If someone could confirm this with a 32-thread CPU, that would be nice. As for x265, I gave it a try as well and the encoder refuses to go on if more than 16 threads are requested, saying the range must be between 0 and X265_MAX_FRAME_THREADS.

    In short, I reckon both these codecs are cutting off at 16 threads on default settings. If Ian or someone else could test how much extra is gained by manually putting in the count on a 32T CPU, that would be interesting.
  • scott_htpc - Tuesday, March 30, 2021 - link

    Splat. Backporting doesn't really work & dead-end platform.

    What I'd really like to read is a detailed narrative of Intel's blunders over the last 5-10 years. To me, it probably makes a case study in failed leadership & hubris, but I would really like to read an authoritative, detailed account. I'm curious why the risks of their decisions were not enough to dissuade them to take a better path forward.
  • Prosthetic Head - Tuesday, March 30, 2021 - link

    Yes, some sort of post mortem on Intel development over the last few years would be interesting. Once they abandoned the Pentium 4 madness, they did a good job with Core, Core2 and then the early stages of the 'i' series. Because AMD were by that point down their own dead end, they had essentially no competition for about a decade. The tempting easy explanation is that as a de facto monopoly for desktop and laptop CPUs, they only innovated enough to keep the upgrade cycle ticking over, then when AMD made a rapid comeback they got caught with their pants down and some genuine technical difficulties in fab tech.... But the reality could be a lot more complex and interesting than that.
  • Hifihedgehog - Tuesday, March 30, 2021 - link

    > But the reality could be a lot more complex and interesting than that.

    The reality is Conroe was a once-in-a-lifetime IPC improvement, literally 90% better (or nearly double the performance!) clock-for-clock than the ill-fated Pentium 4 (see here: https://www.reddit.com/r/intel/comments/m7ocxj/pen... They are not going to get that again unless Gelsinger clones himself across Intel's entire leadership team. Now, they may get something Zen-like in the ~50% range, but nothing Conroe-like unless ALL the stars align after a decade of complacency.
  • Hifihedgehog - Tuesday, March 30, 2021 - link

    https://www.reddit.com/r/intel/comments/m7ocxj/pen...
  • 29a - Tuesday, March 30, 2021 - link

    Keep in mind that P4 was a piece of shit built for marketing high clock speeds and was easily beaten by Athlon 64 running 1Ghz slower so getting that much IPC wasn't as hard as usual.
  • GeoffreyA - Wednesday, March 31, 2021 - link

    "Keep in mind that P4 was a piece of"

    Not to defend the P4, but Northwood wasn't half bad in the Athlon XP's time, beating it quite a lot. It was Prescott that mucked it all up.
  • TheinsanegamerN - Wednesday, March 31, 2021 - link

    TBF, the only reason it wasnt half bad is AMD's willingness to just abandon XP. I mean, only 2.23 GHz? 3 GHz OCs were not hard to do with their mobile lineup, and those obliterated anything intel would have until conroe. IF they had released 2.4, 2.6, and 2.8 GHz athlon XPs intel would have been losing every benchmark against them.
  • GeoffreyA - Friday, April 2, 2021 - link

    Oh yes, the XP had the higher IPC and would have given Intel a sound drubbing if its clocks were only higher. Thankfully, the Athlon 64 came and turned the tables round. I remember in those days my heart was set on the 3200+ Barton but I ended up with a K8 budget system of sorts.

Log in

Don't have an account? Sign up now