Benchmarking Performance: CPU Office Tests

The office programs we use for benchmarking aren't specific programs per-se, but industry standard tests that hold weight with professionals. The goal of these tests is to use an array of software and techniques that a typical office user might encounter, such as video conferencing, document editing, architectural modeling, and so on and so forth.

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

Chromium Compile (v56)

Our new compilation test uses Windows 10 Pro, VS Community 2015.3 with the Win10 SDK to compile a nightly build of Chromium. We've fixed the test for a build in late March 2017, and we run a fresh full compile in our test. Compilation is the typical example given of a variable threaded workload - some of the compile and linking is linear, whereas other parts are multithreaded.

Office: Chromium Compile (v56)

For our compile test, it would appear that the extra memory width afforded by the quad-channel memory of Skylake-X can have a direct benefit in compile performance.

PCMark 10

PCMark 10 is the 2017 update to the family favorite, PCMark 8. PCMark 8 has been part of our test bed since the latest update in Q1. For the most part it runs well, although for some processors it doesn’t recognize, some tests will not complete, leading to holes in our benchmark data (there’s also an odd directory quirk in one test that causes issues). The newest version, PCMark 10, is the answer.

The new test is adapted for more 2016/2017 workflows. With the advent of office applications that perform deeper compute tasks, or the wave of online gamers and streamers, the idea behind PCMark 10 is to give a better ‘single number’ result that can provide a comparable metric between systems. Single metrics never tell the whole story, so we’re glad that Futuremark provides a very detailed breakdown of what goes on.

Ganesh’s article on PCMark 10 goes into more detail than I will here, but the ‘Extended Benchmark’ runs through four different sets of tests: Essential, Productivity, Creation and Gaming. Each of these have sub-test results as well, including startup performance, web performance, video conferencing, photo/video editing, spreadsheets, rendering, and physics, which you can find in Bench.

Office: PCMark10-1 Essential Set ScoreOffice: PCMark10-2 Productivity Set ScoreOffice: PCMark10-3 Creation Set ScoreOffice: PCMark10-4 Physics Score

PCMark8: link

Despite originally coming out in 2008/2009, Futuremark has maintained PCMark8 to remain relevant in 2017. On the scale of complicated tasks, PCMark focuses more on the low-to-mid range of professional workloads, making it a good indicator for what people consider 'office' work. We run the benchmark from the commandline in 'conventional' mode, meaning C++ over OpenCL, to remove the graphics card from the equation and focus purely on the CPU. PCMark8 offers Home, Work and Creative workloads, with some software tests shared and others unique to each benchmark set.

Office: PCMark8 Creative (non-OpenCL)Office: PCMark8 Home (non-OpenCL)Office: PCMark8 Work (non-OpenCL)

Benchmarking Performance: CPU Encoding Tests Benchmarking Performance: CPU Legacy Tests
Comments Locked

222 Comments

View All Comments

  • Ian Cutress - Saturday, October 7, 2017 - link

    That was a mistake on my part. On that I'm still mentally in an era where 150 MHz is a 10% gain. My quick mental arithmetic failed.
  • ScottSoapbox - Thursday, October 5, 2017 - link

    It's a shame you didn't compare it to the 7820X. I think it was expected that it would better the 7800X at least to some degree, so the more interesting comparison is how much performance does the added cost of 8 cores get you.
  • Ryan Smith - Thursday, October 5, 2017 - link

    The graphs were already getting ridiculously long. For something like that, be sure to look at Bench: https://www.anandtech.com/bench/product/1904?vs=20...
  • realistz - Thursday, October 5, 2017 - link

    AMD panic mode. Price drop imminent.
  • Anonymous Blowhard - Thursday, October 5, 2017 - link

    Price drop already happened. R7 1700X now USD$300 on Amazon.
  • willis936 - Thursday, October 5, 2017 - link

    I'd like to see the memory testing done on Ryzen done on coffee lake as well. It's clear that 2 DDR4 channels is not enough for 8 cores, at least with AMD's memory subsystem. Is it enough for 6 cores with Intel's memory subsystem? Also please be sure to use a GPU powerful enough to warrant even reporting the gaming results.
  • bharatwd - Thursday, October 5, 2017 - link

    Kabylake is faster than Coffeelake. where is the 15% increase? what is the point of + and ++ iteration when there is no improvement in performance? intel is just burning wafers for no reason. Better for them to go back to tick tock clock and stop wasting resources................
  • SunnyNW - Thursday, October 5, 2017 - link

    Honestly I'm not sure why Intel doesn't just keep fab lines for the 7th gen i5s going and just re-label into the 8th gen i3s and just bin differently, ie higher base/turbo.
  • AleXopf - Thursday, October 5, 2017 - link

    Thanks for the review Ian. Just one question. Why do you think power consumption differs so much with the data from techspot, were the 8700k consumes 190w, and it's on par with the 16c32t 1920x?
  • Ian Cutress - Saturday, October 7, 2017 - link

    Are they testing at-wall power consumption at stock? That might add a bunch.

    Our power numbers are just for the CPU, not the at wall - they are derived from the internal calibration tools that the processor uses to determine its own power P-states, which in effect is directly related to the turbo.

    There seems to be a lot of boards that screw around with multi-core turbo this generation, which may also lead to higher power consumption numbers.

Log in

Don't have an account? Sign up now