Benchmark patches - what do you test / what do you use

[gegenlicht]

Hey guys,

ive found this post but its quite outdated https://discourse.vvvv.org/t/benchmark-patch/989

We are in the process of buying some new hardware and would like to somewhat test and rank the hardware for usage in VVVV.

The benchmarks for the systems might be different depending on hardware - examples:
IntelNuc

• Video (dds) and HTML rendering and maybe multiscreen performance.
  Content resolutions by standarts as steps -> 720p 1080p 2k 4k -> 30 -> 60fps

PC with/without GFX Card

• same test for Video(dds)

• DX11 Particles + Count should be going up untill max particles are reached while staying on a stable 60fps

• Instance Noodles with nice PBR like shading instanced + Count should be going up untill max particles are reached while staying on a stable 60fps

Perf logs

For these patches it would be nice if some system monitoring could be saved on disk

• GPU / CPU / MEMORY Usage and Temp.

Currently i just use old project patches as a reference for performance but this is not ideal. I know that every project has its own bottelnecks for the hardware but a dedicated Benchmark patch would still make more sense. Sometimes even just to check if problems are programing or hardware related.

Did someone try to create something like this before? Interested in sharing your approach?

[antokhio]

Hi, there are two types of tests, i can imagine, first one would be for a high-end system (highest specs possible), and second one for a low-end (lowest spec possible).

So for high-end i don’t think you can test more then driver capabilities…
For low-end i’ve would recommend to test amount of possible drawcalls on GPU (with low intel gpu’s that can be a pain in the ass). The test is you create renderer you connect quad and start to spread it till you see a framedrop. That amount would be a total amount of layers you can draw, witch is quite important when you build up something like ui…