Is there a viewport benchmark? It would be nice to do some benchmarking fo non cycles applications to understand how different hardware performs in different dcenarios.
I test CPUs for a living, and a full run of this benchmark is a hotly requested item. But without a CLI interface, I can’t do much. Just poking this request to the top again. https://developer.blender.org/T56383
A CLI interface would certainly be welcomed. I perform benchmark tests for CPU’s and other components as well, people have been asking about running this test specifically. Without a CLI interface I can’t. Any idea on an ETA?
Just a note on my experience with the benchmark, 1.0 Beta 2
First off, nice work.
When running with CPU, everything works fine, but every complete benchmark on GPU fails on the same scenes. A bit annoying, but it’s a special case of a bench test I guess. I get 0!
It’d be nice to have it more resilient, but I’m not overly worried about that. For the record, this is on an Nvidia 960 using latest nvidia drivers (not from the board manufacturer, which I can’t even remember now).
The actual issue I’d like to point out is the Mydata views on the website fail when there’s crash results. Probably an easy fix to add a handler to cope. Example: https://opendata.blender.org/benchmark/86eca928-a161-46eb-b4af-d3b10d3089e3
I would like to know a way to render and see the complete benchmark raw data , a .jsonl file, since the site is showing only the top configurations… Is there a way to open them in libreoffice or something?
The Blender team released an update to the Blender Open Data platform, including a brand new benchmark launcher. This update is meant to address several issues raised during the first beta phase. Check out the full story here https://www.blender.org/press/blender-benchmark-2020-update/ and give it a try!
Okay I downloaded the latest Blender Benchmark, partly because the previous one failed to complete, just stopped after the first test is complete. So with this new download upon starting up you need to select to download all the tests which I’ve done various times but each time it is downloading it would fail after a minute or two. I got the following error code:
Now open data gets interesting. Love the new search field at the bottom! GJ on that one!
Is there a feature planned for comparison of two devices?
Unfortunately the benchmark itself isnt working for me either, Ive send a report via the built in function.
Also I think it would be nice if the error message would be selectable, for copy+pasting… Dunno how VPII did it. :’)
You can compare two devices in the new Query interface. Simply type the name of the two devices you are interested in and filters to make your query more meaningful.
I would really like to have the opportunity to run it with combined hardware mode as well. Two video cards and if possible then gpu + cpu render. Even if I wouldn’t be able to submit the results of that for what ever reason! Just so that I could have an convenient way to say compare 1080ti vs 2070 super+ optix. Otherwise, I ran it on two systems and worked fine. It’s a bit confusing what happens with the downloaded test files after the benchmark tho. (maybe a simple line of text explaining it would be cool?)
Very nice that the benchmark get’s an overhaul. Worked for me so far and I like the possibility to download specific scenes.
I think one important thing is missing, the renderview with it’s buckets.
I don’t know, but I think the biggest joy for custom PC builders or overclockers, is when they actually see the buckets run faster. Especially when amd is comparing a intel cpu with their new ones. I think they also want to showcase the render speed visually and not only in numbers. (Like you could see on the CES Presentation where they used Vray)
The new blender benchmark seems great! However, you should be able to choose a download location for the files.
Extremelly handy for backing up, keeping track of different versions to make sure of testing uniformity, and comparing performance between different versions.
Assuming we get the network rendering feature some time in the future to let us use a render farm as a render device. It would be really good to see benchmarks of blender render farms so users have an official comparison of farm price to performance.
I’ve been putting together an unofficial comparison here:
But it would be good to incorporate this into Open Data some in way.
But I’m trying to test across multiple systems that don’t have online access in an automated fashion.
Why does the CLI version need an authentication token, but the GUI version does not?
It makes no sense.
Please remove the authentication requirement from the CLI version - I have no intention of uploading the data (as I have access to NDA systems that can’t submit anyway).
Edit:
So I tried moving all the relevant files and scripting up a benchmark.
The CLI version still wants to connect to get an auth token. Even though I haven’t used the upload flag.
Then the benchmark won’t run, even using the files copied from a different system, because…
ERROR: Blender version 2.81a seems to be broken. Consider running the ‘clear_cache’ command.
ERROR: Blender version 2.81a with checksum 87355b0a81d48ea336948294b9da8670eaae73667fae028e9a64cbb4104ceea1 is broken: missing valid flag file
This all requires to be online, obviously. I can’t script what appears to be a manual process.
So that’s a negative on being able to use the benchmark on AnandTech