See How Your Linux System Stacks Up To 15 GPUs For 1080p Gaming

[pete910]

Hmmm so locally you have the results all working and good but when you go to upload them, it gives you the existing URL? What happens if you try:

phoronix-test-suite upload-results 1507265-BE-1080PLINU24

exact same thing

http://downloads.unrealnetworks.co.uk/benchmarks/failed-upload2.png

[dungeon]

OK, people seems Michael got low results on Xonotic for some... not sure about reason, does anybody of you maybe use Unity (Ubuntu's DE)

Seems not/ seems yes... but it might be Unity or some other overhead for MIcheal as wee see Mint, Mageia and SteamOS all goes higher

[Passso]

I really think old games like Xonotic, OpenArena etc. should be avoided for benchmarking, considering that at high fps the way drivers and GLs are managed and calculate make this FPS counter have no sense...

Only games below 100fps can be compared seriously.

[dungeon]

I really think old games like Xonotic, OpenArena etc. should be avoided for benchmarking, considering that at high fps the way drivers and GLs are managed and calculate make this FPS counter have no sense...

Those still made sense with opensource drivers testing, not high end cards, on 4K+ resolutions, people running 144Hz monitors, etc...

And also to indicate problem with setup or software or drivers, etc... which in this case with Xonotic Michael has.

Xonotic easely shows CPU bound cases or openarena shows how much some GPU then drivers then setups are bandwidth bound

Only games below 100fps can be compared seriously.

No APU from AMD or Intel still can't reach 100fps/FullHD in openarena phoronix's benchmark.

[duby229]

Those still made sense with opensource drivers teting, not high end cards, on 4K+ resolutions, people running 144Hz monitors, etc...

And also to indicate problem with setup or software or drivers, etc... which in this case with Xonotic Michael has.

Xonotic easely shows CPU bound cases or openarena shows how much some GPU then drivers then setups are bandwidth bound

It seems like nobody knows what the purpose of benchmarking. You benchmark what you play, at the settings you play. Thats the whole point, nothing else matters. Everything else is just irrelevant data. If you have a 60hz monitor that's what needs benchmarked. If you have a 144hz monitor then that's what needs benchmarked. If you are measuring data that will not effect you, then it does not matter.

EDIT: What I'm saying is that benchmarking software you don't use at settings you don't use can -not- be extrapolated for the software that you do use at the settings that you do use.

[dungeon]

Purpose of public benchmarking on phoronix is to compare different software and hardware setups, comparison of GPU, CPU, drivers, etc... That said 'openarena' phoronix's benchmark does made sense for comparions of those.

While in real openarena is nowhere that slow, it does not have for example 'r_bloom_reflection' on by default, but also that option does not exist in current ioquake So all this means, it is setted up to - benchmark hardware, not what people use.

What is purpose of 3DMark or OK here Unigine benchmarks, no one use that really to play anything nor to watch how "beautiful" scenes are - for users it is only usefull for api/hardware/drivers comparison purpose.

Of course beside that people like to see how hardware XYZ cope with popular titles, i agree on that

[duby229]

Purpose of public benchmarking on phoronix is to compare different software and hardware setups, comparison of GPU, CPU, drivers, etc... That said 'openarena' phoronix's benchmark does made sense for comparions of those.

While in real openarena is nowhere that slow, it does not have for example 'r_bloom_reflection' on by default, but also that option does not exist in current ioquake So all this means, it is setted up to - benchmark hardware, not what people use.

What is purpose of 3DMark or OK here Unigine benchmarks, no one use that really to play anything nor to watch how "beautiful" scenes are - for users it is only usefull for api/hardware/drivers comparison purpose.

Of course beside that people like to see how hardware XYZ cope with popular titles, i agree on that

Benching software you don't use on hardware you don't use with settings that you don't use does not tell anything at all.

Software like Unigine benches and 3DMark have 0 value for anyone. You can say that they are useful for regression testing or whatever, but the facts are that those results cannot be extrapolated. If you want to know how your hardware performs with the software you use in the settings that you use, then you need to benchmark that.

[dungeon]

Benching software you don't use on hardware you don't use with settings that you don't use does not tell anything at all.

It tells, if you try to understand particular benchmark purpose. For example i expect openarena to show at its best how AMD Fury is not bandwidth bound thanks to HBM

Software like Unigine benches and 3DMark have 0 value for anyone.

Well i think you are wrong, agian. Does it mean to you that this phoronix's article does not meet any purpose?

[duby229]

It tells, if you try to understand particular benchmark purpose. For example i expect openarena to show at its best how AMD Fury is not bandwidth bound thanks to HBM

How does it matter at all if you dont play openarena? The results can't be extraploated for other software. There are bottlenecks in the hardware, in the drivers, in the operating system, and in the software that all effect the results of a benchmark. The only benchmarks that make sense take into consideration all of that.

[duby229]

Well i think you are wrong, agian. Does it mean to you that this phoronix's article does not meet any purpose?

Many of them don't fill a purpose. Most of it is just filler content for all of those people that think they extrapolate results.