Unigine Is Working On Perhaps The Most Beautiful GPU Benchmark, Coming Later This Year

[dungeon]

Also they update TF2 non stop, so what is purpose of comparison and benchmarking that, as perf can go like this like that any another month

[duby229]

Also they update TF2 non stop, so what is purpose of comparison and benchmarking that, as perf can go like this like that any another month

Which makes it important to benchmark for those players that need to know.

[dungeon]

And what if user wanna know which hardware so GPU is faster, he can't conclude nothing by running TF2 bench

[duby229]

And what if user wanna know which hardware so GPU is faster, he can't conclude nothing by running TF2 bench

Your absolutely wrong. He -CAN- conclude that -for- TF2, when what he is interested in is playing TF2. On the other hand if he is interested in palying TF2 and he's benchmarking Unigine, then he isn't getting anything useful at all. Nothing.

[dungeon]

Let say someone wanna by a new card... last thing on earth would be to look at TF2 benchmark results or anything CPU bounded

But results from Unigine benchmarks i can wholehartly recommend as that scale GPUs well

[dungeon]

Of course, for games you never know could next update screw up something or not

[smitty3268]

If Unigine did have games available, then the it still wouldn't be a benchmark, because then the only thing that would make sense to benchmark would be those games as individual games.

Synthetic benchmarks are a real thing and can be useful.

The key is knowing the context and how to interpret their results. Most people just aren't very good at that.

[duby229]

Let say someone wanna by a new card... last thing on earth would be to look at TF2 benchmark results or anything CPU bounded

But results from Unigine benchmarks i can wholehartly recommend as that scale GPUs well

Except that there is no game that they will ever be able to play that those results would be representative of. None. The results you recommend don't mean a single thing to a single person. Not one.

[dungeon]

TF2 result means even less then that, as on next update your results might be screwed

[duby229]

Synthetic benchmarks are a real thing and can be useful.

The key is knowing the context and how to interpret their results. Most people just aren't very good at that.

Sure, there are scenarios, like furmark if you want to see how hot your gpu will get under the worst case load. Almost every synthetic there should be very clear that the benchmark is designed to filter known bottleneck or for known bugs. That's the point in synthetics. Unigine is neither synthetic or real. It has 0 value.