Tweet
Hello,
I'm finding it more difficult to use the vray benchmark results than the otoy ones, when it comes to renting or buying workstations / render nodes. The Vray benchmark scores page is feeling just a bit too much like an overclocking contest leaderboard (which is cool too) and not as useful when wanting to know the real world performance one can expect from a particular hardware config, at stock.
The first suggestion is to use the AVERAGE of all results in a particular config instead of the MAX result for a configuration. For example if I search for a threadripper 3970x, the result shown is 40,332. This actually is the fastest score out of the 92 scores you have on record. its safe to assume some kind of specialist overclocking has been done. you can see some comments from others in the top 10 results indicating CPU and memory overclocking. I doubt the fastest score is the most generally useful result to show unless you're a competitive overclocker.
The average score for the 3970x based on a brief scan of the results looks to be about 33,000 points. and this is what someone should expect when building or purchasing a system, powering it up and hitting render.
I am guessing that you aren't currently calculating the averages, so maybe you can show the result that is most close to the average? That should be fair in my opinion but maybe if you agree with this wish, you would rather calc and show the actual mathematical average. Either case would be immensely more useful.
I'll put the second suggestion into another topic.
thanks for reading
I'm finding it more difficult to use the vray benchmark results than the otoy ones, when it comes to renting or buying workstations / render nodes. The Vray benchmark scores page is feeling just a bit too much like an overclocking contest leaderboard (which is cool too) and not as useful when wanting to know the real world performance one can expect from a particular hardware config, at stock.
The first suggestion is to use the AVERAGE of all results in a particular config instead of the MAX result for a configuration. For example if I search for a threadripper 3970x, the result shown is 40,332. This actually is the fastest score out of the 92 scores you have on record. its safe to assume some kind of specialist overclocking has been done. you can see some comments from others in the top 10 results indicating CPU and memory overclocking. I doubt the fastest score is the most generally useful result to show unless you're a competitive overclocker.
The average score for the 3970x based on a brief scan of the results looks to be about 33,000 points. and this is what someone should expect when building or purchasing a system, powering it up and hitting render.
I am guessing that you aren't currently calculating the averages, so maybe you can show the result that is most close to the average? That should be fair in my opinion but maybe if you agree with this wish, you would rather calc and show the actual mathematical average. Either case would be immensely more useful.
I'll put the second suggestion into another topic.
thanks for reading
Comment