Guys I just released v1.0.2.1. The user reported issues above that provided enough information should be fixed. The ones that report it doesn’t work with no screenshots or information for me to diagnose, I can’t help you.
Earlier I wanted to be conservative so I had a 2gb vram floor and 4gb sysram floor for autotune to prevent the app from being too aggressive with recommendations. This worked well on my 96gb vram with 96gb sysram hardware but for folks with 16gb vram and 16gb sysram, that means they are left with 14gb vram and 12gb sysram so all the autotune test will automatically fail. I removed the vram buffer but still kept a 1gb sysram buffer for systems with 24gb and below sysram and 2gb sysram buffer for systems with > 24gb sysram.
I also added a cuDNN only test configuration so users with 16gb and especially 12gb vram that want to keep stock SLP settings but still get the speed boost of upgraded cuDNN, can easily test the speed gains without doing the full suite of tests. You can also add and delete rows in my built-in suite to provide the app more datapoints for the recommendation. The algorithm will only take into consideration the successfully tested rows for the recommended conservative and aggressive presets…
I’ve spent dozens if not over 100 hours testing/benchmarking but I only tested it on my own hardware so no matter how many more hours I run tests on my own system, it’s impossible to guarantee it will work the same across other systems with different hardware configurations, different Topaz versions, Windows versions, driver versions, different Topaz installation paths, different output resolutions. If you report issues with screenshots and adequate information, I can try to fix them.
For those who are still questioning whether my Tuner app increases processing speed at the expense of quality, you can use my Video Compare app to definitively compare the results for yourself.
