|
Slight stray from topic to update the last line of my post above "I just which the LN reference generation and local normalisation steps in WBPP were both GPU accelerated - that would be a real boon!" in another thread on my dual Xeon rig - I noted swapping from Windows 10 Pro to Windows 11 Pro - sped up WBPP by a factor of 3.3 times for me - with the major speed gain in LN Reference Generation and Local Normalisation.
Tasks that used to take 10 - 20 minutes per master (and I have many masters in my typical workflow) now take 1-3 minutes - that is an amazing boon that left me delighted.
I really think PI deserves a WBPP benchmark - as well as its normal post processing benchmark - as My WBPP times have now decreased by 3-4 hours to often under an hour - that is a very real win.
Back on topic now - as Peter mentioned the 5XXX series of NVidia cards now do enable GPU acceleration for key tasks within PixInsight - something that for a few months was a worry it would not be enabled. So it is a total win.
Even I have shifted to the single repository line GPU install approach on all my PI rigs - its a great one line solution!
|