Improved win10/11 drivers from Nvidia, vastly improved processing times in PL

((copied from a single post in another thread))

For the past year or so (beginning with driver version 5.11), NVIDIA has been working to implement a vastly-improved OpenCL/CLANG JIT compiler internal to the drivers. This was an end-user opt-in post-install configuration choice for “power users” until the most recent release of the driver packages.

There is some (minor) information available on this in the NVIDIA Driver release notes, section 2.5.3.

Beginning this week with the release of driver pack 5.22.30, the use of the new compiler became standard for Windows and Linux users. There is no indication when/if the improved driver sets will be available for Mac.

The new internal compilers also improve/speeds up CUDA and DX11/12 operations.

Beyond DxO, I have seen a 25-40% improvement in GPU-specific rendering and video stream manipulation work on my Davinci Resolve Studio v18.0.4 installation. I am not a PC gamer so can’t speak to any improvements there :wink:



I’m on Windows 10. I downloaded the latest driver for my GTX 1050ti and immediately saw a 30% improvement in processing time for DeepPRIME and DeepPRIME XD. I’m very pleased.


1 Like

Wow yes, easily a 30% improvement!

1 Like

My GTX 1050ti card is long in the tooth. With this new driver acceleration, I would imagine that even mid-range cards will allow almost anyone to use DeepPRIME XD on a regular basis without slowing down their processing significantly.

When DeepPRIME was originally released in PL 4, I did not have a card which was supported. Average CPU processing time for a single image was around 91-92 seconds!! When I first got my card, processing DeepPRIME dropped to around 30 seconds but went down to around 20-22 seconds after 2 years of tweaks by DxO. Now with this new driver, processing time for the same images has dropped to to 14 seconds which is around 6.5 times faster than the initial PL 4 CPU processing time…

Amazing!!! Really happy about this


1 Like

Sometimes I wonder over these kind of dramatic improvements in the next generations of structural growth and developement. Why do these break throgh steps come far later than at the original release of these cards, processors, auto focus tech or what ever …

Why didn’t these developers see these development paths from the beginning or is the explanation that the marketing people prefer to release these improvements step by step to lure us to buy new bodies and upgrade our software every year and year after year?

Refinements such as improved processing time may take a significant amount of additional effort and testing. In the meantime, if you hold a product back until this is done it may mean a huge delay getting it into production as well as taking resources away from other new features. Who among us would have been willing to wait a year or two more to get DeepPRIME while DxO engineers refined the processing time? In fact over time they may still find more ways to make it even faster.


I have now upgraded my Geforce RTX 3060 TI driver.

Sorry for me: absolutely none improvement with Photolab 6 and Windows 11 in my machine.

I get 3-4 seconds export time with Deep Prime 100% JPEG (uncompressed)
and get 7-8 seconds with Deep Prime XD and the same conditions (24 MP images).

I really don´t complain because I´m more than happy with the processing time I´ve got.

Don’t be sorry, I have also got on my GTX 1660S with PL5.5/W11/522.30SD absolutely no improvement…

Seems AMD has been making strides too.

Just updated AMD Drivers and then processed an image via DP XD and came out at 10s for an AMD Ryzen 5 3600

Under DP it is now 8s too

Reckon about 50% faster on previous drivers

How big are you raw files?

I think that users of Studio drivers may notice a huge improvement because they are updated on a longer schedule.

Users of the Game Ready drivers (like me) are seeing no further improvements because probably the improvements had already been implemented months ago.

1 Like

It’s in the text you commented :slight_smile:

“24 MP images” in this case.

Sounds like a plausible explanation. My computer is brand new.

1 Like

Win 11 here, but same 1050ti card. I couldn’t believe it, but processing of one image takes now 40s instead of 60s on my computer, which is a huge improvement (for a driver update).

@ Sparky2006
Thanks for the info!

As info trickles out of NVidia, it appears as through the new compiler were default-enabled in the game-ready driver releases in early September.

That sounds like a bit more than the 30% improvement I’m seeing, but in the same general range. An early Christmas present! :smiley:


By the way, for what it’s worth, athough I don’t use it much I have a license for ON1 Photo Raw 2023 and when I tested it yesterday I noticed a speed up of some of its functions as well. I suspect that those of us using other processing software alongside PhotoLab may also notice improvements on features using their card’s GPU.


For me also (GTX 1650 MaxQ combined with a 8gen i5) the speed of 20MP files has improved from around 50 seconds to around 35 seconds. For Deepprime the time has gone from 16 seconds to 12 seconds. Seems a 25-30% speed increase. Not bad without buying new stuff :slight_smile:

1 Like

But in the release notes for the Studio driver, 522.30, it states the list of changes, which includes the optimal support for DXO Photo, are since driver 522.25, which is the latest Game ready driver released 6 days previously.

I never tested with the most recent previous driver. Perhaps that would have also improved processing speed.