One question about CS6 - I'm using beta now and really disappointed with speed of saving files in ACR module. If in CS5 it was 4 files in row and my 2600K was loaded at 96 - 98% constantly - in CS6 beta its one file and CPU load jumping from 50 to 90% result - 2 or 3 hours to save 1500 files against one hour or less in CS5 (same PC - near top - 2600K 16 Gb RAM, SSD SLC for system, adobe cache on 10K rpm WD HDD, Win 7 64 bit). What about commercial version?
The new Camera Raw is a little slower to do the actual conversions - no question.
I expect the improved conversion quality (e.g., more perfect CA removal, better shadow/highlight retention, etc.) to come at some computational cost... There's no reason to think that everything that's newer should also be faster, especially not when it has new features and produces better results. You can be sure the previous versions were highly optimized, making them tough to beat.
I have several versions of Photoshop on the same workstation, a dual quad core Xeon x5460 system (8 cores total)... Here's a comparison of my conversion times:
Times to convert 50 Canon .CR2 raw files converted to 6144 x 4096 x 16 bits/channel PSD files:
Each file in the newer version seems to take about 1.5 times as long as the previous version to convert, which is quite acceptable from my perspective given the increases in image quality.
Guess we're going to have to start thinking of dumping these old, slow systems and getting some really powerful computers.
Not entire correctly. I have LCD screen on keyboard, that show CPU load realtime. Looks like new conversion module lost multi-thread compatibility. Main problem - CPU didn't loaded fully - ACR 6.6 was able to work at 4 files - new just 1 file. And performance is jumping from 40 -50% to 90 - 95% And for me it looks like serious problem with software optimization. BTW new CA module without manual correction work like dumb auto red eye removal module - if it didn't detect CA that visible on picture - it didn't remove it at all ;( - like module that couldn't for example detect red eye on close face portrait.