Hello Organorak,
Apart from that, all samples were identical with the exception that when I reinstalled Windows, for some reason my Zwolle configuration of 24bit samples front, 20bit samples rear that had loaded (just) in 23GB RAM on my RAID0 configuration refused to fully load in my reinstalled Windows environment so I had to adjust it and load some of the rear ranks in 16bit sound in order for it to fit. I don't know if that speeds up the loading?
In general, if samples are loaded in a lower resolution then they'll take less disk space and should thus be faster to load, to some extent.
If you format a hard-disk when reinstalling Windows the it will probably be less fragmented, which could also make it perform slightly better.
Having recently reinstalled Win 7 (64 bit) onto my computer I decided to try installing to a RAID1 configuration rather than as previously RAID0 to see what difference it made to load times. I am using two 1GB Samsung Spinpoint F3 harddrives. Best of three load times in seconds follows, starting from clicking on the organ in an open Hauptwerk and stopping the clock when the console is fully open.
_______._____RAID0____.____RAID1
Caen_________.__86_____._____158
Paramount 310____8_____._______6
Salisbury________67_______.___122
Zwolle_____._____91______.____188
Note that for sample set loading times to be measured fully meaningfully you need to reboot the computer each time before loading a sample set (otherwise Windows might still have some or all of the sample set file in its file cache in memory, rather than reading it physically from the disk).
Most (maybe all) RAID controllers built into desktop motherboards will give much higher performance with RAID 0 than RAID 1, with RAID 1 typically only giving about the same performance as just a single disk. However, it's theoretically possible for RAID 1 to give almost double the performance of a single disk (i.e. comparable to RAID 0) when reading large sequential files (because alternate chunks can be read from each of the two disks simultaneously), an optimisation some of the more expensive dedicated RAID controllers probably take advantage of (although I don't know specifically which). (A few years I used to use a 2-disk RAID 1 array via OS X 10.4's software RAID, which appeared to manage about 1.4-1.5 times the performance of a single hard-drive, and so presumably took advantage of that, or a similar, optimisation.)
I'm intrigued to see that whilst most samples take roughly twice as long to load as they did before, one sample (Paramount) loads faster in RAID1 than in RAID0. Are there any ways of optimising load times for the other sample sets?
The slightly lower loading times you saw for the Paramount probably aren't related to the RAID configuration. Perhaps they occurred because Windows had been reinstalled, the drives had been reformated (fragmentation or where they happened to be located on the platters), or Windows happened to have (more of) the file in its file cache in memory, for example.
Hence I don't think you should attach any significance to that particular result.
Hope that helps.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.