Interesting article. Would their conclusions still be valid for the specifics of a GigaDAW? I would think that access times and concurrent reading is more important here and would favour SCSI a tad more than it did in this \"generic\" DAW review.
I\'d love to try out SCSI just because I\'ve heard so much comments that SCSI is much better than IDE, but I just simply can\'t afford buying the good HD + controller.
On the other side, I\'ve just purchased an Athlon T-Bird 900, 384 megs of ram and seagate barracuda 20gig + maxtor 40gig (both UDMA/66 and the giga drive is maxtor). My motherboard is ASUS A7V that has onboard Promise Ultra/100 plus VIA builtin Ultra/66.
When I plug my harddrives into Promise Ultra/100 controller, I was getting less than 80 polyphony. When I plug in my HDs to the builtin VIA Ultra 66, I was getting 160 polyphony using GigaPiano. No sound breakups.
If I copy Gigapiano gig file from one harddrive to another, it takes roughly the same amount of time using VIA or Promise controller.
This leads me to believe that controller that plugs in to PCI bus may add latency that drags Giga performance. But it doesn\'t drag Cubase or other types of tasks that make use of sequential read. What we need for GigaDAW is LOW LATENCY for both drives and controller, and IDE drives (especially maxtor diamond max) is up to the job.
<BLOCKQUOTE><font size=\"1\" face=\"Verdana, Arial\">quote:</font><HR>Originally posted by abi: This leads me to believe that controller that plugs in to PCI bus may add latency that drags Giga performance. But it doesn\'t drag Cubase or other types of tasks that make use of sequential read.<HR></BLOCKQUOTE>
Indeed. I have made the same type of observations with a plug-in Highpoint ATA/66 controller card. One can measure the access time in diagnostic programs like Sisoft Sandra lite (free download). In my case a drive that measured at 5ms via the onboard VIA controller measured at around 30ms via the plug-in controller. This means that the random block-read rate (and consequently the polyphony) radically drops. This is really an issue that people need to be aware of.
Now my question is only: does the same thing not happen when using a plug-in SCSI controller? These cards frequently have designs so close to the IDE controllers (with similar chipsets), that we really need someone to measure this. After all, SCSI does incur quite a price premium!
[This message has been edited by cc (edited 09-09-2000).]