So if I understood you right, your PCI-X SeriTek SATA card is faster than your PCI ACard SATA card, at least according to XBench, and it is also faster than IDE-to-SATA adapters you have used. I have some questions about that:
- Which IDE-to-SATA adapters were used on the MDD? Did you use the same SSD for all the 3 test cases? (PCI-X SeriTek, PCI ACard, SATA-to-IDE adapter)
- When you tested the ACard, were any other PCI devices also connected? (PCI and PCI-X devices slow down with each extra PCI/PCI-X connected)
I find it a bit strange that there could be performance gains by using a PCI-X card on a PCI slot, because it all boils down to the slot it is connected to, which is the same old regular PCI. Now whether or not the card itself has better technology (chips etc.) is something I find more believable, although I still tilt an eyebrow at that possibility, as all devices supposedly just provided the same SATA I interface over the same connector (PCI), over the same technique (SCSI "emulation"). Does the ACard use SeriTek firmware? The SATA card I used is not SeriTek's, but it is flashed with SeriTek's firmware. I do have a real SeriTek card (plain PCI card rather than PCI-X like yours), which I didn't test and compare against yet, but I do plan to do so sometime in the future, which hopefully might clear this up.
In the screenshots, I see the cards working and two benchmark results, which shows write/read speeds being faster in one than the other, but nothing on how fast booting was. In my case, I picked a physical timer and timed it multiple times.
Now, if the only comparison that was made between the PCI Acard SATA card and a "SATA-to-IDE adapter" was by using a different computer (PowerBook instead of MDD), that too with a different SSD model, brand and type/form factor (mSATA vs. SATA), then it is not a very good comparison. To be clear, in my case I used a StarTech adapter in the same MDD machine as the SATA card, and also the exact same physical SSD (not just same brand and model, but literally the exact same unit). My SSD is a 2TB Samsung 860 PRO.
============
Leaving the SATA business aside, I noticed you mentioned the Radeon 9000 as the "max video card" for OS 9 besides the GeForce 4 Ti 4600. Technically, it is the 3rd best card, as the Radeon 8500 (64MB VRAM only though, not sure if a 128MB version of it can be flashed) is considerably better. In fact, when I ran Diablo II with FPS displayed, the Radeon 8500 ran at a much higher FPS than the NVidia card, and I'm not sure why. (
The House of Moth suspected something is wrong with the NVidia drivers we host here, which made me wonder, although it could very well just mean NVidia drivers for OS9 aren't as good as ATI's, and/or older versions of their drivers might be better, and/or something is wrong both with my NVidia card and simultaneously with House of Moth's card coincidentally.)
On the NVidia side of things, there is yet another faster card that we can use (forgot its name... NVIDIA Quadro4 900 XGL, I think?), but it requires resistor modding to change its ID (if not also flashing). But it's a very rare GPU to find... So certainly not convenient, and I'm not sure if the difference in performance is noticeable.
I also wonder if the best Voodoo card, Voodoo5 5500, has some uses over the best ATI and NVidia cards, despite purportedly lower specs. But I got one just in case, to compare it against the ATI and NVidia offerings.