4:28 PM - IDE controller CMD PCI0646U on the Sun Ultra 10 Creator 3D
I've been trying to determine why my FreeBSD Sparc64 install only supports WDMA2 (16mb/s) with the CMD PCI0646U IDE controller. According to various google results, the IDE controller is extremely buggy and does not properly initialize itself. There are several undocumented features and the vendor does not want to release the information. As a result, the card is marked as broken and degraded to WDMA2. It theoretically can support UDMA33. I do not know how fast it runs on solaris. OpenSolaris may provide clues...
NetBSD introduced a flag in the kernel to enable Ultra DMA 33, but cautions against it pointing to the linux developers observations. I've previously hacked on some FreeBSD SATA code and may need to do the same here. Its worth trying UDMA33 before I put the machine into production.
The CMD PCI0646U2 supposedly can be hacked around and made to work properly.
On a side note, it appears solaris 2.6 can't address a 120gb disk even though most suns from the era can handle up to 137gb disks (ultra 5s and higher) . I've got a working FreeBSD sparc64 6.0 install with an 80gb disk and had solaris 10 working fine on my ultra 10.
NetBSD introduced a flag in the kernel to enable Ultra DMA 33, but cautions against it pointing to the linux developers observations. I've previously hacked on some FreeBSD SATA code and may need to do the same here. Its worth trying UDMA33 before I put the machine into production.
The CMD PCI0646U2 supposedly can be hacked around and made to work properly.
On a side note, it appears solaris 2.6 can't address a 120gb disk even though most suns from the era can handle up to 137gb disks (ultra 5s and higher) . I've got a working FreeBSD sparc64 6.0 install with an 80gb disk and had solaris 10 working fine on my ultra 10.
0 comments