> : Well, but I just can't leave my hands away from the keyboard. I've
> : hacked an ISA interrupt handler for the Pica board. Dunno if it will
> : also work for the EISA Jazz machines. Then I took a short look into the
> : old Deskstation Tyne interrupt handler and found the part handling
> : interrupts >= 8 completly buggy. Tried to fix it but without a board
> : I can't test it ...
> : But the really good news is that the Acer now recognices ISA cards -
> : I tried a NE2000. I'm gonna try to boot the machine via TFTP.
> Cool. Did this stuff get uploaded before the great departure to the
> nether regions of Europe?
No, but see below ...
> Now that I'm back from visiting my inlaws for a month this past
> weekend, I should have a little time to spend getting all the latest
> software and hopefully booting it on my Deskstation. Maybe I'll get a
> chance to see if the code that you've done is good for the Deskstation
> or not. In any event, I hope to process this, and the other patches
> that I owe people this week.
I've got even better news for you:
> Console: 16 point font, 400 scans
> Console: colour PICA-S3 80x25, 1 virtual console (max 63)
> Calibrating delay loop.. ok - 66.56 BogoMIPS
> VDMA: R4030 DMA pagetables initialized.
> Memory: 7276k/8192k available (640k kernel code, 276k data)
> Swansea University Computer Society NET3.033 for Linux 1.3.38
> NET3: Unix domain sockets 0.10 BETA for Linux NET3.031.
> Swansea University Computer Society TCP/IP for NET3.032
> IP Protocols: ICMP, UDP, TCP
> VFS: Diskquotas version dquot_5.6.0 initialized^M
> Checking for 'wait' instruction... unavailable.
> Linux version 1.3.48 (ralf@rio) (gcc version 2.7.2) #1 Tue Dec 26 14:25:07
> MET> Floppy drive(s): fd0 is 1.44M
> FDC 0 is a post-1991 82077
> ne.c:v1.10 9/23/94 Donald Becker (email@example.com)
> NE*000 ethercard probe at 0x300: 00 40 95 e1 1d c2
> eth0: NE2000 found at 0x300, using IRQ 12.
> NFS: Got RARP answer from 184.108.40.206, my address is 220.127.116.11
> NFS: Got file handle for /tftpboot/18.104.22.168 via RPC
> VFS: Mounted root (nfs filesystem).
For all the people that are notorious benchmarkers here the results of
Dhrystone (think it's version 1) running on a Acer Pica (R4400 133MHz)
compiled with gcc 2.7.2 and -O3 optimization:
> Dhrystone time for 100000000 passes = 373
> This machine benchmarks at 268096 dhrystones/second
For comparison I benchmarked a board with Trition chipset and a 586 at
100MHz. Same compilers and options:
> Dhrystone time for 100000000 passes = 516
> This machine benchmarks at 193798 dhrystones/second
So far I'm satisfied with the speed. The only problem is that NFS is slow
by principle and really begins to suck when Linux is both server & client
and two NE2000s are in use and all binaries are static & huge ...
I've built some more of the essential software (~70mb) for a Linux system,
including the world's second best editor ed and the world's finest editor
vi. So my Linux/MIPS system starts to behave like a *real* Linux system.
Unfortunately I found that glibc still is quite far from getting 100%
compatible with the Linux libc which makes it sometimes a bit difficult to
build software for Linux/MIPS. Also the kernel is still relativly fragile
so I didn't yet try to rebuild Linux/MIPS on a Linux/MIPS machine. But I'm