Subject: Re: fpu trap with current from 2 Jan
To: David Rogers <davidrogers23@earthlink.net>
From: Kazuyuki Inanaga <happyday@pp.iij4u.or.jp>
List: port-mac68k
Date: 01/16/2004 05:50:03
Hi,

My Macintosh IIci + DAYSTAR Turbo040/40MHz runs tremendously
fast on NetBSD-1.6.1. Compare the benchmarks with normal IIci.

normal IIci('030/25 MHz)
Dhrystone Benchmark, Version 2.1
Microseconds for one run through Dhrystone:    146.6
Dhrystones per Second:                                   6821.3

IIci + DAYSTAR Turbo040/40MHz
Dhrystone Benchmark, Version 2.1
Microseconds for one run through Dhrystone:     25.5
Dhrystones per Second:                                39215.7

bytebench-4.1.0              FINAL SCORE                 4.0

This is scored on NetBSD-1.6.1 with MacOS/CDEV/QuadControl all ON.

NetBSD-1.6ZH (20040112 JST), MacOS/CDEV/QuadControl ON, I get
uvm_fault right after NetBSD copyrights and boot sequence drops
into db>.
If Macos/CDEV/QuadControl/Secondary Cache is turned OFF, 1.6ZH
boots up. This is different from dave's IIci.

At 5:50 PM -0600 04.1.11, David Rogers wrote:
>
>I also tried booting with extensions off.  same result.  then again
>with no nubus cards in.  still no dice.

At 0:41 AM -0600 04.1.12, David Rogers wrote:
>I have a kernel that I built on 12/28/03 that works just fine.  I'm not
>sure what changed between then and now.

I tried to boot 1.6ZG (20031221) and 1.6ZE (20031104) kernels, but
no luck, got same MMU fault and drops into db>. This is different, too.
What does make such differences? Any idea?
My Turbo040 is daughter card type, and QuadControl is version 2.3.
How about yours, dave?

When QuadControl/Secondary Cache is turned OFF, my 1.6ZH runs
much slower.

Dhrystone Benchmark, Version 2.1
Microseconds for one run through Dhrystone:     39.3
Dhrystones per Second:                                25423.7

bytebench-4.1.0              FINAL SCORE                 2.7

I also put Turbo040/44MHz into SE/30 and made it "SE/40". I really
expect to fix this. I'll try to compile 12/28/03 kernel, but I dislike
to see that "BAD SUPER BLOCK: CAN'T FIND SUPERBLOCK" again ;)

Thanks,
Kazu Inanaga