New problem: Apollo 1240/25 accelerator won't boot correctly

[dweomer]

Now that I think I have the problem with the PCMCIA ethernet card licked (spend more money on another card, that oughta fix it), I installed the accelerator and it's acting wierd: Guru errors, Workbench boot-up errors. It keeps trying to boot, but no go.

 I tested the 16mb simm that came installed in the Apollo in something else, to make sure that worked fine, then I tested the Apollo without the memory in the A1200. Boots fine, no problem.

I thought it might be giving me grief because I was using the wimpy original power supply, so I upped the p/s to an A500 35watt unit--no Bigfoots (Bigfeet? Proudfoots/Proudfeet?) to be had at Software Hut, if anybody has a spare, I might be interested), but got the same results as before, with an old simm that I'd used in my A2000 with A2630, and with the simm that shipped with the Apollo.

 :-? Hmmm. Held down mouse buttons for early startup, disabled the CPU caches, and everything boots up fine, all ram is recognized. The thing seems happy as a clam.

How much speed am I sacrificing here with the caches disabled? Is there something loading at boot that's causing this error, that I could alter and still use my caches? It's kind of a pain in the rear to always boot with both fingers down, too.

I have no previous personal experience with the Apollo 1240, but it seems to get mediocre reviews. I just wanted to get some fast ram going, and this was a fairly inexpensive way to do it.

Current system configuration: A1200, 8mb 2.5" ide hard drive, OS3.1 on 3.0 rom, Apollo 1240/25 with 16mb fast, no scsi.

Any assistance is appreciated.

DWMR

[Piru]

You have the proper 68040.library installed?

(Sorry I have no further information what is correct for Apollo, I've only had p5 hw)

[dweomer]

Yes, I think so.  When I installed the OS on the 8mb drive it seemed to be there by default. Unless it's corrupted (seems unlikely) or there's a better version for Apollo boards that I don't know about (entirely possible).

DWMR

[elendil]

Also make sure you have a recent setpatch in your c: drawer.

Both (setpatch and 68040.library) are very likely to cause these errors, afaik.

Sincerely,

-Kenneth Straarup.

edit: I am not sure about this but your memory might also be too slow / too fast for the accelerator. I think some accelerators handle (or perhaps even require?) 60ns edo ram, while others cannot handle such fast (relatively) memory.

[N7VQM]

edit: I am not sure about this but your memory might also be too slow / too fast for the accelerator.

Faster RAM shouldn't be the problem.  If the system needs at least 60ns SIMMs and you put 50ns SIMMs, you shouldn't see a problem since 50ns is faster than 60ns.  Don't put 70ns SIMMs were 60ns SIMMs are required, though.  It's been awhile since I played with SIMMs but I seem to remember issues with putting EDO RAM into systems that don't support EDO.

PC133 can be an exception to the speed rating.  Certain manufacturers PC133 DIMMs simply won't work in PC100 systems.  This is most likely caused by bus impedance issues or a designed-in limitation but, the company I was dealing  with wasn't answering questions.

[dweomer]

The documentation for the Apollo calls for 70ns or faster. It's been
quite a while since I've seen anything _slower_ than 70ns, but I
suppose it's possible. I'm shopping around for some simms that will
fit the slot(chips only on one side), 32mb fast page mode. Everything
I have laying around is sdram.

[csirac_]

Perhaps it doesn't like double sided ram? My CSA 12 Guage would act wierd with a double sided simm with OS3.1 kickstarts; then again it acted wierd with 3.1 kickstarts anyway..

CPU chache does make a big difference (I did a VERY tedious assignment testing performance of CPUs with L1/L2 caches in different sizes/configurations).

I would have to say, read up about this accelerator, find the correct 040 lib and latest setpatch. It's not running/had been running overclocked? Does it boot OK with no startup sequence with CPU cache turned on?

A (very) long shot... clean the SIMM/slot and CPU edge connector with metho, and try (very carefully reseating the CPU.

Short of finding known working 60ns FP ram I don't have any other ideas..

- Paul

[PMC]

Well that to me sounds like a software problem.  I know that the BB2 revision of OS3.9 doesn't like the RemApollo utility, and caused similar issues with my system.  

Can you tell mne what OS you're using and whether you're using any patches etc with it?

[dweomer]

From original post:

"Current system configuration: A1200, 8mb 2.5" ide hard drive, OS3.1 on 3.0 rom, Apollo 1240/25 with 16mb fast, no scsi."

Showconfig reports : Kickstart version 39.106, Exec version 39.47, Disk version 40.42. It also shows the accelerated "CPU 68040/68040fpu." Interestingly enough, it breaks down the 16mb simm into two 8mb nodes, which I don't remember my old A2KHD doing, but it's been awhile.

Also, I'm not using a double sided simm. I tried one out (8mb) and it had the same issues as the single sided 16mb that shipped with the Apollo

The only patch installed is a newer version of setpatch from Aminet, installed based on earlier advice in this thread, which doesn't seem to be causing any new errors, but doesn't solve the problem either.

[JaXanim]

I think the answer to your problem lies in your description of the current system.

I think you'll find that you need the v3.1 ROMs to run Workbench 3.1.

Cheers,

JaXanim
http://waveguide.v21hosting.co.uk

[dweomer]

Swing and a . . .  miss.

3.1 runs fine on 3.0 roms, in my experience. 3.5 _requires_ 3.1 roms, but you only get a few cosmetic diffs with 3.0 roms and 3.1 on top.

Anyone care to correct me? That's a cheap fix if it's the real problem, but I've always (for years) heard that 3.0 roms ran fine with the upgraded disk versions.

by the way, I booted w/out startup sequence, got to the prompt, and genetic species runs very nicely, so I'm tinkering with the startup-sequence now, to narrow it down. Is there a good debug util for this process on aminet--i'm looking, but if anyone can save me spending precious brain juice, that'd be great.

[dweomer]

The memory turned out to be the problem.

I was rebuilding a couple of 486 based systems, digging around in my storage boxes for parts, and happened upon my box of jumbled ram that I thought I'd given away. I think I actually did offer it to someone, but they just sniggered and said, "No thanks, man."

I had been calling around to various memory companies looking for a single sided simm that met my spec, and was having no real luck, so I decided to try all the 72 pin simms in the Apollo. One at a time. To see what happened. Boy, I  need a real life.

The last simm I tried was the keeper. 16mb single sided were plentiful in the box, but apparently none of them were high quality enough until the very bottom of the stack. I'd still like a 32, just to max it out, but I can hold off and browse a bit before buying now.

Interesting tidbit: I'd had the 1200 null-modem cabled to my pc to use the Amiga Explorer software bundled with UAE. It doesn't work with the accelerator plugged in. Ah well. I'm better off than I was before.

[olegil]

I had 32MB on a dual-sided stick until I soldered in an old SIMM slot salvaged from a PC motherboard and upgraded to 2x32MB. Never looked back :-)

For some reason I often need to touch the chips and memory sticks before booting, or it won't find any memory at all. Have never found out excactly what is causing it, but when it boots it seems to work ok.

I'm using a kickstart 3.9 image made by the author of Remapollo (been using it on UAE as well)