Furia MAPROM / SHADOWROM issues

[amigasith]

Hi all,

since a couple of weeks, I'm the proud owner of a Furia Rev2b with FPU - the red one. In general, I think the Furia is a great little card that has a lot of horsepower, but there is one thing that really drives me nuts with it: I cannot get it stable with either MAPROM or SHADOWROM enabled As soon as I start using one of those two features, I get instabilities.

Here is what I have done so far / tried so far:
- The A600 is re-capped
- Cleaned all 68000 pins before fitting the Furia
- Tried different PSUs - ATX and original ones
- Added an earth cable - to the floppy bracket - in order to ground the Furia
- Tried an SD card
- Tried with the RAM wait state jumper open / closed
- Tried different parameter combinations of furiatune

The problem I am facing is that I get a red screen (Kickstart error) when I reset the A600 after letting it work work for a while. With "work" I mean running an LHA loop into RAM: for ~30 minutes. If I don't reset the Amiga, I can let it run for hours with the LHA loop without any issue. It's really the reset that causes the red screen. When the A600 hasn't worked for some time, I can reset it for 20+ times without getting a red screen.

If I don't use MAPROM or SHADOWROM, everything is fine. However, I would really really like to use MAPROM, since it provides such a nice speed boost - the entire system feels much snappier with it.

Is there anything else that comes to your mind that I could try

[project23]

Welcome to the perilous world of Furia ownership. It actually sounds like yours is more stable than a lot of others.

Grab yourself a copy of MBRTest-2, which is the software Lotharek uses to test the cards. At the very least you'll be able to stress test the memory and deduce whether or not there are any repeatable issues with particular memory data/address lines.

It would seem that these things are incredibly sensitive to working conditions. I say that as Lotharek has tested 'bad' cards and found them fine. It's possible that like me, you actually have a bad ram chip or two, but to be honest I find this unlikely given how stable you're reporting it with those features disabled.

I did find that the PSU I used made a minimal difference. An ATX supply was often worse than a good old A500 one - I found that for some reason running the Amiga loaded the 5v rail down to about 4.6v, while the A500 supply stayed stable around 4.9v. It could be that Lothareks power supply is absolute rock solid and this is why he's not experiencing problems - but this is all mere conjecture.

Now - I do note that MAPROM/SHADOWROM run in an area of memory beyond the 8mb addressable fastmem. If you can get a hold of MBRTest-2 as i say, it will allow you to specifically test those memory areas. If there is a problem, it'll show up as consistent failure on particular data lines. On the off chance that it is memory related, you're probably lucky that it's up in the SHADOWROM range - but at least you'll know either way.

If the ram is fine, then its simply the stability/instability of the memory/cpu - see my thread here on the proliferation of counterfeit CPU's of this type - but don't be so quick to cite this as the problem, as don't forget Lotharek personally tests all these cards before they go out. It could however be that the counterfeit chips (if you have one) are much more sensitive to fluctuating supplies, or heat, or something like that.

Counterfeit CPUs and the Furia

Hope that's of some help - and don't be afraid to contact Lotharek he's a helpful guy - far better help than amigastore or the like.

John

[amigasith]

Hi John,

thanks for your comment! Of course, before creating my own thread, I read yours with great interest. I will try to run some tests with MBRTest-2 as you suggest, however I somehow doubt that I will find bad data / address lines.

Nevertheless, I am almost 100% sure that the issue I described above is somehow memory-related. I just finished an endurance test where I did *not* hammer at the memory: I let the Amiga run for a couple of hours with just the screen saver "Blanker" from "Tools/Commodities" and the result was that I could reset the machine 10+ times after the test without any problems at all - with MAPROM enabled, by the way.

Besides memory, the only other thing I could imagine is some logic in the Furia firmware code related to the reset operation that gets confused under specific circumstances?

Cheers,
Marc

[project23]

Can you do something CPU intensive for any reasonable period of time?

I found that the frontier intro would run for at best 20 mins before crashing out. I can't imagine that's thrashing the memory too hard, but it is very CPU intensive.

Maybe that can help pin down if its memory or CPU a little? (With me I think it may have been a combination).

EDIT: On second thought iirc Frontier would always guru at the same instruction addresses, with illegal instruction errors - so I dunno, maybe it was an intermittent memory fault after all?

Either way i'd try and do both Frontier and MBRTest-2. Do all of the tests in MBR, and in particular i'd look at the shadowrom area. Something of note was that for me at least diagrom would always pass flying colours on all memory tests. I have no explanation for this or what it points towards.

We need to start a Furia Owners Relief Fund or something

John

[Mark sealey]

Hi

With my Furia i could not get both Maprom and ShadowRom to work with OS3.1....When i upgraded to OS3.1.4 there was a HUGE improvement in Stability.....it was like a Magic wand had been waved over my A600.

Before with os 3.1 i would lock up.....this was also with a modded PC psu.


With OS3.1.4 burnt to a rom....Everything WORKS......i have Furiatune shadowrom in my startup sequence....been like that for 5 months.....not tried MapRom yet.

Im now using a "light weight" A500 psu...and cannot remember the last time i locked up or crashed.

https://www.facebook.com/AmigaDudeA600/?ref=bookmarks

[project23]

The white boards with the 'flat' ram chips seem to be doing better than the more recent red boards. A few people with the older white ones have spoken up on the stability of their cards. It seems to be people who purchased around this time last year that are having the problems.

That being said, 3.1.4 would be worth a try. Can't remember if i was running that - think i was. I did try 3.9 too (slow as hell, like), with no improvement.

[amigasith]

Originally Posted by project23

Can you do something CPU intensive for any reasonable period of time?

Alright, instead of messing around with Frontier, which I have never tried in the past, I just let my A600 play an mp3 for ~2 hours in a loop. After that, I could reset the machine without any problems for more than 10 times with MAPROM enabled. This tells me it really must be memory-related, since playing mp3's is hammering hard at the CPU and only uses a buffer of 16KB if I remember correctly. MBRTest-2 is still on my list, and I will update this thread as soon as I have some news on this front.

Originally Posted by Mark sealey

With my Furia i could not get both Maprom and ShadowRom to work with OS3.1....When i upgraded to OS3.1.4 there was a HUGE improvement in Stability.....it was like a Magic wand had been waved over my A600.

Now that's really an interesting find! I am also running OS 3.1. Admittedly with quite a bunch of patches, but still OS 3.1. Before I go down the 3.1.4 route, I would really like to have some additional confirmations of owners who are running 3.1.4 with their Furia without any issues.

--> Anybody out there?

[amigasith]

Before somebody concludes that SHADOWROM didn't work for me, I think I have to add some additional bits of information here: SHADOWROM itself worked without any problems, but I tried using LoadModule on top of SHADOWROM in order to load a couple of resident modules and *this* didn't work at all. I have used this method successfully for all my other Amiga models that have MAPROM support, so I don't really see why this shouldn't work for the Furia.

With LoadModule from Thomas from aminet here I always got the same Guru after resetting (Error: 8002 8001 Task: 002032B0). Sometimes right after the first reset, sometimes after my LHA loop stress test. Over on a1k.org, Thomas confirmed that this Guru is actually an error message from LoadModule saying that one or more modules got corrupted.

I also tried another LoadModule variant from aminet from here and I also got module checksum errors. So all in all, my conclusion was that something was overwriting the modules in memory - why should they become corrupted otherwise?

I tried all sorts of command line parameter combinations for furiatune - with addmem, without addmem, with ide on, without ide on, etc. - but nothing really helped. On top of that, I also tried many combinations of command line parameters for both LoadModule versions, but none of them helped either.

It were my experiences with LoadModule that actually led me to try MAPROM ...

[EzdineG]

Originally Posted by amigasith

Alright, instead of messing around with Frontier, which I have never tried in the past, I just let my A600 play an mp3 for ~2 hours in a loop. After that, I could reset the machine without any problems for more than 10 times with MAPROM enabled. This tells me it really must be memory-related, since playing mp3's is hammering hard at the CPU and only uses a buffer of 16KB if I remember correctly. MBRTest-2 is still on my list, and I will update this thread as soon as I have some news on this front.



Now that's really an interesting find! I am also running OS 3.1. Admittedly with quite a bunch of patches, but still OS 3.1. Before I go down the 3.1.4 route, I would really like to have some additional confirmations of owners who are running 3.1.4 with their Furia without any issues.

--> Anybody out there?

My (white) Furia’s IDE absolutely does not work with any device if I install a 3.1.4 ROM. I have to get into the early startup menu and disable caches, after which it then will.

Reverting to 3.1 I have no IDE issues with caches enabled, but i am unable to use any of the later 3.5+ ROM updates because they will then behave similarly to the 3.1.4 ROM with regard to the IDE.

The behavior is that I will remain at a black screen indefinitely if the hard drive boots, or the machine will crash if the drive is accessed if booting from floppy.

Everything works great if I avoid the newer ROM or updates.

[project23]

Sorry Marc but i'm yet to get to a point with the Furia where I can mess around with maprom or shadowrom. I'm about 80% done moving the components over to a fresh PCB. If it works i'll follow what you've done so we can compare results - maybe that will help you in some way?

It seems you really are quite lucky to have it as stable as you currently do.

John

[amigasith]

For the records: I just finished a 3+ hours test cycle with MBRTest-2 with all RAM test cases enabled. As expected, MBRTest-2 didn't find a single error. Of course, the A600 didn't lock up or anything similar.

Now what's really puzzling is that I could reset the machine 20+ times after the 3+ hours MBRTest-2 test run without any problems - again MAPROM was enabled I don't get it

[Mark sealey]

Yes as above i have to "double mouse button" to boot to early startup menu and DISABLE CPU Cashe....or the Furia Locks up....Did also Disable CC0 card slot to improve things...further....But I can Confirm my Furia Works fine with OS3.1.4 Shadowrom...IDE speed up...etc....I might make a .HDF of my boot partion ;-)

I made a Modded os.3.1.4 rom that had Card.device removed so i did not have to disable CC0

I did hear the Expansion.Library Chokes the Furia so may make another Rom with the OS3.1 Expansion.library on it.

Also With My Rasberry Pi1 Floppy Emulator i can boot direct from DF0/DF1 without going into the Start up menu....so i thought about making a A1000 style boot disk.

[amigasith]

Hmm... Based on your comments above, I tried again with disabling caches. First try was via "furiatune cache off" in Startup-Sequence and then "C:CPU >NIL: NOCACHE". Both variants didn't work Instead of getting a red screen after reset, however, my LHA loop stress test crashed with a "program failed..." error while running. Meh.

[Mark sealey]

Originally Posted by amigasith

Hmm... Based on your comments above, I tried again with disabling caches. First try was via "furiatune cache off" in Startup-Sequence and then "C:CPU >NIL: NOCACHE". Both variants didn't work Instead of getting a red screen after reset, however, my LHA loop stress test crashed with a "program failed..." error while running. Meh.

did you turn CC0 and cpu cashe off via the early startup menu....?

[amigasith]

Nope... Entering the early startup menu each time I reset the machine is something I would consider highly impractical. In addition, at least "C:CPU >NIL: NOCACHE" should do exactly the same as turning caches off in the early startup menu.

[Torti-the-Smurf]

I get a lot of Video Noise when i use the ShadowRom/Map Rom Function.
Tested with 2 Amiga600-Motherboards and 2 different power supplys.
So, no ShadowRom function for me because i like my video clean

[Mark sealey]

Originally Posted by amigasith

Nope... Entering the early startup menu each time I reset the machine is something I would consider highly impractical. In addition, at least "C:CPU >NIL: NOCACHE" should do exactly the same as turning caches off in the early startup menu.

Just tried this option in the startup-sequence and the Furia don't boot....I don't even think it gets to the startup-sequence before it locks up.

[amigasith]

Phew - that's tough I guess I really have to consider myself lucky as John pointed out. My Furia at least doesn't do things like this to me.

[majinga]

I had some problems to boot from a DOM with the KS 3.1
Then I switched to the KS 3.x and an IDE2SD adapter, and everything works fine now.

I have the white version of the furia, and I have no issue with the shadowrom with the KS 3.x

[amigasith]

Originally Posted by majinga

...and I have no issue with the shadowrom with the KS 3.x

Hi majinga,

thanks for sharing your experience! However, as I described above, I had no issues with plain SHADOWROM. The problems started when I used some modules on top of SHADOWROM by using LoadModule and then running an LHA loop for some time.

It would be interesting to see what results you get when you repeat my experiments

Cheers,
Marc