Sonnet 7200 PCI programming

[drluking]

I use initppc via script after workbench is up

[drluking]

I have time to test more yesterday and today:

1) gear full screen demo 68k with Warp3D: 8 hour working perfectly
2) gear full screen demo PPC with Warp3D: 30 minutes then crash and reboot
3) gear full screen demo 68k with Warp3D: other 8 hour working perfectly
4) gear full screen demo PPC with Warp3D: 39 minutes then crash and reboot
5) gear full screen demo 68k with Warp3D: other 8 hour working perfectly
6) gear full screen demo PPC with Warp3D: 25 minutes then crash and reboot

so there is something only with PPC

where Can I find BlitzquakeWOS 6.0 version ?

[grelbfarlk]

I can't find it at the moment, I'll dig it up.
But I mean how are you "hiding" the powerpc.library from the system during bootup? Is it in a non path directory and you are doing an assign with ADD?

[drluking]

Originally Posted by grelbfarlk

I can't find it at the moment, I'll dig it up.

ok thanks

But I mean how are you "hiding" the powerpc.library from the system during bootup? Is it in a non path directory and you are doing an assign with ADD?

yes, I have powerpc.library in

libs:powerpc

directory, ad I launch it with this script file:

assign LIBS: libs:powerpc ADD
InitPPC

[drluking]

#double post

[drluking]

Originally Posted by drluking

I have time to test more yesterday and today:

1) gear full screen demo 68k with Warp3D: 8 hour working perfectly
2) gear full screen demo PPC with Warp3D: 30 minutes then crash and reboot
3) gear full screen demo 68k with Warp3D: other 8 hour working perfectly
4) gear full screen demo PPC with Warp3D: 39 minutes then crash and reboot
5) gear full screen demo 68k with Warp3D: other 8 hour working perfectly
6) gear full screen demo PPC with Warp3D: 25 minutes then crash and reboot

so there is something only with PPC

If I try a way to don't crash the gear full screen demo PPC with Warp3D, sure I have solved the problem !


my ideas:
- problem with PSU (now I have +4.93V on the 5V pin of the Core Logic osscillator); I try to iniect an other +5V in the PCI slot ?
- or problem with position of PCI cards in PCI slots (card PCI that works correctly only in a particular PCI slots ..etc..)
- or a particular terminator of PCI bus needed (no idea, if not put eventually an usb card pci ?)
- or a software problem of powerpc.library (not yet ready enought, not stable enought)


It's not a Voodoo3 card problem (no problem in 68k) and same proprblems in PPC mode with an other Voodoo3
It's not a PPC card problem I have same proprblems with K1 and M1 cards
It's not a mainboard problem (same problems with an other a4000 mainboard)
It's not a CPU card 68k 040 (same problems with an other 060 card)
It's not a IDE unit/controller (same problem with SCSI unit/controller)
It's not a Mediator problem (same problem with an other Mediator)
It's not a problem of temerature (fully ventilated and all components are not hot (many fans make the temperature perfect at 20° C)

[spudje]

This actually looks a lot like my behaviour from a year ago when I struggled with a Sonnet, gave up and sold the card

[Hedeon]

Add the possibility of incompatible patches / libraries / devices to the list.

[esel]

Originally Posted by spudje

This actually looks a lot like my behaviour from a year ago when I struggled with a Sonnet, gave up and sold the card

Same here with the K1/M1 never got it 100% stable

[trixster]

It's a shame to hear that some people have had issues getting K1/M1 boards working stably. Getting all the elements of a complicated system working in unison to achieve stability is pretty challenging. The powerpc.library is proven to be stable though, over months of testing on many different pci and pmc boards by many different people, so i suspect frustrating instability is down to a combination of different hardware, power supplies, temperature and the patchwork nature of most peoples' workbench setups.

[drluking]

I use basic OS 3.9 boingbag 2 (without boingbag 2 update ROM) and no other (no patchs, nothing only OS3.9 basic reinstalled)

Today I have seen that if I try to use only Vodoo3 and K1 (or M1) the system i completly unstable !!! (guru, freeze ...etc... a nightmare !!!), but if I add again all other three cards (SB128, ethernet and USB) The system retrurn to more stable ...

So probably there is something with the position, or numbers of cards connected, or terminator of PCI bus ... I don’t know

[drluking]

is there a software to debug ? something that I can run in background to register in the hd or via serial nullmodem, when powepc crash ?
Many times the amiga don’t reboot, but freeze (and I need manually to reboot, or It guru in the workbench itself...)

I'm using this powerpc.library #426 with K1/M1 https://jenkins.sakura-it.pl/job/sonnetamiga/426/ , is this the best choice or should I use this aminet version http://aminet.net/package/driver/other/SonnetLibrary ?
or ... ?

I really don't know what to invent to be able to make these K1 / M1 cards work, I've been going crazy for months ... without getting any decent results

[grelbfarlk]

With just the Voodoo and K1, are you installing them in Slot 3 Voodoo and Slot 4 K1?

Something to consider is that the Elbox quality control is not very good lately. The last Mediator I received had a capacitor on the bridge that was cocked at a 45° and the pin was slightly ripped out of the capacitor, it was also crushed against the socket. Also one row of the mach chips was a few degrees off of flush, but it seems to work otherwise. I did replace the capacitor.

You did plug in the P9 connector from your ATX adapter?

Also what Hedeon was saying about incompatible modules, my story for that recently goes, occasionally I was getting PFS read errors on my SSD. Some random applications (some not finished/in development and final releases) would cause random read errors. By chance on one of these test programs, I found that it didn't cause the read errors if I had an IDE drive plugged in and mounted. Turns out there is some issue on my system relating either directly to or touching scsi.device 44.20. Unfortunately that is burned on my kickstart- but downgrading it or disabling scsi.device completely did not resolve the issue. So I for some reason have to mount a Fat95 volume or insert an Amiga volume on the IDE bus.
As far as using BB2 without the ROMUpdate, that's -to me- unfamiliar territory. Most of the people who have working systems are using some flavor of BB2 Kickstarts or later.

[Hedeon]

Regarding debug I recommend http://aminet.net/dev/debug/COP.lha

Advanced users only.

Just ran QuakeGL for an hour. Followed by some Encore demos and then gearsppc. for a total of 2 hours.

Try running just QuakeWOS (not the Warp3D version)

[grelbfarlk]

One other thing to add that at least one other has noticed for hard to diagnose instability. Some 5V fans are causing instability issues, for me it has been 12V fans which I thought 'too loud and still does the job at 5v.'

[drluking]

Originally Posted by Hedeon

Regarding debug I recommend http://aminet.net/dev/debug/COP.lha

Advanced users only.

thanks, if not solved, I'll try this debug tool

Just ran QuakeGL for an hour. Followed by some Encore demos and then gearsppc. for a total of 2 hours.

Try running just QuakeWOS (not the Warp3D version)

problem is It guru/reboot/freeze before an hour, usually some minutes, or 10 minutes ... other time more

[drluking]

Originally Posted by grelbfarlk

With just the Voodoo and K1, are you installing them in Slot 3 Voodoo and Slot 4 K1?

at the moment the Voodoo3 is in slot PCI fourth, and K1 in slot PCI fifth ; always problems

Something to consider is that the Elbox quality control is not very good lately. The last Mediator I received had a capacitor on the bridge that was cocked at a 45° and the pin was slightly ripped out of the capacitor, it was also crushed against the socket. Also one row of the mach chips was a few degrees off of flush, but it seems to work otherwise. I did replace the capacitor.

mine is ok; and 68k software works correctly (2D and 3D)

You did plug in the P9 connector from your ATX adapter?

sure !

Also what Hedeon was saying about incompatible modules, my story for that recently goes, occasionally I was getting PFS read errors on my SSD. Some random applications (some not finished/in development and final releases) would cause random read errors. By chance on one of these test programs, I found that it didn't cause the read errors if I had an IDE drive plugged in and mounted. Turns out there is some issue on my system relating either directly to or touching scsi.device 44.20. Unfortunately that is burned on my kickstart- but downgrading it or disabling scsi.device completely did not resolve the issue. So I for some reason have to mount a Fat95 volume or insert an Amiga volume on the IDE bus.
As far as using BB2 without the ROMUpdate, that's -to me- unfamiliar territory. Most of the people who have working systems are using some flavor of BB2 Kickstarts or later.

the problems are with and without the "AmigaOS ROM update BB2" of OS3.9 BB2 ; I'm using "AmigaOS ROM update BB1"

Originally Posted by grelbfarlk

One other thing to add that at least one other has noticed for hard to diagnose instability. Some 5V fans are causing instability issues, for me it has been 12V fans which I thought 'too loud and still does the job at 5v.'

mine are all +12V and I use an other PSU only for them

Today I have tried to inject +5V also in first PCI slot (and kept also the inject +5V in Zorro slot) ; in this way I have about +4,94V / 4,95V on 5V pin of the Core Logic mediator osscillator, and about +4,96V in P9 AT connector in mediator board ; but even so it has not changed ...

I could try with an AT 200W PSU (I'm using a quality 550W old model with -5V) that time ago It gave me about a plus 0.2 / 0.3 V or about on +5V VS the 550W one

Are you absolutely certain that the powerpc.library is perfectly stable with the K1 and M1 boards ? Because I've been going crazy with these cards for months ... I wouldn't want to spend months trying to solve the problem when maybe It's because these cards are not yet fully supported

[trixster]

Well, there are plenty of people using k1/m1 successfully. I would be happy to test your k1 and m1 boards if you would like an independent check that they are working ok?

[deling]

Originally Posted by trixster

Well, there are plenty of people using k1/m1 successfully. I would be happy to test your k1 and m1 boards if you would like an independent check that they are working ok?

where are they ?

[amigoun]

@drluking: have you changed the voltage regulator on Mediator?