Sonnet 7200 PCI programming

[johnnybarker]

Unfortunately having issues with my a3000 overheating a little so it is currently in pieces again. It works fine if the cover is off and there is an extra fan on the PSU. I think the heat is causing voltage instability so I am reworking the PSU but just waiting on some bit to arrive.

On the plus side I put the A4000 back together and managed to run a DVD on the Apocalypse straight from disc using AMP2. Not 100% smooth but surprisingly good (thanks Buli for the idea). The DVD drive I used was a pioneer SCSI drive through the Cyberstorm MkII.

[BULI]

Originally Posted by johnnybarker

Unfortunately having issues with my a3000 overheating a little so it is currently in pieces again. It works fine if the cover is off and there is an extra fan on the PSU. I think the heat is causing voltage instability so I am reworking the PSU but just waiting on some bit to arrive.

On the plus side I put the A4000 back together and managed to run a DVD on the Apocalypse straight from disc using AMP2. Not 100% smooth but surprisingly good (thanks Buli for the idea). The DVD drive I used was a pioneer SCSI drive through the Cyberstorm MkII.

[Hedeon]

I think I blew up my AA3000+ within 15 minutes of powering it on. So I guess it will take a bit of more time for Prometheus support.

[fitzsteve]

Originally Posted by Hedeon

I think I blew up my AA3000+ within 15 minutes of powering it on. So I guess it will take a bit of more time for Prometheus support.

Sorry to hear that (The damage not the delay!)

I hope it's not a big problem

[starship]

Originally Posted by Hedeon

I think I blew up my AA3000+ within 15 minutes of powering it on. So I guess it will take a bit of more time for Prometheus support.

Sorry to hear that. I hope you fix it quickly.

[DarrenHD]

Originally Posted by Hedeon

I think I blew up my AA3000+ within 15 minutes of powering it on. So I guess it will take a bit of more time for Prometheus support.

Oh dear...what happened? I've been very careful with my AA3000+...I've blown up other things in the past myself....I feel your pain.

I had a Super DMAC chip just stop working recently in one of my original A3000 motherboards...

Darren

[BULI]

Originally Posted by Hedeon

I think I blew up my AA3000+ within 15 minutes of powering it on. So I guess it will take a bit of more time for Prometheus support.

Uuuu- it must hurt
That why I don't change anything in my A4000D with Mediator and Ragnarok

[johnnybarker]

Oh, no that is a shame about your AA3000+. Hope it isn't too damaged.

@DarrenHD, I also had a DMAC die on my a3000 when using an a3660. It ran much hotter after fitting the CPU card and I think it must have overheated. I put a heat sink on the replacement one and thankfully not had that particular problem again.

[spudje]

I know, that's too bad to hear Hope it can be easily fixed.

Also curious what happened.

[DarrenHD]

Originally Posted by johnnybarker

Oh, no that is a shame about your AA3000+. Hope it isn't too damaged.

@DarrenHD, I also had a DMAC die on my a3000 when using an a3660. It ran much hotter after fitting the CPU card and I think it must have overheated. I put a heat sink on the replacement one and thankfully not had that particular problem again.

Hmm maybe I should put a heatsink on my remaining (alive) DMACs....it's not like you can buy them easily anymore In fact if you know of anyone selling a Super DMAC I'd be interested so I can replace my dead one.

[Hedeon]

1) I did not RTFM and firstly placed my 3.1.4 ROMs wrongly in the sockets (the sockets have 1 row more than the ROMs itself). During removing of the ROMs I saw 1 leg being very loose already. I placed the ROMs correctly but then after a few boots and fiddling with the floppy drive the AA3000+ refused to boot (black screen, but light went from dim to bright).

2) After tanking some confidence I removed the ROMs again and then saw that one pin was broken. I could not find the pin. I removed my custom 3.9 ROMs from my A3000D board and placed them in my AA3000+ board. This resulted in a black screen again. I somehow mixed LO and HI while I checked it twice.....

3) Removed and placed the ROMs correctly. All seemed fine.

4) Added CSMK2 -> Yellow(? not sure, my monitor does not really like the native output) screen and reboot, over and over). If the Amiga booted, it runs stable for hours until reboot, then same issues. The debugger which comes up the first cycle gives a bus error on the first address of FAST RAM (so either 0x700000 when the SIMMs are installed or 0x800000 when the CSMK2 was installed with RAM).

5) Lots of fiddling later it even now cycles with no CSMK2 installed. With no CSMK2 it is yellow screen, reboot then black screen (no cycling).

6) Lots of shaking of the MB and now it runs stable for a day. Not sure if something is shorting something (the pin I could not find again? I saw nothing in the socket and the 3.9 ROM went in without problems). With no root-cause it can happen again.

Current casualties:

1) 3.1.4 ROMs
2) SCSI cable connector came off during pulling it out of the connector on the MB
3) SCSI connector on the SCSI2HD broke (at one of the sides) while removing the SCSI cable.

The 3.1.4 floppy also makes a noise as if corrupt (stepping of the heads). WB floppy booted just fine.
The CSMK3 I removed from another Amiga first went from dimmed to bright LED but during latest testing stayed dimmed. Not sure if I broke that one too, but haven't tested it since the Amiga became stable with the CSMK2. (And yes, I set QUADCLK and CPUCLK to EXT and BOARDCLK to INT).

[Crumb]

@Hedeon

Scary stuff, I hope your CSMK3 is fine and you fix the other issues.

[DarrenHD]

Originally Posted by Hedeon

1) I did not RTFM and firstly placed my 3.1.4 ROMs wrongly in the sockets (the sockets have 1 row more than the ROMs itself). During removing of the ROMs I saw 1 leg being very loose already. I placed the ROMs correctly but then after a few boots and fiddling with the floppy drive the AA3000+ refused to boot (black screen, but light went from dim to bright).

2) After tanking some confidence I removed the ROMs again and then saw that one pin was broken. I could not find the pin. I removed my custom 3.9 ROMs from my A3000D board and placed them in my AA3000+ board. This resulted in a black screen again. I somehow mixed LO and HI while I checked it twice.....

3) Removed and placed the ROMs correctly. All seemed fine.

4) Added CSMK2 -> Yellow(? not sure, my monitor does not really like the native output) screen and reboot, over and over). If the Amiga booted, it runs stable for hours until reboot, then same issues. The debugger which comes up the first cycle gives a bus error on the first address of FAST RAM (so either 0x700000 when the SIMMs are installed or 0x800000 when the CSMK2 was installed with RAM).

5) Lots of fiddling later it even now cycles with no CSMK2 installed. With no CSMK2 it is yellow screen, reboot then black screen (no cycling).

6) Lots of shaking of the MB and now it runs stable for a day. Not sure if something is shorting something (the pin I could not find again? I saw nothing in the socket and the 3.9 ROM went in without problems). With no root-cause it can happen again.

Current casualties:

1) 3.1.4 ROMs
2) SCSI cable connector came off during pulling it out of the connector on the MB
3) SCSI connector on the SCSI2HD broke (at one of the sides) while removing the SCSI cable.

The 3.1.4 floppy also makes a noise as if corrupt (stepping of the heads). WB floppy booted just fine.
The CSMK3 I removed from another Amiga first went from dimmed to bright LED but during latest testing stayed dimmed. Not sure if I broke that one too, but haven't tested it since the Amiga became stable with the CSMK2. (And yes, I set QUADCLK and CPUCLK to EXT and BOARDCLK to INT).

I have a MK3 working in my AA3000+ so I can tell you that they work.

Good luck I hope you get it fully repaired. I'm sure hese would fix it for you if you need him to.

Darren

[Hedeon]

The AA3000+ is up and running sort of. It went into cycling again when I unplugged and replugged the VGA connector from the motherboard to the Voodoo card. It's working again.

Some observations:

1) Looking at RTG screen refresh I think the mediator was faster, but I am not sure. It is especially evident in testing/editing screens in P96Mode.

2) Regarding PCI PPC cards: the upper memory of the 512MB space is used for I/O and config space. The mediator has a separate 64MB ZorroIII space for this. It means (maybe) that MPC107 can only do 128MB and the other cards a max of 256MB. I have to see if PPC memory can overlap I/O and config space and then just mark it allocated as from the Amiga side it cannot be addressed anyway as accesses to these addresses go to I/O and config space.

3) Regarding the DSP. I have only read the most basic docs on the AA3000+ and they state there are 2 registers controlling the DSP. A read register (ULONG) at $dd005c and a write register (UBYTE) at $dd0080. I can make the Amiga hang with certain combinations writing to the write register (Hard hang, no guru nothing). Or make it throw a bus error by reading from it. However, nothing I do reflects in the read register. Also according to the docs, some bits are low and some are high after reset. I just see $FF. I do have a DSP32 assembler, but if the DSP doesn't react correctly or at all, it stops there.

[DarrenHD]

Originally Posted by Hedeon

The AA3000+ is up and running sort of. It went into cycling again when I unplugged and replugged the VGA connector from the motherboard to the Voodoo card. It's working again.

Some observations:

1) Looking at RTG screen refresh I think the mediator was faster, but I am not sure. It is especially evident in testing/editing screens in P96Mode.

2) Regarding PCI PPC cards: the upper memory of the 512MB space is used for I/O and config space. The mediator has a separate 64MB ZorroIII space for this. It means (maybe) that MPC107 can only do 128MB and the other cards a max of 256MB. I have to see if PPC memory can overlap I/O and config space and then just mark it allocated as from the Amiga side it cannot be addressed anyway as accesses to these addresses go to I/O and config space.

3) Regarding the DSP. I have only read the most basic docs on the AA3000+ and they state there are 2 registers controlling the DSP. A read register (ULONG) at $dd005c and a write register (UBYTE) at $dd0080. I can make the Amiga hang with certain combinations writing to the write register (Hard hang, no guru nothing). Or make it throw a bus error by reading from it. However, nothing I do reflects in the read register. Also according to the docs, some bits are low and some are high after reset. I just see $FF. I do have a DSP32 assembler, but if the DSP doesn't react correctly or at all, it stops there.

Great you have it working again!! In my AA3000+ I haven't really noticed a significant difference in speed between the Mediator in my A4000 system and the Prometheus in my AA3000+. I'm sure they aren't exactly the same speed, but I think they are close enough for the average user not to notice. I did not really take much notice when I was testing/editing in P96Mode, but I will take a look. My other A4000 system with a CSPPC and GREX 4000 is noticably faster than both.

That is too bad about the DSP - it would be really cool to get this working. Thank you for trying Maybe eventually you or someone else will get it working. I did ask Randall Jesup and Dave Haynie on Facebook about it, to see if they could contribute any original code or insight, but I got no response.

Darren

[BULI]

Originally Posted by Hedeon

Some observations:

1) Looking at RTG screen refresh I think the mediator was faster, but I am not sure. It is especially evident in testing/editing screens in P96Mode.

2) Regarding PCI PPC cards: the upper memory of the 512MB space is used for I/O and config space. The mediator has a separate 64MB ZorroIII space for this. It means (maybe) that MPC107 can only do 128MB and the other cards a max of 256MB. I have to see if PPC memory can overlap I/O and config space and then just mark it allocated as from the Amiga side it cannot be addressed anyway as accesses to these addresses go to I/O and config space.

Very interesting

[matt3k]

One observation I had with my AA3000+ is that it is very picky how I mount the accelerator card. I needed to push down the exact amount of pressure and I could NOT use the screws to mount it.

Looking forward to the progress with the PCI board on the AA3000+ and maybe the onboard DSP, thanks Hedeon!

[Hedeon]

More sad news maybe.

I cannot seem to start up the G4 card with the Prometheus/Firebird. The Harrier refuses to shadow the configuration register block to PCI memory as I can with the Mediator.

[johnnybarker]

Originally Posted by DarrenHD

Hmm maybe I should put a heatsink on my remaining (alive) DMACs....it's not like you can buy them easily anymore In fact if you know of anyone selling a Super DMAC I'd be interested so I can replace my dead one.

Oh, sorry meant to reply last week but forgot. I bought mine from tbtorro on amibay- not sure if he has any left though. The ones I bought from him (also DMAC-2s not 4s) were as an experiment to see if they sorted the SCSI errors with the a3660, which they did. They also seem to work fine with Ramsey 7 (at least on my current configuration with a cyberstormPPC). I have the original DMAC-2 I removed but I am saving this for an AA3000+ build. Just waiting on Mouser to deliver to components.

I also bought 2 DMAC from vesalia but the chips never arrived and they have not replied to at least 10 email I sent to all their different email addresses. Not sure if this is normal behaviour from them but I find it hard to recommend their services in future.

[DarrenHD]

Originally Posted by johnnybarker

Oh, sorry meant to reply last week but forgot. I bought mine from tbtorro on amibay- not sure if he has any left though. The ones I bought from him (also DMAC-2s not 4s) were as an experiment to see if they sorted the SCSI errors with the a3660, which they did. They also seem to work fine with Ramsey 7 (at least on my current configuration with a cyberstormPPC). I have the original DMAC-2 I removed but I am saving this for an AA3000+ build. Just waiting on Mouser to deliver to components.

I also bought 2 DMAC from vesalia but the chips never arrived and they have not replied to at least 10 email I sent to all their different email addresses. Not sure if this is normal behaviour from them but I find it hard to recommend their services in future.

Thanks. Yeah I asked tbtorro and he has DMACS but is only selling them in complete chipset kits. He's not selling individually. I haven't tried Vesalia, but I guess with your experience better I don't

Haven't seen any pop up on eBay....so my A3000 will continue to be without a DMAC and thus broken At least it's my spare/old board and I have a couple others that work.