RadeonHD.chip (0): ERROR: PCI BIOS ROM signature not found RadeonHD.chip (0): ERROR: Could not obtain the AtomBIOS RadeonHD.chip (0): RadeonHD card closed
That's not a good sign. Does this card work when you plug it into a standard PC?
Quote:
PCI BIOS ROM signature... as in legacy parallel PCI?
No. Graphics cards have a ROM onboard. That error message says that the ROM signature doesn't match what it should be for a Radeon HD card.
Sorry for the lack of information regarding the atombios
The old PC I have for testing is unstable and it does not have any free space on the C drive for me to install the drivers, so video card dianogstic programs such as GPU-Z won't function.
I thought it would be good anyway to let you know while I'm looking for a 32 bit CMD program to scan the atombios. ________ *Edit* Removed preview images because they started to break this forum page
The old PC I have for testing is unstable and it does not have any free space on the C drive for me to install the drivers, so video card dianogstic programs such as GPU-Z won't function.
So you're using the VESA drivers. That means there must be something in the ROM, or the VESA drivers wouldn't work. It would still be useful to confirm that the ROM is a valid AtomBIOS one and isn't partially corrupt.
I tested my 7750 at a friend's place on his 64 bit windows 10 PC and my 7750 is working correctly from bios/boot to OS with drivers installed.
and we tested a couple hours of Tux kart. for science.
Next time I'm there I will dump the bios and compare it to the bios from TechPowerUp, that didn't spring to mind then but for now I think that this points toward that the bios is likely intact.
Okay, so the graphics card's ROM is fine. Something is going wrong with reading the ROM. Very weird, as that's been working reliably for years on all other platforms, and I started developing the driver on an A1-XE.
What version of AmigaOS 4.x are you using?
EDIT: Can you also check the state of the PCIe-to-PCI bridge in Ranger? EDIT2: If you do get the driver running, then you'll need to disable interrupts for the board. Ranger shows that the IRQ's line hasn't been set (it's 0xFF which is invalid).
Hans
Edited by Hans on 2018/5/12 8:40:37 Edited by Hans on 2018/5/12 8:42:07
I'm using the latest version with all Amiupdates from Hyperion and UBoot 1.1.1 from March 3 2005.
(Sysmon)
I got off guard by the RadeonHD chip being in there but to reaffirm the bios check failed and my 7750 doesn't show up as a graphics board.
Here is the PEX8111 in Ranger
Quote:
If you do get the driver running, then you'll need to disable interrupts for the board. Ranger shows that the IRQ's line hasn't been set (it's 0xFF which is invalid).
Hmm, would I have to do that via UBoot preferences?
So the PEX8111 used to work, maybe it is something specific with Southern Islands cards?
(On firefox that page displays the PEX8111 as a massive backdrop, but not on OWB)
Edited by Helloworld on 2018/5/12 22:02:54 Edited by Helloworld on 2018/5/12 22:41:51 Edited by Helloworld on 2018/5/12 22:47:13 Edited by Helloworld on 2018/5/12 23:00:29 Edited by Helloworld on 2018/5/12 23:19:08
I got off guard by the RadeonHD chip being in there but to reaffirm the bios check failed and my 7750 doesn't show up as a graphics board.
Here is the PEX8111 in Ranger ...
That screenshot shows what's wrong. the "Command" section should have "IO" and "Mem" as well. Without those, reads/writes to the graphics card's registers and VRAM/ROM never get through. I thought we'd fixed that already (PCIGraphics was supposed to handle that), but obviously not.
I'll see what I can do...
Quote:
Hmm, would I have to do that via UBoot preferences?
No. Interrupts are disabled via the monitor tooltypes in DEVS:Monitors/. You can also use the ScreenMode prefs' Monitor tab to edit those settings.
Quote:
If I boot without the startup-sequence via the early startup menu, the 7750 will gain two seemlingly invalid I/O flags (0xFFFFFFF8)
That's because the OS doesn't know about 64-bit PCI addressing. Resource ranges 1 and 3 are actually the upper 32-bits for 0 and 2.
That screenshot shows what's wrong. the "Command" section should have "IO" and "Mem" as well. Without those, reads/writes to the graphics card's registers and VRAM/ROM never get through. I thought we'd fixed that already (PCIGraphics was supposed to handle that), but obviously not.
I'll see what I can do...
I hope this isn't too difficult to fix. Being so close to getting the SI cards running on the XE is really exciting even though I sold mine on after getting the X1000.
Interesting, although somewhat expected. UBoot has to enable the bridge in order to initialize the graphics card. However, it then disables the graphics card's memory and IO access, and does the same to the bridge.
I haven't had a chance to look at this problem further. I'll try to get to that next week.
Unlikely. AFAIK, the problem of UBoot disabling the bridge is specific to the A1-XE (and possibly the uA1).
From memory, we have had RadeonHD cards work in Sam440's, just not Southern Islands cards. That's caused by a different problem (UBoot not initialising the card properly in a different way).
If you don't have a PEX adapter then I could donate you money to get yourself the same adapter.
Which I know does not warrent me to a working driver nor do I hold you accountable if it can't be fixed, I'm just trying to improve the odds of getting the 7750 to work, I can't wait for tabor anymore.
I'm rather overwhelmed with work right now (really need to get the current contract finished and shipped ASAP). Once that's over I'll be able to have a go at adding the workaround that's needed.