Sonnet 7200 PCI programming - Page 231

Apollon · 21 March 2021, 05:18

Quote:

Originally Posted by grelbfarlk

Yes always and still. For $40 you could buy a Killer NIC and replace the RAM with the correct type and like get 128MB. For $60 you could get an Apocalypse with 128MB or 256MB and upgrade it to 256MB or 512MB for another $20-$120. There are so many RAMs.

really? WOW who makes this and where is there info about it?

grelbfarlk · 21 March 2021, 05:36

Check the internet, most things you could want are on it! If you want RAM the internet is the best place to get it.

jbenam · 21 March 2021, 11:57

Quote:

Originally Posted by Hedeon

Another bug fix. Try with yes and no. I still recommend no as yes takes away precious video RAM.

Sadly still getting Error during MMU Setup of PPC all the time even with this new version.

Sometimes I also get a ramlib #8000000B, #80000004 or #80000003 crash after that message.

I've noticed you have put two envs in the archive, EnStackPatch and DisHunkPatch, both set to 1. I have tried mixing and matching all kinds of setups with them and it seems they make no difference.

When booting with no startup sequence and minimal config (as described by wrangler in the wiki) I never get any ramlib issues but I still get the Error during MMU Setup of PPC all the time.

I can feel we are getting closer to solving this one

Hedeon · 21 March 2021, 14:29

Quote:

Originally Posted by jbenam

Sadly still getting Error during MMU Setup of PPC all the time even with this new version.

Did you try with voodoomemos no already? And ramlib errors are mostly stack related. And I need the debug output.

jbenam · 21 March 2021, 15:16

Quote:

Originally Posted by Hedeon

Did you try with voodoomemos no already? And ramlib errors are mostly stack related. And I need the debug output.

Yep, nothing changes between the two modes.

Yeah the ramlib errors aren't really bothering me anymore since I'm not getting those in "minimal" mode. I am not getting them most of the time in AmigaOS either. Most of the times I just get the PPC MMU error and it quits cleanly.

I haven't attached a new debug output since it's mostly the same as the one I have posted yesterday (the CRT picture took with a phone).

Anything else I can try?

EDIT: Debug output, for sake of completeness.

Started Library Init routine
Version: $VER: powerpc.library 17.14 beta (21.03.21)

Detected supported Elbox Mediator PCI bridge
Opening of support libraries completed
3DFX card detected, Gfx address at 62000000, config address at 60000000
Detected Killer M1/K1 PPC PCI card
Setting up 64MB of PPC memory at 68000000
Setting up PPC card config base at 64100000
Setting up VGA PCI window at C0000000 located from 60000000 with size 800F8000
Setting up VGA config block for 3DFX Voodoo located at 60000000
Setting up extra BAT for extra PCI memory
Waiting on PPC card to respond

Anyway yes, it quits cleanly most of the time in AmigaOS as well now.

My config right now:

VoodooMem 13
VoodooMemOS No
DisHunkPatch 1
EnStackPatch 1

EDIT2: I do sometimes get a crapload of hits on MuForce when I get the Error during MMU Setup of PPC.

They're all "READ from 00000000 (INST)" on ramlib. I suppose these are just the stack issues getting trapped by MuForce?

Hedeon · 21 March 2021, 20:36

Another update. I cannot fix the ramlib errors, if the powerpc lib is to blame, with this kind of info. You really need to show the output. I think it can even show the instructions around the culprit. Or there is a support program that adds it to MuForce (MuGuardianAngel?)

If you want it fixed, you need to give more info.

jbenam · 21 March 2021, 21:14

Quote:

Originally Posted by Hedeon

Another update. I cannot fix the ramlib errors, if the powerpc lib is to blame, with this kind of info. You really need to show the output. I think it can even show the instructions around the culprit. Or there is a support program that adds it to MuForce (MuGuardianAngel?)

If you want it fixed, you need to give more info.

Please do not misunderstand me - if I am not attaching enough info it is not because of a lack of want to collaborate, it's just that debugging on the Amiga is kinda new to me and I am unsure on what to do next.

I am 100% willing to help you by doing as many tests as needed.

Tested with this new lib - I still get Error during MMU Setup of PPC.

I've got a debug log with MuForce + MuGuardianAngel. Sashimi Output attached.

MuForce + MuGuardianAngel

Code:

MuForce 46.4 by Thomas Richter, Copyright © 1998-2021,
based on the Enforcer by Michael Sinz, Copyright © 1992-1998
All Rights Reserved.

Contact: thorfdbg@alumni.tu-berlin.de           Enforcer@sinz.org

MuForce is installed.


Front Mung-wall at 07874dd8 is damaged.
PC  : 00fe6648  USP: 07874010 ( )( )  Name: ramlib
PCa : 00fe65ee  Name: ramlib
By  : AllocMem(00000960,00000000)
Data: 00000960 00000000 078757d4 00000000 07874c64 07874af4 07874bac 00000004
Addr: 07875728 07874dd8 075a55b0 00000000 07875728 070195c4 SysBase  --------
Stck: 070195c4 075a5786 07874108 00fce7cc 00fe4e8c 00000000 07874034 00fcea44
Stck: 00fdbb8e 8000006b 00031046 80000064 00000000 80000065 00000000 80000076
PC-8: 2f2051c8 fffc2060 4e902400 204ccb4e 4eaefd24 2254203c 00000960 4eaeff2e
PC *: cb4e60ac 2f0e2c6e 05644cef 03000008 4eaeff10 2c5f4e75 2f0e2c6e 05644cef
W:07874db8 ed<>07 87 4e dc 00 00 04 d0 00 00 df 84 07 5b c0 70 ..N..........[
.p
  07874dc8     00 00 df b0 07 5b c0 7a 00 00 df dc 07 5b c0 84 .....[.z.....[
..

Sashimi

Code:

Sashimi installed ([Ctrl]+C or "Break 3" to remove)
Started Library Init routine
Version: $VER: powerpc.library 17.14 beta (22.03.21)

Detected supported Elbox Mediator PCI bridge
Opening of support libraries completed
3DFX card detected, Gfx address at 64000000, config address at 62000000
Detected Killer M1/K1 PPC PCI card
Setting up 64MB of PPC memory at 68000000
Setting up PPC card config base at 60000000
Setting up VGA PCI window at C0000000 located from 64000000 with size 800F800
0
Setting up VGA config block for 3DFX Voodoo located at 62000000
Waiting on PPC card to respond
Started Library Init routine



Version: $VER: powerpc.library 17.14 beta (22.03.21)

Detected supported Elbox Mediator PCI bridge
Opening of support libraries completed
3DFX card detected, Gfx address at 64000000, config address at 62000000
Detected Killer M1/K1 PPC PCI card
Setting up 64MB of PPC memory at 68000000
Setting up PPC card config base at 60000000
Setting up VGA PCI window at C0000000 located from 64000000 with size 800F800
0
Setting up VGA config block for 3DFX Voodoo located at 62000000
Waiting on PPC card to respond

Please let me know if that's enough or if I need to enable some specific debugging mode. Thanks again for your time.

grelbfarlk · 22 March 2021, 01:29

Have you tried with the old version of the library?
Wondering if the K1/M1 is actually known to be working or not.

jbenam · 22 March 2021, 10:36

Quote:

Originally Posted by grelbfarlk

Have you tried with the old version of the library?
Wondering if the K1/M1 is actually known to be working or not.

With the old version M1's Memory gets added (I can see around 58MBs being added to FAST) then everything goes very wrong very fast with MuForce literally getting drowned in hits. I get the usual ramlib #80000004.

Since everything freezes as soon as I click anything, I can't save the console's full output. I have attached a picture of my screen.

That happens with stackattack2 set to Auto. If I set a massive stack (131072) it just valiantly freezes before even adding memory.

The M1 works on my Lab PC.

trixster · 22 March 2021, 12:23

I think worrying about stuff like stackattack and mmu-configurations is a red-herring at the moment. If it’s going to work it should just work straight away (and allow you to run basic WOS stuff). Only then do you have to start messing with stack and mmu to get stuff stable and working reliably with games.

Case in point - when I was testing Torsti’s image I could get wos and warp3d running from a zero startup-sequence and then minimum boot config - no mmu or stack trickery required.

Hedeon · 22 March 2021, 15:18

Quote:

Originally Posted by trixster

I think worrying about stuff like stackattack and mmu-configurations is a red-herring at the moment. If it’s going to work it should just work straight away (and allow you to run basic WOS stuff). Only then do you have to start messing with stack and mmu to get stuff stable and working reliably with games.

Case in point - when I was testing Torsti’s image I could get wos and warp3d running from a zero startup-sequence and then minimum boot config - no mmu or stack trickery required.

I tend to agree. There were some errors in the library regarding Mediator + Voodoo + VoodooMemOS = yes, but those are unrelated at least to the ramlib issues. Plus, as I understand it, also 17.13 was tried and that version did not have that error.

I would look more in a HW kind of direction (power, the mediator etc). I also run Enforcer type of programs and I don't see anything. It appears that the memory of the M1 gets corrupted? Maybe a power issue?

jbenam · 22 March 2021, 17:28

Quote:

Originally Posted by Hedeon

I tend to agree. There were some errors in the library regarding Mediator + Voodoo + VoodooMemOS = yes, but those are unrelated at least to the ramlib issues. Plus, as I understand it, also 17.13 was tried and that version did not have that error.

I would look more in a HW kind of direction (power, the mediator etc). I also run Enforcer type of programs and I don't see anything. It appears that the memory of the M1 gets corrupted? Maybe a power issue?

17.13 crashes straight with a ramlib error. 17.14 beta just hangs at the MMU on PPC side error.

Power should be okay - I have recapped the entire PSU a year or so ago (hoping to solve these issues) and I have also got a self-built ATX adapter which sadly didn't improve the situation. I guess I will give it another try since I have got it laying on the shelf next to the Amiga. I also did try injecting 3.3V directly into the mediator from the ATX power supply with no apparent improvements.

I can't at this point exclude the possibility that the issue lies in the Mediator itself. Everything else works well (RTG, Warp3D, Audio, Network) so it might be some kind of incompatibility with the code on the CPLDs. I do know that there was some kind of rework Elbox was doing on first generation Mediator board - not sure what kind of issues it fixed and if I do already have it applied to mine.

I guess the only sure-fire way of excluding an hardware issue would be to send everything to a trustworthy member of EAB with a known working config who can then proceed on systematically excluding a fault with either the Mediator or the M1.

grelbfarlk · 22 March 2021, 18:39

Just to sound like a broken record, in my experience the KillerNIC K1 didn't like the exact same slot order as some of the other PPC boards. In an A4000d I was testing it with it ended up working when it was in the last or top slot. Whereas usually I would be putting the prpmc card in the third or next to top slot.

Hedeon · 22 March 2021, 19:13

Quote:

Originally Posted by grelbfarlk

Just to sound like a broken record, in my experience the KillerNIC K1 didn't like the exact same slot order as some of the other PPC boards. In an A4000d I was testing it with it ended up working when it was in the last or top slot. Whereas usually I would be putting the prpmc card in the third or next to top slot.

That the M1/K1 card is at the start of PCI space could be it. I think for the Prometheus, it depends on the VGA card being first in PCI space (due to 24 bit DMA stuff). For mediator I think it should not matter, albeit that a lot of ppl complained about getting no sound from SB cards when placed in a certain order.

I am not fond of fiddling with the mediator code atm as I don't have a machine to test. Could you put the cards in such an order that the vodoo comes first in PCI space (in your case at $60000000/$62000000)

trixster · 22 March 2021, 21:17

Your 3000 is out of commission?

Hedeon · 22 March 2021, 21:35

Quote:

Originally Posted by trixster

Your 3000 is out of commission?

It doesn't have a Mediator.

jbenam · 23 March 2021, 00:01

Quote:

Originally Posted by grelbfarlk

Just to sound like a broken record, in my experience the KillerNIC K1 didn't like the exact same slot order as some of the other PPC boards. In an A4000d I was testing it with it ended up working when it was in the last or top slot. Whereas usually I would be putting the prpmc card in the third or next to top slot.

Yeah, I have fiddled with slots a lot and nothing much changed. Sometimes it isn't detected at all or I get a "card crashed during setup". I think I usually put in the top slot as well.

Quote:

Originally Posted by Hedeon

I am not fond of fiddling with the mediator code atm as I don't have a machine to test. Could you put the cards in such an order that the vodoo comes first in PCI space (in your case at $60000000/$62000000)

I have already tested that as I thought of same thing, sadly it seems it didn't make a difference.

I will try to get you some logs tomorrow as right now the A4000 is strewn on the bench while I am testing with an ATX power supply as I want to make sure again that it's not a power issue (spoiler: from the bit of testing I have done tonight it seems it didn't improve the situation).

grelbfarlk · 23 March 2021, 02:00

Quote:

Originally Posted by Hedeon

That the M1/K1 card is at the start of PCI space could be it. I think for the Prometheus, it depends on the VGA card being first in PCI space (due to 24 bit DMA stuff). For mediator I think it should not matter, albeit that a lot of ppl complained about getting no sound from SB cards when placed in a certain order.

I am not fond of fiddling with the mediator code atm as I don't have a machine to test. Could you put the cards in such an order that the vodoo comes first in PCI space (in your case at $60000000/$62000000)

I had generally associated the behavior with power slot order as far as certain cards seem to only work on certain slots or maybe that's a manifestation of the mythical termination issue.

As far as power issues go, it sounds like jebdebman is using a 5V only Mediator and trying to pull 3.3V off of the power supply for the Killer NIC. This gets pretty sensitive about power supply choice.
Also using an ATX power supply which is also suspect, is he measuring something close to 5V on the PCI slots? What's the 3.3V reading?
I had problems with the K1 and a Radeon on the same Black Mediator, switching either the K1 or the Radeon to 5V tolerant boards seemed to be the difference to getting it stable.

grelbfarlk · 23 March 2021, 02:09

I just swapped out the Ragnarok and installed the K1. With the old asm version of the library it started right up and ran GLQuake, 40.5 FPS @ 640x480. Swapped in clib version and rebooted and got "Error during MMU setup of PPC", using the last posted version.
K1 is modified with a VRM to work in 5V only mediator.

Code:

Sashimi installed ([Ctrl]+C or "Break 8" to remove)
Started Library Init routine
Version: $VER: powerpc.library 17.14 beta (21.03.21)

Detected supported Elbox Mediator PCI bridge
Opening of support libraries completed
3DFX card detected, Gfx address at 64000000, config address at 62000000
Detected Killer M1/K1 PPC PCI card
Setting up 64MB of PPC memory at 68000000
Setting up PPC card config base at 60400000
Setting up VGA PCI window at C0000000 located from 64000000 with size 800F8000
Setting up VGA config block for 3DFX Voodoo located at 62000000
Waiting on PPC card to respond
Sashimi removed.
8.Sark:sonnet501/Libs/powerpc> pciinfo

Pci Device 1:
   VendorID: $1033
   VendorName: NEC Electronics
   DeviceID: $35
   DeviceName: udp9210 Dual OHCI controllers plus Single EHCI controller
   ClassCode: $c0310
   RevisionID: $41
   InterruptPin: 1
   SubsystemVendorID: $1033
   SubsystemID: $35
   MemSpace0: $60000000 - $60000fff

Pci Device 2:
   VendorID: $1033
   VendorName: NEC Electronics
   DeviceID: $35
   DeviceName: udp9210 Dual OHCI controllers plus Single EHCI controller
   ClassCode: $c0310
   RevisionID: $41
   InterruptPin: 2
   SubsystemVendorID: $1033
   SubsystemID: $35
   MemSpace0: $60100000 - $60100fff

Pci Device 3:
   VendorID: $1033
   VendorName: NEC Electronics
   DeviceID: $e0
   DeviceName: uPD720100A USB 2.0 Host Controller
   ClassCode: $c0320
   RevisionID: $2
   InterruptPin: 3
   SubsystemVendorID: $ee4
   SubsystemID: $89e
   MemSpace0: $60200000 - $602000ff

Pci Device 4:
   VendorID: $1319
   VendorName: Forte Media
   DeviceID: $801
   DeviceName: FM0801 PCI Card MediaForte made in singapore
   ClassCode: $40100
   RevisionID: $b2
   InterruptPin: 2
   SubsystemVendorID: $1319
   SubsystemID: $1319
   IOSpace0: $40c06000 - $40c0607f

Pci Device 5:
   VendorID: $1319
   VendorName: Forte Media
   DeviceID: $802
   DeviceName: FM801-as Xwave PCI Joystick
   ClassCode: $98000
   RevisionID: $b2
   InterruptPin: 0
   SubsystemVendorID: $1319
   SubsystemID: $1319
   IOSpace0: $40c06100 - $40c0610f

Pci Device 6:
   VendorID: $121a
   VendorName: 3dfx Interactive Inc
   DeviceID: $5
   DeviceName: Voodoo3 All Voodoo3 chips, 3000
   ClassCode: $30000
   RevisionID: $1
   InterruptPin: 3
   SubsystemVendorID: $121a
   SubsystemID: $57
   MemSpace0: $62000000 - $63ffffff
   MemSpace1: $64000000 - $65ffffff
   IOSpace2: $40c06200 - $40c062ff
   RomSpace: $60300000 - $6030ffff

Pci Device 7:
   VendorID: $1957
   VendorName: Bigfeet
   DeviceID: $86
   DeviceName: Lilweenie
   ClassCode: $b2000
   RevisionID: $30
   InterruptPin: 4
   SubsystemVendorID: $1a56
   SubsystemID: $201
   MemSpace0: $60400000 - $604fffff
   MemSpace1: $68000000 - $6bffffff

Pci Device 8:
   VendorID: $10ec
   VendorName: Realtek Semiconductor
   DeviceID: $8029
   DeviceName: RTL8029(as)pci ethernet nic windot XPse
   ClassCode: $20000
   RevisionID: $0
   InterruptPin: 1
   SubsystemVendorID: $10ec
   SubsystemID: $8029
   IOSpace0: $40c06300 - $40c0631f

jbenam · 23 March 2021, 08:59

Quote:

Originally Posted by grelbfarlk

As far as power issues go, it sounds like jebdebman is using a 5V only Mediator and trying to pull 3.3V off of the power supply for the Killer NIC. This gets pretty sensitive about power supply choice.
Also using an ATX power supply which is also suspect, is he measuring something close to 5V on the PCI slots? What's the 3.3V reading?
I had problems with the K1 and a Radeon on the same Black Mediator, switching either the K1 or the Radeon to 5V tolerant boards seemed to be the difference to getting it stable.

Yes, I am using a 5V-only mediator. It has been modded with a VRM on card just like yours. I don't use an ATX power supply, I have an original Lite-on PSU which I have recapped around one year ago. I will get more readings soon, but AFAIR, 3.3v was pretty good.

I have put in an ATX PSU yesterday because a few people suggested it might've been a power issue.

Quote:

Originally Posted by grelbfarlk

I just swapped out the Ragnarok and installed the K1. With the old asm version of the library it started right up and ran GLQuake, 40.5 FPS @ 640x480. Swapped in clib version and rebooted and got "Error during MMU setup of PPC", using the last posted version.
K1 is modified with a VRM to work in 5V only mediator

Okay, so K1 support is actually broken in the beta and it's just not my setup

22 March 2021, 01:29	#4608
grelbfarlk Registered User Join Date: Dec 2015 Location: USA Posts: 2,902	Have you tried with the old version of the library? Wondering if the K1/M1 is actually known to be working or not. Last edited by grelbfarlk; 22 March 2021 at 02:03.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
WANTED: PCI Sound Card VORTEX2 SuperQuard / SQ2500 or Turtlebeach SantaCruz pci card	keropi	MarketPlace	12	15 March 2017 02:48
Wanted:Sonnet Technologies QuadDoubler 50 (Doubler 4000)	kronuz	MarketPlace	0	16 June 2013 00:29
eBay : Sonnet Quaddoubler	alexh	MarketPlace	3	07 May 2013 00:50
Overlay problem on GeForce 7200	Leandro Jardim	support.WinUAE	1	10 August 2010 15:50
Prometheus PCI & Voodoo 3 PCI GFX Card	Slayer	support.Hardware	21	05 September 2006 10:57

21 March 2021, 05:36	#4602
grelbfarlk Registered User Join Date: Dec 2015 Location: USA Posts: 2,902	Check the internet, most things you could want are on it! If you want RAM the internet is the best place to get it.

22 March 2021, 12:23	#4610
trixster Guru Meditating Join Date: Jun 2014 Location: England Posts: 2,337	I think worrying about stuff like stackattack and mmu-configurations is a red-herring at the moment. If it’s going to work it should just work straight away (and allow you to run basic WOS stuff). Only then do you have to start messing with stack and mmu to get stuff stable and working reliably with games. Case in point - when I was testing Torsti’s image I could get wos and warp3d running from a zero startup-sequence and then minimum boot config - no mmu or stack trickery required.

22 March 2021, 18:39	#4613
grelbfarlk Registered User Join Date: Dec 2015 Location: USA Posts: 2,902	Just to sound like a broken record, in my experience the KillerNIC K1 didn't like the exact same slot order as some of the other PPC boards. In an A4000d I was testing it with it ended up working when it was in the last or top slot. Whereas usually I would be putting the prpmc card in the third or next to top slot.

22 March 2021, 21:17	#4615
trixster Guru Meditating Join Date: Jun 2014 Location: England Posts: 2,337	Your 3000 is out of commission?

Currently Active Users Viewing This Thread: 3 (0 members and 3 guests)