GOLD2.10 core released for Vampire V2 - Page 2

kolla · 25 July 2018, 14:46

Quote:

Originally Posted by flype

In your case it seems to hang after the cold reboot.

Right. As I said, I will debug this, since it is annoying, I was just looking for potential known culprits. One obvious thing I will try is to disable Poseidon and other rom modules that I load using LoadModule early in s-s.

TuKo · 25 July 2018, 16:54

Quote:

Originally Posted by xanderbeanz

Hi, as a casual observer of the team’s progress, does anyone know, extremely roughly, when the Gold 3 release is planned for? The one that will let you play AGA games on an A500/600?

Maybe 6monhs, a year?

Thanks.

Gunnar announced it for this summer.

kolla · 25 July 2018, 17:49

Quote:

Originally Posted by kolla

Right. As I said, I will debug this, since it is annoying, I was just looking for potential known culprits. One obvious thing I will try is to disable Poseidon and other rom modules that I load using LoadModule early in s-s.

OK, my hunch was spot on...

Removed loading of ENVARC:PsdLoadStack and poseidon modules as rom modules fixed the problem. Just having poseidon modules loaded and then running PsdLoadStack causes trouble for VControl (and c:reboot for that matter), but running PsdLoadStack without having modules resident works fine. Since the kickstart that comes with the core already supports pressing "help" to enter early startup menus, it is not really a big deal, though I will have to navigate them using the keyboard. So, I removed loading of poseidon modules, and run PsdLoadStack from s-s instead.

If there is any interest, I could look further into it and find out if it is exactly one module or combo that causes the problems.

flype · 25 July 2018, 20:24

@kolla very good

Out of curiosity do you need a feature in vcontrol to detect if fpu in On so that you could have a line in your SS to disable only if enabled, some sort of permanent disabling. That would save you to do the operation by hand. Still would need a reboot of course.

Something like this early in the ss :

VControl hasfpu >nil:
If $rc eq 1
VControl fpu 0
Endif

kolla · 25 July 2018, 20:30

Sure, I was a little puzzled that "vcontrol fp" (without argument) doesn't already report 0 or 1

This said, it looks like disabling the FPU now is "reset proof", at least c:reboot doesn't reset the cpu back to default (can't recall if I tried with ctrl-a-a).

EDIT: yes, FPU state also survives ctrl-a-a now, good... I now see a chance to have the FPU permanently off by default

EDIT2: Second thought, why not have VControl check the state of the FPU already, so that "VControl FP 0" does nothing if it is already off?

flype · 25 July 2018, 20:36

Ok. Deal. In some days, miggy is far right now.

kolla · 25 July 2018, 20:49

Quote:

Originally Posted by flype

Ok. Deal. In some days, miggy is far right now.

No rush here.

In general, VControl has a lot of "knobs" for turning features on and off or set a value, but I don't see a way to print out what current settings are for all those "knobs". But it happens sometimes that I miss the obvious.

flype · 25 July 2018, 21:25

'Vcontrol cpu' output the DFP bit among others such as the ESS bit (Enable SuperScalar). Also the ATTNflags etc... I recon it is quite lowlevel but it helps. Some others switches being write only registers, it is not always obvious to output current status. This is why the vampire.resource module is intended for. Nice to have feedback. I intend to put more efforts into this tool also a gui.

For your EDIT2: Indeed i came to same conclusion.

systmcrsh · 26 July 2018, 23:38

in the meantime, if you want to disable the fpu by default i wrote a small program that i put in my startup sequence:

disablefpu -> will check if an fpu is detected and if it is will run "vcontrol fp 0" (which will reboot the system). if no fpu is detected the program doesn't do anything.

disablefpu will only work if vcontrol is somewhere in your path (ie, Workbench:C)

attached, source included

flype · 27 July 2018, 15:15

@systmcrsh

Nice. Thanks.

Maybe better use AFF_FPU40 rather than AFF_68882 ** in our case **. It is guaranteed it will always be set.

kolla · 28 July 2018, 19:50

It just occurred to me that the sources of PPT, one of the pieces of software that just crash on the AC68080 with FPU enabled, is available.
If there is any interest for the Apollo Team (or others) to figure out why....

http://aminet.net/package/gfx/conv/PPT_dist
http://aminet.net/package/gfx/conv/PPT-source

EDIT: I meant to write that it only crashes on startup with FPU enabled, but forgot to mention that little vital detail.
I suspect it tries to detect what FPU, if any, that is available, and then fail miserably.

flype · 29 July 2018, 11:53

@kolla

Interesting.

About our previous discussion :

VControl 1.8

Changelog:

Now, the "VControl FPU 0 or 1" feature reboots only if the requested FPU mode is different than the current mode.

In other words, one can put "VControl FPU 0 >NIL:" as first line in the startup-sequence to permanently disable the FPU.

http://www.apollo-accelerators.com/f...ver.v1.4.2.lha

kolla · 29 July 2018, 12:21

Excellent!

flype · 29 July 2018, 12:50

@kolla

Does PPT works OK when FPU is disabled ?

kolla · 29 July 2018, 13:24

Quote:

Originally Posted by flype

Does PPT works OK when FPU is disabled ?

Yes, albeit slowly.

flype · 29 July 2018, 17:22

@kolla

I tried PPT with FPU enabled. Trying reproduce and locate the issue(s).

It works here, without using the installer, this in order to not overwrite my installed BGUI libs.

I wonder if the guilty files are related to BGUI.

The way i did (as super quick test, i must say), it miss some 'gadgets' so it is incomplete but for what i tried, the program works, it opens his screen, the child windows, and the render effects worked.

Maybe you could try by using the final public release of BGUI, over the ones installed by PPT (which are quite unusual, seems to me) http://aminet.net/package/dev/gui/bgui

I could be totally wrong, just trying locate issue & reproduce by elimination.

kolla · 29 July 2018, 17:42

Sure, will have a closer look.

EDIT: Yes, it was bgui.library 41.08 that was the culprit... wonder what it was trying to do using FPU.

systmcrsh · 29 July 2018, 22:21

Quote:

Originally Posted by flype

@kolla
In other words, one can put "VControl FPU 0 >NIL:" as first line in the startup-sequence to permanently disable the FPU.

awesome, thanx!

flype · 29 July 2018, 22:46

Quote:

Originally Posted by kolla

Yes, it was bgui.library 41.08 that was the culprit... wonder what it was trying to do using FPU.

Very good

That 41.08 seems quite old hacky hacky beta.

grond · 30 July 2018, 22:43

@kolla:

So now with the updated library you can use the Vampire FPU without side effects?

25 July 2018, 20:24	#24
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	@kolla very good Out of curiosity do you need a feature in vcontrol to detect if fpu in On so that you could have a line in your SS to disable only if enabled, some sort of permanent disabling. That would save you to do the operation by hand. Still would need a reboot of course. Something like this early in the ss : VControl hasfpu >nil: If $rc eq 1 VControl fpu 0 Endif Last edited by flype; 25 July 2018 at 20:39.

25 July 2018, 20:30	#25
kolla Banned Join Date: Nov 2007 Location: Trondheim, Norway Posts: 1,893	Sure, I was a little puzzled that "vcontrol fp" (without argument) doesn't already report 0 or 1 This said, it looks like disabling the FPU now is "reset proof", at least c:reboot doesn't reset the cpu back to default (can't recall if I tried with ctrl-a-a). EDIT: yes, FPU state also survives ctrl-a-a now, good... I now see a chance to have the FPU permanently off by default EDIT2: Second thought, why not have VControl check the state of the FPU already, so that "VControl FP 0" does nothing if it is already off? Last edited by kolla; 25 July 2018 at 20:44.

25 July 2018, 20:36	#26
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	Ok. Deal. In some days, miggy is far right now. Last edited by flype; 25 July 2018 at 20:43.

25 July 2018, 21:25	#28
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	'Vcontrol cpu' output the DFP bit among others such as the ESS bit (Enable SuperScalar). Also the ATTNflags etc... I recon it is quite lowlevel but it helps. Some others switches being write only registers, it is not always obvious to output current status. This is why the vampire.resource module is intended for. Nice to have feedback. I intend to put more efforts into this tool also a gui. For your EDIT2: Indeed i came to same conclusion. Last edited by flype; 25 July 2018 at 21:37.

28 July 2018, 19:50	#31
kolla Banned Join Date: Nov 2007 Location: Trondheim, Norway Posts: 1,893	It just occurred to me that the sources of PPT, one of the pieces of software that just crash on the AC68080 with FPU enabled, is available. If there is any interest for the Apollo Team (or others) to figure out why.... http://aminet.net/package/gfx/conv/PPT_dist http://aminet.net/package/gfx/conv/PPT-source EDIT: I meant to write that it only crashes on startup with FPU enabled, but forgot to mention that little vital detail. I suspect it tries to detect what FPU, if any, that is available, and then fail miserably. Last edited by kolla; 28 July 2018 at 21:26.

27 July 2018, 15:15	#30
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	@systmcrsh Nice. Thanks. Maybe better use AFF_FPU40 rather than AFF_68882 in our case . It is guaranteed it will always be set.

29 July 2018, 11:53	#32
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	@kolla Interesting. About our previous discussion : VControl 1.8 Changelog: Now, the "VControl FPU 0 or 1" feature reboots only if the requested FPU mode is different than the current mode. In other words, one can put "VControl FPU 0 >NIL:" as first line in the startup-sequence to permanently disable the FPU. http://www.apollo-accelerators.com/f...ver.v1.4.2.lha

29 July 2018, 12:21	#33
kolla Banned Join Date: Nov 2007 Location: Trondheim, Norway Posts: 1,893	Excellent!

29 July 2018, 12:50	#34
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	@kolla Does PPT works OK when FPU is disabled ?

29 July 2018, 17:22	#36
flype Registered User Join Date: Dec 2014 Location: France Posts: 104	@kolla I tried PPT with FPU enabled. Trying reproduce and locate the issue(s). It works here, without using the installer, this in order to not overwrite my installed BGUI libs. I wonder if the guilty files are related to BGUI. The way i did (as super quick test, i must say), it miss some 'gadgets' so it is incomplete but for what i tried, the program works, it opens his screen, the child windows, and the render effects worked. Maybe you could try by using the final public release of BGUI, over the ones installed by PPT (which are quite unusual, seems to me) http://aminet.net/package/dev/gui/bgui I could be totally wrong, just trying locate issue & reproduce by elimination.

29 July 2018, 17:42	#37
kolla Banned Join Date: Nov 2007 Location: Trondheim, Norway Posts: 1,893	Sure, will have a closer look. EDIT: Yes, it was bgui.library 41.08 that was the culprit... wonder what it was trying to do using FPU. Last edited by kolla; 29 July 2018 at 22:12.

30 July 2018, 22:43	#40
grond Registered User Join Date: Jun 2015 Location: Germany Posts: 1,918	@kolla: So now with the updated library you can use the Vampire FPU without side effects?

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Vampire Gold Core v2.9 released	Sinphaltimus	News	39	11 June 2018 11:03
GOLD core released for Vampire 600 V2	guibrush	News	23	16 September 2016 04:34
SILVER9 core released for Vampire 600 V2	TuKo	News	1	04 August 2016 13:13
SILVER8 core released for Vampire 600 V2	TuKo	News	3	02 August 2016 04:13
SILVER3 core released for Vampire 600 V2	TuKo	News	4	03 April 2016 14:01