Fatdog64-630rc1 (16 Oct 2013) [CLOSED]

A home for all kinds of Puppy related projects
Message
Author
LateAdopter
Posts: 361
Joined: Fri 27 May 2011, 17:21
Location: Reading UK

#21 Post by LateAdopter »

Hello kirk & jamesbond

I didn't trust myself, so I repeated the experiment more systematically, rebooting each time. Then running the Hardinfo GPU Drawing benchmark.

In the order I did them:

5982 - Profile Default/high
3772 - Profile low/mid

2180 - DPM default (streaks)
5061 - DPM force high
4994 - DPM force low
5082 - DPM force auto
4968 - DPM default (no streaks)

I tried a few more times with DPM default but it didn't go into the broken mode again.

So when it is working, all of the DPM modes are the same. The differences are the normal variability of the test.

chapchap70
Posts: 210
Joined: Thu 18 Nov 2010, 05:39
Location: The Island Of Long (NY, USA)

#22 Post by chapchap70 »

Make sure to use a new savefile or save folder while testing.
Well that's no fun :lol: :P


Good Work! The pull down menu (is that what it's called) looks good and there is no "other" category anymore.

The video editor icon was gone because of the new video editor but all I had to do is replace "avidemix2" with "Shotcut" in the edit item; clicking item opens field. ( I thought Kino was the new editor by looking at the screenshot. :? )

I think the clock setting is UTC by default because it was 4 hours ahead of US Eastern after installation. (save folder) So that fix seems to work.

Nothing broke that I know of. The musescore I compiled myself still works & my Firefox and Tbird work as before.

Edit: Clicking on the thunderbird icon launced the new Seamonkey 2.21. I had thunderbird set up as default email. thunderbird-spot still worked in terminal although I have to reconfigure some things.
Last edited by chapchap70 on Fri 18 Oct 2013, 05:50, edited 1 time in total.

User avatar
Billtoo
Posts: 3720
Joined: Tue 07 Apr 2009, 13:47
Location: Ontario Canada

Fatdog64-630rc1 (16 Oct 2013)

#23 Post by Billtoo »

Installed to an 8gb SDHC card using fatdog installer.

video-info-glx 1.5.3 Thu 17 Oct 2013 on Fatdog64 630 Linux 3.11.4 x86_64
0.0 VGA compatible controller:
Advanced Micro Devices [AMD] nee ATI Manhattan [Mobility Radeon HD 5400 Series]
oem: ATI ATOMBIOS
product: PARK 01.00
X Server: Xorg Driver: radeon
X.Org version: 1.12.4
dimensions: 1600x900 pixels (423x238 millimeters)
depth of root window: 24 planes
direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.4
OpenGL vendor string: X.Org
OpenGL renderer string: Gallium 0.4 on AMD CEDAR
OpenGL version string: 2.1 Mesa 9.2.1
# glxgears
Running synchronized to the vertical refresh. The framerate should be
approximately the same as the monitor refresh rate.
302 frames in 5.0 seconds = 60.333 FPS
300 frames in 5.0 seconds = 59.954 FPS
300 frames in 5.0 seconds = 59.954 FPS
#
GPU Drawing
This Machine 1199 MHz 8265.520
Intel(R) Core(TM) i5 CPU M 430 @ 2.27GHz
Core 0: @1199 1: @1199 2: @1199 3: @1199 MHz
Atheros Communications Inc. AR928X Wireless Network Adapter (PCI-Express) (rev 01)
------------------------------------------------------------------
Low
GPU Drawing
This Machine 2267 MHz 3998.513
-----------------------------------------------------------------
Auto
GPU Drawing
This Machine 1199 MHz 8697.118
------------------------------------------------------------------
High
GPU Drawing
This Machine 1199 MHz 8653.813
------------------------------------------------------------------
I played a video at cnn set to low and then again set to high, the
quality of the video was fine both times, I could here the fan running
at both settings while the video played.

Gobbi
Posts: 255
Joined: Fri 09 Mar 2012, 14:01

#24 Post by Gobbi »

Thank's kirk , Jamesbond and everyone following the project for going forward with development of Fatdog .
I tried first a frugal install , but it didn't work on my desktop so I burned a DVD RW and chose ' problematic Radeon cards ' and it booted fine .
I've changed theme and font from Control Panel/Desktop/Chtheme GTK theme chooser and lxpanel crashed , that is it's appearance changed with strange colours in it .
I've set a new wallpaper and after that due to lxpanel crash I restarted X . Lx panel turned normal again but the wallpaper turned back to the default one .
These two bugs happend just once , in the beginning.
Trying to change values in lxpanel settings ( icon size , height or font custom colour ) made crash lxpanel but it recoverd after a few seconds with the initial settings .
The fact that lxpanel returned to Puppy version could explain a few things because changing values in lxpanel was always a pain until 610 ( I had to use the complete pet from the repository to do it ) .

LateAdopter
Posts: 361
Joined: Fri 27 May 2011, 17:21
Location: Reading UK

#25 Post by LateAdopter »

Hello kirk & jamesbond

I thought I might test glamor next, and see what it breaks...

The Arch Wiki says:

Since xf86-video-ati driver-1:7.2.0-1, glamor is automaticaly enabled with radeonsi drivers (Southern Island and superior GFX cards); with other graphic cards you can use it by adding the AccelMethod glamor to your xorg.conf config file in the Device section:

Code: Select all

 Option	"AccelMethod"           "glamor"
However, you need to add the following section before:

Code: Select all

Section "Module"
	Load "dri2"
	Load "glamoregl" 
EndSection
But Fatdog does not use xorg.conf. I see that dri2 and glamoregl are loaded in xorg.log but it isn't used.

If I create an xorg.conf, will fatdog use it? Or is there a better way of enabling it?

Thanks

jamesbond
Posts: 3433
Joined: Mon 26 Feb 2007, 05:02
Location: The Blue Marble

#26 Post by jamesbond »

DPM
----
Billtoo, thanks for testing. The results are strange doesn't really follow the claim that low-->auto-->high power also means low-->auto-->high performance; and from LateAdopter's result the performance definitely isn't higher than the "high" profile.

Note: if you want to use glxgears for testing with radeon, you have to specify vblank_mode=0 like this:

Code: Select all

vblank_mode=0 glxgears
otherwise radeon driver always sync the drawing to the vertical sync so the FPS is almost always 60 or 50 (that's 60Hz or 50Hz, depending on where you live).

The conclusion I get from the tests kindly done by Billtoo and LateAdopter is that for DPM, low/auto/high doesn't make any difference in terms of performance; but it does make a difference in terms of power consumption (=temperature). This is odd (I would expect better performance = more power) but that's what the result speaks. Alex (agd5f) claimed that DPM has better performance because of the ability to ramp up the GPU clock, but I haven't seen it on mine, I don't see it on LateAdopter's and I don't see it on Billtoo's. Its interesting to note from LateAdopter's comment that DPM doesn't always work too :)

I tested using glxgears (admittedly a poor test tool) and 3D game Scorched3D. One can also test using other first-person-shooter games like Urban Terror, etc. However kirk and I would also like to suggest that if 3D performance is important, one can't go wrong with AMD proprietary driver (fglrx aka Catalyst) - that's the only way to go; and this is true for all GPUs except Intel.

VDPAU
-----
Billtoo wrote:I played a video at cnn set to low and then again set to high, the
quality of the video was fine both times, I could here the fan running
at both settings while the video played.
I suppose you play the video using Flash. I just found out yesterday that while the DPM settings doesn't matter, installing VDPAU *does* make a difference. Not on video quality, but more on temperature (=power consumption). Apparently for my radeon the VDPAU works, because without it my temperature will be 10degrees higher when playing Flash video.

Savefile
---------
chapchap70 wrote:]
Make sure to use a new savefile or save folder while testing.
Well that's no fun :lol: :P
Well I don't say that using existing savefile won't always work. The savefile I use today harks back from the days of 600 alphas. But when you're doing testing, if you encounter breakage it will be difficult to pinpoint the cause; because old savefile *may* contain cruft that stops the newer Fatdogs from working (remember, the savefile has higher priority than the basesfs, whatever override/configuration settings and/or newer libs you put into the savefile will hide and/or disable files of the same name from the basesfs). As for me, I happen to know exactly what has changed, so I can anticipate this kind of breakage by removing offending old files - but most people won't know how to do this. So the accurate note is that "using old savefile with newer version may or may not work, depending on various uncontrollable conditions", but this is too long for most people; it is easier to just say "don't use old savefile when you upgrade" :)
I thought Kino was the new editor by looking at the screenshot.
Thanks, that screenshot needs updating :)
Edit: Clicking on the thunderbird icon launced the new Seamonkey 2.21. I had thunderbird set up as default email. thunderbird-spot still worked in terminal although I have to reconfigure some things.
Hmm, need to look again at that that thunderbird pet is doing. Thanks for reporting.

Lxpanel
-------
Gobbi wrote:Thank's kirk , Jamesbond and everyone following the project for going forward with development of Fatdog .
I tried first a frugal install , but it didn't work on my desktop so I burned a DVD RW and chose ' problematic Radeon cards ' and it booted fine .
You can get these special settings by looking at isolinux.cfg and copy them over for your own grub/syslinux/grub4dos installation so that they will work with your frugal install menus too.

That being said, what radeon card do you have? What kind of crash do you get? I remember we had similar problems during 620 release, and I though we fixed it with the "radeon workaround" stuff.
I've changed theme and font from Control Panel/Desktop/Chtheme GTK theme chooser and lxpanel crashed , that is it's appearance changed with strange colours in it .
I've got to try this; but lxpanel has always been a pain. It is nice going if you can get it set up properly, but to reach that ... :)
I've set a new wallpaper and after that due to lxpanel crash I restarted X . Lx panel turned normal again but the wallpaper turned back to the default one .
Odd. How did you set-up the wallpaper? Do you use the RAM layer?
These two bugs happend just once , in the beginning.
Trying to change values in lxpanel settings ( icon size , height or font custom colour ) made crash lxpanel but it recoverd after a few seconds with the initial settings .
That's the magic of Fatdog :lol:
No, seriously lxpanel crashed very often in our earlier tests (and when we make some changes that it doesn't like) that we decided to implement some sort of "lxpanel watchdog" that will automatically restart it if it crash. It doesn't happen with kirk, but for me, lxpanel will crash 3 out of 5 times if I close Firefox. Yes, after I'm done with Firefox I close it; and lxpanel goes down with it for no reason ... hence the watchdog :) Yeah when it crashes it doesn't have the chance to save its settings; so when the watchdog restarts it lxpanel reverts back to the previous setting, which in retrospect is good because if it keeps the settings that causes it to crash in the first place, that wouldn't be nice ...

EDIT: Why don't we ditch lxpanel and replace it with something else? We actually tried this in one or the earlier internal alphas (between me and kirk) - can't remember in 610 or 620 days. We replaced lxpanel+openbox with xfce and found that it was no match in terms of speed and memory/cpu consumption compared to openbox+lxpanel. It had its share of crashes too. So despite its shortcomings, lxpanel stayed, for now.
The fact that lxpanel returned to Puppy version could explain a few things because changing values in lxpanel was always a pain until 610 ( I had to use the complete pet from the repository to do it ) .
No we didn't revert lxpanel to an older version. The lxpanel in 630 is the same as in 621/620; the difference is that we fix the XDG menu following the discussion here: http://www.murga-linux.com/puppy/viewto ... 710#706710 to show the same menu structure as earlier Fatdogs and thus Puppy too. lxpanel in 610 and older were version 0.3.8.1 and the menu structure was fixed - you can't add new categories as such. The lxpanel in 620 onwards are fully XDG compliant and you can add new menus, new categories etc. Assuming it doesn't crash first :lol:

Xorg.conf
-------------
LateAdopter wrote:I see that dri2 and glamoregl are loaded in xorg.log but it isn't used.
That's interesting, I thought it would be automatically used if the driver is available. At the very least dri2 should be automatically used.
But Fatdog does not use xorg.conf.
If I create an xorg.conf, will fatdog use it? Or is there a better way of enabling it?
You can create xorg.conf in /etc/X11 and Fatdog will use it; but the recommended way of doing this is to run xorgwizard (you can run this from existing X session, no need to quit to terminal) and specify the radeon driver. After you save, you will get a file called 20-gpudriver.conf in /usr/share/X11/xorg.conf.d. Edit this file and add the appropriate options - AccelMethod in your Device section; and add additional Module section to load both modules.
Fatdog64 forum links: [url=http://murga-linux.com/puppy/viewtopic.php?t=117546]Latest version[/url] | [url=https://cutt.ly/ke8sn5H]Contributed packages[/url] | [url=https://cutt.ly/se8scrb]ISO builder[/url]

User avatar
Ted Dog
Posts: 3965
Joined: Wed 14 Sep 2005, 02:35
Location: Heart of Texas

#27 Post by Ted Dog »

OK I'll spill the beans on what I strongly expect is the root cause of these unrelated glitches since my hunch is playing out as we test it.

PLEASE if you have a slow machine or a machine with a AMD family chipset. USE

expand base to ram boot switch!

base2ram=expand

and run same software with and without it boot switch. Yes it will be faster, but that is not what I'm pointing out.

the sfs decompression routine can only be run on one core for the format used, and is blocking file related IO, since puppyliux and Fatdog are really ran with hundreds of tiny text scripts, and it can be a nest of tiny text scripts. I suspect the odd root cause is a side effect of a long history that small files load time is tiny, but the opposite is true in our case. the time to load a tiny file is the same as a large file, and the more tiny files needed the worse the effect.
The packing format of SFS files group tiny files together and compress them as a set, but no decompress buffer exists like a read-ahead ext3 at the kernel driver level.
So each tiny file gets unpacked each and everytime time called and all the other tiny text files it was packed with are as well, but that effort is wasted.

One solution I found was to create a standard ext3 block-device file and have that as a single file within a SFS. This way the read-ahead buffer IO of ext3 at the kernel level takes care of the repetitive decompression problem of small files within the same directory structure.

OK it's out there... I do not have the skills to due the sfs and double loop mount ext3 inside at the init level, but main developers can quickly

gcmartin

Decompression subsystem utility during system operations

#28 Post by gcmartin »

Reasonable.
... the time to load a tiny file is the same as a large file ...
This would be the case especially for a RAM based system (similar for those of us running via DVDs instead of full installs) which uses RAM as an I/O device via an architected I/O subsystem.

Base2RAM removes the SFS need, once loaded, for this segment of system operations.

Also to note: AMD processors ahve a differen hardware design from Intel. Its instrution staging is based upon that design.

LateAdopter
Posts: 361
Joined: Fri 27 May 2011, 17:21
Location: Reading UK

#29 Post by LateAdopter »

Hello jamesbond

Thanks for the suggestion, but...

I tried xorgwizard but it didn't work. I got one of those annoying message boxes where the answer is NOT OK, but it only has an OK button.

After I selected Radeon and OK it said: Cancelled, nothing changed.

Also the terminal window said:

Code: Select all

/usr/sbin/xorgwizard: 1: eval: Syntax error: "(" unexpected
rm: cannot remove `/usr/share/X11/xorg.conf.d/20-gpudriver.conf': No such file or directory
I wonder what I did wrong!

kirk
Posts: 1553
Joined: Fri 11 Nov 2005, 19:04
Location: florida

#30 Post by kirk »

LateAdopter,

You can force glamor to be used by editing /usr/X11R7/lib64/X11/xorg.conf.d/glamor.conf. Make it look like this:

Code: Select all

Section "Module"
	Load  "dri2"
	Load  "glamoregl"
EndSection

Section "Device"
  Identifier             "Device0"
   Option  "AccelMethod"  "glamor"
EndSection
Glamor still has it's share of problems. On my laptop with intel ironlake graphics, openbox/lxpanel causes a segfault with glamor on. JWM does seem to work, but be warned, we don't really support JWM. It's just there for trouble shooting. I'm still doing some testing on this. Thanks.

User avatar
Ted Dog
Posts: 3965
Joined: Wed 14 Sep 2005, 02:35
Location: Heart of Texas

Re: Decompression subsystem utility during system operations

#31 Post by Ted Dog »

gcmartin wrote:Reasonable.
... the time to load a tiny file is the same as a large file ...
This would be the case especially for a RAM based system (similar for those of us running via DVDs instead of full installs) which uses RAM as an I/O device via an architected I/O subsystem.

Base2RAM removes the SFS need, once loaded, for this segment of system operations.

Also to note: AMD processors ahve a differen hardware design from Intel. Its instrution staging is based upon that design.
yes you understand this, also neat plus if ext3 block-file is copied to a harddrive, we could have a win-win type of frugal install, not really INSTALLED per say but that would free up RAM weak machines and remove the sfs problem as a bonus.

gcmartin

Re: Decompression subsystem utility during system operations

#32 Post by gcmartin »

... that would free up RAM weak machines and remove the sfs problem as a bonus.
The early 2006 vendor PCs are the only 64bit platforms which had merely 1GB of RAM (excepting for some 64bit netbooks). Most all other vendor desktops-laptops since had 2GB+ RAM. (this was a vendor mandated standard)

Just a note.

gcmartin

#33 Post by gcmartin »

Radeon in FATDOG
While there is discussion on install, setup, and use of these cards, the community seemingly can be broken into several groups:
  • Desktop only users
  • Multimedia users
  • Gaming users
  • 3D user needs
Question:
Is there anyone who is experienced enough that can setup a thread, document, or wiki which guides simple clear-cut instructions for system setup for persons falling into either of these groups?

This is, "time-permitting", after GA release of FATDOG630 and the Radeon issues are understood-addressed.

Loving this FATDOG...
Edited: Changed video to multimedia in outline
Last edited by gcmartin on Mon 21 Oct 2013, 06:19, edited 1 time in total.

User avatar
Ted Dog
Posts: 3965
Joined: Wed 14 Sep 2005, 02:35
Location: Heart of Texas

#34 Post by Ted Dog »

Yes and 2G is plenty large enough to try expand to ram.

2G -sfs file (200M) /2 = ~875M user RAM

same

2G - unpacked 700M /2 = ~ 700M user RAM

Which is why I wanted an auto-expand if RAM>= 10x sfs size. but no... :cry:

chapchap70
Posts: 210
Joined: Thu 18 Nov 2010, 05:39
Location: The Island Of Long (NY, USA)

#35 Post by chapchap70 »

Quote:
Edit: Clicking on the thunderbird icon launced the new Seamonkey 2.21. I had thunderbird set up as default email. thunderbird-spot still worked in terminal although I have to reconfigure some things.

Hmm, need to look again at that that thunderbird pet is doing. Thanks for reporting.




Hi Jamesbond. Don't go crazy looking into the Thunderbird Pet. I installed Thunderbird 24.0 myself and changed the default icon.

It turns out that I couldn't send or receive email because of the firewall (don't know why specifically) but when I turned it off than back on, sending and receiving worked.


When I opened my netbook lid back up, my arrow no longer moved and pushing the power button made my puppy bark and exit X. I put the old suspend.sh file back and renamed the one you rewrote. Maybe there is a logfile somewhere; I don't know how to figure out what exactly caused the glitch.

Gobbi
Posts: 255
Joined: Fri 09 Mar 2012, 14:01

#36 Post by Gobbi »

Thank's for the shortcut Jamesbond :!:
Adding the setting blacklist=radeon from isolinux.cfg into grub made the frugal install work indeed ...
My graphic card is a Radeon HD 7870 XT (Tahiti) and 'radeon workaround' solved the initial problems in FD620 . As then , after loading the kernel everything stopped , freezing the screen after the line where it was loading the kernel modules. I think with the proper Catalyst pet everything will go fine.
I don't remember if I set the wallpaper before I closed the RAM-save... Anyway returning to default wallpaper happened only that once time. I couldn't repeat that.
I could repeat an error though : Changing the Gtk theme ( Gradent Brown ) and then from the same Gtk window changing font (bold 12) then pushed 'Apply' button and then pushed 'Ok'button gave me a black screen with the mouse pointer functioning and after 10 seconds lxpanel ( strange coloured ) appeared . No wallpaper. Restarted X and the wallpaper I've set was on the desktop.
I even repeated changing the font size from 12 to 13 and then --> Apply --> Ok --> No more wallpaper again and lxpanel issue again . Restarting X put everything in order .

A second issue : Control Panel/Desktop/Fatdog64 Event Manager. Canceling the initial 30 seconds of interval for saving , without putting a zero (0) there gives an error and an increase of CPU activity which grows if I repeat trying to acces that Event Manager from Control Panel/Desktop/ .
Previously I remember the zero (0) value was added by default even if someone forgot to put a value...
Attachments
Gtk_server_err_1.jpg
(103.48 KiB) Downloaded 1032 times
Black wallpaper - lxpanel error.png
(29.5 KiB) Downloaded 1007 times

LateAdopter
Posts: 361
Joined: Fri 27 May 2011, 17:21
Location: Reading UK

#37 Post by LateAdopter »

Hello kirk

Thanks for the instructions.
You can force glamor to be used by editing /usr/X11R7/lib64/X11/xorg.conf.d/glamor.conf.
It worked, in the sense that glamor acceleration was enabled and reported in the xorg.log. But it turnout to be slower than the default method.

VLC gave lots of picture too late warnings with my test video. So I reverted to normal and there are no warnings now.

That could be why it is not enabled by default.

User avatar
L18L
Posts: 3479
Joined: Sat 19 Jun 2010, 18:56
Location: www.eussenheim.de/

Fatdog64-630rc1 (16 Oct 2013)

#38 Post by L18L »

Fatdog64-610 (another xorg) was ok for my asrock AD2550-ITX board

But now no go with boot parameters:
blacklist=radeon loadmodules=gma500_gfx loglevel=7 savefile=none

last line I could read before no signal input and black screen was something containing:
drm
error

What could I do to get the exact error messages please?

kirk
Posts: 1553
Joined: Fri 11 Nov 2005, 19:04
Location: florida

#39 Post by kirk »

L18L,

Sometime when one thing gets fixed something gets broke. Take a look at http://distro.ibiblio.org/fatdog/web/faqs/screen.html and see if that helps.



I've noticed that if you enable glamor, lxpanel will cause xorg to segfault. JWM does work. I'm working on an alternative.

User avatar
Ted Dog
Posts: 3965
Joined: Wed 14 Sep 2005, 02:35
Location: Heart of Texas

#40 Post by Ted Dog »

Video Editor crashes and disappears on all videos at the file section stage. (granted it was being pulled from NTFS.)

Was able to use FD621 to 'fix-usb' to fat32 for FD630rc, I think the recognition of fat32 is not working, from programs that are looking for it by partition, FD621 saw that the first attempt to construct fix-usb worked in that the partition was there, but unformat, FD630rc1

Also oddly I was unable to change the permissions as root even when root owner/group on a file on said fat32 yet I could copy it fine.

Have my testing box set back up, so I can try burning a few bluerays/DVD multisessions later this weekend.

Post Reply