IcePuppy-014 for newer and Ice Puppy-015 for older computers

[pemasu]

Beem, 003 was the conservative version, 486 optimized which means automatically that it is without highmem+pae property.
If this latest lupe23 seems good enough, I will make next that conservative kernel.
So there will be version for older hw also.

I have compiled some stuff for spup just for fun of it.

Now I am updating Snow Puppy. Next will be newer conservative kernel.

[Béèm]

Beem, 003 was the conservative version, 486 optimized which means automatically that it is without highmem+pae property.
If this latest lupe23 seems good enough, I will make next that conservative kernel.
So there will be version for older hw also.

I have compiled some stuff for spup just for fun of it.

Now I am updating Snow Puppy. Next will be newer conservative kernel.

OK not so easy to change the cpu chip then.
Maybe I'll look at snowpuppy then.

Yes I saw you compiled for spup as well. Nice.

[gcmartin]

@Béèm opens an interesting point. I may be wrong, but, the glxgears program will "lock" on a single process as the system time-slices workload (Linux architecture). I don't believe this program was designed as Multi-Processor, multi-threaded test application.

That, I believe, tis the reason it is a reference tool as opposed to a tool with any accuracy for video performance on a given system.

See this discussion.. Contribute if you have additional findings.

Pemasu has brought 2 new things to bear in Puppy with his work on these new kernels: the ability to better exploit the number of processors available to the system as well as to make ALL RAM available for Linux support of running applications in the OS. This is obvious to us because of the behavior and responsiveness we see in ICE when we compare it against all other 32bit Puppy distros. It flat out is significantly different. And, in my visual looks, (even though I know he doesn't support it), I do not seem to suffer in comparing behavior in ICE versus the behavior in SNOW (where, to me, this is a more apple-to-apples comparison).

The balancing we may see referenced in the GUI CPU meters for glxgears, I believe, is a distortion of what really is happening in the execution path of its running.

With what he has enabled for us, we really do need a better tool for a more accurate measure of video performance. This is not only necessary here in 32bit environments, but as well, in 64bit arenas, as well. Until then, our ONLY video measure is the emotional one

Hope this helps

[jim3630]

Pemasu has brought 2 new things to bear in Puppy with his work on these new kernels: the ability to better exploit the number of processors available to the system as well as to make ALL RAM available for Linux support of running applications in the OS. This is obvious to us because of the behavior and responsiveness we see in ICE when we compare it against all other 32bit Puppy distros. It flat out is significantly different. And, in my visual looks, (even though I know he doesn't support it), I do not seem to suffer in comparing behavior in ICE versus the behavior in SNOW (where, to me, this is a more apple-to-apples comparison).

Hope this helps

gcmartin, thanks for the insight and link.

In that vein having both IP12 and SP18 and only doing everyday stuff, movies streaming and downloading, email, and no number crunching or heavy processor apps or games, etc seems the only difference can discern is the download speed. Faster in IP

[James C]

Ice Puppy 012 live pfix=ram on my backup Windows box. Everything good on boot.

-Computer-
Processor : 2x Intel(R) Pentium(R) 4 CPU 2.60GHz
Memory : 2592MB (185MB used)
Operating System : Unknown distribution
User Name : root (root)
Date/Time : Sun 29 May 2011 01:35:01 AM CDT
-Display-
Resolution : 1440x900 pixels
OpenGL Renderer : Mesa DRI Intel(R) 865G GEM 20091221 2009Q4 x86/MMX/SSE2
X11 Vendor : The X.Org Foundation
-Multimedia-
Audio Adapter : ICH4 - Intel ICH5

-Version-
Kernel : Linux 2.6.38.4 (i686)
Compiled : #1 SMP PREEMPT Thu May 26 16:10:39 EEST 2011
C Library : GNU C Library version 2.11.1 (stable)
Default C Compiler : Unknown
Distribution : Unknown distribution


2.0 VGA compatible controller: Intel Corporation 82865G Integrated Graphics Controller (rev 02)
oem: Intel(r)865G Graphics Chip Accelerated VGA BIOS product: Intel(r)865G Graphics Controller Hardware Version 0.0

X Server: Xorg Driver: unknown
X.Org version: 1.7.6
dimensions: 1440x900 pixels (380x238 millimeters)
depth of root window: 24 planes

direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.2
OpenGL vendor string: Tungsten Graphics, Inc
OpenGL renderer string: Mesa DRI Intel(R) 865G GEM 20091221 2009Q4 x86/MMX/SSE2
OpenGL version string: 1.3 Mesa 7.7.1
sh-4.1# glxgears
1836 frames in 5.0 seconds
2326 frames in 5.0 seconds
2384 frames in 5.0 seconds
2365 frames in 5.0 seconds
2408 frames in 5.0 seconds

Looking good.

[James C]

Beem, 003 was the conservative version, 486 optimized which means automatically that it is without highmem+pae property.
If this latest lupe23 seems good enough, I will make next that conservative kernel.
So there will be version for older hw also.

I have compiled some stuff for spup just for fun of it.

Now I am updating Snow Puppy. Next will be newer conservative kernel.

Personally looking forward to a noin-PAE build. My older hardware will thank you.
Contrary to some, there is hardware that will not run on the PAE kernel and on lower ram machines the extra ram usage is noticeable.I'm not even going to test PAE on a box with less than 1 GB of ram anymore, too much extra overhead, but Ice runs great over 1 GB though.

You're doing some excellent work.

[James C]

Not bad performance with several Firefox tabs open.......

[Béèm]

I see weird results on glxgears.
Running now icepuppy and snowpuppy on the Medion 8818 and snowpuppy on the R40 laptop.
The Medion has dual processors of 1,86 each and the R40 1,400Ghz single processor.
The MD8818 has a nVidia Gforce 7650 video controler, the R40 a radeon 7500 one.
And yet on the R40 (icepuppy FAE doesn't run) show much more frames per 5 seconds.

Code: Select all

Medion 8818, CPU 1,86Ghz
Thinkpad R40, CPU 1,400Ghz

Booting MD8818 Icepuppy 011 without nolapic, thus dual processor
sh-4.1# glxgears
1848 frames in 5.0 seconds
1833 frames in 5.0 seconds
1848 frames in 5.0 seconds
1837 frames in 5.0 seconds
1833 frames in 5.0 seconds
1834 frames in 5.0 seconds
1839 frames in 5.0 seconds
1835 frames in 5.0 seconds
1844 frames in 5.0 seconds
1825 frames in 5.0 seconds
1833 frames in 5.0 seconds
^C
sh-4.1# 
Booting MD8818 Icepuppy 011 with nolapic, thus single processor.
sh-4.1# glxgears
1235 frames in 5.0 seconds
1211 frames in 5.0 seconds
1212 frames in 5.0 seconds
1210 frames in 5.0 seconds
1228 frames in 5.0 seconds
922 frames in 5.0 seconds
1121 frames in 5.0 seconds
1201 frames in 5.0 seconds
1295 frames in 5.0 seconds
1264 frames in 5.0 seconds
1299 frames in 5.0 seconds
1282 frames in 5.0 seconds
1290 frames in 5.0 seconds
1270 frames in 5.0 seconds
1307 frames in 5.0 seconds
1261 frames in 5.0 seconds
1304 frames in 5.0 seconds
1277 frames in 5.0 seconds
1305 frames in 5.0 seconds
1238 frames in 5.0 seconds
1312 frames in 5.0 seconds
1228 frames in 5.0 seconds
^C
sh-4.1# 

Booting MD8818 Snowpuppy 018 without nolapic, thus dual processor
sh-4.1# glxgears
1857 frames in 5.0 seconds
1837 frames in 5.0 seconds
1858 frames in 5.0 seconds
1822 frames in 5.0 seconds
1660 frames in 5.0 seconds
1847 frames in 5.0 seconds
^C
sh-4.1# 
Booting MD8818 Snowpuppy 018 with nolapic thus single processor
sh-4.1# glxgears
1251 frames in 5.0 seconds
1269 frames in 5.0 seconds
1227 frames in 5.0 seconds
1275 frames in 5.0 seconds
1251 frames in 5.0 seconds
1271 frames in 5.0 seconds
^C
sh-4.1# 
Booting R40 Snowpuppy 018 no kernel parameters
sh-4.1# glxgears
4048 frames in 5.0 seconds
3498 frames in 5.0 seconds
3498 frames in 5.0 seconds
3496 frames in 5.0 seconds
3498 frames in 5.0 seconds
3493 frames in 5.0 seconds
3490 frames in 5.0 seconds
^C
sh-4.1# 

[pemasu]

01micko suspected it might be woof installed script problem, but I dont have a glue what is the real reason that pmedia=usbflash parameter does not work.

I suspect also that is due to some change in these newer kernels.

If you have some words of wisdom, I am gratified if you post them.

Barry Kauler responded:

Did you apply the 'usb-storage.patch'?
See here:
http://bkhome.org/sources/kernel-2.6.35.7/patches/

Well, I just checked mine and wuxiandianzi applied patches in 2.6.38 kernels. That wasnt applied.

So, it is time to to the drawing board and apply that patch to the kernel.
I hope it resolves the problems with usb booting.

[pemasu]

Well, this means major overhaul. I will stick with 2.6.38.4 kernel, because I have made about working configuration for it.
But, I will disable the minor numbering from the kernel, so the kernel will be called 2.6.38. The update to the newer 2.6.38 might be bossible without unknown symbols error in kernel module compiles when using older 2.6.38 version. That means that the kernel specs are about the same. So upgrading to the newer version should be possible later. But right now I dont want to test newer version with my previous DOTconfig file.
But...the third party modules compiled so far wont be usable. I need compile them again but hopefully then this is the last time.
Usb_storage patch was overlooked even that ist existence was easily found when I knew that it would be needed. Iguleders new kernel kit also lacks it. So if this fixes the usb booting, that kernel kit needs updating also.
I have now created that patch and inserted it to the Iguleders kernel kit platform.
Now the new kernel is at the creating process.....

[Béèm]

I don't know much about those different kernel versions.
If I understand well, most often a new versions come out (at an amazing speed) with a new version of kernel, but without some change in the packages.

Now I wonder if it would be possible to change a kernel in an existing version by means of loading with sfs on the fly of a new_kernel.sfs

[pemasu]

Beem. No you cant. It doesnot change in the fly initrd.gz and booting vmlinuz.
That handy solution is not applicable....yet...lol

[Béèm]

Beem. No you cant. It doesnot change in the fly initrd.gz and booting vmlinuz.
That handy solution is not applicable....yet...lol

Thanks.
So developers will not be out of work soon then.

[pemasu]

Success with booting usb stick without pdev1.

Kernel repatched with usb_storage.patch and new rebuild of Ice Puppy.
I used Iguleders kernel-kit with needed modifications.

Posting now from usb stick boot of 2.6.38.4 kernel named as 2.6.38 and not using pdev1 at menu.lst.

My notices. at kernel row pmedia=usbflash still didnt work. I formatted the stick from ex4 to ext2, reinstalled the puppy files and installed grub4dos again, reboot, no luck...I remember that I hadnt ticked the boot flag with gparted...oh well...reboot...no luck....frustrated i tried pmedia=hddflash...IT BOOTS !!!! I want to see this again.....reboot......
Next finding...searching for puppy files....pausing......and then boot again.

But it took about 1o seconds before it found the puppy files from the stick.
So...there could be timing issue also why the booting from usb stick has failed.

So for me pmedia=usbflash didnt work but pmedia=hddflash worked. I had removed all lupu 013 sfs files from hdd to be sure it didnt use any of them.

Uploading now Ice Puppy-013. For me acpi_call module worked from Ice Puppy-012 folder. I would like to get feedback about those other drivers if anyone has possibility to test some of them.

Also I would like to get feedback can others usbboot from usbhdd or usbstick or usbnosering or whatever with pmedia=usbflash or pmedia=hddflash without that pdev1 parameter.

Anyway I opened bottle of sparkling......, namely big coke bottle because the new kernel works, it might boot with pmedia parameter and the kernel is now without minor numbering it is build using Iguleders kernel-kit first time and it is patched freshly using needed patches and it might be compatible with previously compiled kernel module drivers.

[pemasu]

IcePuppy-013 is now in its own folder.

Download link: http://smokey01.com/pemasu/IcePuppy-013
Previous drivers are in Ice Puppy-012 folder atm: http://smokey01.com/pemasu/IcePuppy-012

Those who are familiar with usb booting, please test booting with pmedia=usbflash, pmedia=usbhd or even as I did pmedia=hddflash.

Well, you need usb bootable media of course.

[pemasu]

It looks like pmedia=ataflash works for me, but only with pausing. Pmedia=usbflash or pmedia=usbhd does not boot. If I left pmedia parameter out, it boots from usbstick.
Oh well, it looks like init script still does not boot with usb storage definitions.

[jim3630]

Success with booting usb stick without pdev1.

Kernel repatched with usb_storage.patch and new rebuild of Ice Puppy.
I used Iguleders kernel-kit with needed modifications.

Posting now from usb stick boot of 2.6.38.4 kernel named as 2.6.38 and not using pdev1 at menu.lst.

My notices. at kernel row pmedia=usbflash still didnt work. I formatted the stick from ex4 to ext2, reinstalled the puppy files and installed grub4dos again, reboot, no luck...I remember that I hadnt ticked the boot flag with gparted...oh well...reboot...no luck....frustrated i tried pmedia=hddflash...IT BOOTS !!!! I want to see this again.....reboot......
Next finding...searching for puppy files....pausing......and then boot again.

But it took about 1o seconds before it found the puppy files from the stick.
So...there could be timing issue also why the booting from usb stick has failed.

So for me pmedia=usbflash didnt work but pmedia=hddflash worked. I had removed all lupu 013 sfs files from hdd to be sure it didnt use any of them.

Uploading now Ice Puppy-013. For me acpi_call module worked from Ice Puppy-012 folder. I would like to get feedback about those other drivers if anyone has possibility to test some of them.

Also I would like to get feedback can others usbboot from usbhdd or usbstick or usbnosering or whatever with pmedia=usbflash or pmedia=hddflash without that pdev1 parameter.

Anyway I opened bottle of sparkling......, namely big coke bottle because the new kernel works, it might boot with pmedia parameter and the kernel is now without minor numbering it is build using Iguleders kernel-kit first time and it is patched freshly using needed patches and it might be compatible with previously compiled kernel module drivers.

finally got ip13 frugal install on usb stick. have problem with universal installer since would not install spup to hdd 2 days ago with error "cannot be mounted currently mounted read-write does not have lupusave.3fs.file in it" Doesn't mention what cannot be mounted.

ip13 and spup burnt to cd, mounted and files copied for frugal install.
ip13 needed pmedia=usbflash and the Broadcom wo pet 22, was the last one you made for IP think is 22. connected to internet straight away.

acpitool -s works without issue.

Sun 29 May 2011 Operating System: Ice Puppy-013 Linux 2.6.38
2.0 VGA compatible controller: Intel Corporation Core Processor Integrated Graphics Controller (rev 02)
oem: Intel(R)Ironlake Mobile Graphics Chipset Accelerated VGA BIOS product: Intel(R)Ironlake Mobile Graphics Controller Hardware Version 0.0

X Server: Xorg Driver: unknown
X.Org version: 1.7.6
dimensions: 1024x768 pixels (270x203 millimeters)
depth of root window: 24 planes

direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.2
OpenGL vendor string: Tungsten Graphics, Inc
OpenGL renderer string: Mesa DRI Intel(R) IGDNG_M GEM 20091221 2009Q4 x86/MMX/SSE2
OpenGL version string: 2.1 Mesa 7.7.1

Intel(R) Pentium(R) CPU P6100
Core 0: 1999 1: 1999 MHz

...the above also recorded at '/tmp/root/report-video-glx'.

# glxgears
3563 frames in 5.1 seconds
5446 frames in 5.0 seconds
5515 frames in 5.0 seconds
5528 frames in 5.0 seconds
5524 frames in 5.0 seconds
5527 frames in 5.0 seconds
5562 frames in 5.0 seconds
5522 frames in 5.0 seconds
5542 frames in 5.0 seconds

spus still haven't got connected. Odd the pet loads and configures. eth1 goes up but won't stay up. Micko PostPosted: Fri 20 May 2011, 19:42 Post subject: Reply with quote
Very Happy Hello,
Frisbee works well...
Only one dilemma which is why I didnt include it...
If I want to switch connections via the traditional wizard....
I have to ifconfig wlan0 down first, or any changes that the wizard makes are ignored....
It is awesome if you use only one connection....

pemasu do you have a frisbee pet for spup? wonder if that would be the fix? thanks jim

[James C]

Ice Puppy 013 live muti-session cd.Everything working as usual, no problems.

# glxgears
2192 frames in 5.0 seconds
2206 frames in 5.0 seconds
2202 frames in 5.0 seconds
2200 frames in 5.0 seconds
2198 frames in 5.0 seconds
2192 frames in 5.0 seconds

Sun 29 May 2011 Operating System: Ice Puppy-013 Linux 2.6.38
d.0 VGA compatible controller: nVidia Corporation C61 [GeForce 7025 / nForce 630a] (rev a2)
oem: NVIDIA product: MCP61 - mcp61-86 Chip Rev

X Server: Xorg Driver: unknown
X.Org version: 1.7.6
dimensions: 1366x768 pixels (413x232 millimeters)
depth of root window: 24 planes

direct rendering: Yes
server glx vendor string: SGI
server glx version string: 1.2
OpenGL vendor string: Mesa Project
OpenGL renderer string: Software Rasterizer
OpenGL version string: 2.1 Mesa 7.7.1

AMD Athlon(tm) II X4 620 Processor
Core 0: 2611 1: 2611 2: 2611 3: 2611 MHz

...the above also recorded at '/tmp/root/report-video-glx'.

[pemasu]

would not install spup to hdd 2 days ago with error "cannot be mounted currently mounted read-write does not have lupusave.3fs.file in it"

James C. I saw samekind report in spup devel thread by Jim3630. And I had that problem also. There seems to happened something with puppyinstaller.

ip13 needed pmedia=usbflash and the Broadcom wo pet 22, was the last one you made for IP think is 22. connected to internet straight away.

If I understand right, pmedia=usbflash worked for you. And previous wireless worked also with this wholly recompiled new kernel. These are really good news.
That puppyinstaller problem needs to be checked next. There is possibility that new woof has even problem.

[James C]

would not install spup to hdd 2 days ago with error "cannot be mounted currently mounted read-write does not have lupusave.3fs.file in it"

James C. I saw samekind report in spup devel thread by Jim3630. And I had that problem also. There seems to happened something with puppyinstaller.

ip13 needed pmedia=usbflash and the Broadcom wo pet 22, was the last one you made for IP think is 22. connected to internet straight away.

If I understand right, pmedia=usbflash worked for you. And previous wireless worked also with this wholly recompiled new kernel. These are really good news.
That puppyinstaller problem needs to be checked next. There is possibility that new woof has even problem.

Wrong Jim..... That problem belongs to jim3630.

However, I will test a hdd install just to check things out.

Kinda like running the multi-session discs.....been working perfectly so far.