T2-puppy for other CPUs

A home for all kinds of Puppy related projects
Message
Author
User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#16 Post by cb88 »

double post ... heh i even refreshed the page before posting and didn't see it...
Last edited by cb88 on Fri 18 Jan 2008, 17:42, edited 1 time in total.
Taking Puppy Linux to the limit of perfection. meanwhile try "puppy pfix=duct_tape" kernel parem eater.
X86: Sager NP6110 3630QM 16GB ram, Tyan Thunder 2 2x 300Mhz
Sun: SS2 , LX , SS5 , SS10 , SS20 ,Ultra 1, Ultra 10 , T2000
Mac: Platinum Plus, SE/30

User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#17 Post by cb88 »

"And finally: We had several feedbacks that PPC was still working (Yes working!). If you are a PPC user, please do keep checking for regressions as we need to be notified ASAP about PPC regressions. "

THat is a quote from the nouveau open source nvidia driver wiki:

probably of interest to you since the PS3 has an Nvidia RXS GPU also i think they are starting to look into shader support and 2D support is improving ie better Xv support


so how hard is a livecd to boot on a PS3?

what about crosscompiling x86 puppy on the PS3!!!
Taking Puppy Linux to the limit of perfection. meanwhile try "puppy pfix=duct_tape" kernel parem eater.
X86: Sager NP6110 3630QM 16GB ram, Tyan Thunder 2 2x 300Mhz
Sun: SS2 , LX , SS5 , SS10 , SS20 ,Ultra 1, Ultra 10 , T2000
Mac: Platinum Plus, SE/30

Leachim
Posts: 229
Joined: Sun 27 May 2007, 23:04

#18 Post by Leachim »

I would be very interested in a x86-64 version!

I dream of CPU auto-detection at startup and starting a x86-32 or a x86-64 based Linux depending on the detected hardware. :-)

User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#19 Post by cb88 »

hmm are you talking about an autobuild T2?

that would be cool with fixes to each architecture so the build goes correctly... would need a target platform though like i586 or i686 or PS3 :twisted:

as far as the kernel detecting the cpu... i think there will allways be limitations good to detect x86 based cpus but an x86 kernel wouldn't even start up on say a sparc or PPC would it....
Taking Puppy Linux to the limit of perfection. meanwhile try "puppy pfix=duct_tape" kernel parem eater.
X86: Sager NP6110 3630QM 16GB ram, Tyan Thunder 2 2x 300Mhz
Sun: SS2 , LX , SS5 , SS10 , SS20 ,Ultra 1, Ultra 10 , T2000
Mac: Platinum Plus, SE/30

Leachim
Posts: 229
Joined: Sun 27 May 2007, 23:04

#20 Post by Leachim »

I just meant autodetection for the x86-architecture. We could move the autodetection into the boot-loader.

I have some experience with writing boot-loaders.

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

Lowest common clean cross compile -T2- dingo

#21 Post by Ted Dog »

I am still in the discovery phase, I really can't answer anything yet. So here is what I pretend to know.

livecd on ps3 looks very possible, but I am going to start with usb flashdrive. The blu-ray is very strongly locked down. Can't even read DVD's vob within linux.
The PS3 should only need a single (but less than 4M) boot kernel like thing and the single SFS file. Due the fact that the wireless,eth, sound and graphics kernel modules do not change from PS3 to PS3 those can be directly combined within the kernel and all the other removed.
This uniformity will make the complex post boot yet pre JWM destop much more easy.

T2 auto-build sounds interesting but very far off, guilded T2 build sound more promising like what BarryK has started. My feedback should help put together a few get started html guides for other CPUs. T2 is not that hard its the long time between start and end of compile, to see where the mistakes are being made in the process.
Since I also have access to the puppy source files I plan to post the CPU unique T2 build configs there as well.
I do not have access to any x86-64 type machines so I would need feedback on problems after all the cross-compile issues are worked out.

Point to remember T2 just handles program compiles and does very little file movement, and no puppy custom scripts [all that glue].
And more than a handful of programs used in puppy [like JWM] are not found in the T2-dingo package list.

Custom boot guy hang around I'll be asking some questions, just not there yet!

"that would be cool with fixes to each architecture so the build goes correctly... would need a target platform though like i586 or i686 or PS3"
----Exactly my idea, cb88!!!!



Ted

ArneBab
Posts: 6
Joined: Wed 20 Aug 2008, 18:48

News?

#22 Post by ArneBab »

Hi,

I just searches for Puppy Linux on PPC for again for some time till I found this thread.

The Puppy for ps3 looks very promising, but the last post was a few months ago, so I'd like to ask for the current state of affairs of Puppy for PPC.

Also I'd like to ask:
Is it possible to compile it for 32bit PPC, too? (like an old G3 ibook)?
Last edited by ArneBab on Wed 20 Aug 2008, 22:01, edited 1 time in total.

User avatar
prehistoric
Posts: 1744
Joined: Tue 23 Oct 2007, 17:34

more modest project?

#23 Post by prehistoric »

I've been waiting for some time to hear about progress on this thread as well. What little I know about the PS3 appears to make it a difficult target architecture, filled with proprietary hardware and firmware.

How about a more modest project as a stepping stone?

Here's something I came up with after I tracked a problem to my router while configuring my system to run in what "Shields Up!" calls stealth mode.

How about converting this ActionTec Wireless Router to a more recent kernel with an open source module for the Ralink RT2500 in place of the proprietary code they originally used? At present, it is possible to convert this to run DD WRT by substituting a different mini-PCI card. The advantage comes from the flexibility of DD WRT, more like an expensive CISCO router than a router for consumer use.

Other routers which can be converted are on the market, but you must be careful, as manufacturers frequently change designs without changing the consumer part designation. Some Linksys wireless routers with the number WRT54G which was easy to use with DD WRT have now been "improved" so they only have 2 MB of flash memory. NetGear has offered an Open Source version of their WGR614 for sale. (I suspect the ones offered as refurbished were returned because they were "bricked" by experimentation.) The ActionTec router above has 16 MB. flash memory and 32 MB RAM, leaving lots of room for experimentation. I found one for $16.

Is this plausible? More than I want to get into? Completely irrelevant and belongs in a separate thread?

ArneBab
Posts: 6
Joined: Wed 20 Aug 2008, 18:48

#24 Post by ArneBab »

My main interest is an old ppc ibook (the one from my wife :) ), so I can't help with routers...

Still I wish you good luck, should you want to tackle it!

amigo
Posts: 2629
Joined: Mon 02 Apr 2007, 06:52

#25 Post by amigo »

For someone who has never done a cross-compile -even for a single architecture, dissapearing for a few months is not uncommon. Just pray that the guy comes back someday with any of his wits left...

User avatar
prehistoric
Posts: 1744
Joined: Tue 23 Oct 2007, 17:34

cross compiles and wits

#26 Post by prehistoric »

Good call, amigo! There speaks the voice of experience.

I've done worse than cross-compiles, even bootstrapped a compiler onto a new architecture. As for wits remaining, there is some dispute on the subject.

I intend to take the router project slowly and husband my remaining resources.

User avatar
prehistoric
Posts: 1744
Joined: Tue 23 Oct 2007, 17:34

Alpha 400 MIPS laptop

#27 Post by prehistoric »

Here's a machine without an x86 processor which sounds like it could use Puppy. Anyone know more?

User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#28 Post by cb88 »

$200 new for a 400mhz cpu? looks like a waste of money to me

....www.openpandora.org <<< 600mhz 128mb ram 10+ hours battery and dual SDHC ... the smaller than the original nintendo DS.. also it has built in #d acceleration and a dedicated DSP for multimedia acceleration for programs that support it (ie modified mplayer VLC maybe some games emulators etc... rather difficult to program for though)

or an eee $400 but you are getting a whole lot more horsepower for the price i don't think a 400mhz mips is even as fast as a 400mhz pentium 2 but i could be way off on that

BTW: i got to the part in the T2 compile where perl breaks is there a way to permanently fix that so every thing will compile perfectly first try other than what barry suggests?
Taking Puppy Linux to the limit of perfection. meanwhile try "puppy pfix=duct_tape" kernel parem eater.
X86: Sager NP6110 3630QM 16GB ram, Tyan Thunder 2 2x 300Mhz
Sun: SS2 , LX , SS5 , SS10 , SS20 ,Ultra 1, Ultra 10 , T2000
Mac: Platinum Plus, SE/30

User avatar
prehistoric
Posts: 1744
Joined: Tue 23 Oct 2007, 17:34

not rushing out

#29 Post by prehistoric »

@cb88,

I'm not rushing out to buy things, just keeping my eye on the field. My own experience with MIPS is that you get roughly 1 instruction/cycle which, as you said, would fall behind the superscalar architecture of a Pentium II. I don't know about the particular derivative they're using.

That's a good link above. People can click through to Open Pandora here. I'd have to hold one of those to see about the keyboard and display size.

User avatar
Aitch
Posts: 6518
Joined: Wed 04 Apr 2007, 15:57
Location: Chatham, Kent, UK

#30 Post by Aitch »

Hi Ted Dog

Question:

This distcc you mention, could it be used in combination with rdesktop or some such, to make for distributed compiling via network?

That way, maybe some of our 'heat' could be utilised, like FAH, perhaps?

I'm sure distributed compiling would have advantages for many puppy devs

yes?

possible?

If not, no worries, not meaning to sidetrack you :D

Aitch

User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#31 Post by cb88 »

the arm in the pandora gets 2 instructions per cycle but is not super scalar like the PII .... should whip a MIPS... and yeah the screen is really small comparatively but slightly larger than the nokia n800 screen

@Aitch..... the compile environment must be identical... also the config options would need to be as well.... distcc would not be good from over the internet T2 does compile slowly but not THAT slow... the piglett001 profile for T2 took 3.5 days or so on a dual PII at 300mhz.... a quad core could do it in a few hours i would think....
Taking Puppy Linux to the limit of perfection. meanwhile try "puppy pfix=duct_tape" kernel parem eater.
X86: Sager NP6110 3630QM 16GB ram, Tyan Thunder 2 2x 300Mhz
Sun: SS2 , LX , SS5 , SS10 , SS20 ,Ultra 1, Ultra 10 , T2000
Mac: Platinum Plus, SE/30

User avatar
Aitch
Posts: 6518
Joined: Wed 04 Apr 2007, 15:57
Location: Chatham, Kent, UK

#32 Post by Aitch »

Hi cb88

Thanks for that, I just thought, having an interest in clustering, that it might be possible to put together a clustered puppy environment, maybe even virtualised, which could be tailored to need for any one job, as I am just about to bring into use, an old server, twin 1.2ghz P3 dell which will probably be sitting near idle for the sort of use I put it to, and I have other p3s with vast scsi arrays, I am working on too, and am curious about what better use I can put all this processing power/storage media to?

Aitch

User avatar
cb88
Posts: 1165
Joined: Mon 29 Jan 2007, 03:12
Location: USA
Contact:

#33 Post by cb88 »

virtualised? no you only use that for testing software installation

to get an improvement in proformance you would need a real cluster and anyway your 1.2ghz system could probably compile puppy in a day don't forget to setup T2 to compile multiple things at the same time in the config or you won't get a proformance improvement
Taking Puppy Linux to the limit of perfection. meanwhile try "puppy pfix=duct_tape" kernel parem eater.
X86: Sager NP6110 3630QM 16GB ram, Tyan Thunder 2 2x 300Mhz
Sun: SS2 , LX , SS5 , SS10 , SS20 ,Ultra 1, Ultra 10 , T2000
Mac: Platinum Plus, SE/30

Post Reply