Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
Unix

Journal Journal: Darwin PPC Emulation on Debian Linux with PearPc

Well, I have almost got PearPC running with OpenDarwin on my Debian box... the only downside is that I have not got network activity, which means I have a very basic install.

Though, I am impressed with the efficiency of the PearPC emulation. Though currently no graphics are being drawn (and this appears to be where most of the performance issues in Bochs originates from).

But as I donot have X11 installed on my darwin install on PearPC, and I really do not have any idea what is going on under the hood in PearPC, I really can't give a good comaprison between PearPC and Bochs. It would be comparing apples and oranges.. haha.

I did notice on the PearPC website, that it can emulate[?] two kinds of CPU. A generic PPC and an x86 PPC [?] where the x86 is faster (at about 1/15th the speed of the host CPU) where the generic cpu is at 1/500th the speed. But the x86 is only available on x86 systems. However, if the CPU I am emulating translates to x86 instructions... and Darwin can be installe don an x86 machine, then is it running as a PPC emulation at all? I suppose it is, however it is running terribly fast for an emulation of a completely different architechture.

If it is an accurate emulation, then what is PearPC doing so well that Bochs could build from? Or is it that PearPC is concerned solely in producing a platform that can run some Mac variants (Some PPC Linux distros also run reportably well, which would contradict this hypothesis. Though not many PPC distros including OSX run well, which may indicate the limited testing resources of the development team)... Or is Bochs too concerned with accurate emulation of all x86 architectures?

 

Unix

Journal Journal: Frenzy on Bochs on Debian

I was reading this /. article http://bsd.slashdot.org/article.pl?sid=05/02/28/1230206&tid=190&tid=7/ and wanted to make a post, but it had been archived.

Anyway, I have recently been playing with Bochs on my new Debian PC. I was originally interested in
a way of running Windows applications on my linux machine. And had already had limited success with Wine. However the ability to emulate an entire machine is attractive. Especially since the Bochs project does such an amazing job of hardware emulation... It would be great to get an old version of linux or BSD running in it and actually start playing with OS code (kernel sabotage is what another /.er called it...).

Again I have digressed, the point is I have the latest version of Frenzy. And it does have a huge range of applications on the disk. Including an XServer. However, I have 3 machines at home (1 laptop, 1 store build PC with common midrange hardware - of its period, and 1 home built PC with high end hardware)... and I have not been able to get Frenzy to boot on any of these machines.

It did boot on Bochs. I couldn't get the XServer to start, but this was possibly because I hadn't compiled in VBE support to Bochs on the build I tried to start Frenzy with.

Now its possibly particularly obvious, but it struck me as odd that a large portion of the messages, informational, warning and error level are in Russian. As the Frenzy project is russian. However it does give you an option to use english at the start... I am not sure if this applies to all messages or just some? I have compiled in VBE now, so I should find out how the Xserver runs!

On a different topic, Bochs collects substantial abuse from people who think that it is possible for an emulator to emulate hardware and run just as quickly. Especially those MAC users who have been converted to MAC OSX (as opposed to the linux and BSD users who are attracted to the pretty colours running on a system they know and love). Many of these users expect Bochs (and variants of Bochs, based on Bochs, such as OpenOSX WinTel) to work at the same speed as the machine they are running.

Emulated speed being equivalent to hardware speed simply is not possible. However optimisations and forwarding of instructions are possible. And this is where running Bochs on a PC to emulate a PC is beneficial, as it is like most of the instructions can be forwarded to the host machine. Emulation of a PC on a PPC is going to make it difficult to do this.

However, when you limit the variation of the instructions you are going to handle, then you simplify this task greatly. For example, on a MAC OSX machine, it is likely that virtual PC doesn't pretend to be a PC... it would make much more sense to provide a native PPC library that supplies the API that a Microsoft Windows based application would expect if running in MS Windows on a PC. This way you don't need to emulate a hardware layer, all you need to do is supply an interface, which can be backed with native code. Similar to Wine. Of course Microsoft have simplified this further, as they have versions of virtual PC to match specific versions of windows.

I have actually got very respectable performance out of Bochs by enabliing the optimisations when I compiled my own version. I also enabled the X11 hack, so it didn't burn CPU cycles when it wasn't doing anything. However I wouldn't know how many of these optimisations would be of any use when you were running Bochs on a PPC machine?

Now I am keen to download PearPC and see what the performance is like running that on my Debian Linux box!

Please excuse the spelling, I studied computer science.

User Journal

Journal Journal: i-mate smartphone 3

I have scored this smartphone through my employer... one downside is that it runs windows. However it did look like the best of the options I was given.... And M$ does have a free development environment for the phone.

What I would really like to do is investigate developing mozilla for smartphone. I see that there is a minimo project, but this is only for linux and GTK. Is it possible to reduce the footprint of GTK for windows, for use with smartphones? Is it possible to take the Linux Minimo project and port it to Windows smartphone 2003?

Update Wednesday, 13 April 2005
Since I posted this journal item, I discovered that there was a port of the minimo browser to windows ce. Check out the google group here.
User Journal

Journal Journal: J-Boss UCL

I have been having issues with a customised preferences library that my company has developed... in particular, using it as an fresh resource.

It seems that the JBoss UCL is the problem and it seems that I am not the only person in the world that has run into this issue.
http://www.jboss.org/?module=bb&op=viewtopic&t=41379
http://jira.jboss.com/jira/browse/JBAS-630

Though this doesn't seem to be the only problem... It also seems to be impossible to get my preferences library to give my the values I want.. or even admit to hold any... even though I can see them being read.
User Journal

Journal Journal: Alpha FooJournal

foobar journal... if you expect something intelligent, then there is nothing to see here.

Slashdot Top Deals

Getting the job done is no excuse for not following the rules. Corollary: Following the rules will not get the job done.

Working...