There's been so many discussions of so many different methods, it's hard to keep track. So, one big question:

[Kernel]

JamesH said:


Yes, you could run an emulator on the Raspi to run x86 code. It already works - see Dosbox. No need to write another emulator.

However, with the figures above of about 20% of native, and given that the Raspi is about the equivalent of a 300MHz Pentium 2, it woudl be approximately the speed of a 66Mhz Pentium. That's about the year 1993.

Not sure the effort would be worth it to emulate a nearly 20 year old machine.



shirro said:


Joe Schmoe said:


In fact, given enough time and patience, we should be able to run any version of Windoze on the Pi.


Well, no, because we will die before Windows 7 would boot up. Since we would need to swap off disk to supply memory needs there is a possibility civilisation will have crumbled and been replaced by apes before a login prompt appears and we would be racing the exhaustion of the Sun's fuel to get to a desktop. If aliens should find our experiment and wonder what Adobe Photoshop does the race would be on to find out before all the stars start to go out. That is assuming a $35 computer made in China can last that long.



Hahahaha - both great posts - having a good laugh now

[abishur]

Joe Schmoe said:



I take your point about DosBox, though.  BTW, why did you go with DosBox instead of QEMU for your experiments?



I'm more familiar with dosbox and I was already using it to do other things, I wanted to play chip's challenge and it seemed like the simplest method for getting it all to work

[Joe Schmoe]

OK - sounds good to me.

[tritonium]

JamesH said


However, with the figures above of about 20% of native, and given that the Raspi is about the equivalent of a 300MHz Pentium 2, it woudl be approximately the speed of a 66Mhz Pentium. That's about the year 1993.


A 486 sx25 has all the power I need (with the Pi's graphics)

I want an Arduino on steroids.....(running Basic would be good like Duinomite)

http://www.olimex.com/dev/duin.....-mini.html

Dave H

[abishur]

Joe Schmoe said:


OK - sounds good to me.


Does qemu tend to run faster than dosbox or is it more a 6 to 1 half doezen the other?

[Joe Schmoe]

I think that QEMU could very well run faster, because I think one of the purposes of DosBox is specifically to run slow.  I.e., to make old DOS games playable on modern hardware.  I know that DosBox is significantly slower running DOS games than is, say, VMWare (and/or, presumably, VirtualBox).  But then again, this argument may not be compelling, since VMWare runs things "almost natively" when running x86 under x86.  It would be different if you're running x86 under ARM (which VMWare can't do).

Also, QEMU is more "hip" than DosBox, so you may get more mindshare using it than using DosBox (if you see what I mean...).

[bobc]

tritonium said:


JamesH said


However, with the figures above of about 20% of native, and given that the Raspi is about the equivalent of a 300MHz Pentium 2, it woudl be approximately the speed of a 66Mhz Pentium. That's about the year 1993.


A 486 sx25 has all the power I need (with the Pi's graphics)

I want an Arduino on steroids.....(running Basic would be good like Duinomite)

http://www.olimex.com/dev/duin.....-mini.html


Does it have to be DMBasic? I am sure there is already a Basic for Linux which runs on the Pi.

Sometime I might look at doing a native port of DMBasic to the Pi. Unfortunately the license got a bit messy with the original MMBasic having an open source license which was then withdrawn.

[jamesh]

tritonium said:


JamesH said


However, with the figures above of about 20% of native, and given that the Raspi is about the equivalent of a 300MHz Pentium 2, it woudl be approximately the speed of a 66Mhz Pentium. That's about the year 1993.


A 486 sx25 has all the power I need (with the Pi's graphics)

I want an Arduino on steroids.....(running Basic would be good like Duinomite)

http://www.olimex.com/dev/duin.....-mini.html

Dave H





The real problem then is the use of high performance graphics. Even if 486 levels of performance are OK for the CPU, there is no link from emulators to the acceleration libraries, so you cannot take advantage of the GPU power. Also, code written in 1993 for that sort of machine would not be written to use acceleration. I think even OpenGL was only a glint in Khronos' eye back then.

[AndrewS]

JamesH said:


I think even OpenGL was only a glint in Khronos' eye back then.


2D sprites FTW!

[nimitz]

JamesH said:


[...]and given that the Raspi is about the equivalent of a 300MHz Pentium 2[...]


I think this is the second time I read a similar statement on these forums, I just had to register and call you out on that one.  Granted, I do not have my Pi yet so I can"t run actual tests. But making an educated guess with my decent knowledge of assembly, cpu architectures and embedded systems. I would expect the Pi to perform on the level of a ~1000mhz Pentium III, considering the differing architectures and their cycle efficiency, cache size, ram speed, etc…

[Jim Manley]

JamesH said:


Also, code written in 1993 for that sort of machine would not be written to use acceleration. I think even OpenGL was only a glint in Khronos' eye back then.


In 1992, SGI led the creation of the OpenGL architectural review board (OpenGL ARB), the group of companies that would maintain and expand the OpenGL specification for years to come. On December 17, 1997, Microsoft and SGI initiated the Fahrenheit project, a joint effort with the goal of unifying the OpenGL and Direct3D interfaces (and adding a scene-graph API), and in 1998, Hewlett-Packard joined the project.  On July 31, 2006, it was announced at SIGGRAPH that control of the OpenGL specification was being passed to the Khronos group.

One day in 1994, when I was putting gas/petrol in my car at the Shell station on North Shoreline Boulevard in Mountain View, which was the closest to Silicon Graphics' headquarters and other office buildings, I happened to notice the license plate on a yellow Ferrari being fueled by its driver on the other side of the pumps.  It read "OpenGL", and when I looked at the driver, I realized it was none other than SGI founder and CEO Jim Clark.  I guess it was his way of promoting OpenGL, although it is possible that some licensing fees had been paid to SGI to integrate with the proprietary precursor SGI Iris GL APIs, and that's where he "invested" his share of the fees.

Despite people confusing "open-source" with "free", this is an example of how one can make quite a nice living (and then some) developing and supporting open-source software and standards.  Many people have no idea that virtually every standard used in our computing devices today produces at least a small amount of revenue from the sale of each device for its inventors/assignees, e.g., USB (Intel), Firewire (Apple), WiFi (Commonwealth Scientific and Industrial Research Organisation in Canberra, Australia), etc.