shouldn't the topic read "TI-85 emulator for Psion5 released"?

I was just thinking the same thing. Wouldn't it be pointless to have an 85 emulator on the 85?

Maybe it is for the 85 and it's written by the same guy who wrote Virtual Virtual TI.

For those of you who don't know, Virtual Virtual TI is a TI-89 program that emulates Virtual TI. It was written by Richard Short after all his friends complained that they couldn't run a calc emulator on their calculators.

sounds cool, where can I find it?

Someone must have changed it for me.... What on earth did the title *used* to say? :)

It used to say "TI-85 Emulator for TI-85 Released"

That would have been a great idea: I've always wanted to be able to play TI-85 games on the TI-85.

hahahahaha

Would be nice to have an emulator for PALM os devices. There are certainly la lot of those out in there.

This is really quite an achievement! I don't actually have a Psion 5 so I can't test the programs' speed, but I do know that the Psion hardware (including the processor) is quite different from that of the TI-85, so the emulator must be quite sophisticated if it emulates it well. The Psion's quite fast, so I'd imagine it could achieve close to TI-85 speeds, especially as it's written in the fast C++ language. And the screenshots look really good, too...

Erm, just to point out a slight error in the article title - shouldn't it be "TI-85 Emulator for the Psion 5 Released" ?

YES! finally I can play ti-85 games on my ti-85! i've been wait..oh wait.. on a psion 5? that's cool 'cept i droped my psion 5 and the screen cracked. anyone wanna buy a working (though needing new screen) psion series 5? no really though.. the psion kicks butt... it's fun to play with and much better than a palm or dunce.. err make that wince device....

Well, This is probably really super nifty, but why wouldja wanna lug about one of these (click URL above) when the 85 is a lot easier to fit in your pocket? (:^)

it's not much bigger than the 85 when it's folded up, and plus, it's got a nifty little keyboard on it. it's probably better than a TI-92 in some respects, i.e. thinner, lighter.....

well, it's probably plagued by the Windows Virus

The Psion handheld computers run their own proprietary OS, so they are fast, relatively crash free, easy to use, and very efficient, basically everything Windows CE/Handheld PC is not. The Psion computers also predated the Palm and Windows CE/Handheld PC by many years.

Um...Nathan, you must not be a multi-platform programmer (or a programmer for that matter). C++ is fast as hell, especially on handheld devices. I'm a Palm OS programmer and I can tell you, althought it may not be fast on the TI (they're calculators, for heaven's sake), it sure is speedy on handheld devices. TIs are devices that aren't capable of high-level languages such as C++. They are made for low-level assembly languages and BASIC. So learn something before you go around making assumptions as to the speed of the language.

Frequently, when emulators are written "for speed," the core of the virtual microprocessor is written in assembler. Now obviously for a processor like the Z80 which can be emulated easily by modern processors this type of effort is not required, but it is commonly important when you seek to emulate more comparable systems. For example, David Ellsworth wrote an Apple II emulator that ran at full speed on a 486 sx. The only way this was possible was using assembly.

I have an Apple ][ emulator that runs at full speed on a 12MHz 286. I know, I had used the thing back when all I had was a 286 (now I have a Celeron 500). The thing was written entirely in assembly and came with source. It is a DOS program (of course, if it ran on a 286). I have yet to find an Apple ][ emulator that is that efficient. It seems that the CPU emulation part of the emulator was written by another person, and the Apple ][ emulation part was added to it. I don't know how the author of the CPU emulator got it to run that fast, but somehow he did. It was written back in the late 80s for MASM. E-mail me if you want me to send it to you.

TI calculators are just as "capable" of running programs written in C++ as other computing devices. In case you didn't know, the C++ code is translated into low-level assembly language code by the compiler. C++ programs will _always_ be slower than well-written assembly language programs, because the compiler isn't as good at optimizing as a real human. If a C++ program runs fast on a handheld computer, it's because the computer is fast, not the C++. Make sure you know what you're talking about before you go around telling people to "learn something."

hey dude, actually modern compilers are (dunno if we can use html here but we're gonna try it, sorry if it doesn't work but we're going for italicizing) <i>extremeley</i> good at compiling efficient code. If you write your code correctly, they can use MMX and 3DNow! extensions even, with minimal effort on your part. However, the c(++) compilers for calcs are generally not as good as for PCs / Modern systems.

Actually, the TI-GCC compiler is the *same* compiler (GCC) which is the standard compiler on Linux, and also the basis of the DJGPP compiler for DOS and CygWin. However, it does use a slightly older version of GCC than the most current one.

that doesn't make sense. Optimizations are different for different processors. And I don't think the computer I'm running right now is a z80 or a 68000 since its running windoze. So it can't be the *same* one, can it? Or are the GCC compilers like visual C++, where it can compile for both 68K and x86?

GCC is capable of generating code for many CPUs, including x86, 68K, PowerPC, SPARC, DEC Alpha, and many others (but not the Z80). The compilers for different targets are of course slightly different in that the final output is for a different processor. However, most optimizations really are the same on different processors. GCC first converts the code into an internal format and optimizes that, then generates assembly code for different processors. Rather than having different code generation routines for each processor, the characteristics of each processor are defined by symbolic expressions describing each instruction. There is of course some machine-dependence involved (different optimization strategies for different groups of processors) but GCC does not have completely (or even largely) separate routines for each processor. See GCC documentation for more details.

By the way, I wonder where you got the idea that Visual C++ generates code for 68K processors.

I could be wrong, but somewhere there is some sort of #define or something that lets you generate for the Mac, which would include both 68K class and powerpc class computers, right?
I'm almost 100% sure of it. I saw it many moons ago, so I don't remember where. I don't use any other compilers though, so it had to have been vcpp

not the Z80 ? I though that the Z80 was fully compatible with intel 8080....maybe this one is not a x86 family.... ok.

Yes, modern compilers are good at optimizing. But they can not come close to an actual human. Human-written assembly-language code is faster because humans can optimize better than compilers. Don't get me wrong, though, high-level languages are very useful. Without them, most of today's software couldn't exist, because programming in assembly language is extremely impractical for large- to medium-sized programs.

P.S. The only C++ compiler for calculators that I know of is a port of the GNU C++ compiler, the same compiler used for many high-powered modern computer systems and PCs. (however, I think the port has been discontinued... I can't find the page about it anymore :( ). TI-GCC is also a port of the GNU compiler, but only the C part.

I have to wonder if you've ever actually examined the output of a modern optimizing compiler for quality. My best guess would be that you haven't. However, as someone who actually has seen lots of GCC code, I can say that it is generally very good, though it may be possible to improve it in some cases (human-written code is rarely perfect either). If you don't believe me, you can always test it yourself, and compare it to the work of whichever assembly programmer you think is best. TI-GCC does unfortunately suffer the disadvantage of stack-based parameter passing, which results in some performance loss, but not much if functions are large. Also, this can be solved by configuration changes.

For more information on this, you can click the web page link above, to see what Linux's Assembly HOWTO has to say about the issue (of course, it should be noted that on the calculator, one is always programming for a particular hardware setup).

Well, you've got me there. I've never personally examined the output of a compiler. Maybe I should :) What are those configuration changes to TI-GCC you mentioned? I'm interested in anything that might help TI-GCC be more efficient.

Thank you, I AM a programmer, and I am very aware of the use and function of C++. It's just not the kind of thing one typically writes emulator CPU cores in. I stand by my original statement, and thanks to Andy for defending me.

David Ellsworth's ApplePC emulator had an ASM CPU core, and it ran nearly full-speed on a 386. Not bad at all. It wouldn't have been possible in C++, I'm sure.

Hi,

It surprises me how much interest this emulator has created! For those of you that don't know, the Psion 5 is a handheld device sold primarily in the UK. I agree with comments that say the emulator would be a lot faster written in assembler, but Psion only release a C++ software development kit. I don't think there's any way of writing in assembler for the Psion.

Also, I'm not a software programmer (although I work in the hardware industry, which isn't all that different), and I wanted to learn some C++ skills. Even just using the Psion SDK was difficult because of the weird techniques used - see www.epocworld.com.

I'd estimate the emulator runs at 50%+ speed on a Psion 5, although I'd be interested to hear how it fairs on a Psion 5MX, which has a 36 MHz ARM processor rather than the 18 MHz one on a Psion 5.

If anyone wants to port the program to other handheld devices, e.g. Palm, I may consider releasing the source code. However, the C++ "wrapper" around the emulator core is very Psion-specific.

Glad this has created so much discussion!

Regards,

Daniel Rigby

Oops, forgot to say - there is another way of programming on the Psion, and that is using a language called OPL. This is an interpreted language like BASIC (i.e. slow), so that's why I wrote "written in C++ for speed"!

Daniel

Hey, I just wanted to say how cool it was that you wrote this. And i'm sorry people are ragging on you for using C++. I would be willing to bet that most people couldn't write one that even ran at 50% in any language. Unfortunately I can't use it, on account of my Psion 5 doesn't exist;-) I've never even heard of one before now. But still, i'm impressed

Would it be possible to take a look at the code? I would like to see if I can spot anymore was of squeezing more performance out of it =)

regards

boris

Daniel,
your emulator looks great so I´ve downloaded it to my 5mx with a rom image, but when I try to run it I get a pop-up saying <program closed, reason code: kern-exec, reason number 0>
Maybe it´s not compatible with ER5 for some reason...I don´t know
If you have any idea and want someone to test on a 5mx, just let me know

Well,...I feel so stupid now... I forgot to rename the rom image...I´ve done it now and it works fine. So I´ll try it tomorrow as it´s 1:20 a.m. now and I´m quite tired and ashamed.
Regards,
rrovaz

Hi,

Is it now working? What's the speed like on a 5MX?

Daniel

The speed on my 5mx is actually pretty fast. It's not full speed, though. At least, I don't think it is. Ah drat. Now I'm going to have to get my 85 and compare speeds. Even if it's not as fast, it's still fast enough. Woohoo, I just got this 5mx yesterday and it's already useful :-)

Chris the MathFreak

There are many emulators written in C++, and some of them run quite fast. I use Basilisk II (a Macintosh emulator, written in C/C++) on my Celeron 500 and it runs about as fast as most real 68040 based Macintosh computers. I estimate that I have the equivalent of a 40MHz (or faster) 68040 using that emulator. The Texas Instruments TI-81 emulator for Mac OS, written in C/C++ as far as I know, runs quite a bit faster (about least twice as fast) than a real TI-81 on my PowerBook 140 (16MHz 68030 CPU). Although I don't have a Psion 5 to test the TI-85 emulator on, I must say that achieving the equivalent of a 3MHz Z80 on an 18MHz CPU is quite a great achievement.

Oh great, not the Windows Virus, but the deadly Mac-O-Death