Index
> Main > fasm-based code generation libraryGoto page Previous 1, 2, 3 Next |
| Author |
|
|
Tomasz Grysztar 09 Aug 2005, 18:26
Yes, I plan the library to be quite modular and allow things like use assembler without preprocessor etc. Perhaps even customizable formatter (with callbacks to handle the external symbols, relocations, etc.).
|
|||
09 Aug 2005, 18:26 |
|
|
f0dder 09 Aug 2005, 18:32
That sounds very good - I guess you'll (re?)implement fasm in terms of this library? Whatever you do, good work
 |
|||
|
09 Aug 2005, 18:32 |
|
|
Tomasz Grysztar 09 Aug 2005, 18:36
Another interesting thing would be to reimplement fasm on the x86-64 architecture, taking advantage of the SSE et al. - but I have to wait till I get any 64-bit machine for such development.
|
|||
|
09 Aug 2005, 18:36 |
|
|
THEWizardGenius 09 Aug 2005, 19:10
Quote:
Agreed
Holy Wars are for USENET... And I do think the GNU website is a little too dogmatic on the GPL, like it's the best thing ever. Open Source is great, but GPL causes problems sometimes. Still, my OS will be GPL'd. Tomasz: It would be great to have a 64-bit part too. But be sure to keep a 32-bit part, too, since most of us don't have a 64-bit computer yet...  |
|||
|
09 Aug 2005, 19:10 |
|
|
Tomasz Grysztar 09 Aug 2005, 19:19
Quote: since most of us don't have a 64-bit computer yet... Just like me.  |
|||
|
09 Aug 2005, 19:19 |
|
|
Raedwulf 09 Aug 2005, 19:43
Yeah nor do I...
but I'm hoping to get one next year  |
|||
|
09 Aug 2005, 19:43 |
|
|
THEWizardGenius 09 Aug 2005, 20:31
I'm hoping to get one either the year after next or the year after that
it'll be a LONG time, since I still have to make the upgrade from 486 to Pentium 4 (sometime this year, hopefully). I would get a 64-bit computer, but I'm making a 32-bit OS, so that wouldn't work too well... |
|||
|
09 Aug 2005, 20:31 |
|
|
f0dder 10 Aug 2005, 01:33
I don't think a 64bit port of fasm itself is too important, as the current 64bit systems have fine "legacy" 32bit modes. It would probably be a lot of work(?), without too much gain. But it might be a fun project nonetheless
Quote:
Latest model P4's are also 64bit. And if you haven't got one yet, I'd say go for an AMD64 instead, it's a better price/performance ratio as things are now (not to mention less power consumption and less heat). Both processors work quite fine in 32bit mode as well, so consider it a future investment. XP64 does seem to run pretty nicely. |
|||
|
10 Aug 2005, 01:33 |
|
|
Raedwulf 10 Aug 2005, 06:24
I agree with f0dder ...i see most of the benchmarks going to amd's way
.
But you could ge a P3 and notice the speed difference compared to your 486 ...no offense meant.
So any old P4 will feel a lot different
32bit or 64bit....I'll go 64bit because 1.) It has new instruction sets. 2.) You need to know the compatibility...64bit runs 32 bit most f the time but just in case - but anyways did 16bit programs have much difficulty with 32bit machines? Not really....it was only the operating system ie. winxp which got some 16bit programs not to work. So I'd go with f0dders suggestion |
|||
|
10 Aug 2005, 06:24 |
|
|
Raedwulf 10 Aug 2005, 06:37
66Mhz -> approx. 2ghz-3ghz
30x to 45x faster computer....WOW! Your next computer after this new one. 3ghz * 45 = 135ghz ......hmmm j/k Besides computer speed isn't really about the clock speed anymore
It's about latencies etc. so this is purely just for fun . You will notice that P4 3.0ghz can perform less quick than amd's that run at 2ghz.
Different architecture |
|||
|
10 Aug 2005, 06:37 |
|
|
Tomasz Grysztar 10 Aug 2005, 06:58
Quote: I don't think a 64bit port of fasm itself is too important, as the current 64bit systems have fine "legacy" 32bit modes. But some features could be easily added when making such port. Like 128-bit calculations (as currently they are 64-bit ones in 32-bit) etc. |
|||
|
10 Aug 2005, 06:58 |
|
|
Tommy 10 Aug 2005, 10:39
I think it would be great to see a 64-bit port...
128-bit calculations, even faster compilation (that's not a problem with fasm though, but it would be cool to see how fast it can get).. :p
One question Tomasz: will you implement "float" calculations in the future? |
|||
|
10 Aug 2005, 10:39 |
|
|
Tomasz Grysztar 10 Aug 2005, 10:56
Didn't plan it - the only solution I would consider would be to do high-precision calculations in the style of current string to floating point conversion routine, so the results would be always the same (independent of environment), and this would need quite a complicated set of routines. Also there would be problem with distinguishing where to use floating point and where integer operations - currently everything is simple and clear, floating point values are just converted to binary, and all the calculations are on the binary integers. Thus you can even implement floating point operations yourself, using the existing features.
|
|||
|
10 Aug 2005, 10:56 |
|
|
f0dder 10 Aug 2005, 11:41
Quote:
That's true... guess it might be useful to somebody. Of course this could be implemented on 32bit architectures as well, but requires more work
Quote:
That does seem to require qutie some code - just look at how much work it requires in NASM. |
|||
|
10 Aug 2005, 11:41 |
|
|
Tomasz Grysztar 10 Aug 2005, 11:55
Quote: That does seem to require qutie some code - just look at how much work it requires in NASM. AFAIK, NASM does just as much as fasm does - converts the decimal floating point values to their binary representation, and doesn't allow any calculations. |
|||
|
10 Aug 2005, 11:55 |
|
|
Reverend 10 Aug 2005, 22:04
Tomasz Grysztar wrote: But some features could be easily added when making such port. Like 128-bit calculations (as currently they are 64-bit ones in 32-bit) etc. |
|||
|
10 Aug 2005, 22:04 |
|
|
Tomasz Grysztar 10 Aug 2005, 22:09
I meant implementing 128-bit calculations in 64-bit mode analogously to the current implementation of 64-bit ones in 32-bit mode. Without writing much new (and perhaps complex) code.
BTW, bignum addition is an easy excercise to write, it's the division with remainder that may be a bit of challenge (at least if you want to have it fast). I would recommend the classic AOP by Knuth, there are some good solutions nicely presented. |
|||
|
10 Aug 2005, 22:09 |
|
|
Tomasz Grysztar 13 Aug 2005, 18:25
After contacting the support at PayPal I've found out it's hard to tell when they allow me to receive money. For now I found the other such service - the moneybookers.com. Do you think it's popular enough I would use it for the purpose of flat assembler donations?
|
|||
|
13 Aug 2005, 18:25 |
|
|
THEWizardGenius 14 Aug 2005, 18:12
I doubt it, but I could be wrong. PayPal seems like the most popular, but if you can't get it there, you may have to use something less popular...
|
|||
|
14 Aug 2005, 18:12 |
|
| Goto page Previous 1, 2, 3 Next < Last Thread | Next Thread > |
Forum Rules:
|