now that the ATi app is a stock app (and no more only a gift from Gipsel - thanks man!), could we have the source code, owing to your GPL promise?

now that the ATi app is a stock app (and no more only a gift from Gipsel - thanks man!), could we have the source code, owing to your GPL promise?

ok!

I asked it because I want to see your magic!

I asked it because I want to see your magic!

The PTX code can be generated from the source, but I'm not well versed in how exp() would be calculated, what method was used to calculate it in the ATI version above?

The PTX code can be generated from the source, but I'm not well versed in how exp() would be calculated, what method was used to calculate it in the ATI version above?

and after this you still say there's not magic?

Thanks for sharing a bit of your code. It's great to see how you can really optimize apps.
I'm not really used to work on nvidia [and on ATi, too!], but I will ask to some friends (and colleagues) if someone knows how they implement the exp

For a detailed discussion about exponentiation algorithms, it might be interesting to look for papers on Pippenger's exponentiation algorithm. Of course, you won't be able to affect ATI's implementation of the exp() function, but perhaps you can further improve how it is used.

now that they're releasing the new app, maybe it would be useful if you Gipsel release the whole source code of your old app... So that they can implement the best way of doing some calculations!


btw, why CAL is being deprecated in sdk 2.5 (the next sdk)? I cannot fully understand that... Does it mean that all your IL code won't work anymore with newer drivers (11.7+, maybe?) ??

now that they're releasing the new app, maybe it would be useful if you Gipsel release the whole source code of your old app... So that they can implement the best way of doing some calculations!

This won't happen and it won't be useful. He never gave us the source, so I had to rewrite it. I'm pretty sure there's a GPL violation here also. I am not at all happy about it. I've spent a great many hours of my life that could have been spent on something new.

He didn't use any of your code, and on top of it at the beginning Travis code was not GPL if I remember well.

That's not true; there is most definitely code from milkyway@home in there. If it wasn't GPL, and if the code didn't have a license he couldn't use it at all. He even mentions in this thread http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=576 that he's violating it.

Sorry, I don't want to be pedantic, but I don't see anywhere in that thread an admission that his code is stolen from milkyway's code.

Here http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=576&nowrap=true#14639: "I know. "There is still the copyright issue. From a legal standpoint the MW project could demand to stop the distribution." I interpreted this as violating the GPL. In any case, if the original code wasn't GPL, and there wasn't a license, then it couldn't be used. On a second reading I guess he could mean using licenseless code.

I don't know when did you start to work in this project, but the first GPL release benefitted SO MUCH from Andreas Przystawik's work that maybe you're not enough thankful. Original Travis code (not GPL) was so bad that it wasn't even useful to optimize it, he just rewrote it from scratch and for CAL/IL ATi Hardware structures.

I started at the very end of last May. Other people were trying for a while to get the source from him, but after various ignored requests (after various promises for it) it gradually became one of my jobs to rewrite it.

Of course, the only one with a clear right to complain (or file suit :P) is Travis, and he obviously won't do that, since he (now) agrees with the work being done, which I guess counts as having given permission to Physik to do what he's doing.

You got it ;)

It's probably a bit late to answer here. But I want to explain a bit the course of events from my point of view.

To say the truth, the whole thing started indeed in a quite grey area. The source I began with (not only me, there were at least one other guy before) had no license, so I was bound only by copyright issues. Under those terms, the copyright holder could have demanded, that I stop what I was doing. But that simply never happened.

Quite to the contrary, an exchange via PMs and email developed, where I shared code fragments as well as ideas for optimizations. For the Radeon versions, I basically got requests from Travis/Anthony to look into some issues and provide fixes. There was indeed some cooperation going on, which ended up in a publication the MW team and me authored together (and I've written quite a bit of the text in there). Also one shouldn't forget that the MW project itself distributed my applications as stock apps for Radeon GPUs. Claiming afterwards, I was violating their copyrights by compiling those versions sounds not only a bit ridiculous. At this point I had at least a silent consent/tacit approval for my work (which is legally relevant, just to mention it).

But I understand, that Matt wasn't happy about writing a CAL version of the program on his own. And admittedly, I said I will provide the project with the complete source code after a cleanup (it looked really messy as all the changes, fixes, and additions were basically done on the fly). In fact I had already started with that as you can see at the nicely commented CAL kernel I posted at the beginning of this thread. Unfortunately I had not much time back then (I communicated that to Travis and Anthony), as my PhD defense was just two weeks after the just mentioned post. So I was really busy for my actual work. After all, this just started as a small hobby project and was never meant to take so much of my time. And when I had to decide between doing my job (after my defense there were some urgent projects to finish before my leave) as well as finding a new position afterwards on one hand and doing some voluntarily and time consuming stuff for MW@home, I opted for my job.
The last communication I received from the MW project was on the very day before I started with my new position (in another city, so I had to find a new flat, move and all that stuff on top of getting to grip at work). So while I'm sorry I didn't contribute anymore, there was nothing I could realistically could do back then.
And afterwards I simply lost track. It isn't that way that I was still around and simply didn't want to do something. I also lost the connection to my team (P3D, I'm still a member, but I was not around in their forum and didn't contribute there neither).

And frankly speaking, it isn't that way the source code hadn't to be rewritten either way. As said, the host code was quite convoluted and the GPU code written on a really low level (I was thinking about using calpp, the same what is apparently used now, too). So what would have been the alternative to Matt rewriting it as part of his work as a student on the project? Me as an alleged copyright infringer rewriting it just for fun/free of charge for the project? Yeah. But anyway, I would have done it under different personal circumstances/at another time.

By the way, I just had a glimpse over the current source code. For me it looks like Matt was able to "borrow" quite a few ideas from the stuff I shared (some codeparts are simply back-translated from CAL-IL to CALPP, it is virtually identical), same as Travis did before with the CPU versions (and even the nvidia versions). Is that infringing on any license? Of course not, the code came without a license. Is that infringing on any of my copyrights. No, I posted it in the (not explicitly stated) intent that others/the project uses it/learns something from it/get the underlying ideas to improve on the project. Of course they had my implicitly given consent.

thanks a lot, Gipsel!

Is there someone with licensed ICC (Intel C++ Compiler) who could compile CPU app MW_0.91_intel_sse3 into an ATOM-optimized code? It would be 31 percent faster on ATOM, because it is old fashioned "in-order-processor"...

I could really use MW_0.91_intel_SSE3_ATOM , as here is lots of ATOM-based computers here at home... and there are around 30 milion of them still operating around the world... ;)

Thanx for any information on whom to write this request. :)

you can download intel compilers for free for non-commercial and academic uses.