Pianoteq on iPad ? (Page 2 of 2)

I really did try to avoid being insulting, certainly did not intend to, truly sorry if I was. I even am not sure that I am disagreeing with you.

My point was, however, is that it is perfectly reasonable to ask for an ios version of pianoteq, and you simply can not expect your average customer like me to have an exact idea of the implications of ARM architecture for modeling pianos with algorithms that we do not even know. To amateurs like me the newest generations of ios devices seem pretty powerful, but you may be right, of course.

I think the definitive answer is with Modartt, and since we do not have an ios version yet, I guess that question is already answered; it would perhaps help if the answer was a bit less implied.

The A7 and A8 are indeed impressive processors. I just looked at a few benchmarks, and the A8 seems to be roughly comparable to a Celeron G530. For a mobile device this is indeed very impressive, I'll gladly accept that Apple seems to have made great strides indeed. The performance of an A8 device should be about 50% higher than e.g. that of the Samsung Galaxy S4 and a few percent higher still than the S5. If this raw power were to be used absolutely perfectly, it should give enough 'juice' for a polyphony of 32, maybe even 64 or higher.

I am still skeptical about the porting effort, however. And Pianoteq is almost definitely not written in C but in C++, which is a completely different beast. Yes, it should be possible to interface Objective C and C++ via a plain C interface, but this feat is a tricky one at best. As I said: all quite possible, but very probably still a lot of work, which was my actual point.

And my last point still stands. Yes of course, Modartt can set any price they want. But people are simply not used to things costing hundreds of euros on iOS and/or Android. So technically possible: yes. But... are there really that many people out there who would buy an iOS Pianoteq for e.g. EUR 200 who have not bought it already for MacOS or Windows or Linux? I.e., how big would the number of customers willing to pay full-price really be? I am still most skeptical about this point. And the iOS version would in theory have to be even more expensive than the x86 versions, since those are sold directly without Apple taking their (rather significant) due.

How does A7/A8 CPU fare with SSE2 and/or AVX optimizations that Pianoteq most definitely uses?

Those reviews don't really answer my question.

Reason being - if a program depends very much on SSE2/AVX registers that a CPU offers (Intel/AMD), and the ARM CPU doesn't have those registers, it's definitely NOT the case of a simple port, because the whole DSP part would have to be recoded to account for the differences in the CPU architecture. Which is something I don't think Modartt would be willing to do, since it would mean maintaining two separate codebases.

GPUs aren't really THAT good at serial processing (which audio is in 99.99% of cases all about). They are made for parallel processing, which is great for graphics and some other calculations on many vectors at once. Audio is processed in blocks. So, even though there is a lot of power in those GPUs, the critical thing that makes them not so usable for audio purposes is the latency. It's ok for FFT stuff like convolution (which can benefit from parallel processing a lot), but not really for calculating a complex physical model in realtime (meaning as little latency as possible, and currently there's too much latency with GPGPU as far as audio is concerned).

You may disagree, but polyphony does not equal parallel processing in case of a physical model, so you are wrong on this account.

Not to go into too much detail, but I am studying computer science and have a particular interest in DSP, so I have informed myself about GPGPU quite extensively. It is not a good solution for realtime playable instrument, period. Otherwise we would have already seen some actual plugin synthesizers utilizing GPGPU by now - the technology itself has been available for quite a while now. But, nope, didn't happen. Why? Because latency issues make it unplayable in realtime. Case closed.

Yes. But they almost certainly do not keep a revenue share of 30% either (plus taxes). Either way, the main problem as you have already noted is the relation of expected revenue to needed investment, and that is something we can only speculate about. I have my doubts, you apparently haven't. But revenue on mobile devices is not all it's cracked up to be if you listen to anyone actually selling software on those platforms. And the death of the personal computer has been prophesied many times already, and yet... e.g. Apple still makes and sells Macs.

As a developer myself, I know a thing or two about cross-platform development. If Modartt know what they are doing, and I suppose they do, and assuming a codebase of at least a few hundred thousand lines of code for Pianoteq (more likely much more)... if they use proper design and cross-platform libraries, the Windows, Mac and Linux versions share probably about 99% of their codebase, meaning a difference of a few thousand lines for the low-level stuff. So maintaining those three ports does not mean much more work than maintaining only one.

In the iOS case, were this one percent even to rise to maybe 5 percent, this would mean that the iOS port is more work to maintain than the other three combined. And I guess we would be talking of 10, maybe 20 or even 30 percent of code that has to be ported and maintained specifically for iOS and the ARM architecture. If it even works. As I said before, the A8 is only powerful enough if you manage to completely use its potential... and as EvilDragon correctly points out, there's only so much to gain by parallel processing in solving this particular problem.

So that would still mean a decent amount of rewriting happening, no?

I think if you ask yourself the reciprocal question (i.e., why would virtually every reseller of musical instruments and software offer this product in their catalog?), then you'll have your answer.

OK, but I find it really hard to believe that there are that many dealers willing to chase 10-12% margins.  Even 20% would raise my eyebrows for a specialty item like this.

I was looking at the registered user list in the forum.  There are less than 2500 users registered over a 7 year period.  I wonder how that number correlates to number of units sold.