From: Yao Konan on 2 Sep 2006 10:36 Hi Michael, Michael Kuyumcu a écrit : > Hi, > > o the taylor series development didn't yield a result after 253 > minutes, when I cancelled. So i think that not only it does use the same algorithm of the TI92+(which returns a memory error after 50 min) but this alogrithm is useless on this kind of problem if the TI-NSpire fails with both much greater speed and much more memory. > o randmat(60,60) works Sorry,i meant approx(randmat(60,60)) or randmat(120,120) which will surely exceeds the 32 KB limit. > o any integer may not be longer than 614 digits. I find that limit > downright silly. > o 300! fails (the nspire responds with "300!"), and 450! fails,too. > That's very limited. It is exactly the same limitations as all calculators of the TI92 series. > o the (seq...) constructs yields an error message: "Too few arguments" Curious i will try to rewrite it: seq(seq(1/(sqrt(2*i+1)+I*sqrt(2*j+1)),i,1,4),j,1,4)^-1, But i think that it is better to use the following one: seq(seq(1/(sqrt(2*i+1)+I*sqrt(2*j+1)),i,1,3),j,1,3)^-1 as the former has taken over an hour on my TI92+ without completing. This problem underlines the difficulty that the TI92 series has with matrix incorporating both complex and square root expressions. > What do you mean by "managing variables" within a document? You can > define them, undefine them, edit them, redefine them, delete them, view > their contents. You can have as many notes, graphs, calculators and > spreadsheets per document as you like. I wondered if there was a variable manager availlable for each document. However what you wrote is very interesting. So it is possible to have as many document as possible, And within each document it is possible to have as many problems as one want and one can add up to 4 applications to a problem,right ? Best regards. > Regards, > Michael Kuyumcu > > > > > Yao Konan schrieb: > > > Ok i have just a few tests which take a long time or fail on the the > > TI92 series: > > 2 speed tests: > > --------------------------- > > *taylor(tan(tanh(x))-tanh(tan(x)),x,7) > > *(seq(seq(1/(sqrt(2*i+1)+I*sqrt(2*j+1)),i,1,4),j,1,4))^-1 > > > > A memory test: > > ---------------------- > > *randmat(60,60),to see if it can handle algebraic objects bigger than > > 32 KB > > > > Limit of the exact mode for integer: > > --------------------------------------------------- > > *300! and 450! > > > > I is the imaginary number > > sqrt stand for the square root function. > > > > > > Btw,you said that one can only save documents,does that mean that one > > can not manage variables and functions within a document ? > > Does that also mean that you can save only one note session or > > spreadsheet or plotting session per document ? > > > > Thanks. > > > > Michael Kuyumcu a écrit : > > > > > No speed tests here, sorry. I think I'm not interested in them. I want > > > to see if the device meets the curricular requirements for my math > > > classes. But if you post a few representative timing challenges here, I > > > will be glad to have them executed :-) > > > > > > Regards, > > > Michael Kuyumcu > > > > > > From: Yao Konan on 2 Sep 2006 15:08 > Michael Kuyumcu wrote:> > > Interesting: fib(10000) produces an overflow, even in Exact mode, > > Rather disappointing. It seems to have maintained the limitations of > the TI89/v200 - it's probably an exact recompile of the core OS used in > those calculators. I think that it is not an exact recompile of the core OS but certainly a recompile of the PC version because it seems that some limitations have been removed such as the 8 characters limit for variables names and some of the memory limitations. However it is more than sure that some limitations remain. > > while the CAS seems to be capable of handling multiple-precision > > numbers otherwise with no problems.fib(1000) works, and does so > > within 2 seconds. > > Takes 0.05 seconds on the HP50G. It takes 0.05 seconds with a C program ? Because i doubt that it would take 0.05 s in RPL. > It would be cool if the NSpire could > be coded in C or C++ for sure, but I believe the OS is too limited to > realize the full potential of such programming. We shall see. > Regards > Steen From: Steen Schmidt on 2 Sep 2006 16:20 Yao Konan wrote: > However it is more than sure that some limitations remain. Most of the limitations it seems. Directory structure for one, as well as all the math limitations. > > Takes 0.05 seconds on the HP50G. > > It takes 0.05 seconds with a C program ? > Because i doubt that it would take 0.05 s in RPL. The C program probably executes in 0.01 seconds or less - the rest is the RPL wrapper. The FIBNUM function executes out of the emulator completely transparent to the user - i.e. it works as any other function in the Saturn environment. But for your analogy to work we assume the NSpire emulates the 68k like the HP49G+/50G emulates the Saturn? Since I expect the OS to be recompiled for the processor used in the NSpire, I'd expected the performance to be nearer what we achieve in C on the HPs. I'd have expected something like FIBNUM(10000) to run in 0.5 seconds or thereabout. But in UserRPL the recursive Fibonacci formula traverses to FIBNUM(1000) in 4 seconds on a HP50G. > > It would be cool if the NSpire could > > be coded in C or C++ for sure, but I believe the OS is too limited > > to realize the full potential of such programming. > > We shall see. For sure. Cheers, Steen From: Jean-Yves Avenard on 2 Sep 2006 23:07 Yao Konan wrote:> Even for BCD floating point computations ? > Because both of the speed test with the 20*20 matrix involved intensive > BCD floating point computations unless TI use binary floating point > computations and in that case the speed of the TI-NSpire would be a bit > disapointing. Neither the 68K nor the ARM9 are great for doing BCD calculations, actually there are no processors these days able to perform BCD on more than 8 bits register... So BCD calculations are only made using software. TI has never used binary floating point in the past (they surely didn't with the TI89 or 92) why would they change that in any new calculators ? JY From: Jean-Yves Avenard on 2 Sep 2006 23:08 Steen Schmidt wrote:> Takes 0.05 seconds on the HP50G. It would be cool if the NSpire could > be coded in C or C++ for sure, but I believe the OS is too limited to > realize the full potential of such programming. Why that? You can do C/C++ without any OS if you wanted to. JY First  |  Prev  |  Next  |  Last