Re: A89: TI89 ROM vs. HP49 ROM
[Prev][Next][Index][Thread]
Re: A89: TI89 ROM vs. HP49 ROM
Yes, the TI89's ROM is written in (slow and bloated) C, with the CAS core
coming from the commercial software Derive. The HP49's speed is comparable
because it was written in almost all Saturn ASM.
Almost nothing is known about Advanced Mathematicals Software v2.00, other
than the fact that it is geared towards enhanced external application
support. It must be somewhat optomized, since TI claims that it'll free up
another few sectors (I think it's 2 64k sectors?) of User Data Archive
memory. I know someone who claims that he'll be seeing a beta copy of AMS
2.00 soon; we'll see. . .
On a side note, a lot of those things are like the Inverted Matrix test -
test geared towards the abilities of the HP49 (one of its specialties
happens to be its 3D grapher). Keep in mind that the 3D graphing is also on
a MUCH smaller screen (hence less time to draw), although it is still
supposedly faster, and that the 100! * 100! time is almost exactly the same
(both calculate and display time. . .)
>Hi, I recently got my 89, and was wondering a couple of things. First I
>like to know whet is going on on the other side of the road. When HP came
>out with the 49, I was curious. In the FAQ at hpcalc, it said that the 3d
>graphing was faster then the 89. I wondered how this could be, considering
>that the 49's processor is only a 4bit, 4mhz Saturn. I found it hard to
>believe, so I emailed the author of hpcalc's FAQ. Here is his response:
>>>me
>>response
>
>>Sorry for the delayed response, but I had to get some figures from HP
(they
>>had to try your example on a >prototype) in order to answer your question.
>
>>>First, for number 3.1, you said that the 49 can plot 3d graphs faster
than
>>>any other calc. I was wondering >>exactly what equation you used?
>
>>It's faster for ANY equation that the HP49 can plot. You have to see it
to
>>believe it...it's amazing what a 4 >MHz CPU can do with optimized
software.
>
>>>z1=(x^3*y-y^3*x)/390
>
>>That's a rather flat surface...maybe you shouldn't have divided by 390 and
>>it would look cooler :)
>
>>About 6 seconds to calculate, and then it rotates it in real time around
10
>>frames per second on any of the >X, Y, or Z axes.
>
>>>When calculating 100!*100! were you using the calculator in RPN or
>>>Algebraic mode?
>
>>It doesn't matter: same speed. The 2.5 seconds figure is to calculate
>>100!, then calculate 100! a second >time, and then multiply the two
>>numbers. Of course, the quickest way to do it would be *100! DUP* but
>> >that would be cheating (only calculating 100! once). The 2.5 seconds
>>includes the time it takes for the >display to update with the number. It
>>actually only takes about 1.5 seconds for the computation.
>
>>Regards,
>
>>Eric Rechlin
>>Bismarck, ND, USA
>>eric@hpcalc.org
>>http://www.hpcalc.org/
>
>As you can see, the graphing is considerably faster, go ahead and try the
>sample equation on your 89.
>Obviously HP has had some excellent programmers working for them, but
enough
>to make up 28bits and 6mhz? My next thought was that since the major speed
>improvements on the HP49 came from rewriting the ROM completely in
Assembly,
>maybe TI's ROM wasn't written in Assembler. Dux Gregis said:
>
>>What they would most likely have to do is look at the assembly code
>>generated from their C source
>
>So the 89's ROM was written in C?!
>
>OK, here is my main questions: was the 89's ROM written in C? Will Version
>2.00 be written in assembly? What will be the major changes in Version
>2.00? If Version 2.00 isn't going to be written in Assembly, will there be
>some functions written in assembler i.e. 3d graphing.
>
>Thanks for your time
>
>Josh Cunningham
Follow-Ups: