From: fractint-owner@xmission.com (fractint Digest) To: fractint-digest@xmission.com Subject: fractint Digest V1 #25 Reply-To: fractint@xmission.com Sender: fractint-owner@xmission.com Errors-To: fractint-owner@xmission.com Precedence: fractint Digest Wednesday, October 8 1997 Volume 01 : Number 025 In this issue: (fractint) Re: (fractint) dmj-pub.frm Re: (fractint) the death of DOS fractint [none] (fractint) Re: your mail [none] (fractint) Re: (fractint) MFC vs. OWL Re: (fractint) MFC vs. OWL Re: (fractint) the death of DOS fractint (fractint) Real-vs-Integer Re: (fractint) MFC vs. OWL (fractint) RE: removing integer code from fractint (fractint) Re: Which compiler(s)... Re: (fractint) RE: removing integer code from fractint (fractint) IFS Formulas (fractint) Remove Int math ? Re: (fractint) Real-vs-Integer (fractint) Remove integer math? Re: (fractint) Remove Int math ? See the end of the digest for information on subscribing to the fractint or fractint-digest mailing lists and on how to retrieve back issues. ---------------------------------------------------------------------- Date: Tue, 7 Oct 1997 09:13:34 -0700 From: "Mike or Linda Allison" Subject: (fractint) Re: Hi, Peter! I don't think I ever saw an answer to your question, so I'll give it a go! When you are in Fractint, load one of your favorite fractals. Then hit "b". On the first line, give your par file a name (pfj.par ?). On the second line, give your fractal a name (fract001 ?). Add any comments (copyright Peter Jakubowicz ?) and hit enter. You just generated a par file! Load another favorite, hit "b," leave the par file name the same, and on the second line enter the second fractal's name (fract002 ?). Hit enter. You have two par files (fract001 and fract002) in the file named pfj.par When you exit DOS back to Windows, you can open pfj.par as a text file, see the par file you generated, copy it, upload it, whatever! Very simple! Linda - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 07 Oct 1997 15:33:18 -0400 From: "Damien M. Jones" Subject: (fractint) dmj-pub.frm Howdy folks, Just wanted to let everyone know that an update version of my FRM for new coloring methods has been placed on my web site. This contains slight adjustments to the formulae to address the color drop-out problem Kerry Mitchell pointed out, and a few tweaks to the smooth coloring algorithm (and all formulae that incorporate it). I have also included a PAR file that makes getting started with these formulae easier. Rather than select a new fractal type, and having to remember which other settings need to be used (outside=real, decomp=256, periodicity=no), you can select the corresponding PAR file entry and have everything set for you. There are also a few more examples in the PAR file. You can download the ZIP file containing FRM and PAR from this URL: http://www.geocities.com/~fractalus/misc/dmj-pub.zip If you don't use this URL, the file is a little hard to find--there IS a link, but it's obscure. I intend to fix this, but wanted to let interested folks know the new file is available. Those who do not have web access but would still like the file can e-mail me and I will send it as an attachment. If you have trouble with attachments, I can send it as plain text--but the plain text is 200K, and the ZIP file is only 15K. Damien M. Jones / temporary sanity designs / http://www.emi.net/~dmj/ dmj@emi.net / my gallery: http://www.geocities.com/SoHo/Lofts/2605/ - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 07 Oct 1997 10:33:55 -0600 From: Rich Thomson Subject: Re: (fractint) the death of DOS fractint In article <34397EE0.7A511611@studbox.uni-stuttgart.de> , Dirk Meyer writes: > DJGPP is also 32 bit, and there would be no "resource eating" windows > in background (if you kept an old DOS 6.2 version on your PC...). I didn't say windows was less resource hungry than DOS; I said that 32-bit code is likely to be faster than 16-bit code. This is my understanding for modern processors anyway (pentium). - -- ``Between stimulus and response is the will to choose.'' -- Steven Covey =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= 3D Paint: The Power to Create in 3D; Rich Thomson email me for more info rthomson@ptc.com - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 7 Oct 1997 16:33:15 -0500 From: newstedclan@juno.com Subject: [none] Want a novice opinion? Pleeese write a 32 bit "windows" version of Fractint. I don't care what you call it! It seems that ediingt formulas and such could be done alot easier and faster using pop-up windows for entering new information. Plus, I wouldn't have to close fractint and edit my formula (simple as they are) and re-open fractint, select display, type etc... I could just leave Fractint open and maximize, edit,minimize, then try my changes. This would be great for us who use trial and error more than mathematical planning. ANNNND. What is the big deal with snowflakecolor with float=yes. I tried it with my pentium 166 and it only took about 2.5 minutes at the sixth iteration. And I didn't see any missing pieces. I didn't go any higher. ANNND since you said it was o.k. to ask simple questions here's one: Do you mean delete integer math in favor of real numbers ( 1,2, 3 versus 1.0, 2.0, 3.0)? I use real numbers all the time when I play around, is fractint just ignoring me? - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 7 Oct 1997 18:02:48 -0400 (EDT) From: ijk@sas.upenn.edu (Ian J Kaplan) Subject: (fractint) Re: your mail newstedclan@juno.com wrote: > ANNND since you said it was o.k. to ask simple questions here's one: > Do you mean delete integer math in favor of real numbers ( 1,2, 3 versus > 1.0, 2.0, 3.0)? > I use real numbers all the time when I play around, is fractint just > ignoring me? In the bad old days, most computers lacked a math-coprocessor (or floating-point processor) optimized for real numbers instead of integers. If you passed an instruction down to your processor that basically said "compute 3.5*2.0", it would run whole orders of magnitude slower than "compute 4*2". A sorry, sorry state of affairs for fractals. Fractint slipped around this speedbump by converting all floating-point (or 'real') numbers to integers before doing the math. Of course this limited the precision of the operation and hence the maximum zoom before you switched back into floating-point mode; if you dig up an old PC and start zooming in on the M-set, you'll see a sudden slowdown at a certain magnification. It introduced a few artifacts of the technique and added a great deal of nonportable, complicated DOS code. But it ran much, much faster than other fractal engines of the time. With the advent of the 486DX, math-coprocessors became built-in hardware; in a modern pentium, fractint doesn't really gain speed this way, to my knowledge. So if we consider only what's best for relatively new PCs, and if we don't worry about the very very large workload of virtually rebuilding the fractint source code, it's definitely time to do away with integer math. On the other hand, there are a lot of 486SX or less users out there. On the gripping hand, they can still use old versions of fractint if they want. Me, I think it's time to move forward... assuming we find some willing victims to plow through every last bit of code... BTW, for those who've made it this far, can we PLEEZE promise to base this code in GNU C++ and not proprietary Microsquish stuff? and consider writing win95 interfaces with Borland or someone, just to preserve the free and open tradition fractint represents against the evil onslaught of microsoft Everything? - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 07 Oct 1997 16:06:22 -0600 From: Rich Thomson Subject: [none] In article <19971007.163316.10118.0.newstedclan@juno.com> , newstedclan@juno.com writes: > ANNNND. What is the big deal with snowflakecolor with float=yes. I tried > it with my pentium 166 and it only took about 2.5 minutes at the sixth > iteration. And I didn't see any missing pieces. I didn't go any higher. With L-systems, the computation time tends to go up exponentially, rather than linearly. That is why at order 7 you should expect it to go significantly longer than 2.5 minutes. (I think people reported it going at about 12 minutes on a P166). > ANNND since you said it was o.k. to ask simple questions here's one: > Do you mean delete integer math in favor of real numbers ( 1,2, 3 versus > 1.0, 2.0, 3.0)? > I use real numbers all the time when I play around, is fractint just > ignoring me? Ultimately everything in the computer is either a 1 or a 0. "Integer math" refers to using instructions that manipulate integer quantities. If you take the number 1.5 and multiply it by a scale factor, like 10000, then you have 15000, an integer. Using integer arithmatic with a scale factor in this fashion is called "fixed point", because of the scale factor, you have a fixed number of digits to the right of the decimal point when you do computations in integer arithmatic. Fractint uses integer math and fixed-point arithmatic when it can to avoid the more costly floating point instructions. Until the 486 generation of machines, floating point instructions were emulated in software unless a hardware coprocessing chip was installed in the machine. Thus, if an integer add operation had a normalized cost of 1, a floating-point add operation would have a normalized cost of 10 without the coprocessor. (The numbers here are for the purposes of explanation and have no bearing on the actual cost of such operations on a specific processor.) "Floating point" operations carry around two quantities for each number -- the mantissa and the exponent. The mantissa maintains the significant digits of the quantity and the exponent is used as a scale factor to increase the dynamic range of the representable quantities. Think of a number in "scientific notation", like 6.023 x 10^23; the mantissa is "6.023" and the exponent is "23". Most modern microprocessors have hardware accelerated floating point operations, and so the advantage of using fixed-point arithmatic and integer instructions isn't as great as it used to be. However, once you get to arbitrary precision calculations, they tend to be implemented as a chain of integer math instructions operating on the separate digits of the arbitrary precision quantity. Most CPUs have integer performance (in terms of calculations/second) that exceeds their floating point performance. So fractint isn't ignoring your fractional numbers when you input them, but if the numbers are within the range where fractint can use its integer math code to perform the computation (and float=no), then it will use a fixed-point algorithm. - -- ``Between stimulus and response is the will to choose.'' -- Steven Covey =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= 3D Paint: The Power to Create in 3D; Rich Thomson email me for more info rthomson@ptc.com - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 07 Oct 1997 17:12:37 -0500 From: "Justin A. Kolodziej" <4wg7kolodzie@vms.csd.mu.edu> Subject: (fractint) Re: newstedclan@juno.com wrote: > ANNNND. What is the big deal with snowflakecolor with float=yes. I tried > it with my pentium 166 and it only took about 2.5 minutes at the sixth > iteration. And I didn't see any missing pieces. I didn't go any higher. The big deal WAS with float = no. As we've found out, snowflakecolor works perfectly with float = yes, even at levels 7 and 8. (Level 8 took 9 hours 14 min 40.02 seconds on my Pentium Pro 180 at 1024x768x256!) Just use float=yes and you'll be fine. > ANNND since you said it was o.k. to ask simple questions here's one: > Do you mean delete integer math in favor of real numbers ( 1,2, 3 versus > 1.0, 2.0, 3.0)? > I use real numbers all the time when I play around, is fractint just > ignoring me? Not really. Fractint uses real numbers all the way through. It's just when Fractint goes to calculate the image, it does some programming hocus-pocus to represent real numbers (1.24...) as integer types in the program (74593 or whatever it comes out to). This was great when math coprocessors were expensive and not widely used, as integer math ran MUCH faster on the 386 than floating-point math did (calculating numbers while keeping track of a decimal point). However, times change, and the Pentium series does floating-point math as fast as (if not faster than) integer math with its built-in math coprocessor. So integer math doesn't help very much now, and that's why the programmers are thinking about dumping it. I hope this cleared things up, and if it doesn't, or I've made some serious error somewhere, Tim should be able to help much more than I can. After all, he's one of the programmers! Justin Kolodziej "I only use windows because: 1. Solaris doesn't run on PCs :( 2. I can't afford a Sun and I'd have to buy a workstation AND a server :( :( :( 3. When's the last time YOU saw an X-Windows app in your local software store ?! ;)" - -- Me - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 07 Oct 1997 23:05:26 -0400 From: "Damien M. Jones" Subject: (fractint) MFC vs. OWL Ian, - BTW, for those who've made it this far, can we PLEEZE promise to base - this code in GNU C++ and not proprietary Microsquish stuff? The core of FractInt--the fractal calculation routines, speedup algorithms, coloring techniques, formula parser, etc.--can be written in 32-bit code without dependence on *any* particular compiler, even Gnu C++. Interface stuff and screen-access stuff will of course be different on each platform. - and consider writing win95 interfaces with Borland or someone, just to - preserve the free and open tradition fractint represents against the evil - onslaught of microsoft Everything? So rather than use proprietary Microsoft interface class libraries (MFC), you'd rather use proprietary Borland interface class libraries (OWL), to make a *political* statement? Sorry, this seems rather silly. If you want to write the interface with OWL instead of MFC, that's fine, but I would hope you have a better reason than "I hate Microsoft". Damien M. Jones / temporary sanity designs / http://www.emi.net/~dmj/ dmj@emi.net / my gallery: http://www.geocities.com/SoHo/Lofts/2605/ - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 7 Oct 1997 23:26:32 -0400 (EDT) From: ijk@sas.upenn.edu (Ian J Kaplan) Subject: Re: (fractint) MFC vs. OWL Damien M. Jones wrote: > > Ian, > > - BTW, for those who've made it this far, can we PLEEZE promise to base > - this code in GNU C++ and not proprietary Microsquish stuff? > > The core of FractInt--the fractal calculation routines, speedup algorithms, > coloring techniques, formula parser, etc.--can be written in 32-bit code > without dependence on *any* particular compiler, even Gnu C++. Interface > stuff and screen-access stuff will of course be different on each platform. > Granted. Looking back, I'm not quite sure what I was referring to. gotta stop posting when half-asleep... > - and consider writing win95 interfaces with Borland or someone, just to > - preserve the free and open tradition fractint represents against the evil > - onslaught of microsoft Everything? > > So rather than use proprietary Microsoft interface class libraries (MFC), > you'd rather use proprietary Borland interface class libraries (OWL), to > make a *political* statement? Sorry, this seems rather silly. If you want > to write the interface with OWL instead of MFC, that's fine, but I would > hope you have a better reason than "I hate Microsoft". > Well, if you grant that there's no really non-proprietary way to code a win95 interface, I'd rather use Borland sort of the way you support your local grocery man, all other things being equal. I don't have direct evidence that their windowing libraries are sounder than microsoft's, but I do feel I'm on good ground claiming that other, more easily measurable aspects of Borland's compiler are stabler, better designed and better optimized, and that the average microsoft product has an awful lot of bugs that look like they're of the 'software team just didn't care' variety, as opposed to the 'really hard bug to get rid of' variety. That's somewhat shaky argument, of course. But I wasn't really volunteering for the thankless task of writing win95 interfaces, though if my bosses stop keeping me at work eleven hours a day, I might sign up to do XWindows... at any rate, my thanks for pinning me when I was talking out of my arse. I deserved it. - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Tue, 7 Oct 1997 23:38:16 -0400 (EDT) From: ao950@freenet.carleton.ca (Paul Derbyshire) Subject: Re: (fractint) the death of DOS fractint >I didn't say windows was less resource hungry than DOS; I said that 32-bit >code is likely to be faster than 16-bit code. This is my >understanding for modern processors anyway (pentium). 32 bit code != Windows code. I have written 32 bit apps in DOS myself. DJGPP is the key. Also, modular code and interface libraries does not mean Windows either. (See other tangent threads.) - -- .*. Where feelings are concerned, answers are rarely simple [GeneDeWeese] -() < When I go to the theater, I always go straight to the "bag and mix" `*' bulk candy section...because variety is the spice of life... [me] Paul Derbyshire ao950@freenet.carleton.ca, http://chat.carleton.ca/~pderbysh - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 10:59:05 -0500 From: leenewsted@juno.com (Merle L Newsted Jr) Subject: (fractint) Real-vs-Integer Thanks for the answers to my question concerning integer math! it actually helped me understand it! Did anyone try my deranged m-set NukerBrot? later, Nuke - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 13:24:05 -0300 (ADT) From: Michael C Taylor Subject: Re: (fractint) MFC vs. OWL On Tue, 7 Oct 1997, Damien M. Jones wrote: > - BTW, for those who've made it this far, can we PLEEZE promise to base > - this code in GNU C++ and not proprietary Microsquish stuff? > > The core of FractInt--the fractal calculation routines, speedup algorithms, > coloring techniques, formula parser, etc.--can be written in 32-bit code > without dependence on *any* particular compiler, even Gnu C++. Interface > stuff and screen-access stuff will of course be different on each platform. I think that a possible goal for fractint 30.0 :) would be to have one common source code tree, a la gcc or most gnu products, which could help keep development of multiple targets closer (DOS, MS-Windows, X-Windows). > So rather than use proprietary Microsoft interface class libraries (MFC), > you'd rather use proprietary Borland interface class libraries (OWL), to From a development pov, current Borland and Watcom C++ compilers ship with MFC libraries, so using MFC does make more sense. Though if the 'fractint developer's compiler of choice' was a freely available compiler such as djgpp it has the added advantage of younger programmers (i.e. students) would be more likely to contribute since often they cannot afford commerical compilers such as Visual C++ or C++Builder. I know that as a student I couldn't even afford the academic pricing of Visual C++, but could afford to FTP (free) djgpp. Since I'm not actually a fractint contrib, I don't hold my own opinion as very meanful. - -- Michael C. Taylor Programmer, Computing Services, Mount Allison University, Canada sci.fractals FAQ fractal and cryptography archive - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 17:49:49 +0100 From: Edward Avis Subject: (fractint) RE: removing integer code from fractint IMHO removing integer calculation from fractint would be a really bad idea. Over the years, processor manufacturers had put a great deal of effort into impoving the floating point performance of their chips. For example, all Intel x86 chips now have an FPU as standard whereas ten years ago it would have been a costly extra. From the 486 to the Pentium, integer performance was only doubled at most (for the same clock speed) but floating point was more than three times as fast. However, the boom predicted some years ago in floating-point heavy applications has not materialized. CAD, computer animation and so on use just as much floating-point muscle as they ever did, but for the vast majority of users almost all applications are integer-based. Chip manufacturers are beginning to realize this and refocus on integer performance rather than floating-point. Already, we have cheap Pentium-alikes which skimp on the FPU to cut costs - yet for almost all everyday use, they are just as fast. What I am saying is, just because right now it is faster to use floating-point on a Pentium II than to use integers, in five years' time integer speed may have caught up again. So don't burn your bridges. - -- Ed Avis - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 10:50:54 -0600 (MDT) From: Jason Hine Subject: (fractint) Re: Which compiler(s)... >On Tue, 7 Oct 1997, Damien M. Jones wrote: > >> - BTW, for those who've made it this far, can we PLEEZE promise to base >> - this code in GNU C++ and not proprietary Microsquish stuff? >> (snipped much of Michael's comments...) > >Though if the 'fractint developer's compiler of choice' was a freely >available compiler such as djgpp it has the added advantage of younger >programmers (i.e. students) would be more likely to contribute since often >they cannot afford commerical compilers such as Visual C++ or C++Builder. > >I know that as a student I couldn't even afford the academic pricing of >Visual C++, but could afford to FTP (free) djgpp. In my opinion, this is a valid consideration. I'm dismayed at how few of my fellow students are aware of Fractint, but it would be great if those who are could use the free DJGPP compiler (which is what I'm using) to write their contributions... oth the other hand, perhaps there aren't enough financially-disadvantaged-students-who-would-be-contributors out there to make it worthwhile, if there is some _great_ advantage to using a Visual package (not compatible with DJGPP) intead... my three dracmas' worth... PS - anyone have any ideas on my truecolor autokey problem? Anyone??? Peas, Jason (Spidey) Hine _ __ ___ ____ _____ ______ _______ ______ _____ ____ ___ __ _ _ __ ___ ___ __ _ _ __ ___ Jason N Hine ___ __ _ _ __ ___ GIS Specialist ___ __ _ _ __ ___ Colorado State University ___ __ _ _ __ ___ Pedology and Soil Information Systems Lab ___ __ _ _ __ ___ (970) 491-6832 ___ __ _ _ __ ___ http://boralf.agsci.colostate.edu/~jason ___ __ _ _ __ ___ jason@cnr.colostate.edu ___ __ _ _ __ ___ ____ _____ ______ _______ ______ _____ ____ ___ __ _ - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 13:17:00 -0400 (EDT) From: ijk@sas.upenn.edu (Ian J Kaplan) Subject: Re: (fractint) RE: removing integer code from fractint Edward Avis wrote: > Chip manufacturers are beginning to realize this and refocus on integer > performance rather than floating-point. Already, we have cheap > Pentium-alikes which skimp on the FPU to cut costs - yet for almost all > everyday use, they are just as fast. That's a bit of a stretch. The K6, say, does test slower than the equivalent pentium II on floating-point, but it's still running floating-point faster than any earlier-generation chip; and I'd really guess that while the exact performance ratio between the two modes will vary over time, I think we can continue to expect them to advance at similar rates. Considering the sheer overhead associated with int math, I don't think that there are many situations where people will want to run in float=no, any more than they do now... - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 12:48:48 +0500 (GMT) From: Ramiro Perez Subject: (fractint) IFS Formulas Hello Here is an example of a small utility (20k zipped) that I write for fractint 19.6, for inverting the ifs attractors defined in the fractint ifs files, and rewritting it for fractint formula types and for fractint par files. It takes a fractint ifs file, scan it, then if there is an ifs definition, and if it isn't 3D ifs and has less than 9 transformations, the program writes the formula and its correspondient par file.. The program is in my web pages, in the gallery section, but if you wish to go directly, here is the address: http://www.utp.ac.pa/~rperez/ifsinv.html By the way, this program is a new release of an old one that I have for the same work, but this one if far more simpler and this time the program optimizes the formulas, so it runs faster than the old one.. Any comment will be greatly appreciated.. Ramiro Perez ps. sorry for the bad english.. - -------------------------------------------------------------------------- frm:f2_J{ a0=0.01000,d0=0.16000, a1=0.20000,b1=-0.26000,c1=0.23000,d1=0.22000, a2=-0.15000,b2=0.28000,c2=0.26000,d2=0.24000, a3=0.85000,d3=0.85000, al0=0.00160, al1=0.10380,k1=-0.13000,l1=-0.10000, al2=-0.10880,k2=0.14000,l2=0.07500, al3=0.72250,l3=-0.59500, z=pixel: x=real(z) y=imag(z) o0=(d0*x)/al0+flip((a0*y)/al0) op0=|o0-p2| o1=(d1*x-b1*y+k1)/al1+flip((-c1*x+a1*y+l1)/al1) op1=|o1-p2| o2=(d2*x-b2*y+k2)/al2+flip((-c2*x+a2*y+l2)/al2) op2=|o2-p2| o3=(d3*x)/al3+flip((a3*y+l3)/al3) op3=|o3-p2| IF (op0 Subject: (fractint) Remove Int math ? Hello World. I'm using Fractint 19.6 on a 486dx33/CGA. I have make a few tests on my configuration. In general, float is faster or equal than integer, but there was differences between the two modes. Recently M.L. Newsted Jr. have posted a little formula who behave differently in int or in float. And I'm working on a long term project with this formula : - ----------------------------------------------------------- frm:rand_generic { ; **** PARAMETER 1 MUST BE > 0 **** ; Thierry Boudet 101355.2112ompuserve.com ; 19 Juillet 1997 z = rand: z = fn1(z) + fn2(pixel); |z| <= real(p1) } - ----------------------------------------------------------- who works very fine (for me :-) in integer mode, but is completly broken in float mode... Here is a sample parameter. Try it in real and integer mode, and play with the formula. rand seems broken in float mode. - ----------------------------------------------------------- demo_of_rand_bug { ; created Oct 08, 1997 in Toulouse reset=1960 type=formula formulaname=rand_generic function=recip/sqr passes=1 center-mag=0.0260181/0.0209436/0.5956599 params=6/0 inside=zmag outside=real biomorph=64 } - ----------------------------------------------------------- May be with only one math mode in Fractint, this type of problems will go in a "bug-eater-black-hole" ... Best regards from Toulouse, France. Thierry Boudet. - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 14:32:39 -0500 From: leenewsted@juno.com (Merle L Newsted Jr) Subject: Re: (fractint) Real-vs-Integer On Wed, 8 Oct 1997 14:27:47 -0400 (EDT) aq936@freenet.carleton.ca (Michael Traynor) writes: >>Did anyone try my deranged m-set NukerBrot? > >I tried it a bit, and found that it is very like a mandebrot with >inversion radius 1 (on the "Y" screen). It appears just the same at >low >zooms using FP. It is more different with integer math, but that >seems to >be an integer math artifact. If you have any images from this fractal >which are truly different from the mandelbrot, in FP, could you post >the >parameters to the list (or send them to me). I don't have the time, >or >recall my complex math well enought to figure it out, but the Nuke may >be >arithmetically the same as the inverted Mandelbrot. Maybe someone >else >will let us know. > >-- >Mike Traynor > >People who like this sort of thing will find this the sort of thing >they like. > Abraham Lincoln > Honestly, I am just now beggining to wrote my own fractals, and obviuosly I based that one on the Mandelbrot. But I thought it "was" quite different than an ordinary Mandelbrot. Especially when I zoomed. It seemed to me more chaotic than the "smaller mandels in smaller mandels in smaller mandels etc.." My fractal and some more that I have writen seem to me to not be quite so symetrical. And isn't that the point? Still, I will certainly not try to correct anyone on this subject. I am still learning. thanks, Nuke - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 19:09:45 -0400 (EDT) From: Jim Muth Subject: (fractint) Remove integer math? I vote yes. I've been following the thread about the future of Fractint with interest, especially about whether to retain the integer code. My feeling is to go ahead, dump the integer code, and make room for more advanced features, but keep the DOS version as long as DOS lives. (And as long as a program such as Fractint for DOS exists, DOS will live.) I run Fractint on a 486-100mhz machine, and I notice very little speed advantage when integer math is used. In fact, on many fractals, floating math is quite a bit faster. I realize that some fractal artists have created images based upon the artifacts of integer math, and would not be able to reproduce these images if integer math were not available. These artists could merely keep an old version of Fractint with integer math capability on their disks, and use it whenever they wished to work with their integer fractals. I admit to being a DOS fan, (I'm writing this in WordPerfect 5.1). I stayed with DOS when everyone was turning to Windows because it is *fast*! When I open WP51, I need only blink my eyes and I'm ready to go. When I open a Windows app, I have time to go to the next room, pour a drink, and return to my desk before I'm ready to work. Also, when I'm running DOS, I feel as though I am actually con- trolling the computer, while Windows makes me feel as though someone else were doing it for me. But these are my personal opinions, and I realize that DOS, with all its limitations, just can't compete with the windowed environments. I therefore do use Windows in my business, but I do so only because the programs I need are not available for DOS. That said, I must report a small bug I have found in version 19.6 of Fractint. I have noticed that whenever I try to set the ranges to a value greater than 32767, I get the error message, "I couldn't understand the argument" Maybe I'm doing something wrong, but I don't think so. . . Jim Muth jamth@mindspring.com - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ Date: Wed, 8 Oct 1997 20:27:17 -0500 From: newstedclan@juno.com Subject: Re: (fractint) Remove Int math ? Hey Thierry. I checked out your fractal you sent out. It "was" different in FP! I also used Bio-morph (256) and it looked pretty cool. I'm not sure what you mean by posting the parameters, maybe you could explain it or send an example. Anyway, thanks for trying the one I sent out. Here is another: Egg (xaxis) { ; M.L. Newsted Jr. z = 0 c = pixel d = pixel: z = (z*z + c) / d } Well, see ya. Hey What's the weather like in Toulouse? "...God is in the details..." Linus C. Pauling Nuke On Wed, 8 Oct 1997 15:10:49 -0400 Thierry Boudet <101355.2112@compuserve.com> writes: > Hello World. > > >I'm using Fractint 19.6 on a 486dx33/CGA. I have make a few >tests on my configuration. In general, float is faster or >equal than integer, but there was differences between the >two modes. Recently M.L. Newsted Jr. have posted a little >formula who behave differently in int or in float. > >And I'm working on a long term project with this formula : > >----------------------------------------------------------- >frm:rand_generic { ; **** PARAMETER 1 MUST BE > 0 **** > ; Thierry Boudet 101355.2112ompuserve.com > ; 19 Juillet 1997 > z = rand: > z = fn1(z) + fn2(pixel); > |z| <= real(p1) > } >----------------------------------------------------------- > >who works very fine (for me :-) in integer mode, but is >completly broken in float mode... > >Here is a sample parameter. Try it in real and integer >mode, and play with the formula. rand seems broken in >float mode. > >----------------------------------------------------------- >demo_of_rand_bug { ; created Oct 08, 1997 in Toulouse > reset=1960 type=formula > formulaname=rand_generic function=recip/sqr passes=1 > center-mag=0.0260181/0.0209436/0.5956599 params=6/0 > inside=zmag outside=real biomorph=64 > } >----------------------------------------------------------- > >May be with only one math mode in Fractint, this type of >problems will go in a "bug-eater-black-hole" ... > > > Best regards from Toulouse, France. > > Thierry Boudet. > > > > >------------------------------------------------------------ >Thanks for using Fractint, The Fractals and Fractint Discussion List >Post Message: fractint@xmission.com >Get Commands: majordomo@xmission.com "help" >Administrator: twegner@phoenix.net >Unsubscribe: majordomo@xmission.com "unsubscribe fractint" > - ------------------------------------------------------------ Thanks for using Fractint, The Fractals and Fractint Discussion List Post Message: fractint@xmission.com Get Commands: majordomo@xmission.com "help" Administrator: twegner@phoenix.net Unsubscribe: majordomo@xmission.com "unsubscribe fractint" ------------------------------ End of fractint Digest V1 #25 ***************************** To subscribe to fractint Digest, send the command: subscribe fractint-digest in the body of a message to "majordomo@xmission.com". If you want to subscribe something other than the account the mail is coming from, such as a local redistribution list, then append that address to the "subscribe" command; for example, to subscribe "local-fractint": subscribe fractint-digest local-fractint@your.domain.net A non-digest (direct mail) version of this list is also available; to subscribe to that instead, replace all instances of "fractint-digest" in the commands above with "fractint". Back issues are available for anonymous FTP from ftp.xmission.com, in pub/lists/fractint/archive. These are organized by date.