<> <> <>¢ The Battle of The Basic's¢ (C) 1992 By Tom Hunt¢ All rights reserved¢ Reprinted by OL'HACKERS AUG¢¢ This article may be freely¢ distributed, and included in user's¢ groups newsletters, and electronic¢ magazines.¢ There seems to be a universal¢ language among computer users. It's¢ the basic programming language.¢ Almost everybody had dabbled with¢ basic, at one time or another. There¢ are alot of other programming¢ languages available for the Atari¢ 800/XL/E. Like C, Action!, Pascal,¢ Logo, Pilot, Assembly, and AVUE. But¢ Basic remains popular, even today.¢¢ Basics come in two major¢ flavors. The first being the¢ interpretive type, and the second¢ being the type that compiles the¢ Basic source program. For those of¢ you who are unfamiliar with these¢ terms, an interpreter gets a chunk of¢ tokenized program, and interprets it,¢ one chunk at a time. Almost like a¢ human language interpreter, it must¢ translate the source code into¢ something the computer's CPU can¢ understand. And more importantly, it¢ does this while your program is¢ running! A compiler, on the other¢ hand, will translate your Basic¢ program into something your¢ computer's CPU can understand, called¢ machine language. It does this once,¢ and saves the compiled Basic program¢ out to disk, usually with a ".COM" or¢ ".OBJ" file extension. Occasionally¢ you will find a compiler which¢ generates what is known as pseudo-¢ code, or "p-code". But since none of¢ the Basics reviewed here uses pseudo-¢ code, we won't cover that here.¢¢ The Basic's reviewed in this article¢ are as follows:¢¢ Interpreters:¢ Atari Basic Revision C¢ Basic XE ¢ Turbo Basic XL 1.4¢ CTH FastBasic 2.0¢ Turbo Basic 3.2q¢¢ Compilers:¢ MMG Compiler¢ Turbo Basic XL¢¢ In this article I will attempt¢ to touch upon several of the most¢ popular Basic language programs, and¢ briefly describe what I feel are each¢ one's strong points. I will also be¢ describing how each of them performed¢ a simple benchmark program I made.¢ First of all, about the benchmark¢ program.¢ The benchmark program I made was¢ no scientific marvel. It made no¢ attempt to measure "drhystones" or¢ "whetstones", which are pretty alien¢ to most Basic users. Instead, I made¢ my program access certain functions¢ of Basic, which your own programs¢ would most likely be accessing. The¢ program also benchmarked how long it¢ took each Basic to perform a GOSUB¢ loop, and a FOR/TO/NEXT loop. Here's¢ a list of the functions and tests¢ which my benchmark program tested¢ these Basic's with.¢¢ LOOP TEST, 500 ITERATIONS¢ GOSUB TEST, 400 ITERATIONS¢ SIN(X) TEST, 50 ITERATIONS¢ COS(X) TEST, 100 ITERATIONS¢ SQR(X) TEST, 30 ITERATIONS¢ ATN(X) TEST, 30 ITERATIONS¢ 2^X TEST, 30 ITERATIONS¢ X/.2 TEST, 200 ITERATIONS¢¢ The benchmark program used the¢ real-time clock inside my 800XL to¢ time how long it took for each Basic¢ to perform each test, along with the¢ amount of time to run all the tests¢ together. The total time to run all¢ the tests is interesting, because¢ since the results of each test were¢ printed out to the screen, it¢ indicates how efficient each Basic's¢ print-to-screen routine is. The¢ following table is formatted for 80¢ columns, and shows the average total¢ time each Basic used to run through¢ the entire benchmark program. The¢ time recorded is in jiffies, or 60ths¢ of a second.¢¢ -------------------------------------¢ Average Total Time Test For¢ Interpreted Basic¢ -------------------------------------¢ BASIC USED| ATARI| BASIC XE| CTHFB|¢ TB XL| TB 3.2Q |¢ | 1889| 455| 778|¢ 364| 364|¢¢ -------------------------------------¢ Average Total Time Test For Compiled¢ Basic¢ -------------------------------------¢ COMPILER USED | MMG | TB XL|¢ | 1416| 300|¢¢ Now for a brief examination¢ about each interpreted Basic¢ performed.¢¢ ATARI BASIC¢ The version of Atari Basic used¢ was revision C. Results of the¢ benchmark program shouldn't differ if¢ ran on revision B. Atari Basic came¢ in last in the benchmark test. Didn't¢ surprise you, did it? However, I¢ don't want to "beat up" on good old¢ Atari Basic too bad. You have to¢ remember when it was designed, and¢ that it fits neatly into 8k of¢ memory. ¢¢ BASIC XE¢ This is a Basic sold by ICD, and¢ before that by OSS. It has many fine¢ features which weren't accessed by¢ the benchmark program. One of these¢ features is the fact that it can use¢ banked ram on machines with more than¢ 64k. For running the benchmark¢ program, it had all extension loaded¢ in, and FAST was set to ON. It¢ finished second.¢¢ CTH FASTBASIC¢ This is a shareware Basic, sold¢ by my company, CTH Enterprises. It¢ came in third. It features a built-in¢ screen accelerator, and a built-in¢ fast floating point math package. It¢ should be noted that the numerical¢ results printed out as the result of¢ each individual test varied from the¢ other Basic's. This is because of the¢ custom fast floating point package.¢ It is more accurate than any of the¢ other Basics tested.¢¢ TURBO BASIC XL, and TURBO BASIC 3.2q¢ These two Basic's tied for first¢ place! Turbo Basic XL will only run¢ on XL/E computers. It uses the ram¢ that is underneath the operating¢ system rom, in the $C000 through¢ $FFFF address area. Unfortunately,¢ Sparta dos 3.2d also wants to use¢ this area for itself, making Turbo¢ Basic XL incompatible with it.¢¢ Turbo Basic 3.2q is a variation¢ of Turbo Basic 400/800, originally¢ written for the old Atari 400's and¢ 800's. The 3.2q version was made so¢ that it is fully compatible with¢ Sparta dos 3.2d. It supports all the¢ same functions that Turbo Basic XL¢ has, except for the BRUN and BLOAD¢ functions. However, these can be¢ easily done with XIO's.¢¢ Now for a brief examination¢ about how the compiled Basic's¢ performed.¢¢ MMG COMPILER¢ When program compilation starts,¢ you have the option of including¢ either the integer or the floating¢ point math package into your program.¢ Integer math is much faster than¢ floating point math. However, integer¢ math cannot handle functions that¢ require a floating decimal point,¢ such as many of the individual tests¢ within the benchmark program require.¢ To be fair, if I had made a benchmark¢ program based upon integer math, and¢ had included the integer math package¢ into the compiled code, the MMG¢ compiled Basic program would have¢ fared much better. But in real world¢ programs, floating point math is used¢ more often than integer math. That's¢ why I included floating point math¢ functions into the benchmark¢ program.¢¢ One of the big advantages of¢ using the MMG compiler, is that it¢ produces object code which will run¢ under any dos. Another advantage is¢ if you actually DO have a program¢ which only uses integer math. You can¢ include the integer math package into¢ your compiled code, producing a very¢ fast running program.¢¢ TURBO BASIC XL¢ This is the winner, in the¢ compiler category. It came in at a¢ blazing 300 jiffies for the average¢ run time of the entire benchmark¢ program. There is one major¢ detraction from this otherwise¢ excellent compiler. Neither the¢ compiler, or the compiled code it¢ produces, will run under Sparta dos¢ 3.2d. It functions just fine under¢ Sparta dos X, MyDos, and Atari dos¢ 2.0/2.5. ¢ So, which basic is the best? It¢ depends upon what you want to do. I'm¢ sure that it is fairly obvious that¢ each Basic has its own strong points,¢ and their own weaknesses. But for all¢ out speed, Turbo Basic wins, hands¢ down!¢ ****¢ If you would like to download¢ the benchmark program, you may do so¢ from the bbs listed at the end of¢ this article.: ¢ Closer To Home BBS- 419-368-4413¢ 24 Hours/7 Days- 300/1200/2400¢ << END >>
