Mirror sites Mailing list Have this site translated  Datasheet Archive  Embedded Systems

Compiler

Links

Discussion

Date:         19971030
From:         "Batke, Keith" <Keith.Batke@inteq.bhi-net.com>
To:           "'Mot-68332-Apps@freeware.mcu.motsps.com'"
Subject:      Microtec Compiler?

Does anyone use Microtec's compiler?
How do you like it?

Keith.Batke@inteq.com

Date:         19971031
From:         Shay Collins <scollins@onramp.net>
To:           Mot-68332-Apps@freeware.mcu.motsps.com
Subject:      Re: Microtec Compiler?

It's a good compiler, but I can't say the same for their tech support. If that's important to you, you should probably look elsewhere. I've been using the Microtech tools for about 5 years.

Date:         19971031
From:         Dietmar Millinger <dietmar@vmars.tuwien.ac.at>
Organization: Department of Real-Time Systems, TU-Wien, Vienna
To:           Mot-68332-Apps@freeware.mcu.motsps.com
Subject:      Re: Microtec Compiler?

At our department the microtec compiler (dos and nt versions) is in use. we developed software for the 68332. it worked fine, although the software licence is too expensive in my opinion. So check the gnu compilers too.

Date:         19971031
From:         "Jim M Hartmann"<Jim_M_Hartmann@cidcn00002.cidc.cummins.com>
To:           Mot-68332-Apps@freeware.mcu.motsps.com
Subject:      Re: Microtec Compiler?

I use it. The compiler seems pretty good. It optimizes well and is pretty fast. It has some limitations, for one, the section control is limited. Some other compilers allow you to specify what memory section individual objects will go in, Microtec only allows section control on a whole storage class -- all constants for example. Section control turned out to be very important in my project. I was very disappointed with Microtec's Xray debugger for BDM and am now using SDS Singlestep.

I hear lots of good things about the Gnu compiler but have never tried it myself. There are many others that offer evaluation copies, for example SDS and Diab Data. Try compiling some code with each and look at the assembly output. Look seriously at section control if you think it might be important.

Good luck,

Jim Hartmann, Speaking for myself
Onan Corporation

Date:         19971031
From:         MikeTurner@kemet.com (Mike Turner)
Organization: KEMET Electronics
To:           Mot-68332-Apps@freeware.mcu.motsps.com ('Mot-68332-Apps@freeware.mcu.motsps.com')
Subject:      Re: Microtec Compiler?

The compiler seems to work pretty well. I evaluated Microtec, SDS, Diab and gnu by running a rather large complex machine control program through them. All generated pretty good code although gnu did not optimize nearly as well as say SDS. The only problem that was a killer for me was Microtec's generating compiler errors for some interesting preprocessor statements that I have spent significant time developing. Microtec said the code was correctly being parsed. Microsoft, SDS, Diab, and gnu compiled everything and the code ran correctly (well, I could not run Microsoft since I could not target the 68K). Outside of this problem, which most people will never encounter, Microtec seemed to be pretty solid.

Date:         19971031
From:         "Rajeev Rohatgi" <Rajeev@srsys.com>
Organization: Stanford Research Systems
To:           Mot-68332-Apps@freeware.mcu.motsps.com
Subject:      More on Compilers

I see a compiler thread here. Perhaps it's time to share my shopping and post-shopping experience. (Back in September I had asked for compiler recommendations...)

SUMMARY

In a nutshell, I was disappointed by the sales attention I received (or didn't receive) from the various compiler vendors, esp. the industry standard ones. I've worked as a physicist for 6 yrs and building scientific instruments for 5 and over the years have done a lot of varied shopping. The C compiler vendors _collectively_ were singularly uninterested in making the sale, and remarkably careless in whether they created a good impression or not. It felt very much like dealing with a monopoly vendor rather than a competitive marketplace. Enough vague generalities.

I had my sales lead dropped (ie lost) by 2 industry standard companies. One company took 2 weeks to mail me their introductory advertising, after 6 weeks they have told me that their evaluation CD-ROM is on its way. A number of companies asked what I felt were interrogatory questions, ie out of line for a beginning sales lead. Another company published an 800 number in a leading journal and interrogated me, then made me make two toll calls (not 800 numbers) just to get the _price_ of their product ! Is there anybody reading this who would expect a company to provide better response after the sale than before the sale ?

The following companies gave me the worst sales experiences (alphabetical order): Diab, Green Hills, Microtec.

IN CONTRAST,

Non U.S. compiler suppliers showed far superior interest in making the sale, no pressure but prompt courteous answers to all my questions, very open with price information. I can specifically recommend both Crossware (U.K., www.crossware.com) and Hi-Tech (Australia, www.htsoft.com, US rep is CMX www.cmx.com) for having had good interactions. Plus their low-end compilers are way cheaper than the industry standard US companies (though at the low end you won't a Windows interface. If you care.) One US company to keep an eye on is Metrowerks, hopefully they'll have an embedded 68332 package in less than a year.

Talking about price, I also think that $2000 is too expensive for a 68332 C compiler, considering that the 68332 sells for $15 these days, no longer $70 ! and considering what compilers for other processors can be had for.

DECISION

I ended up buying Hi-Tech, for a couple of different reasons -- I wanted a complete solution (HiTech includes a source level serial port debugger for which you incorporate their code into your ROM -- I haven't done this yet) -- the Crossware compiler can be run with the Huntsville BDM debugger for a solution that's pretty comparable total cost -- but Huntsville gave me misleading info at the time, so I ruled Crossware out prematurely. I'm not competent to make comparative judgements of output code quality, but I've been totally satisfied in this respect with the HiTech output.

If you're shopping now, you might ask whether the package you're considering includes support for the TBLU (table look-up) instructions -- this was added to some later 68k derivatives incl the '332 and isn't in a lot of the mainline 68k processors. (My colleagues using other processors drool when I mention what it can do !) Crossware does support TBLU, HiTech gave me a simple workaround so I can use it.

MORE ABOUT HITECH EXPERIENCE

I've been very satisfied with the Tech Support I've received from HiTech. They appear to be a small outfit and may not have a lot of patience for beginner level questions, (99% of what you need to know is in the manual or source code anyway), but they have come through for me more than once when I've needed it.

Be prepared though that the interface looks 10 yr old (DOS mode, single edit window) but _it_does_ everything_you_need ! I have found very satisfactory operation with 4 DOS windows open under Win3.1 : one for _my_ editor, one for the HiTech integrated development environment (their editor), one for my ROM burner, and one for my RS232* program so I can talk to my board. I can turn around a new pair of ROM in about 3 minutes, and have been very pleased with rate of code development.

I also think their manual could do with a revision, some features are not mentioned ('persistent' attribute) and there are other small holes that I wish had been more fully explained (I had some headaches with sscanf(), though they're all resolved now.)

I doubt anybody's manual or support would be perfect though.

Best wishes to all out there !
Rajeev Rohatgi
Stanford Research Systems
408-744-9047
408-744-9049 FAX
rajeev@srsys.com

Date:         19971031
From:         "Batke, Keith" <Keith.Batke@inteq.bhi-net.com>
To:           "'Mot-68332-Apps@freeware.mcu.motsps.com'"
Subject:      RE: Microtec Compiler?

Section control is exactly what is causing me to inquire about other people's experience with Microtec. We are also disappointed in our evaluation of XRAY.

We are initiating eval of Green Hills Multi and SDS SingleStep.

Keith.Batke

Date:         19971031
From:         Frank Henriquez <frank@ucla.edu>
To:           Mot-68332-Apps@freeware.mcu.motsps.com
Subject:      Re: More on Compilers

Metrowerks is Canadian, although they have a branch in the US. I have a lot of experience with Metrowerks on the Macintosh, going back to their early C/C++ and Pascal compilers for the PowerPC and 68K. Their compilers and IDE are top notch, and the company is fantastic; you can post a message on one of the related Usenet newsgroups and often get a response from a Metrowerks programmer within hours or a day - the same goes for direct e-mail. In some cases, even the CEO will jump in! Their customer support is extraordinary, and they are *VERY* responsive to user comments, suggestions, etc.

They have also released the API for their compiler plug-ins, so it's relatively simple to code up a compiler that works within their excellent IDE. I'm working on a simple C compiler for the MCF5206 ColdFire, based on a public domain DOS compiler. The port hasn't been difficult so far - the hardest part has been to modify the compiler to produce ColdFire friendly mnemonics!

They currently have a Windows compiler..in fact, they seem to have an IDE for just about every high performance device in existence, including one for the Sony Playstation.

I'm not sure what their embedded compiler prices are, but the standard Mac* package is about $400 for a year's "subscription" (two CDs) that includes their IDE, debugger, 68K/PPC C/C++ compiler, Pascal compiler and a Java compiler.

When they entered the Mac* development market, they took the market away from Symantec within a year...it'll be interesting to see what happens when they enter the embedded market in full force.

I don't work for Metrowerks - just a satisfied customer. Here's their URL:
www.metrowerks.com/

Frank Henriquez Programmer/Analyst Jules Stein Eye Institute, UCLA
frank@ucla.edu