If AVR is claming things that is not true time after time I do not blame Microchip to write something to their defence. In this truth.pdf file there are both god and bad things about the PICmicro and my view is that it is written in a objective manner. 
I have Heard AVR seminars where they talked bad about the PICmicro instead of concentrate on their own products. If someone is telling bad things about their competitor they probably don have much self-esteem. Some people like the AVR and some the PICmicro, so why argue and why just don't leave it at that. 
I like the PICmicro and that's why I'm on this PIC mailing list. 
Regards
Niklas

----- Original Message ----- 
From: Graham Daniel <gdaniel@paradise.net.nz>
To: <PICLIST@MITVMA.MIT.EDU>
Sent: Saturday, October 02, 1999 12:43 PM
Subject: Re: SV: [OT] Atmel AVRs and ICE200


> Niklas Wennerstrand wrote:
> 
> > I have stumble on a pdf file on the new Microchip CD-ROM under Download\lit\truth.pdf
> > This is a comparison between the AVR and the PIC if any one is interested.
> > Regards
> > Niklas
> 
> I've read it, it's a very misleading  document.   The very fact of it's existence
> indicates that uChip consider Atmel to be a major threat.
> 
> I just looked at again and now that I have actually worked with the Mega, 1200, 2313,
> 8515, 8535 and 8533, I can tell you from experience that most of the above "TRUTH.PDF" is
> absolute male bovine fecal matter.
> 
> There is repeated mention of Atmel's usage of No of bytes to indicate size of program
> memory and those who have not grasped the basic concept that all AVR instructions are 16
> bit multiples might be missled (for example) by Atmel's true claim for the AT90S1200 of
> 1k bytes program memory.
> 
> If you do not appreciate the above difference then it is easy to underspecify required
> memory for doing a job with Atmel AVR parts.
> 
> As an example I used a '1200 to engineer an eight channel PWM controler with custom bit
> banged serial protocol and reprogrammable network I.D.  The code just fit and the '1200
> was just fast enough running with a 4mHz xtal.   Using 16F84s I would have had plenty of
> code space but spent almost twice as much for the uC, had to run it at 10MHz and still
> probably required 4 x 16F84s to control the same No of channels due to a much lower
> effective speed of instruction execution relative to clock speed.
> 
> If you need to use more than the 32 bytes (Number of cache accumulators on the AVR) then
> your program will start to use program memory at a higher rate due to the extra overhead
> but then.... PIC bank swapping, INDF addressing etc required to control I/O etc is just
> as bad if not worse.   The large No of vectored interupts in conjunction with 32 cache
> accumulators means that you can customise a rapid interupt system such as the Z80's
> shadow registers.   Addressable push/pop stack on all but the '1200 give flexibility as
> do many special addressing instructions.
> 
> Down side is that the Atmel free assembler is very rudimentary (ie not allowing nested
> macros or conditional directives)
> 
> I suggest you invest an Atmel STK200 kit (cheap) and start reaping the benefits, Last
> project I priced using both uChips new analogue flash parts and Atmels (flash) analogue
> parts, the distributors quoted me half *again* the price for uChip's slower non vectored
> interupt chips !   While people are hooked to uChip there is (I believe) no chance of
> serious price drop, uChip  share investors are getting value for money.
>