are dsPIC's viable for new product designs?

[pmuldoon]

This is a question for all of you hardcore guru's out there.
I've been using CCS & microchip products for many years and many many projects. I've been extremely satisfied with both companies and the CCS compiler has saved me a lot of time.

My experience with the PCD and dsPIC30Fxx has been completely the opposite. I've spent tons of hours only to find out there is some compiler issue behind the various problems. Some of the issues seem to be very fundamental to a good compiler. I realize that this chip family is extremely complex compared to the old 18F chips.

My question/delemma is this:
Is it beyond CCS's skill level to write a solid compiler for that family?
Would other compilers be any more solid?
Would other chip mfrs/compilers provide a more stable environment for designs requiring the more advanced features of the dsPIC type devices?

Right now I intend to not use anything beyond the 18F family in any new designs. But when the time comes and I need more performance, should I look at other chips? other compilers? Or should I just ride it out with CCS?

This isn't a slam towards any company. I intend to stay with CCS/Mchp thru the 18F stuff and below. I have lots of product to support and have no reason at all to change. I'm very happy with what I've been able to accomplish with those families of devices.

I was going to send an email to both CCS & Microchip, but I thought I'd start here first. I've learned a lot over the years by reading this forum and respect the opinions of you guys.

Thoughts?