Mike Fulton wrote:
In *HINDSIGHT*, I would say that it's possible if Eric been pulled off his other projects (multiple) and been tasked to work on the compiler stuff, he might have been able to come up with something. The same is probably true of Ken Rose and a number of other engineers. But the conventional wisdom within Atari at the time was that there wasn't any way to create a reliable workaround for some of these issues. The fact that we had Brainstorm working on the project in the first place was something of a "Hail Mary" effort, hoping they'd be able to come up with something nobody else had thought of.
Speaking of hindsight, back a few years ago when Ken Rose was hanging out in the community he was thinking about rewriting the BIN utilities and modernizing them. He asked us if we needed anything as far as tools go for him to work on short of a new compiler. He said he had a custom compiler he had made he wanted to target to the Jaguar that he would do in his own good time.
At this time I don't believe anyone mentioned the risc gcc to him. Everyone disregarded it because we believed it was SO broken on the software side that a new compiler was needed. In hindsight if we had known it was hardware bug causing the problem we could have sicced him on it. I think if he had even known there was an risc gcc he probably would have been curious enough to take a peek no matter what state it was in. If he had even known that he probably would have seen and realized what was going on. This whole situation probably would have been solved for the public by now. I knew even less about the subject than I do now but in our own arrogance and Brainstorm bashing we all figured they just weren't competent enough to retarget a compiler for the GPU. Hindsight hindsight hindsight.
A wasted opportunity and a shame.