21 Jun
2024
21 Jun
'24
4:06 p.m.
On 2024-06-21 05:17, Eric Smith via cctalk wrote:
A pipelined FP operation immediately followed by a
store will
not store the result of that operation. Needless to say, programming in
pipelined FP mode is challenging, but it's the way to get the highest FP
performance out of the i960. Math libraries used hand-written code to do
that, but I don't think any of the i960 compilers were smart enough to do
it for you.
I guess, you're still talking about the i860, not the i960? ;-)
Anyway, that could be the reason, why most high performance examples in
the application notes are hand crafted assembler code ...