<div dir="ltr">Thank you. How about float vs. integer? Is gmp<div>used only for integers, or is it used for floats as well?</div><div><br></div><div>In Radium, I think only calculations with floating</div><div>points are important to be fast. And only</div><div>integer calculations needs to be accurate. Big-number</div><div>accuracy for floating points are probably not important,</div><div>and normal 64 bit "double" should be good enough.</div><div><br></div><div>Would it make sense to enable gmp for s7 in Radium</div><div>before the optimizer is rebuilt?</div><div><br></div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Apr 14, 2017 at 1:41 PM, <span dir="ltr"><<a href="mailto:bil@ccrma.stanford.edu" target="_blank">bil@ccrma.stanford.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">In the gmp version, it's probably safest to assume<br>
everything is going through gmp. In cases where the<br>
optimizer can ascertain ranges, s7 will use<br>
faster code, but I just tore out half the optimizer,<br>
and am rebuilding it. So I don't know how the<br>
in-between stuff interacts with gmp -- give me a<br>
month or two! My old timings indicated that gmp<br>
could be 10 to 50 times slower than straight C (but at<br>
least it gave the right answer).<br>
<br>
<br>
</blockquote></div><br></div>