RafaGago

Using 3 of the DC blockers named by "mystran" on the thread below and double precision seems to make things much better, even at 16x. https://www.kvraudio.com/forum/viewtopic.php?f=33&t=545280#top One after each tanh...

BTW, I didn't open this as a bug, as I'm not sure if it's one.

Thosa are resources from the "rules_haskell" project, but I'm only using "rules_nix" in isolation. I did take a look at that script and the only obvious (for me) thing missing...

Thanks. Now after the fact I see it was a pretty obvious mistake. My idea if this had worked, was to try to geneate a `nix.shell` file via a Bazel...

Although I'm not working with Bazel any longer (I'm between jobs now), #183 looks awesome.

Maybe adding a new constraint parameter saying how many times an operator can be used could help. Alternatively being able to specify complexity weights for each operator. It would be...

Reading BL_HAS_CPU_TIMEPT and failing gracefully (#error) or providing a fallback from malc would probably have been preferable. You seem to be mostly right. I'm not planning it to add it...

Thanks for debugging. I don't have time now to check, but that code seems to be doing some heuristics to detect clock latencies. On a quick test I couldn't reproduce....

This is old code, so I don't even myself remember what it does :) A quick read of the code it seems that it first tries to calculate the "clock_gettime"...

Notice the "bl_fast_timept_to_nsec" call on "src/malc/serialization.c:510". It is translated afterwards. If my memory serves me right on all modern enough Intel processors the clock runs at a constant frequency and...