Jacob Lifshay
Jacob Lifshay
Copying a comment from the nextpnr-xilinx issue that applies to VPR too: Luke (the person at Libre-SOC who originally encountered this bug) thinks this bug is best fixed fixed by...
a minimal solution to allowing multiple cases is to allow something like: ```python with m.Case(1, 2, 3): do_stuff() ``` Edit: turns out I misread the issue title :P
that doesn't actually work, there can be a lost wakeup: assuming little endian: | thread 1 | thread 2 | |-|-| | write 0x100000000 | | | start i64.atomic.wait 0x100000000...
Can someone review this? I needed it today and had to use a workaround.
Same in Linux
it works because stdcall doesn't do anything on 64-bit windows.
not at the moment. Also, all the functions that vulkan api users can directly call need to be `extern "system"` as well
decimal conversion takes so long because bigint doesn't use an efficient algorithm. when using bc from GNU coreutils (which I think uses libgmp), printing `2^166096` (close to `10^50000`) takes about...
> > decimal conversion takes so long because bigint doesn't use an efficient algorithm. > > when using bc from GNU coreutils (which I think uses libgmp), printing `2^166096` (close...