Bob Cao
Bob Cao
There is actually hardware support on both Nvidia and AMD for native bit extract (a lot faster than using lower level bit operations). Maybe we should keep them not demoted?
> > There is actually hardware support on both Nvidia and AMD for native bit extract (a lot faster than using lower level bit operations). Maybe we should keep them...
> Ok, edited the CMakeLists.txt and built again with clang++14. Gives a different error. https://pastebin.com/Z4Tp3t90 Should I try again with a lower version of clang? Yes, maybe clang 10 or...
Our installation document provides a clang10 tool chain that you can use
ref: In SPIR-V there is `FrexpStruct`. In metal there is `frexp`
What happens if you add a `ti.sync()` after each run?
This is a static range case, considering N is 32 bit, the Ndrange should know whether it requires 64 bit indices or not statically. i64 is quite slow and not...
> > This is a static range case, considering N is 32 bit, the Ndrange should know whether it requires 64 bit indices or not statically. i64 is quite slow...
What's the typical shape of Ybus and V?
> Hi, would you mind update the full code s.t. we can also run some tests locally? Or I think a sample data or data generator of Ybus and V...