12/31/2025 at 4:48:45 PM
I don't understand on one of the later graphs the core to core latency for strix halo goes out to 32 cores but he says only has 16 cores?by Neywiny
12/31/2025 at 4:56:25 PM
AMD's cores have SMT, allowing them to run two threads at a time and appear to the OS and its scheduler as two logical cores despite being implemented as a single physical core.by wtallis
12/31/2025 at 5:23:26 PM
What pattern in the data shows that's what's being measured? I would expect to see basically 0 latency between adjacent "cores" then since L1 is shared per thread?by Neywiny
12/31/2025 at 6:30:12 PM
Co resident threads might not get any speed up here since coherency instructions are functionally operations on the L2 cache.by monocasa