My CPU is Xeon platinum 8124m * 2, 36 core 72 threads, 64g of running memory. In the windows 10 workstation version system, the Houdini rendering CPU can be fully loaded normally, but the actual rendering speed is the same as blinding a CPU, and even the two-way rendering is a little slower than the single way, and the power supply of the two-way rendering is equivalent to wasting a CPU, but the rendering window is indeed 72 Small block, just this small block, is twice as slow as the small block with 36 threads open. It is found in the actual measurement that because additional CPU memory needs to be called, there is no problem in the use of E5 detail two-way before. I don't know why series 8 has this bug now, is there any God to know? Thanks for answers