confuse about Xeon Scalable 8173M v.s Xeon E5 2679V4

I have 2 servers. RAM/HDD are 32G/1T on both servers.
CPU are as following: Intel(R) Xeon(R) Platinum 8173M CPU @ 2.00GHz x112, Intel(R) Xeon(R) CPU E5-2679 v4 @ 2.50GHz x80@ 2.50GHz x80 The vraybenchmark scores are 19s for 8173M and 24s+ for 2679v4: V-Ray Benchmark | Chaos
but when I use vray to render same scene, 2679V4 is about 7% faster than 8173M

There is probably something in the scene that requires synchronization between the cores and reduces the scaling factor. It’s hard to say without looking at the actual scene.

Best regards,
Vlado

I know thread_num is for working thread num, and num_pass is for lighcache count unit. Is there any other setting relates to cores and the scaling factor?