With the new Intel 11 Series about to be released there have been some leaked benchmarks.
I have a 5950X so I decided to disable one of the CCDs to see how an eight Core/16 Thread AMD 4th Gen Ryzen CPU would stack up against the leaked benchmark of the 11900K.
The first benchmark I tested was TimeSpy Extreme and during the CPU portion of the benchmark run I wanted to monitor the temp of the CPU with Ryzen Master.
I discovered something very peculiar.
My Ryzen CPU was being throttled and even though the CPU was configured to run all core at 4.775 GHz the maximum that any core would ever achieve is 4.624 GHz.
The following is an example of a run at 4.775 GHz:
I decided to downclock the CPU to 4.7 GHz and the loading of the cores went down proportionately:
It appears that there is a negative offset applied to the Ryzen CPU.
There is no conceivable reason why the clockspeeds should go down by approx. the same amount that I lowered the clockspeed of the cores even though that was still ABOVE the maximum that the benchmark had reached previously.
AMD should look into this and ask what 3DMark thinks it is playing at - if necessary with lawyers.
Same thing happens to my 3700X manually set to 4.25ghz, it's running 120-150mhz slower than peak, so it's not a Ryzen 5000 series issue. I would ask anyone with an Intel CPU to see if the same thing holds true with them, but it's a Futuremark issue, and likely not one that would require legal bantering because 3DMark isn't a workload application, just a program which generates a meaningless number.
