Intel Turbo Boost doesn't work

g00ey

Distinguished
Aug 15, 2009
470
0
18,790
I have tried different benchmarks, both multi-threaded and single-threaded benchmarks (AIDA Stress test, Passmark tests, SuperPI) but the frequency of the CPU never goes beyond the stock frequency of 3.7GHz. The max Turbo Boost frequency of the Intel Xeon E5-1620v2 is 3.9GHz, which it never reaches.

The Intel Turbo Boost monitor does show Turbo Boost enabled during load but it never goes beyond the stock 3.7 GHz. I get similar readings of these clock frequencies in AIDA64. I have checked that the "Turbo Technology" is enabled in the BIOS CPU clock settings (Asus P9X79 WS UEFI BIOS).

I also checked that the temperatures never are exceeded critical levels. The T_case temperature on the Intel Ark shows 70 degrees Celsius and AIDA64 never reports temperatures higher than 60 degrees under load. So the cooling appears to be adequate.

I have also checked that the CPU is not throttled by the power management settings of the Win7x64 operating system, i.e. the power management allows the CPU to run at 100%.

What could be the culprit, and what further steps can I take to enable this his super power mega multi incredible Turbo Boost 2.0 tech Intel is promising in the advertisement?
 

g00ey

Distinguished
Aug 15, 2009
470
0
18,790
Yes, Passmark has a single-thread benchmark and SuperPI was one of the results I got when I searched the net for a single thread benchmark. I also monitored the load of each core through the "task manager -> performance" window. When running the single threaded benchmarks, the Turbo Boost wasn't even activated according to the Turbo Boost monitor.

According to Intel, the Turbo Boost 2.0 allows to boost the CPU a number of base clocks above the stock frequency depending on how many cores that are being active as long as the CPU runs below its 130W TDP. The Xeon E5-1620v2 is a quad core CPU but I have failed to find the 'frequency bin' information when the CPU is running 4/3/2/1 cores respectively, but I'd expect it to be something like 0/0/1/2 or better which means that there should be a boost above stock frequency at least at some point during those benchmarks which I could not see.
 

g00ey

Distinguished
Aug 15, 2009
470
0
18,790
I have now tried Prime95 running "Torture tests" with 1 and 2 threads. I tried the torture test with 8 threads also. But in neither of then did the clock frequency go above the 3.7 GHz.