Finally I take some time for testing lovely Visheras chip. This is first part, because I have still many other work to do.
First one is 1327PGS
Cooling AIO AMD Cooler liquid - so a bit better than Noctua NHD14
FX-9590 is highend CPU with high TDP and of course high stress for motherboard VRM (CPU want more current for right working :-) ). The clocks are 4700 MHz and up to 5000 MHz for light stress. I used again older Asus Crosshair V Formula. Formula-Z is a bit better. At C5F is not official support, but CPU working OK. Only the one issue is CPU ratio * is set default at 23.5, not auto. So your CPU is not underclocking here. Thats mean a bit higher idle power consumption (+20-30W I mean)
So, what about power consumption at stock settings?
As I mention, the CPU ration in idle is still at 23.5 (4700 MHz), the power consumption of PC is:
In practice load in Cinebench render is:
In stress stability test is:
stress in x264FHD (Cinebench description is mistake in editor)
So, we can try this CPU undervolting? Yes
I tried few steps in undervolting. First one in stability test for 30 minutes. 1.44 V in idle, 1.416V in load. LLC for CPU and CPUNB set as HIGH. Ussually in practice use will be power consumption around 30 W lower than in stability stress test...
Second step is a bit better, nice settings with 1.428 V and 1.404V in load.
Third step of undervolting in stress test, next settings is and now "only" 301W in load with 1.38 V in load! Idle jumping between 1.404-1.416
PS: Im still testing next steps in udnervolting. In next week we will see some results and OC resutls as popular benchmarks and some OC stability tests of first 9590. Later second CPU...
Next part of undervolting
First under 300W PC in stress test, 1.392 V in idle and 1.368V in load
This part is 1.38 V in idle and 1.356V in load!
Final max stable undervolting settings for FX-9590 stock clocks 1.368V in idle and 1.344V in load.
There was error after 20 minutes :-/, but still not bad result
So, first part of FX-9590 is done. Undervolting at stock clocks. Its awesome to know, there is difference around 70 W in power consumption for load!!!
Bookmarks