Personal computing discussed

Moderators: renee, Flying Fox, morphine

 
ronch
Graphmaster Gerbil
Topic Author
Posts: 1142
Joined: Mon Apr 06, 2009 7:55 am

Impact of Vishera Undervolting

Sun Jan 27, 2013 12:00 pm

Hi, Gerbils. I've been tinkering with my month-old rig, trying to undervolt my CPU to see how much impact such a thing would have. My specs:

AMD FX-8350
MSI 990FXA-GD65
2 x 4GB G.Skill DDR3-1866
Powercolor HD7770 1GB
TP-Link TL-WN723N USB 2.0 Wifi Adapter
Hitachi 7K1000.C 1TB HDD
LG ODD
Cooler Master GX650 PSU
Windows 7 64-bit Ultimate
AMD Catalyst 13.1 + latest chipset drivers

In the BIOS there's an 'Auto' setting for the CPU voltage. I tested stability at the lower voltages using BurnInTest from Passmark and the lowest I could go without getting errors is 1.317v (at least that's what the BIOS says). I monitored voltages within Win7 using CPU-Z. At stock, the CPU voltage at idle is 0.920v, load at 1.4v. Set to 1.317 in the BIOS, CPU-Z reports idle voltage to be 0.864v, load volts at 1.28v. BurnInTest tests stability with all cores at 100%, so it's not gonna be pretty on my power bill. Tests run for 15 minutes. I did two runs, so that's 30 mins.

To test performance I haven't used any benchmarks, but rather, decided to find any perceived differences in actual gameplay between the 'Auto' voltage setting and 1.317v in the BIOS using UT2004. UT2004 generally runs very smoothly, and it's set to run at just 1280 x 1024, a resolution that I believe the HD7770 should be comfortable with. I specifically use the Osiris2 map because using diff. maps could make the test muddy; not all the maps are smooth. Osiris2 is somewhat between a silky smooth map and a difficult map; it's silky smooth but sometimes there are hiccups, specifically at 1.317v. I go around the map, no bots around, to see if there will be high latency frames, a 'hiccup', if you will.

What I found out is that at 1.317v, I seem to be having a hiccup or two than if the voltage is set to 'Auto'. At 'Auto', I can hardly see any hiccup, if at all. I know undervolting isn't supposed to impact performance at all. Plus, with the lower power consumption at 1.317v, Turbo Core should kick in more readily because there will supposedly be less heat. UT2004 is pretty much single threaded so it should invoke Turbo Core when it can and the benefits will be more useful than if a game is heavily threaded, I believe.

I'm not sure if something else is messing with UT2004 when CPU voltage is 1.317, but if true, they sure are a bunch of weird coincidences. I ran the map several times using stock voltage and 1.317. I don't know if this bit of testing on my part will prove to be useful to Vishera owners, or it's just me being too OC or something.

Any input would be appreciated. I think this deserves some investigation given how undervolting can benefit Vishera owners by decreasing power consumption especially at load.
NEC V20 > AMD Am386DX-40 > AMD Am486DX2-66 > Intel Pentium-200 > Cyrix 6x86MX-PR233 > AMD K6-2/450 > AMD Athlon 800 > Intel Pentium 4 2.8C > AMD Athlon 64 X2 4800 > AMD Phenom II X3 720 > AMD FX-8350 > RYZEN?
 
anotherengineer
Gerbil Jedi
Posts: 1688
Joined: Fri Sep 25, 2009 1:53 pm
Location: Northern, ON Canada, Yes I know, Up in the sticks

Re: Impact of Vishera Undervolting

Sun Jan 27, 2013 3:15 pm

Indeed, my gigabyte mobo used to run my AMD 955BE at about 1.40V in auto mode at stock speeds, well all the time since I disabled CnQ and C1E and all that jazz.

I undervolted it to 1.250V and it was stable 24hrs prime95, so that was good enough for me. Idle and load temps both dropped also.

I think I am going to try out one of those Richland APU's when they come out to upgrade the HTPC and reuse the HTPC parts for a PC for the garage.

I will post back when I have info on undervolting that chip.
Life doesn't change after marriage, it changes after children!

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On