Observations with an FX-8350

[yournamehere485]

I hardly ever post, unless I have something to ad. However, all power dissipated by the CPU will ultimately be converted into heat. Heat is simply the transfer of energy. Since the processor is stationary, energy is not transferred from it by either work, or a mass flow. (Energy transfer from the heatsink is a different matter). Therefore it is safe to say that all energy is transferred via heat.

This can included, radiation from the processor to the surroundings, convection from the processor to its surroundings, and the dominate mode, conduction of the energy to the heatsink.

 

[SlowSpyder]

Holy crap, you have a big set. There's no way I'd crank the voltage up to 1.5!

I've gone to 1.65v before, 1.5 is what the FX 9xxx CPU's are set at from the factory. I've gotten this CPU to eat over 400 watts just by itself. But that's at the edge, not stable, and just for limited runs. I have a solid water cooling set up, I would probably run this thing at or near factory clocks with an undervolt if I was on air cooling.

It just occurred to me as odd, but your turbo clocks work with an overclock and are possibly editable? When I make a change, turbo gets disabled and I get the clock I put in. So when I put in 4.2Ghz it is my max clock at load, nothing more. Not that I'd really want any higher due to the power draw. Maybe this is a quirk for getting a 970 based board? I've got the Biostar TA-970.

My bios has a base frequency multiplier and a turbo multiplier. Both top out at 6300MHz. I have a second 990FX board, can't remember if that one also allowed for a manual turbo clock change. I'm pretty sure AMD Overdrive allows for a separate turbo clock setting as well. Might be an option for you worth looking at.


Abwx PM'd me about the factory voltage settings for the FX 8350:

http://forums.anandtech.com/showpost...&postcount=620

http://forums.anandtech.com/showpost...&postcount=623

First is from 2012 review, second is from 8370E review where they retested the same 8350 but on another Asus MB, a Sabertooth this time, in both cases the VID is 1.338V, same as with IDC MB but in their case the LLC is set to auto and the delivered voltages are 1.24V and 1.27V on the Asus Evo and Sabertooth respectively, this is defined by the CPU and you cant really reproduce it with a 9370 since this latter has a different VID and power management.

So maybe different motherboards can have an effect on how much power the CPU uses. Changing my LLC settings alone from 'Auto' to '50%' made for a 61 watt difference at 1.35v. How much of that is AMD's fault and how much of that is the motherboard manufacturer's fault, I don't know. I don't think either side is pouring any effort into refining the AM3+ parts anymore.

 

[railven]

So maybe different motherboards can have an effect on how much power the CPU uses. Changing my LLC settings alone from 'Auto' to '50%' made for a 61 watt difference at 1.35v. How much of that is AMD's fault and how much of that is the motherboard manufacturer's fault, I don't know. I don't think either side is pouring any effort into refining the AM3+ parts anymore.

So where does that put us in regards to the claims made in this thread? With your findings, would you conclude that IDC was being "incompetent" "viral marketing", spreading a "fuddy review" or any of that?

Seems what you just said was what I quoted two people concluding 2 years ago, ie might be a board issue as much as a processor issue.

Should someone dig up the manual from IDC's board to see if they have any wording as to what Auto/Manual does for the damn thing? Abwx said 'set it to auto' which you did and only setting it to '50%' didn't result in huge power draw numbers.

 

[SlowSpyder]

So where does that put us in regards to the claims made in this thread? With your findings, would you conclude that IDC was being "incompetent" "viral marketing", spreading a "fuddy review" or any of that?

Seems what you just said was what I quoted two people concluding 2 years ago, ie might be a board issue as much as a processor issue.

Should someone dig up the manual from IDC's board to see if they have any wording as to what Auto/Manual does for the damn thing? Abwx said 'set it to auto' which you did and only setting it to '50%' didn't result in huge power draw numbers.

I don't believe IDC is a viral marketer for a second. Sometimes results come up that someone may not like, but that doesn't mean there is a conspiracy afoot. I'm probably the last person here who would claim IDC is incompetent, too. I've worked with him in the past on some things and always come away a little more knowledgeable myself from my dealings with IDC.

As far as the results, I wouldn't consider them scientific as my power supply is pretty large for these power draws and the efficiency can change quite a bit between ~150 watts at idle and the 400+ watts I saw when I loaded it up overclocked at higher settings.

I really don't know who to point a finger at. But this just goes along with most of my experiences with the FX. And that being that the CPU has a lot of wiggle room for tweaking, and can make it fairly efficient despite it's reputation (especially considering the 220 watt TDP version of the FX I have). I think that's fine for the do-it-yourself crowd who build their own system. However, if your HP or Dell is sucking up ~200 watts in this stress test with an FX 8350, you're not getting what you thought you were paying for based on the specs.