8350 Power Consumption

[Zeus]

so this is the new PS4 amd chip their using

PS4? Forgive my ignorance but do you mean Play Station 4 by chance?

[tbone8ty]

yes I meant PlayStation 4

[tbone8ty]

Got a 1249pgt fx-8350 today. Will test later tonight

Hopefully they weren't already on holiday when they spun the silicon haha lol

[FlanK3r]

where do i find the x246 benchmark?

is this it?

http://www.techarp.com/showarticle.aspx?artno=520

yes

[The Stilt]

Things seem to have improved a bit...
I see up to 15-20% reduction in power consumption on FX-83xx series parts since the launch
The average seems to be around 11%.

Maybe the chip fairies have been working overtime

[Tomasis]

stilt, good to hear

i took out power test meter.

Idle is about 110 watt for the whole case.

At stock (volt 1.425) I get about 230 watt when I run CB benchmark.

Then I oc'ed to 4,6 or 4,7 (i forgot) and raised volt a little and still 250watt.

All this is CNQ and all other enabled, a lot volt going up and down of course (lack of that load line)

Not bad for the Fx

[AMDforME]

Things seem to have improved a bit...
I see up to 15-20% reduction in power consumption on FX-83xx series parts since the launch
The average seems to be around 11%.

Maybe the chip fairies have been working overtime

GloFo's 32nm process is pretty well refined. They have 28nm running and 20nm and 14nm enroute. These will all provide additonal power savings, though 125w TDP under full load at default values isn't an issue for most folks. Lower is always nice but not a deal breaker for most people, especially considering most PCs are under full load for only a small percentage of time in typical use other than stress testing, looking form Primes, etc.

[mongoled]

Well here are my results to add to Daveburt714 thread.

All power saving options enabled in bios.

Check sig for rig details (watercooled etc).

I havnt figured out what is going on with the voltages.

HWMonitor / CPU-Z show no change when the multiplier drops.

AMD Overdrive, shows changes, but the voltages are incorrect.

If anybody has any ideas on what I can do to get the correct voltages (excluding using a DMM let me know).

Idle 2.2Ghz



Idle 5Ghz



Prime Full Load



Dont think I will be keeping the 5Ghz as 24/7 settings, as its not prime95 stable (IBT / OCCT are OK) as power consumption is huge

[The Stilt]

I have been living in a faith that Prime95 is the worst thing you can throw at your CPU.
I was wrong.

AMDs own test program (called "Hotdog") makes the chip draw almost 14% more power than Prime95 LargeFFT (V27.9)
Kabini and Kaveri optimized versions work on 15h Gen1/2 too, however they result a lot lower power draw. Maybe it is a sign

[mongoled]

I have been living in a faith that Prime95 is the worst thing you can throw at your CPU.
I was wrong.

AMDs own test program (called "Hotdog") makes the chip draw almost 14% more power than Prime95 LargeFFT (V27.9)
Kabini and Kaveri optimized versions work on 15h Gen1/2 too, however they result a lot lower power draw. Maybe it is a sign

Any chance we can get access to it

[FlanK3r]

agree with mongoled, I want to have AMD stress program too . AMD could try new program for biggest enthusiast in forums with some FTP link and we could testing new OC*tweaking etc softwares for users .

[The Stilt]

"Hotdog" is part of AMD ThermNow software, which is under NDA.
Besides there is no real use for such a program, as the resulting stress does not correlate by any means with the stress levels generated by the real world applications:

DCR Peak Power by application (during 10 minutes):

100% - Hotdog 15h V0.02
87% - Linpack 11.0.1.005
87% - OCCT 4.40 (CPU OCCT - SDS)
87% - Prime95 V27.9 (LargeFFT)
83.8% - AOD 4.2.6 (Calculation Test)
83.8% - Prime95 V27.9 (SmallFFT)
72.9% - Blender 2.66a (Burning Plane scene)
70.6% - Cinebench R11.5
68.4% - X264 R2273
67.3% - Euler3D Stars (CFD Benchmark)
66.2% - HyperPI 0.99b
65% - AOD 4.2.6 (CPU Test)
62.9% - AOD 4.2.6 (FPU Test)
56.2% - wPrime 2.09

[FlanK3r]

how about AMD overdrive stability test? Can you compare it with others?

[The Stilt]

Added.

83.8% - AOD 4.2.6 (Calculation Test)
65% - AOD 4.2.6 (CPU Test)
62.9% - AOD 4.2.6 (FPU Test)

[cab]

how about AMD overdrive stability test? Can you compare it with others?

It looks like AMD stopped updating AOD a while back, it can't even read the sensors on my sabertooth 990FX R2.0

[FlanK3r]

Last version is 4.2.6
http://download.amd.com/Desktop/aod_setup_4.2.6.exe

I use some 4.2.x. I think, we will se some update with Richland or with Vishera new models at summer or so.

[demonkevy666]

Last version is 4.2.6
http://download.amd.com/Desktop/aod_setup_4.2.6.exe

I use some 4.2.x. I think, we will se some update with Richland or with Vishera new models at summer or so.

yes it's been in dire need of a revamp.

I'd like to see some SSE (1,2,3,3S,4.1,4.2,4.1A), AVX, AES, XOP, FMA load tests

and some kind of way of checking which is being used on the cpu.

[The Stilt]

A bit off topic, but..



http://valid.canardpc.com/show_oc.php?id=2837872

And thats a ultra low-leak part.
The first run too

Should I try 6GHz?

[Evantaur]

Should I try 6GHz?

on water?

edit: nvm noticed temps only now

[FlanK3r]

omg, you say, this is with air/watter ?! My best is 5740...

[The Stilt]

On AMD FX (Asetek 570LC) watercooler
It has only one of the original high cfm fans installed too, the second one is some random 120mm fan.

[The Stilt]

Replaced the exhaust fan on AMD FX CLWC with a Scythe Ultra Kaze.
The ambient was lowered by couple of degrees once the window was opened.



http://valid.canardpc.com/show_oc.php?id=2838300

Now I would say that the 'big nine' is only a matter of helium.

[felix_w]

What was the ambient ? Cold weather still in Finland, right ?

[The Stilt]

What was the ambient ? Cold weather still in Finland, right ?

Now it is a bit colder again, this morning it was around 16 celsius outside.
The ambient was at 24 celsius during the run.

[felix_w]

Impressive...not very low temp as i expected, yet your result is probably one of or the highest with this type of cooling...