Unofficial Opteron 939 Thread

[omi7000]

correction to my 148 from Monarch: CABYE 0535XPMW

[ixtapalapaquetl]

Thanks for your data guys. I have a new theory... but first, regarding OCCT: Just ran 30 min test. At 15 min, I started prime on Core 1. At 22 min, I started another instance of prime on Core 0. At 28 min I closed one instance of prime, and at 29 min I closed the other. These actions are reflected in the temperature graph below:

 

 

 

What about my Vcore during this time?

 

 

 

Needless to say, I CAN NOT recommend using OCCT as an accurate barometer of what is actually happening to your Vcore (sorry about the mixed metaphor). My earlier post clearly shows the above graph has no bearing on reality.

 

 

Here's my latest theory: ncsa had virtually no increase under load with his 146, where EdLSmith did with his 148. This got me thinking...

 

 

As you can see, the nominal current draw of these two chips are 67W and 85.3W respectively, whereas mine draws 110W. Therefore, this phenomenon may be model dependent. I PM'd a guy with a 154 a little while ago. If he has a multimeter, we may have some confirmation on this theory.

 

If you have a Dual Core or high end Single Core Opteron and a multimeter, I would love if you could check your idle/load Vcore. Thanks.

[tatts]

nice work so far ix

as always been something ive been interested in knowing but dont fancy poking my mobo with a multimeter lol

[loc.o]

146 CABYE 0540 FPMW on its way

[ncsa]

Here are the updated results:

 

 

The 136% was not fully tested as it jumped to 1.4v when the calculation was 1.326 .. this I suspect maybe due to the % not liking a base of < 1.00v.

 

Anyway, it appears that 123% & 126% are the most stable, for my MB at least.

 

With 123% set, OCCT was run - 30min test, and according to OCCT the voltage dropped to 1.27v in the display..but this is not reflected in the graphs - see below...

 

 

CPU Temp:

 

Vcore:

 

There are a few bumps along this Vcore graph and I suspect that this maybe due to the low Vcore and high FSB settings ... so maybe it is on the edge of stability for these settings.

 

It will prove itself once the Vcore gets bump further in order to get a higher OC.

 

@ix - we should also run the MaxTCase tool and see how this also reflects in relation to the Wattage theory ...

 

EDIT:

 

[Squid1]

Allright, I tested this:

 

Thanks Martin but I should have been asking for CPU VID multiplier.

ie, 104%, 110%, 133%, 126%, etc.

or voltage readings at 1.5 vs 1.2x126%

trying to understand if the voltage error is different using the VID multiplier.

(long day at work, hope I don't sound more brain dead than usual)

 

It is not different for me with my current Opteron, board and BIOS.

 

I verified that a completely stable overclock(*) at 1.4 Volts is exactly at the same MHz within 20 MHz, no matter whether I compose these 1.4 Volts by using 1.4 * 100% or whether I use lower base volts and a special vid factor.

 

I did the same test at 1.525 Volts and 1.53 volts (from 1.1125 * 136%) and the same results: max overclock is within 20 MHz the same.

 

My multimeter also says I get exactly what I tell the BIOS I want.

 

From what I can see the supposed difference between using special vid and not using it does not exist.

 

%%

 

(*) = the 1 hour variant.

[ixtapalapaquetl]

@ix - we should also run the MaxTCase tool and see how this also reflects in relation to the Wattage theory ...

Thanks for the help ncsa - bet you're lovin that Vapochill! You have clearly demonstrated that your 146 (thankfully) does not experience this phenomenon.

 

I tried MaxTCase before to no avail:

 

 

I tried a few different versions with no help. Have no idea why it doesn't like me. But it is still a very good thought, and may be useful for other users.

 

Two more items: OPPAINTER has reported that he has similar jumps in Vcore under load with his 154, so the high current draw theory is starting to look good. Also, derektm should be reporting back fairly soon with his results on a 170.

[ncsa]

Yes it is a very nice unit which is fun to use

 

I have run MaxTCase and I get temp details will post it up with the previous thread once I get the image off the system a little later on.

 

Thanks for the help ncsa - bet you're lovin that Vapochill! You have clearly demonstrated that your 146 (thankfully) does not experience this phenomenon.

 

EDIT: Updated SS .. interesting that mine shows up as skt 940 but E4 and your is an E6

[Mott13]

Got an opty 146 commin from ZZF with my soon to be ultra-d at some point a powerstream 520 and a 7800gtx, how will my old school BH-5 play with this setup?Verry new to the 64 thang...

 

btw I will be watercooling the cpu and gpu:)

[SubZeroGTS]

so what would you guys say about 'safe' voltages for an Opteron dual core? i'm thinking of setting the bios no higher than 1.5v for my upcoming Opteron 170, since i don't know if load will spike it to 1.55.

 

i wonder if i can hit 2.8+ghz on that

[ViperJohn]

Here's my latest theory: ncsa had virtually no increase under load with his 146, where EdLSmith did with his 148. This got me thinking...

 

 

As you can see, the nominal current draw of these two chips are 67W and 85.3W respectively, whereas mine draws 110W. Therefore, this phenomenon may be model dependent. I PM'd a guy with a 154 a little while ago. If he has a multimeter, we may have some confirmation on this theory.

 

If you have a Dual Core or high end Single Core Opteron and a multimeter, I would love if you could check your idle/load Vcore. Thanks.

 

That is power drawn in watts not the current drawn in amps. To find watts input you multiply the voltage applied (Vcore) by the current drawn (Icore) to get the Power input (Wcore).

 

To get 100W input at 1.40 Vcore applied the current drawn from the Vcore power supply will be 71.43 amps. Assuming an 75% efficiency of the on board Vcore switching power supply that would translate into 11.11 amps (133 watts) drawn the from the computer power supplies 12V rail at 12 volts dead nuts to power the CPU.

 

When over clocking the power input in watts will increase at the ratio of the frequency change.

 

When over volting the power input in watts will increase at the square of the ratio of the voltage change.

 

If your CPU draws 100w input at say 2.4Ghz with 1.40 applied Vcore then that same CPU at 3.2Ghz with 1.60 applied Vcore the watts input will be:

 

3.2 / 2.4 = 1.333

 

(1.60 / 1.40)^2 = 1.306

 

1.333 x 1.306 = 1.740

 

1.740 x 100w = 174w input at 3.2Ghz with 1.60 applied Vcore.

 

Viper

[Squid1]

 

As you can see, the nominal current draw of these two chips are 67W and 85.3W respectively, whereas mine draws 110W.

 

No. This is the thermal design power (TDP) which has very little to do with actual power consumption.