Is Turbo Mode less efficient?

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
Hey everyone. Overclocking my D0 920 and I noticed something. If I enable "Turbo Mode" my TDP in Core Temp is 147.3 watts vs the usual 130 watts.

For example, I'm OCing to 4Ghz.

21x + 191 = TDP 147.3 watts (Turbo Mode on)
20x + 200 = TDP 130 watts
19x + 211 = TDP is 130 watts

Does this mean that Turbo Mode is less efficient?

 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
Okay, as far as I can tell, the 147.3 Watts shown in Core Temp looks to be an anomaly in which the mobo can't figure out how to read the CPU correctly. Correct me if I'm wrong but I figured that if the TDP went from 130 Watts to 147.3 Watts, it was consuming more power.

In fact, my UPS says the exact opposite. looked at the display on my UPS (showing Kw output) under two different settings shown below.

Setting 1

CPU Ratio: 19x
BCLK: 206
TDP: 130 Watts
Speed: 3.9Ghz
Kw Idle: .263Kw
Kw Load: .413Kw


Setting 2 (Turbo Mode)

CPU Ratio: 21x
BCLK: 186
TDP: 147.3 Watts
Speed: 3.9Ghz
Kw Idle: .255Kw
Kw Load: .405Kw


So by using Turbo Mode (or manually setting the CPU Ratio to 21) with a lower BCLK it results in using less power than an equivalent speed using different settings.

Is this because the Motherboard is using less power since it has a lower BCLK?
 

overshocked

Distinguished
Feb 14, 2009
1,999
0
19,960
Lol, we are talking about very small differences in the system power draw. The minimal change you see could be due to a change in voltage (assuming you have vdroop on). Or it could be due to the small increase in ram speed because of the higher base clock.
 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
Oh, I know it's a minimal amount. I was merely worried that the 920 has a TDP of 130 Watts and with Turbo Mode enabled it jumped to 147.3. I just wanted to make sure I wasn't losing efficiency because I plan on running this machine for rendering for hours to days at a time. Any bit of efficiency loss in my own home is understandably a concern of mine.

I was just happy to see that I can achieve the same OC, without changing voltage, using less overall power draw - insignificant as it may be.

I repeated the OC tests/settings. They were consistent every time with power draw.

Oh, and vdroop is disabled so that wouldn't be the cause. I'm sure it had something to do with the lower BCLK for RAM and Mobo. Yes?
 

overshocked

Distinguished
Feb 14, 2009
1,999
0
19,960
Vdroop will drop the voltage depending on the load. So i was thinking there might have been a higher load on the cpu when you read it once than when you read it with the different settings. And thus the power draw was higher the second time.
 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
I know what you were getting at. I would have thought the same but I always disable vdroop.

Now here's something even stranger:

I disabled Speedstep to see if there was a difference in power draw when idle @ 2.4GHz vs idle @ 3.9Ghz.

Power draw:

Idle @ 2.4Ghz (speedstep enabled): .255W
Idle @ 3.9Ghz (speedstep disabled): .248W

Load @ 3.9Ghz (speedstep enabled): .405W
Load @ 3.9Ghz (speedstep disabled): .398W

Again, a small margin but also again (and contrary to logic) what should be using more power is, in fact, using less.

Does this mean Speedstep is a useless technology?
 

overshocked

Distinguished
Feb 14, 2009
1,999
0
19,960
Where are you getting these values again?

I assume you are using a program? A much more accurate way would be to use a kill-a-watt reader with the system under load and not under load etc. I have tested it this way and have had a lower wattage draw with speedstep on
 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280

lol...nice.

Okay, so when you tested yours, was it your i7 system? If so, what was the difference in power draw between having speedstep enabled and disabled?
 

overshocked

Distinguished
Feb 14, 2009
1,999
0
19,960
Dont remember the exact difference, but i know there was a small difference.
Besides, if your running it for rendering, it will always be loaded, so it will be at full speed not stepped down.

Even with speedstep, the wattage difference will be very very small.

BTW, i tested it with my old e8500, not the i7.
 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
Alright bro. Thanks for all the responses. I appreciate the info.

Hey, can you add your specs on my other thread about posting 920 OC BIOS settings? Nobody's responded and I think it could be helpful. You wrote a great 920 OC guide and I figured if everyone posted their stable OC's all in one thread, it could help the cause as well.

Thanks either way!
 

overshocked

Distinguished
Feb 14, 2009
1,999
0
19,960
Aww man, my system isnt stable. I will post my specs for you here though....

Base clock: 226
Multi: x21
Voltage Control
EVGA VDroop Control ( Without VDroop )
CPU VCore ( 1.425 )
CPU VTT Voltage ( +300 )
CPU PLL VCore ( 1.815 )
IOH PLL VCore ( 1.815 )
DIMM Voltage ( 1.65 )
DIMM DQ Vref ( +0mV )
QPI PLL VCore ( 1.4 )
IOH VCore ( 1.225 )
IOH/ICH I/O Voltage (1.6)
IOH VCore ( 1.225)
VTT PWM Frequency ( 250 KHZ )
CPU PWM Frequency ( 800 KHZ )

I wrote another even better guide that is a pretty popular non-stickied thread.

It is in my sig.

There are also many more settings in this thread:
http://www.tomshardware.com/forum/256708-29-intel-overclocking-club
 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
Whoa! 4.75Ghz? How long can you run Prime before it errors/reboots? Do you have a D0? I'll check out the guide. Hey, one last question. From what you've seen, what's the average/typical voltages (just vcore and QPI) needed for a D0 to hit 4Ghz? And is that with or without Speedstep?
 

overshocked

Distinguished
Feb 14, 2009
1,999
0
19,960
Haha, yes i have a d0. I never ran prime because my temps were to high. That was on air dude!

Off hand i have seen anywhere from 1.28-1.34 needed to hit 4 ghz most of them fall between 1.29-1.32v vcore. Vtt and qpi vary greatly. Some take no adjustmants some tak up to 1.45v. My system is without speed step. Most people take off speed step to make the sytem less stable.
 

krazynutz

Distinguished
Jan 30, 2003
1,023
0
19,280
Oh okay. That makes me feel better. Those voltages seem a bit more realistic on average. I read somewhere the average 4Ghz voltage was between 1.2 and 1.25v for a D0. I thought that can't be for a true 24-hr prime stable machine.

Thanks for all the great info and congrats on the OC on air!