Pentium Dual Core e6300 vs Ivy Bridge i7 power consumption

[2is]

This isn't a question, but an observation. The tinkerer in me was feeling neglected so I slapped some parts into a case consisting of

Pentium Dual Core e6300
4x2GB DDR2 800
Radeon HD5870
1 Optical drive
1 7200RPM HDD
500watt 80+ PSU

And compared it's idle power consumption with

i7 3770k @ 4.2GHz
4x8GB DDR3 1600
EVGA GTX 980Ti FTW (factory OC)
GTX 650Ti (dedicated PhysX)
3 SSD's
3 Optical Drives
1 7200RPM HDD
750watt 80+ Bronze PSU

Both systems had all the default power saving features enabled, nothing overclocked on the Pentium machine. Both machines running Windows 10 (no I didn't waste a COA on the old build, used a 90 day evaluation) and the results?

Pentium box = 115 idle watts
Ivy Bridge box = 100-102 idle watts

The strides in efficiency are pretty amazing when you think about it and actually compare two machines several generations apart. In addition to the IB system running two video cards, I also have the intel IGP enabled and driving my second monitor so I can have QuickSync enabled, so that's 3 GPU's vs 1, not to mention everything else and it's still pulling about 15 watts less. I'd imagine Skylake would have been even more impressive.

 

[VirtualLarry]

That is pretty nice.

I've noticed the same thing, and that's one reason why my main machine is a G3258 @ 4.0Ghz, instead of my Q9300.

Last month I was running the Q9300, to refurb it, and mess around with it. I put in one of my R7 250X 2GB DDR3 cards (not that fast, sadly). Crunching BOINC, using both 3 core CPU cores and the GPU, I measured a power draw of around 175W total. I sadly don't remember the idle, but it was above 100W I think.

My G3258 w/7950 3GB GDDR5 card, crunching the same projects in BOINC, my power draw is like 270W. But idle without crunching, is like 78W.

So the newer rig, takes more power when fully loaded, but takes less at idle. (Edit: But for that higher power draw when crunching, I get a lot more points.)

Which kind of makes sense. I've got an EVGA 500B PSU in the G3258 rig, and the GPU takes two 6-pin PCI-E power connectors. The Q9300 rig has a Antec VP-450 PSU with a single 6-pin PCI-E power connector.

Edit: I think that you would see even lower idle wattage, if you disabled the iGPU, and removed one of the two GPUs. (I don't think that the 650ti is all that efficient at idle, it's Fermi, right?)

I should note that my power readings were including a 25-30W 24" HDTV LCD monitor. (Yeah, I used a 24" HDTV with a single HDMI input as my monitor, because I was too cheap to spend the $20-40 more for a "real" monitor. It worked out well though, being an HDTV, the speakers are better than most monitor speakers.)

Edit: Maybe you could sell that old box as an "Older Custom Gaming Rig" on CL for fairly cheap but not too cheap? Someone might want a lower-end gaming rig for Christmas. (If you don't want to part with it, I understand. But that's still competent for older games, IMHO.) I was thinking of unloading my Q9300 w/R7 250X in the same manner.

Edit: Another under-stated point of your comparison, that's a MUCH faster quad-core (8T) CPU, compared to just an older dual-core. So to be taking LESS power, is pretty amazing.

 

[SPBHM]

you forgot to mention the motherboard, which is the most important thing for low power consumption with a 775 system (and still a big factor for the new platforms), some high end boards had ridiculous power draw

with a G31/G41 board and IGP I was getting around 40W idle with a Pentium e5400 and Xeon e5420 (quad core), could probably get under 40 with the right components.

 

[Magic Carpet]

you forgot to mention the motherboard, which is the most important thing for low power consumption with a 775 system

With any system.

 

[Schmide]

5870s were not known for their idle consumption. They were leaps and bounds better than the 4000 series, but still in the 20-40w range.

Slap the GTX 650Ti into the low end system and compare.

 

[SPBHM]

With any system.

yes but I think specially with 775 the variation was far bigger than with newer systems,

5870s were not known for their idle consumption. They were leaps and bounds better than the 4000 series, but still in the 20-40w range.

Slap the GTX 650Ti into the low end system and compare.

yes, the 5850 reference that I have uses 25W with a single monitor, a lot more with another one

 

[2is]

Didn't mean to imply all the extra power was due to the CPU alone, I know it's the system as a whole. 5870 is certainly not very efficient when dual monitors are connected to it, which I do have. Also got duals going on the i7 machine, but running off of two different GPUs. I'll double check the chipset when I get back home, I THINK it was a G43 but not positive.

 

[Magic Carpet]

The G4x line was very power-efficient but also feature-limited compared to previous power houses.

Let's just see how many chipsets LGA775 has lived though:

i845GV/GE/i848P/i865G/GV/P/PE/i910GL/i915G/GL/GV/P/PL/i925X/XE/i945/955/i945G/P/ i955X/i946/946GZ/PL/965/i975/Q965/P965/G965/Q963/i975X/ X35/P35/Q35/G35/P33/G33/Q33/P31/G31/X38/X48/P45/P43/G45/G43/G41/B43/Q43/Q45,

SiS:
SiS 649/649FX/655/656/656FX/662/671/671FX/671DX/672

VIA:
PT800/PM800/PT880/PM880/P4M800/P4M800 Pro/PT880 Pro/PT880 Ultra/PT894/PT894 Pro/P4M890/PT890/P4M900

Ati:
ATI Radeon Xpress 200; ATI Radeon Xpress 1250, ATI CrossFire Xpress 3200

nVidia:
nForce4 Ultra; nForce4 SLI XE; nForce4 SLI; nForce4 SLI X16; nForce 570 SLI; nForce 590 SLI; nForce 610i; nForce 630i; nForce 650i Ultra; nForce 650i SLI; nForce 680i LT SLI; nForce 680i SLI; nForce 730i; nForce 740i SLI; nForce 750i SLI; nForce 760i SLI; nForce 780i SLI; nForce 790i SLI; GeForce 9300; GeForce 9400

 

[Magic Carpet]

yes but I think specially with 775 the variation was far bigger than with newer systems,

Just to give you an idea how power figures can fluctuate between different boards within one product family. Let's use modern tech for this case. In all tests the same 6700K/PSU/RAM/SSD were used. And here are the numbers:

In this case, the board with the most-efficient power delivery system wins. The system board can even define the PSU parameters required for the job.

If you are into building power-efficient rigs, it usually takes more than just looking into processor TDP numbers. Every component must be "right", but mainboard along the power supply unit usually lead the way.