Greetings to all!

I've been lurking around the site and forums here for a long time since I'm content on finding out things myself, but I've finally found something that has truly stumped me! I'm looking into a new system to replace my big Core 2 Quad Q9400 build with similar or better performance, but with lower idle power draw (because the current build draws 95W on an Earthwatts 380D... and it was 115W on the previous Corsair HX620W!).

I've been flip-flopping between an i5-661 and an i5-750. Obviously the 661 will idle much less due to integrated graphics, however I intend to use a discrete video card anyway and thus will probably go for a P55 chipset. And this is where the confusion sets in! I'll simply make a list here:

i5 661 idles lower than the 750:
- SPCR (Intel P55, 6W DC difference): http://www.silentpcreview.com/article1013-page5.html
- TechSpot (both on Asus P55, 18W diff): http://www.techspot.com/review/233-inte ... age15.html
- Hexus (Intel H55, 13W diff): http://www.hexus.net/content/item.php?item=21684&page=7 (though this pits i5-661+GMAHD against i5-750+GT220 so not really a fair comparison)

i5 661 idles EQUAL TO OR HIGHER THAN the 750:
- Bit-Tech (Gigabyte P55, equal): http://www.bit-tech.net/hardware/cpus/2 ... u-review/8
- TechPowerUp (Asus P55, 6W diff): http://www.techpowerup.com/reviews/Inte ... 61/12.html
- Anandtech (Asus H57, 9W diff): http://www.anandtech.com/show/2901/15 (though site acknowledges results could be faulty)
- TomsHardware (Asus H57, 8W diff): http://www.tomshardware.com/reviews/int ... 14-15.html

There are probably more but I figured I wasn't going to find a holy grail of conclusion anytime soon! On one hand, I trust SPCR most and a dual core 32nm logically should idle less than a 45nm quad. On the other, they cost the same so I'd be losing two cores and paying for integrated graphics I wouldn't use (I do some heavy video editing sometimes so the quad would help!).

At the end of the day, I want to reduce my CPU power consumption (the 8800GT graphics is next on the chopping block!) but I would rather not lose too much of my current CPU horsepower. I don't care much for load power draw as the rig will be mostly idling. Is there something I'm missing in regards to the power draw between these two i5s?

Thanks in advance!

(EDIT: Improved readability)

If you can hold your breath for 9-12 months, you can have your cake and eat it with the Sandy Bridge chip, and faster better 750 with an embedded video chip.

Fair enough, but I'm looking to replace it pretty soon as my current system's components still have fairly good resale value. And a year is quite a long wait.

What are your video capability requirements?

I won't be using the integrated graphics - instead I'm aiming to get something like the ATi 5670, which is low-mid range but quite low idle consumption. Should give it decent gaming capabilities!

If you meant video conversion, I sometimes have to work with the AVCHD footage from my camcorder which takes quite a bit of grunt to process. I originally upgraded from a E6750 to a Q9400 and it helped speed things along greatly, so I don't want to drop too much performance for the new rig.

You're talking about 10 watts at most. The video card and support chips are more important. Unfortunately the 1156 and h55 / p55 are being superseded by the 1155 and the H67 / P67 with Sandy Bridge.

So you by saving your current resale value, you will be impairing the resale value of what you are buying now.

If your current setup works, why not wait till the end of the year. You will get sata 3 with the new system too.

I appreciate the suggestions ces; however I do have other reasons to upgrade the system as well, which is why I kept the original question specific!

In brief, this is actually replacing two systems - a low-power mini-ITX system and the big rig mentioned above. If I can achieve optimal efficiency with one system then I can sell off both systems for probably more than the cost of the new system, whereas if I wait for the new bits to come out both systems will quickly lose their value!

(And I can't wait 12 months. )

So anyway, what is going on with the differences in idle power between all the review sites?

If the reviews are split on which has the higher power consumption, then I'd say it's a toss-up. They're close enough to be called the same. They're also the same price.

However, one thing the reviews agree on is that the i5-750 is faster. The faster your processor is, the faster it finishes its work and gets back to idle power. Also, the faster it is, the faster it finishes its work and you can see the results.

Quite frankly, the i5 600 series doesn't make a lot of sense when the i5 750 is available for the same price and a similarly performing i3 530 can be had for almost $100 less. The only thing the i5 600 has going for it is audio bitstreaming which is rendered unnecessary by a discrete HD5670.

My vote's on the i5 750.

I don't know that we can help you any more than we have.

Why don't you just get the 750?

Thanks guys, I think the 750 is to go!

Why I asked in the first place though is because of the rather significant differences in power draw between the two i5s over different reviews. I mean, even if someone had a faulty testing technique it could give wrong results - but to the point where one CPU has lower consumption than the other when it shouldn't? Just thought it was an interesting thing to try and figure out...

Not quite exact: the -661 idles at a mere 2.5w lower than the -750

http://anandtech.com/show/2921/2

About performances, your mileage may (slightly) vary

http://www.anandtech.com/bench/Product/117?vs=109

Regards,
Luca

That's a good find, thanks! Guess they did fix that one up.

Then again, a few more clicks and I found this: http://www.anandtech.com/show/2972/the- ... reviewed/8

...so now its 10W more. They might be using integrated vs discrete here, but the game benchmarks on the previous page show all the CPUs having similar performance which is unlikely if integrated was used.

Ugh, man, so they can't even get consistent power measurements between reviews now...

They're using different set-ups on those reviews. On the i3 530 review, they were using Gigabyte H57 + Radeon HD5870. On the other one (Rest of Clarkdales), they were using Asus H57 + GeForce GTX 280.

Just goes to show how much motherboard selection matters. It also shows that NVIDIA's power consumption at both idle and load sucks.

Using Anand's comparitor I am really surprised at how the i3-530 stacked up against the i5-750. Particularly the number of real world benchmarks in which the 530 significantly outperformed the 750.

http://www.anandtech.com/bench/Product/118?vs=109

In what benchmark (aside from power consumption)? I did a quick check and it looks to me that the i5 750 beats the i3 530 in each one. Albeit, the difference doesn't seem to be very significant. Certainly not in parity with pricing considering the i5 750 costs nearly twice as much. Just a note, some of those graphs are in seconds meaning lower is better.

I stand corrected, I was misreading the bar graphs. In some instances higher is better, in some others lower is better. I am used to organization/coloring to be done differently in benchmarking suites and reports. It would be really nice if these were inverted appropriately.

I agree wholeheartedly! Though I still find it bizarre that the power consumption gap widens between the two by so much on different equipment (2 to 10W!). I guess those differences would be attributed to equipment details far too minute to discern.

Actually, from a silence perspective nothing may worth more than a value line cpu (i.e. Pentium G6950 or Core i3-530), which can be easily be overclocked up to 4GHz effortless (for any performance on demand profile).

If substantial more performance is required (i.e. for some serious gaming, or special applications), then the very first worthwhile option would be a triple channel Core i7-930: but such an option would also be thermally incompatible with a really "silent PC" (even if it could be pretty quiet).

Regards,
Luca

There's always place for a happy medium.

Besides, the Lynnfields are surprisingly conservative when it comes to power consumption, only ramping up during heavy load. The only issue I see is price - the i3 530 can be bought on sale for $100 and a Gigabyte H55+i3 540 combo from Fry's costs even less than the chip alone on Newegg. Hmm, is there a review somewhere that compares power consumption between Lynnfield and Clarkdale when only two cores are used?

By the way, the only benefit with regards to gaming is if doing SLI. Tom's Hardware did a review and determined a Core i3 530 isn't likely to bottleneck a single GPU system (HD 5850).

I wonder, whether the only issue is a tag, what about that happy medium space (I mean for i5, both -66X, -670 and -750).

Regards,
Luca

I dunno. For me, among those chips, the i5-750 is a no-brainer (unless you've got a MicroCenter nearby and can get the i7-860 for the same price ). The i3-530 is basically the same as more expensive Clarkdales and you can easily bring performance on par or even beyond that of the i5-600 series by oc'ing. However, despite hyperthreading, there are still applications (video encoding and 3D rendering) which benefit from the two extra physical cores on the i5-750.

The 750 is usually faster at recoding A/V and often faster with the latest games.

I think you inadvertently confused the "Lower is better" with the "higher is better".