overclocking i5-750 vs. power consumption (article)

[biostud]

http://www.tomshardware.com/reviews/core-i5-750-efficiency,2500.html

 

[ShawnD1]

Interesting article. Seems fairly consistent with what we've seen with other processors - eventually they hit a wall and power demand goes through the roof.

One can use the stock voltage as a reference for where the zone of efficiency is. For all of my processors, power demand goes waaaay up once the frequency limit at stock voltage is reached.

 

[cbn]

http://www.tomshardware.com/reviews/core-i5-750-efficiency,2500.html

Yep, these are good articles. (It looks like they also did power analysis for Core 2 and Core i7)

Thanks for the link.

 

[Munky]

Interesting stuff. I experimented briefly with and overclock keeping dynamic turbo enabled, but testing for stability is more tricky and time consuming that way.

However, from reading this article and seeing how very few of my workloads use all 4 cores/8 threads, I'm tempted to give it another shot.

 

[edplayer]

nice article

Makes me want to move to the i5 750 but I still think I will wait to see what 32nm quad cores pop up and hear more about when native USB 3 and SATA 6Gbps are coming out.

 

[Lonyo]

Would have been interesting to see some underclocking and undervolting too, but can't have everything!

 

[lopri]

It's one of the reasons why I'm waiting for 32nm processors for my next upgrades. I have been at 3.60 GHz for way too long.. 3.60 GHz seems like a sweet spot for 45nm quads when you consider performance/power/heat and overall system longevity.

 

[Tsavo]

Interesting stuff. I experimented briefly with and overclock keeping dynamic turbo enabled, but testing for stability is more tricky and time consuming that way.

However, from reading this article and seeing how very few of my workloads use all 4 cores/8 threads, I'm tempted to give it another shot.

I wonder how to get Turbo to work when ocing...I set the BCLK on my Giagbyte P55 UD3R and Turbo stops working, even if I have it set to enabled in the BIOS.

 

[nyker96]

I think this is true for most generation of cpus, the final 10% of oc headroom can only be reached with significant voltage increase make it not too economical. but for most enthusiasts power usage isn't that much of a concern.

 

[Munky]

I wonder how to get Turbo to work when ocing...I set the BCLK on my Giagbyte P55 UD3R and Turbo stops working, even if I have it set to enabled in the BIOS.

It depends on the board and the bios. On my asus board dynamic turbo still works even when raising the base clock, but I've read the newer bios "fixes" that issue, and that's one reason I'm staying away from the later bios.

 

[lothar]

What cooler did they use?

 

[M0RPH]

Interesting. Took a look at the i7 article and from that it seems pretty dumb to go over 3.8 GHz. So why are so many people here so intent on overclocking the i7 to 4+ GHz? Maybe more folks need to take a look at that article.

 

[MrK6]

It's one of the reasons why I'm waiting for 32nm processors for my next upgrades. I have been at 3.60 GHz for way too long.. 3.60 GHz seems like a sweet spot for 45nm quads when you consider performance/power/heat and overall system longevity.

I agree. Through my own testing I've been trying to find an overclock that maximizes efficiency, and 3.6GHz (give or take 50MHz, I don't have enough time to pinpoint it) seems to be the sweet spot.

One thing I'm surprised they didn't include in the article (maybe they didn't spend enough time with the chip) is that you can increase efficiency once you turn off the i5's power saving features if you enable Vdroop correction. It's still not an efficient way to run the CPU, but if you're going for maximum overclocks, it's the best solution.