Sign in with
Sign up | Sign in

Conclusion

Does Turbo Boost Help Or Hurt Core i5/i7's Power Efficiency?
By

There’s no denying that Intel did a great job of optimizing its Nehalem architecture for the mainstream. System idle power went down considerably, and even though system peak power didn’t decrease much, you get a lot more responsiveness. Intel integrated the PCI Express interface into the processor and adjusted clock speed management for each of the four available cores via its second-generation Turbo Boost technology to maximize performance when needed.

As a result, the Core i5/i7’s efficiency (performance per watt) increased nicely. The new platforms can do more work at slightly reduced peak power consumption levels. You’d suspect that Turbo Boost deserves credits for this development, but this isn’t quite true, at least for the Core i7-870. Turbo Boost's massive clock speed bumps (up to 667 MHz) utilize the available thermal envelope more aggressively, which, in the end, actually decreases the power efficiency of the Core i7-870 system when compared to the same machine not running Turbo Boost. We don’t want to generalize these results, but they’re certainly the case when benchmarking PCMark Vantage.

In everyday life, Turbo Boost will probably kick in for smaller periods of time and we’d assume it typically does so for one or two cores. Scenarios in which three or all cores can be accelerated requires heavily threaded applications, and in such cases you probably want all the performance you can get. Efficiency obviously isn’t a concern at these times. Having one core running at 3.6 GHz will certainly not increase system power consumption enough to significantly impact efficiency.

We will follow up with more analysis soon.

Display all 30 comments.
This thread is closed for comments
  • -8 Hide
    MiamiU , October 23, 2009 6:44 AM
    will they ever make LGA 1156 mother boards with 2 x pci-2 x16? it would make these new cpus so much more attractive for gamers that want a more efficient pc.
  • -4 Hide
    MiamiU , October 23, 2009 6:51 AM
    i meant pci-e* no post edit=ftl

    its nice to have all these new components with high efficiency power. SSD, core i5/i7 cpus, HD 5800s GPUs, even powerful machines can be green for everyday use (e-mail, a/v streams, etc)
  • 0 Hide
    theholylancer , October 23, 2009 7:18 AM
    to avoid the cannibalization of x58s, the chance of those are very slim, especially if intel can't sell you 2 600 dollar GPUs
  • 5 Hide
    Anonymous , October 23, 2009 7:25 AM
    They already have, go have alook at the top of line EVGA P55 boards, they support 2x PCIe @ 16x speed because added a NF200 chip to them.
  • -1 Hide
    Neggers , October 23, 2009 7:34 AM
    MiamiU have alook at the top of the line EVGA P55 boards, they have added the NF200 chip to them, allowing for 2x PCIe @ 16x bus speed.
  • 4 Hide
    pender21 , October 23, 2009 7:51 AM
    MiamiUwill they ever make LGA 1156 mother boards with 2 x pci-2 x16? it would make these new cpus so much more attractive for gamers that want a more efficient pc.


    Remember that the VGA PCI-e controller in on the Lynnfield CPU itself and not on the motherboard chipset. No LGA 1156 chipset will ever have a PCI-e controller for GPU.

    The PCI-e controller on Lynnfield CPU is 1x16/x4 or 2 x8/x8.

    As for Multi CPU scaling in x8/x8 mode vs x16/x16 mode, it is usually 2% slower in framerates, but nothing more. Remember, Lynnfield is the cheaper alternative to X58 Nehalems because of the cheaper motherboard chipset (lacks PCI-e controller for VGA) and Dual Channel DDR3 vs Triple Channel DDR3.
    http://techreport.com/articles.x/17513/8
  • 1 Hide
    pender21 , October 23, 2009 7:55 AM
    Oh, there are some spelling mistakes, in the above post

    Also look at this Lynnfield/Bloomfield bench - In some newer games x16/x16 makes quite the difference
    http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3634&p=9
  • 4 Hide
    pender21 , October 23, 2009 7:56 AM
    Ohh, Foxconn and ASUS (and EVGA) apparently have a solution for x16/x16, but these boards are $220+
  • 0 Hide
    JAYDEEJOHN , October 23, 2009 12:29 PM
    Reading it now, thanks for this, as its been asked by us, the readers, and once again Toms comes thru, with the help of great writing/fact finding and benching.
  • -4 Hide
    JeanLuc , October 23, 2009 1:02 PM
    MiamiUwill they ever make LGA 1156 mother boards with 2 x pci-2 x16? it would make these new cpus so much more attractive for gamers that want a more efficient pc.



    Why should they? x8/x8 is shown to be just as fast as x16/x16 so long as you a good PCI-ex controller (p55 has an excellent controller but on some P45 boards Crossfire isn't that great due to controller issues). Read Toms review on the HD5850, it shows Lynnfield didn't bottleneck Crossfire performance.

    At the moment there is no need to add more lanes and by adding more lanes it will only further increase the cost of an already costly P55/LGA1156 setup.
  • 0 Hide
    jblack , October 23, 2009 1:42 PM
    Quote:
    Ed.: and priced accordingly; that Q9550S is still a $219 part)


    Ummmmmm Where did you see it for that price? --- The cheapest I can find it is $350 at Newegg.
  • 0 Hide
    geofelt , October 23, 2009 1:50 PM
    The conclusion that turbo is not that efficient is not completely accurate. The total wattage used for a job with turbo might actually be lower because the higher speed will get the job done in less time, therefore keeping the cpu at high power usage for less time.
  • 1 Hide
    joejamesatou , October 23, 2009 2:08 PM
    There are a couple of flaws in this test setup in my mind. Testing the new i5/i7's against the Core 2 Quad S series is a bit of a apples to oranges. No one bought the S series unless they were making an HTPC. The Core 2 Quad Extreme also, as the name implies, isn't mainstream either. The article would have been better served using the Core 2 Quad 9550 and maybe even AMD's Phenom 965. Also, no one is buying the i7 870. The i7 860 is much more mainstream. These are more realistic buys.

    X2 jblack. Waaaaaaaay off calling the 9550s a $199 processor. Even the regular non-s version is going for $269 on newegg.
  • 1 Hide
    Anonymous , October 23, 2009 3:23 PM
    Am I missing something obvious here?

    Wouldn't Vantage be at least partially GPU-dependent as well?

    Meaning that the power-efficiency discussion isn't entirely accurate as the test run-time isn't wholly dependent on the CPU horsepower.

    I might be wrong of course but I'd rather see something like a Prime95 run for a static number of loops with 1, 2, 4 and 8 threads as an alternate comparison. Vantage would, in my mind at least, be dependent on too many different aspects of the computer hardware platform to be used for what should be a strict CPU-comparison.

    Well that's my 2c anyway. Feel free to correct me. :) 
  • -1 Hide
    kureshii , October 23, 2009 5:12 PM
    I guess these results aren't that surprising; we've known for quite some time now that moderate overclocking doesn't actually do much for power efficiency, and Turbo mode does exactly that: overclock the processor while staying within the TDP envelope (although I'm not sure if overvolting is involved as well).

    Still, I conjecture that perhaps limited overclocking might actually increase efficiency by a little; maybe around the 266MHz mark (again, just a conjecture). This is the kind of clock speed gain we might see with two or three cores active.

    For this reason alone I do agree with Exodite2 to some extent that benchmarks with applications known to be moderately multithreaded would be interesting. Even if they are not particularly useful in describing typical power efficiency numbers, they certainly contribute to our understanding of Turbo's impact on power efficiency.

    And yes, please do run this test on the i7-860, I believe that is the processor of choice for a budget-conscious Lynnfield-i7 buyer, after all. Looking forward to additional analysis :) 
  • -1 Hide
    kureshii , October 23, 2009 5:15 PM
    geofeltThe conclusion that turbo is not that efficient is not completely accurate. The total wattage used for a job with turbo might actually be lower because the higher speed will get the job done in less time, therefore keeping the cpu at high power usage for less time.

    Wasn't the total wattage already measured in this graph and shown to be higher with Turbo enabled? (At least, just for the i7-870, and only for PCMark.)
  • -1 Hide
    WheelsOfConfusion , October 23, 2009 5:27 PM
    It makes sense, because Turbo is basically overclocking.
    What about the effect of hyper-threading?
  • 0 Hide
    pcman911 , October 23, 2009 6:09 PM
    Would like to have seen the AMD 785G board and Athlon II X4 CPU with the same equipment (GTX 260) thrown into the mix just for results. The AMD 785G and the Athlon II x4 are both fairly efficient and beat on price big time.
  • 0 Hide
    Pei-chen , October 23, 2009 8:37 PM
    pcman911Would like to have seen the AMD 785G board and Athlon II X4 CPU with the same equipment (GTX 260) thrown into the mix just for results. The AMD 785G and the Athlon II x4 are both fairly efficient and beat on price big time.

    Athlon II is designed to counter i3, not i5. PII vs. i5 make more sense.

    BTW, since i3 will have PCI-express lane and graphics on chip, it will be the most efficient desktop setup by far. Athlon II X4's die will be huge next to the 32 nm Clarkdale i3 which is "quad" core with hyper-threading.
  • -1 Hide
    xtc28 , October 23, 2009 10:10 PM
    Well I would say the i5 is certainly a leap in the right direction!!!
Display more comments