TechSpot

Most power supplies won't support Haswell's C6/C7 low-power states

By Shawn Knight
Apr 30, 2013
Post New Reply
  1. Skidmarksdeluxe

    Skidmarksdeluxe TS Evangelist Posts: 3,183   +767

    You must remember that Intel doesn't design chips for the desktop specifically anymore. All Haswell desktop chips are derived and adapted from laptop parts where power states are important.
     
  2. Skidmarksdeluxe

    Skidmarksdeluxe TS Evangelist Posts: 3,183   +767

    Corsair uses a lot of OEM's for their PSU's Seasonic is just one of them.
     
  3. Skidmarksdeluxe

    Skidmarksdeluxe TS Evangelist Posts: 3,183   +767

    If your PSU can't support the new power states then you shouldn't loose any sleep over it. Just disable them in the BIOS if they're not already disabled.
    PSU manufacturers should be paying Intel for introducing these new power states because when people hear about this they'll rush off to buy new units. The R&D costs to implement this in upcoming PSU's will be negligible.
     
    JC713 likes this.
  4. dividebyzero

    dividebyzero trainee n00b Posts: 4,910   +716

    There are a fair few zero load PSUs on the market, it just means that the vendors will now have another bullet point to add the already lengthy list.
    Enermax seem to have the jump on some of the other vendors (so the 23 unit "approved" list is already out of date). Expect a rash of "Haswell Ready" press releases from OEM/ODM's.

    Skidmarksdeluxe
    Might I suggest the edit button. Triple posting :eek:
     
  5. Seventh Reign

    Seventh Reign TS Rookie Posts: 93   +35

    Right .. because electricity is free ... Try having 6 High End Gaming Desktop PCs, all running at the same time, and say there isnt any point in reducing power consumption, after you see your electricity bill.
     
    havok585 likes this.
  6. lipe123

    lipe123 TS Guru Posts: 449   +81

    You know there might be some truth to this but the current facts presented is TOTAL BS. Either the report is worded very wrong or these vrzone guys have no idea what the difference between voltage and current is.

    If something requires less current (0.05 vs 0.5 Amp) than what your current device can supply then there is absolutely no problem. And the current draw is not supposed to affect the voltage output of the psu AT ALL unless you try drawing more current than the device can support which will lead to voltage drops.

    This whole thing seems like its a new publicity stunt from certain PSU manufacturers to bash other companies and promote their products. Just like the "Y2K ready" era was mostly BS.
     
  7. It's amazing how many posters are oblivious to the benefits of low idle. Low idle allows your system to stay on 24x7, making it available for tasks without having to wake from S3 or hibernation. Those of us who run home media servers have been dying for something like this for a while. No more having to choose between high electric bills and having elaborate setups to send Wake on Lan packets to your box when you're not local to your LAN. No more having to fiddle with S3 settings to have your box sleep and wake correctly. (Go to sleep if idle for 10 mins. Crap, but not while a video or music is playing. Oh damnit, an active network connection is keeping my box awake all the time, I'll disable it. Oh crap, now my box shuts off while it's backing up. Ahh!!!) I don't really care about peak power usage, just hate to waste 60 watts for the box to sit and do nothing. Looking forward to upgrading to Haswell.
     
    havok585 and cliffordcooley like this.
  8. Vrmithrax

    Vrmithrax TechSpot Paladin Posts: 1,286   +232

    I can assure you, this is not something that was just made up or misrepresented. The problem is in how the power supply handles the ultra low power draw. I've run into this many times with very expensive industrial power supplies as well as smaller consumer grade units driving electronic control systems.

    If the circuitry of the power supply is too "generic" (for lack of a better term), those ultra low power modes could be electrically considered a "no load" situation. Meaning, you get a complete drop to zero. Sleep state now becomes a coma. Or, even worse, there is no actual stabilization at those super low power states, so you get little constant fluctuations. The fluctuations may be tiny when compared to the full scale of the power supply, but when compared to a 0.05 amp level, they could be huge. Sleep state now becomes a seizure. Hence the warning about possible instabilities.
     
    Burty117 and cliffordcooley like this.
  9. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    Upgrading to more modern components is part of the whole deal. You cant use DDR2 with a modern CPU for instance.
     
  10. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    I hate when my comment generates so much hate lol. I am just saying people, if you can pay for a high end gaming PC, you can pay for electricity.
     
  11. Darth Shiv

    Darth Shiv TS Evangelist Posts: 1,169   +176

    JC713 It's not just about whether you can pay for electricity. Haswell is not using power it doesn't have to when you are idling. Gaming rigs aren't running games the whole time they're on imho. People stop for food, drinks, talking etc. Also less power draw in idle is less heat so I don't know why there is a problem with moving to better PSU tech and allowing ultra low power states?
     
    cliffordcooley likes this.
     
  12. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    I have no problem with it, I am just stating that I think it is unnecessary. I dont really mind saving money on electricity.
     
  13. cliffordcooley

    cliffordcooley TechSpot Paladin Posts: 6,123   +1,523

    In the grand scheme of things from a global perspective, changing one CRT monitor for a LCD is unnecessary. Going green in only one room of your home is unnecessary. Going green in only one aspect of the PC would be unnecessary. The idea is to go green with every possible aspect that presents itself.

    Sometimes I wonder if innovators see their innovations as unnecessary because of such little progression. But yet if these innovations could see the light of day, they would pave the way to a new future. At least in the PC industry things are still moving forward.
     
    havok585 likes this.
  14. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    True. But I still wonder why this low power state was introduced now if Haswell has a higher TDP than Ivy Bridge.
     
  15. Darth Shiv

    Darth Shiv TS Evangelist Posts: 1,169   +176

    Because they can. It's an improvement in an area that can be improved. TDP increase - well I'd hope that any TDP increase has a decent performance increase associated with it!
     
  16. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    It is because of the new graphics.
     
  17. cliffordcooley

    cliffordcooley TechSpot Paladin Posts: 6,123   +1,523

    That would be my guess.
     
  18. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    There isnt any increase in clock speed and nothing new that would eat up power other than graphics.
     
  19. Adhmuz

    Adhmuz TechSpot Paladin Posts: 930   +106

    As long as your overclocking this doesn't concern you. If your concerned with power usage unplug a clock or something, you'd save more power that way. This is such a minuscule change in power consumption I'd be surprised if anyone can tie it in to saving power on their monthly electrical bill. For laptops I guess it makes sense, in a desktop, really not so much. At .6 Watts the CPU is practically off it might as well be, and btw do they mention how much power is required to return to a usable state? I've been disabling Low power states since my first Intel chip, don't see it changing with Haswell.
     
  20. Darth Shiv

    Darth Shiv TS Evangelist Posts: 1,169   +176

    Was just doing some reading on non-GFX related performance. Performance increases of 7-12% from new branch prediction, better cache performance (handles cache misses better), "deeper buffers" (presuming the instruction fetch buffers increased?), more execution units (that is a very interesting one).
     
  21. JC713

    JC713 TS Evangelist Posts: 7,029   +913

    New branch prediction? Wow. Good stuff.
     


Add New Comment

TechSpot Members
Login or sign up for free,
it takes about 30 seconds.
You may also...
Get complete access to the TechSpot community. Join thousands of technology enthusiasts that contribute and share knowledge in our forum. Get a private inbox, upload your own photo gallery and more.