Power Consumption

AMD did a lot of work on reducing idle power consumption with the move to Phenom II and thus we see the 758M transistor X4 940 draw less power than even the Q8400 at idle.

Idle Power Consumption

Under load it's a different story entirely, the Phenom II X4 940 needs another 32.1W. Granted it's a faster chip, but with all of those transistors switching at 3GHz the power consumption is simply higher.

Load Power Consumption (x264 HD 1st Pass)

Gaming Performance Overclocking at Stock Voltages
Comments Locked

60 Comments

View All Comments

  • eXistenZ - Thursday, May 7, 2009 - link

    Obviously, AMD is no good in Far Cry 2 game. K8, K10, K10.5, al these architectures were always slower than intel's competitors. And it is really crappy game, so i don't see any reason why are you testing right on this one. I think, more fair testing is with Crysis or CPU-eaters = RTS...
  • Goty - Thursday, May 7, 2009 - link

    Anand seems to be buying into all the FUD about AMD lately. Sure, AMD's not doing so hot right now, but they're not in much worse a position than they were in the middle of the P4 era (probably about the same position, all told).
  • Anand Lal Shimpi - Friday, May 8, 2009 - link

    I'm not sure I would call it FUD. AMD lost $2.36B before taxes in the last four quarters combined. Their chief competitor made $6.13B. Now Intel has always made more than AMD, but the issue now is that AMD is losing a considerable amount every quarter. That can only continue for so long.

    What I'm more worried about is the impact this is having on the next-generation cores that AMD is developing. While engineering budgets are the last things to go, if you're losing a few hundred million a quarter everyone from marketing to engineering gets hurt.

    Ignoring the problem isn't going to make it go away, I felt that it would be important to at least bring some of this stuff to the table so we can at least be thinking about it.

    Take care,
    Anand
  • microAmp - Thursday, May 7, 2009 - link

    Actually, it's worse, they are running out of cash.
  • ssj4Gogeta - Thursday, May 7, 2009 - link

    "It's the beauty of Moore's Law: with fewer transistors crammed into a much smaller area, we're able to see the same performance."

    Shouldn't it be "MORE transistors crammed into a much smaller area"?

    :)
  • hooflung - Thursday, May 7, 2009 - link

    I am another that buys based on the ability of Virtualization via virt functionality. I got a P2 940 because I wanted the ability to have 4 cores to split up to VM's running Hyper-V, Xen and KVM. I just can't do that on new intel chips that fall in the price range right now.

    My C2D is still rocking a venerable 1ghz OC on a e4300 and P35 chipset. For me to install an OS to do development as the top level is just wasting wattages at my home.


  • snakeoil - Thursday, May 7, 2009 - link

    ''Phenom II Earns a Financially Troubled AMD Less per Chip than Core 2 Quad''

    well you are saying that amd make less money because phenom 2 has a little more area,but in your happy calculations you forgot that bad quad core dies are used to make tricores and soon dual cores phenoms.
    harvesting.

    what are you doing little annand
  • crimson117 - Thursday, May 7, 2009 - link

    One wafer costs a fixed amount to make; let's say $200.

    Let's say AMD can get 10 CPUs made from each wafer, while intel can get 20 smaller CPUs from each wafer. They each sell their chips for $180.

    AMD puts $200/10 = $20 worth of wafer into each $180 CPU.
    Intel only has to put in $200/20 = $10 worth of wafer into each $180 CPU.

    So assuming all other things are equal, Intel makes $5 more on each CPU sale than AMD.
  • crimson117 - Thursday, May 7, 2009 - link

    *clicks Edit button*

    So assuming all other things are equal, Intel makes $10 more on each CPU sale than AMD.
  • mkruer - Thursday, May 7, 2009 - link

    You are also forgetting that AMD and Intel use two different lithphogathy technologies to AMD uses submersion and Intel uses double pattering. The Submersion takes slightly longer then a single pattering, and yeild fewer defects. This meas that AMD should be able to preduce a high volume of chips per platter then Intel. Adding to the confusion, is that neiter intel nor AMD releases what there yeilds are and as such, too comepare based upon die size alone is folly. People can crunch the numbers anyway they want, but in the end it should be a, for more or less, wash.

Log in

Don't have an account? Sign up now