Page 9 of 18 FirstFirst ... 7891011 ... LastLast
Results 81 to 90 of 180

Thread: Anand: Bobcat "full performance-preview" coming "before the end of the week"!

  1. #81
    8-bit overflow
    Join Date
    Oct 2010
    Posts
    383
    Quote Originally Posted by accord1999 View Post
    ... and a lot faster, but still uselessly slow GPU.
    I am driving two 1920x1200 screens with a uselessly slow HD4350 GPU.
    I have found it to be fast-enough for all practical pruposes _including_ ocassional gaming (CS2/UT2004/etc).

    Tried 780G vanilla for a quarter, then 5650, and settled for a passive 4350 that fits the bill.
    Don't have a patent granted for being right. Would love one though!

  2. #82
    Quote Originally Posted by accord1999 View Post
    It'll probably means that AMD's test platform was a notebook platform that has an unfair advantage when tested against low cost desktop Atom products. Against other notebooks, it's power consumption is not so special.

    And when actual products come out, you'll probably find Bobcat has near Intel CULV power efficiency with modestly better than Atom performance and a lot faster, but still uselessly slow GPU.
    You need to make some more documentation for your claims. PC perspec have the facts, and they say you probably is wrong. They explained the methology pretty good in my view. We will know but it looks good from the facts we have.

    If bobcat is near Intel CULV 32nm power efficiency, its the biggest win in CPU history. Its not like CULV is cheap, its damn expensive, and to achieve that,- or just near that, on cheap 40nm bulk, is impressive - to say the least.

    Modestly better Atom performance you say. 65% better single threaded perf, and that what count when the single core perf, is the weakest part of Atom (well all is weak, but this is the weak of the weak)

    A lot of people buyed Atom integrated thinking the could watch HD. They could simply not do it. It cant be used for anything but painfully slow office use, no multimedia. Bobcat can play sims3 and most of the basic stuff that is sold today. We will be surprised to know how much gaming will be done on this bobcat fellow.

  3. #83
    640k who needs more?
    Join Date
    Aug 2010
    Posts
    783
    Until we have a Zacate in a platform very similar to CULV, can't really compare but I'd be willing to bet they're going to very close at maximum CPU/GPU cycles ie 28-33watts total.

    The 45nm Intel CULV should get destroyed by Zacate on GPU benchmarks, probably equivalent on CPU benchmarks (Intel CULV is 1.3ghz), Intel die size is at least 3x if you include the IGP so overall its a major technology win for AMD.

  4. #84
    8-bit overflow
    Join Date
    Nov 2010
    Posts
    343
    Quote Originally Posted by accord1999 View Post
    It'll probably means that AMD's test platform was a notebook platform that has an unfair advantage when tested against low cost desktop Atom products. Against other notebooks, it's power consumption is not so special.

    And when actual products come out, you'll probably find Bobcat has near Intel CULV power efficiency with modestly better than Atom performance and a lot faster, but still uselessly slow GPU.


    It has near Atom power, it has near ULV cpu performance it has near sb desktop gpu performance.

    So wth are you babbling about?

  5. #85
    Lets have the facts again:
    75mm2 for dualcore+gpu
    Aprox 6,3 usd per die (5000k wafer, 800 dies)
    Marginal production cost including packaging and shipping aprox. 9 usd.

    413 pins = cheap integration for the OEM, - cheaper than Atom + DX 11 label

    If these cost numbers does not spell it. Here it is. CULV and whatever i3-5 is not a competitor for bobcat. Atom is. Intel needs a new out of order Atom. Period.

  6. #86
    Quote Originally Posted by krumme View Post
    You need to make some more documentation for your claims. PC perspec have the facts, and they say you probably is wrong. They explained the methology pretty good in my view. We will know but it looks good from the facts we have.
    It's not the methodology, it's the choice of components. Single-core Atom desktops idle at 20W, dual-core Atom netbooks load at <15W.

    If bobcat is near Intel CULV 32nm power efficiency, its the biggest win in CPU history. Its not like CULV is cheap, its damn expensive, and to achieve that,- or just near that, on cheap 40nm bulk, is impressive - to say the least.
    I don't think it'll be close to a 32nm CULV in pure CPU performance/watt, it'll be close to a 45nm Penryn CULV with comparable GPU performance.

  7. #87

    Bobcat beats CULV hands down for efficiency

    Quote Originally Posted by accord1999 View Post
    It's not the methodology, it's the choice of components. Single-core Atom desktops idle at 20W, dual-core Atom netbooks load at <15W.


    I don't think it'll be close to a 32nm CULV in pure CPU performance/watt, it'll be close to a 45nm Penryn CULV with comparable GPU performance.
    What choice of components?

    Here is from the article:

    "The AMD E-350 is able to sit idle at less than half of the SU2300+ION combination and 76% less power than the Atom D510 - very impressive results."

    "This is where things look even better for the E-350 as it is not only able to offer playable, but modest, gaming experiences on its platform but it is doing it at a lower cost and lower power draw than the CULV+ION combination as well. "

    http://www.pcper.com/article.php?aid...e=expert&pid=8

    And you know, all this for say aprox 25% of the die area (cpu+gpu). AMD wins 400% on the cost side, how impressive is that? When is the last time the difference was 400%?

  8. #88
    8-bit overflow
    Join Date
    Aug 2009
    Posts
    416
    Let's see... from PC Perspective's review

    idle power

    E350: 9.3W
    CULV: 21.2W
    Atom: 16.4W

    Looks good, but read on

    load power Cinebench
    E350: 19.2W(+9.9W)
    CULV: 29.2W(+8W)
    Atom: 20.1W(+3.7W)

    load power L4D2

    E350: 28.8W(+19.5W)
    CULV: 37.1W(+15.9W)
    Atom: 23.3W(+6.9W)

    What is that mean? Idle power is higher on CLV and Atom because they are not on mobile. But both CULV and Atom systems are adding less watts when loaded compared to idle. Put them on a proper mobile platform and Zacate might be on par with CULV with power consumption.

    413 pins = cheap integration for the OEM, - cheaper than Atom + DX 11 label
    45nm CULV chips won't be expensive at all to make because its already on an obsolete platform and process technology.

  9. #89
    Quote Originally Posted by DavidC1 View Post
    Let's see... from PC Perspective's review

    idle power

    E350: 9.3W
    CULV: 21.2W
    Atom: 16.4W

    Looks good, but read on

    load power Cinebench
    E350: 19.2W(+9.9W)
    CULV: 29.2W(+8W)
    Atom: 20.1W(+3.7W)

    load power L4D2

    E350: 28.8W(+19.5W)
    CULV: 37.1W(+15.9W)
    Atom: 23.3W(+6.9W)

    What is that mean? Idle power is higher on CLV and Atom because they are not on mobile. But both CULV and Atom systems are adding less watts when loaded compared to idle. Put them on a proper mobile platform and Zacate might be on par with CULV with power consumption.
    That's an interesting take on it. On the other hand



    The atom didn't run as usual, so its load power under gaming doesn't really mean a great deal. The CULV was slower at 40% higher power draw.

    Don't forget Zacate isn't being benchmarked on a proper mobile platform either.

  10. #90
    640k who needs more?
    Join Date
    Aug 2010
    Posts
    783
    Quote Originally Posted by DavidC1 View Post
    Let's see... from PC Perspective's review

    idle power

    E350: 9.3W
    CULV: 21.2W
    Atom: 16.4W

    Looks good, but read on

    load power Cinebench
    E350: 19.2W(+9.9W)
    CULV: 29.2W(+8W)
    Atom: 20.1W(+3.7W)

    load power L4D2

    E350: 28.8W(+19.5W)
    CULV: 37.1W(+15.9W)
    Atom: 23.3W(+6.9W)

    What is that mean? Idle power is higher on CLV and Atom because they are not on mobile. But both CULV and Atom systems are adding less watts when loaded compared to idle. Put them on a proper mobile platform and Zacate might be on par with CULV with power consumption.



    45nm CULV chips won't be expensive at all to make because its already on an obsolete platform and process technology.
    Can't compare desktop and mobile as power supply effiiciency could be vastly different. My CULV is 31watts at maximum CPU/GPU but its a 90% efficient Delta AC adapter. If it was on a 70% effiicent cheap PC power supply, the wattmeter would report 40W

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
WordPress Appliance - Powered by TurnKey Linux