45nm Phenoms to show up in the third quarter after all?

AMD’s latest official statements regarding versions of processors based on 45nm process technology is that shipments should kick off in the second half of this year. The company has yet to disclose a more precise time frame, and recent rumors have suggested that we may not actually see 45nm Phenoms hit stores until the fourth quarter—roughly a year after the launch of Intel’s first 45nm desktop CPU.

However, the folks at Fudzilla have published a new report that casts doubt on those rumors. According to Fudzilla, AMD has informed its staff that 45nm Stars processors will launch in both the third and fourth quarters of 2008. If true, that statement means we could see 45nm Phenoms start rolling out anytime between July and September. Fudzilla says the 45nm Phenom lineup will include Deneb FX and Deneb quad-core models with both L2 and L3 cache as well as Propus quad-core models with only L2 cache. Triple- and dual-core 45nm Phenoms aren’t due until early 2009.

Judging by a presentation recently leaked by Sun, Intel anticipates that AMD’s 45nm quad-core Opterons will run at up to 2.8GHz, so similar speeds on the Phenom side should be possibile. In that case, Phenom could put up a decent fight against Intel’s 45nm Core 2 Quads.

Comments closed
    • VILLAIN_xx
    • 12 years ago

    im plugging my eyes and ears, and screaming LALALALA!

    until i see a product from AMD thats on time with no excuses lol.

      • pluscard
      • 12 years ago

      The HD3870X2 was dead on time – does that not count for some reason?

      ps: it’s also the industry leader in single card GPU’s.

        • VILLAIN_xx
        • 12 years ago

        Yes .. i guess it really does count for something!

        :o)

        i havent been avid on release dates for GPUs as of late since I went back to gaming consoles.

        my bad!

        Edit: But im still not crossing my fingers hehe.

    • pluscard
    • 12 years ago

    Today is March 2nd, and at least on pricewatch.com I find only one 45nm quad desktop part, the QX9650 – priced over $1000.

    While I can’t say INTC is having yield problems, they are certainly having “some” problem.

      • green
      • 12 years ago

      alternatively, intel under estimated demand
      (like the recent xbox360 shortage)

    • xtreme2k
    • 12 years ago

    my take on this is

    release first
    then talk

    • clone
    • 12 years ago

    it’s all smoke and mirrors atm….. AMD needs product out now.

    future promise doese not a sale make.

    • mackintire
    • 12 years ago

    First, Intel is having problems with yields on 45nm.

    Second the TDP goes up dramatically as you overclock your CPU. My QX9650 chips is nice and cool at 3.0hz 95 watts at 3.4Ghz its at 140Watts at 3.8Ghz its near 200watts!

    So can intel release faster chips, YES they can. But they will be hot.

    If AMD can release their 45nm CPUs at 2.6-3.0 Ghz and with a competitive TDP then and only then will Intel will have competition for the normal user CPU buying market.

    The 45nm AMD chips are 15-20% faster then the B3 versions due out next month. That wil put AMD’s new chips +- 10% of Intel Core 2 chips clock for clock.

    Of course this depends on AMD actually releasing a part on time.

    Another possiblity is if Intel gets its yield issues under control, Intel could hurt AMD by lowering prices.

      • green
      • 12 years ago

      l[

      • kilkennycat
      • 12 years ago

      Gee, I (and The Tech Report and Anandtech) must have missed the news conference where AMD publicly-demonstrated fully-functional 45nm processors with speed and power-consumption benchmarks…

      URLs please…………………………

      I suppose it’s always nice to day-dream about an alternate reality.

      • kilkennycat
      • 12 years ago

      l[

      • smilingcrow
      • 12 years ago

      “the TDP goes up dramatically as you overclock your CPU. My QX9650 chips is nice and cool at 3.0hz 95 watts at 3.4Ghz its at 140Watts at 3.8Ghz its near 200watts!”

      Anandtech looked at the power consumption of an over-clocked QX9650 and it scaled in a fairly linear fashion to about 3.6GHz and only got out of control above 4GHz. Either they got a good chip or you got a bad one. What voltage did you use at 3.8GHz?

      §[<http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3184&p=2<]§

        • pluscard
        • 12 years ago

        Per your link to Anand – a 50% overclock of the 9650 would cause the power consumption to increase by 500%.

        Pretty much right in line with what #15 is saying.

        (Note Anand also states on the graph that he used the minimum achievable supply voltage – likely well below the motherboard default).

          • green
          • 12 years ago

          yeah except that his temps are off anands by around 40, 60 and 80 watts respectively (if i’m reading it right anyway)

          but it’s indicative of intels’ 45nm process which is still clinging on to a similar style of 65nm
          intel is the only one to go 45nm without using immersion lithography
          mostly as they’re too cheap to buy the new equipment
          investors will be a little happy with the short term benefit

          can’t seem to find how many of intel’s fabs are 45nm at the moment (cbf)
          can’t be many though. they’re still producing 65nm/90nm stuff in bulk

    • kilkennycat
    • 12 years ago

    That’s odd.

    Intel demonstrated a bunch of fully-working 45nm processors running several OSs to the press over a year ago, so where are the AMD public demos of fully-working 45nm silicon, if they expect to ship volume in the 3rd quarter ??? Lots of process-related questions too, anyway. The AMD 65nm process doesn’t bring much in either speed or lower-power improvements over their 90nm. So what will be their process answer to Intel’s hafnium-gate exercises to get power-consumption better under control ??

    When is AMD/ATi going to “walk the walk” again instead of “talk the talk” ?

    Meanwhile, I have a very nice spare bridge near Manhatten that I would like to sell. Any offers??

    • lolento
    • 12 years ago

    If the 45nm chips are ready to ship in July – September time frame, then samples of the chips should already been leaked. Or at least they should have samples at trade shows.

    Board manufacturers should already be gearing up for the launch.

    But if Fudzilla say so, then it must be true. Since AMD had contacted their staff.

      • Gerbil Jedidiah
      • 12 years ago

      I think we’ll just have to agree to agree. I’m not saying anything that is in conflict with your assessment.

      Oops meant for #12

      Guess I’ve had too much wine=)))

      • green
      • 12 years ago

      amd traditionally keep things under wraps til very close to launch (~1 month or so?)

        • echo_seven
        • 12 years ago

        Just wanted to update this thread, as we now know that AMD is showing 45nm wafers of Shanghai(although not samples, then again we’re still at least 4 months away) at CeBIT.

    • Prototyped
    • 12 years ago

    Ugh, more Fraudzilla. Is there a similar report from a more reputable source?

    • charged3800z24
    • 12 years ago

    You never know.. These chips could over clock well. Tho we will have to wait on sample to see how they will perform. If AMD can hit the schedules right on they look to have a pretty good outlook. I have never bought the newest fatest CPU. And probably never will. So if these are priced right I will move to get one

    • pogsnet
    • 12 years ago
    • pogsnet
    • 12 years ago
    • Gerbil Jedidiah
    • 12 years ago

    2.8GHZ is not decent if Intel decides to stop sandbagging their clock frequencies and shipping stock clocked Quads in the 3.5+ range.

      • Flying Fox
      • 12 years ago

      I’m not sure if they are entirely sandbagging. Looks like they are still keeping volume and GHz low while still perfecting the manufacturing process as well. This gives them room to maneuver should they need higher GHz while keeping the power down, as it gains more experience for fabbing the next gen stuff on the same process. This is the key to this whole tick-tock strategy.

        • Gerbil Jedidiah
        • 12 years ago

        Or they could be sandbagging their 45NM tech to keep their 65NM tech relevant while they get rid of inventory and change their manufacturing process.

          • charged3800z24
          • 12 years ago

          Word is they are having production issues with thier 45nm process… Saw this on the fudzilla.

            • Flying Fox
            • 12 years ago

            As I was using the term “not entirely”, which means sandbagging may not be the only reason. As most decisions go there are often multiple factors involved.

            • Gerbil Jedidiah
            • 12 years ago

            I’m not assuming such a narrow view as to suggest sandbagging is the only factor affecting intel’s production. I’m merely suggesting that intel has parts in the market that are clocked lower than they could be. The 45NM parts on the market now have been shown very overclockable, which indicates intel could offer the CPUs it currently has in the market at a higher clockspeed if it had so desired. i.e. 3.2 or 3.4 instead of 3.0GHZ… If intel is sandbagging, this switch in clockspeed could quickly send AMD’s quad cores downmarket in the bargain section.

            Me personally, I hope AMD comes out with a part that smacks intel in the head=)

            • Flying Fox
            • 12 years ago

            And I was suggesting that this “could be” thing is not completely correct. Look at the new HSF shipped with E8400, it’s puny. So they are saving costs. Sure they can put out faster GHz models now but they may have to stick with the old copper core HSFs (actually that will be a good thing for overclockers). I’m pretty sure that power, heat and costs concern is the bigger part of the equation here, because that makes more sense than always thinking about the rivalry. Core 2 CPUs are leading so far ahead now Intel doesn’t really need to think about AMD all the time. They just need to keep executing their plans and that should do.

            Once they get the process more matured and the yields up, may be fix a few errata here and there, plus even cooling running steppings, then they can think about faster clocks. Meanwhile they need to juggle the conversion from 65nm to 45nm for the remainder of their fabs. You don’t switch all of those overnight in one shot.

Pin It on Pinterest

Share This