AMD Phenom II Available To Distributors This Week 114
jdb2 writes "Fudzilla reports that AMD's Phenom II is already available to distributors, and will be available to sell to consumers in the week of the 29th of December. The Phenom II is AMD's consumer version of its 'Shanghai' 45 nanometer SOI process Quad-core Opteron chip and will reportedly ship in 3 and 2.8 gigahertz flavors corresponding to the model numbers '940' and '920' respectively. This first release will be packaged as a Socket AM2+ part which only supports DDR2 memory. The following month AMD is reportedly going to release a new '9x5' series of Socket AM3 versions which support DDR3 memory — these will be backward compatible with Socket AM2+ . This may be an inflection point for AMD if the Shanghai architecture lives up to the performance numbers from preliminary reports and if so it will no doubt also be a welcome belated Christmas present for the already salivating hordes of Tech Junkies."
Re:125 watts! (Score:3, Insightful)
How much power does the off-chip memory controller draw on that 95-watt core 2 quad?
Re:FAR better option than anything AMD has to offe (Score:1, Insightful)
You clearly lack reding comprehension skills.
Mod parent "+1 Ironic"
Re:FAR better option than anything AMD has to offe (Score:3, Insightful)
um, if you don't want integrated graphics, you can just get the 790X--it's cheaper.
you might want to get a clue first before ranting off incoherently.
Re:It will come down to clock speed. (Score:3, Insightful)
You don't have to use the RAID on your board either. It sucks, too. Would you refuse to buy a board simply because it had the option of using RAID? Do you want an MB with just the number of PCI or RAM slots you plan to use, too? What if it has PS2 and a serial port you don't want?
You're just being argumentative. Get a life.
Re:It will come down to clock speed. (Score:5, Insightful)
"It easily smokes most Intel quad boxes I've seen that cost several hundred more."
Except, it doesn't smoke any Intel Quad, at all. That CPU is about the same speed as the *2 year old* Q6600 at stock speed (couple hundred extra MHz but less cache). Both overclock VERY similarly too. Almost no differences in benches all-around. Unless you were comparing your OC'ed CPU to a stock Intel, which is just as true as saying a Q6600 (once OC'ed) smokes every Phenom out there at any price point (at stock speed). In fact, it's more like your box gets smoked by an Intel that costs one hundred more...
The only real difference here, is that your new CPU uses significantly more power. The $20 you saved, you'll pay 10x back in extra power used.
It's kinda sad to see AMD's latest best offering that's 2 months old barely managing to compete with a 2 year old chip...
"I'll always run Intel on servers and what not - but for a workstation/desktop you can't go wrong with AMD."
Actually, it's EXACTLY the inverse! Opterons have always been much nicer: HT bus, NUMA architecture and everything else, up until now, with Intel's QPI bus. The funny thing is, now that it's going to start not sucking (Xeons), you're switching away from it? And on the desktop, Core 2 is just much better. You got it all COMPLETELY backwards!
Re:It will come down to clock speed. (Score:1, Insightful)
I have 12 USB ports. I could lose half of those and still be quite happy because I'm only using 3 (that's including occasional plug-ins - I only use 1 most of the time, and that's my mouse which could go to PS2 if necessary). I have no firewire or eSATA devices nor do I know anyone that does.
To me this argument looks like: "I want to spend $100 on a board with X set of features that I won't use, but for $100 they'll only sell me a board with Y set of features that I won't use, and they want me to pay $150 if I want to not use X features".
The original point was whether or not you could get a board for $100 - if you're building on a budget then onboard graphics is a useful feature, because it immediately saves you from having to cough up for a stand-alone graphics card. OTOH firewire, eSATA and oodles of USB ports are probably features which you don't want. (And when I say 'you', I don't actually mean you personally; I mean a consumer who isn't being deliberately obtuse)
so wrong (Score:1, Insightful)
I've been studying advanced mathematics ( at the graduate/doctoral level now ) for over a decade.
Most graduate programs I've seen require completion of the degree for which one is a candidate within 5 years, maybe 6. The reader may reasonably infer from your claim that a) you've had enough time to earn such a degree twice over, and b) that you have not done so because having done so would give you an honest claim even greater than the one you've actually made here. I fully realize that the question of having such a degree is orthogonal to the question of possessing competence-- period--, and this illustrates in part why one should never argue from authority.
But we're talking about elementary calculus here, not topology or algebras or anything so rarefied. There's no reason to berate the claim about the [local] minim[um] value of h(t):
...for the past year or more the derivative has been negative, and growing more negative.
(Emphasis mine.)
I disagree: h(t) has units of "performance somehow compared to Intel's performance", and AMD's xx50 series have been competitive at their and profitable at their position in the mid-range market as with the X2s in the low-end market. h'(t) has been negative because overall marketshare loss driven by Intel's aggressive product releases (5200/7200/8x00), but your claim that h''(t) is negative is too strong. But this is all an aside; let's say you're right, for the sake of your argument:
If the Phenom II launch is indeed the catalyst for an AMD "turn around"...
Yes, by definition "turning around" means changing from decreasing to increasing, and not merely decreasing the rate of change of decrease. Therefore, the claim that h(t) will experience a local minimum [that h'(t) = 0 is coming] was true.
It's not very interesting to note that there will be an inflection point before h'(t) goes positive if h''(t) is negative, because even though that "if" is actually *false*, it's the slope of h(t) we're talking about in the first place! Any niggling disagreement about the present nature of h''(t) is irrelevant.
The momentum analogy is predicated on your claim that AMD's suck has been increasing locally, so 1) this is disputed, and 2) either way, the analogy is inapt because momentum changes can exhibit behavior you don't seem to allow: while it is well-understood by this late date that there is "no" [i.e. fanboys only] brand loyalty in computing, whether AMD-Intel or ATI-Nvidia. We're not guaranteed that h(t) is even differentiable for t corresponding to product launches. In your analogy, this is a perfectly elastic 1D collision of a point particle.
Human intuition about momentum is (clearly!) insufficiently general for momentum to make a good analogy for reasoning about non-physical phenomena.
Recapping:
1) the claim you were refuting was correct
2) you used a spurious and non-generalizable argument in your attempt at (1)
3) your claim of expertise in mathematics is likely false, as evidenced by (2)
4) you don't understand the aspect of the computer industry about which you are making a claim here and in the article summary
5) likely due to (3) and (4), your particular claim about the computer industry is trivial, wrong, or both
And using an emoticon-smiley doesn't absolve you of being a jerk about it all above and beyond being right or wrong. The comment to which you were responding should have been modded insightful, not funny. Don't pull punches that need to be thrown, in the first place don't throw punches that shouldn't be thrown.