To 'us', yes probably. To the Dells and HPs of the world, perhaps not. One-stop-shopping for a majority of the important parts should be pretty inciting for them. It's not like this thing totally sucks, and even if it did, that wouldn't necessarily stop the major OEMs from being interested if the price is right.
Phenom video leaks from Lake Tahoe
- Thread starter digitalwanderer
- Start date
Well, maybe not buying a rig today, as sale quantities are expected limited and apparently nonexistent until at least the end of the week.
Those Intel parts are already out, and the AMD chips that will be selling supposedly have a TLB issue where exceeding current clock rates leads to the processor crashing the system under load.
This is by AMD's own admission.
Overclocking Phenom at this stage seems to top off at 2.4-2.6 GHz if you don't want your system imploding at random.
Another stepping on top of the previous two another steppings will be needed to correct this.
It also throws the AMD 3.0 GHz demo into even more doubt, as nobody was allowed to seriously test it.
AMD's competitiveness is rather inconsistent. The tests I've seen thus far show it is either massively behind, somewhat behind, relatively equal, or a sliver faster than the Q6600
I noted that in my edit. MSI makes shit motherboards for enthusiasts. They haven't put out a true enthusiast mobo since the NF3 days (Neo 2 Platinum). Their NF4 & NF5 boards are an absolute joke. Their entry-level boards such as those based on the 690G chipset are a fine choice however. I don't want to buy an entry-level board for an enthusiast rig though.
The price is not right though. This round AMD decided to supply the channel first. They hyped the massive shipments of HD3800 for this quarter, and you can bet that if they had anything even remotely similar going on for Spider they'd have mentioned it by now. OEMs are not interested in Spider because it is not competitive, thanks mostly to the joke that Phenom currently is. Drop the prices by 50% and then we'll talk. Of course, if they did that they'd be bleeding money even faster than they already are. At least then maybe they could move some chips somewhere other than the low-end though...
This review of Phenom at Anandtech has a pretty bleak tone.
Man, what happened to them ? I used to be all over their CPU's in the Athlon/Athlon XP/Athlon 64/X2 days, and now they can't get a single chip out there that takes a decisive lead in the performance arena (either CPU, GPU or chipset)...
This might even hurt Intel customers, because they might stop cutting prices of their CPU's, due to lack of meaningful competition.
Man, what happened to them ? I used to be all over their CPU's in the Athlon/Athlon XP/Athlon 64/X2 days, and now they can't get a single chip out there that takes a decisive lead in the performance arena (either CPU, GPU or chipset)...
This might even hurt Intel customers, because they might stop cutting prices of their CPU's, due to lack of meaningful competition.
Which Phenom did Kyle B see go to 3.6?The engineering sample that had some parts downclocked?Really?What I've read from him states that currently available B2 steppings are likely to top out at 3Ghz at best. Most reviewers have gotten fairly crap overclocks, in the realms of 100-200 mhz.
Do you really think that AMD has loads of CPUs that clock great, but they'd rather sell them as lesser specced ones in order for enthusiasts to unlock their potential?Is this like the R600 story where they didn't want to compete at the top, but rather wanted to provide affordable performance to everyone?That one didn't workout all that well, looking in hindsight.
What I saw and consider fairly scary is an AMD slide suggesting that the TDP for the 9900 2.6Ghz chip will be friggin 140W!Come again?That's fairly huge...and doesn't bode all that well for higher clocked parts on 65nm.
Looking back, maybe AMD would've been better off dumping the whole "true multicore" approach in the desktop area, and go the bolt-together route that Intel chose. They would've had quads earlier, and I think that would've helped their bottom line somewhat, not to mention market perception.
The true multicore thingie is irrelevant for most desktop apps-server stuff is another story. I simply don't see those IPC and architectural improvements between Phenom and the oldish Athlon to be all that gush-worthy, and it's fairly safe to assume that Phenom cost them quite a lot in terms of R&D, costs that'll be mildly hard to recoup.
Bouncing Zabaglione Bros.
Legend
A year or two ago when it was just USB devices and the occasional external USB hard drive, yes, but in today's market with SATA disk writers arriving and SATA hard drives all over the place and pretty much standard, it is a big deal. If you're buying a performance gaming machine, the last thing you want is your hard drive interface to be slow because of bugs.
AMD know this, and it's been on the list of important fixes for SB700, and if they'd been able to execute properly (suprise surprise), SB700 would be on the launch motherboards, instead of a few months down the line.
2900 XT's market penetration is practically non-existant compared to *any* 8800 sku (even the Ultra). I'll bet they sold as many Pros & GTs as they did XTs.
My point being, since there were some people who thought R600 was a worthwhile purchase, there will be people who, despite having expected Core 2 killer will talk themselves into still getting Phenom, actual results notwithstanding.
*Shanghai* is the 45nm version of Agena with larger L3 (6MB). AMD needs higher clocks to compete with Intel. Do you think AMD's manufacturing prowess is superior to Intel's? If you do, I've got a bridge for sale in Brooklyn, really cheap too.
I think you've bought their marketing bs hook line and sinker. Wake up man! Don't throw away your time defending AMD's failures! I used to be like that, until they disappointed me one time too many.
No shit AMD's world-champion overclocker can o/c to 3GHz, whoopdy-doo! You cannot buy a Phenom right now that will reliably overclock to beyond 2.6GHz, according to all the reviews.
790FX is a fantastic chipset. Too bad it's priced way too high, and the flagship CPU it supports is a joke compared to the year old+ competition.
I agree. I don't think there will be enough of those people for AMD to return to profitability on the back of Spider platform or Phenom sales though.
Ppl as in DIY enthusiasts?Most certainly not...but if they can snag some good OEM deals(mostly due to the 38XXs being nice and cheap, and the 7 series of chipsets being very nice), things won't be so gloomy for em'.
Bouncing Zabaglione Bros.
Legend
If only it was available for Intel processors!
The only hope for Spider until they upgrade the southbridge and make faster CPUs for it is to sell it with a juicy platform discount to OEMs who can then flog it out to customers cheaper than an Intel/Nvidia platform.
They don't have any OEM design wins for Spider the last I heard. Thanks mostly to Phenom, you can bet.
True that.
Agreed. Unfortunately, there is no indication this has happened or is likely to happen. You can bet that's what they're shopping around, but apparently none of the OEMs are biting.
I don't know if I feel sorry for Charlie for buying AMD's marketing BS, or if I'm pissed at him for spreading the lies and trying to improve public perception of Phenom. I suppose it's a mix of the two.
I think AMD should try to slap a couple of these on a die for the first 8-core architecture. Its every bit as valid an Octo-core as Kentsfield/Yorkfield is a quad core and on paper it would offer more power than Intels highest end CPU ata potentially much lower price point.
Obviously very few benchmarks would demonstrate the potential of 8 cores but i think they would get a lot of buyers just on the strength of having theoretically more power. Plus there must be some niche areas that really could benefit.
Would it be possible to do this on AM2? If so AMD might be able to salvage something afterall!
Obviously very few benchmarks would demonstrate the potential of 8 cores but i think they would get a lot of buyers just on the strength of having theoretically more power. Plus there must be some niche areas that really could benefit.
Would it be possible to do this on AM2? If so AMD might be able to salvage something afterall!
At 65nm?
It seems difficult to fit that much silicon onto a package that will fit in current sockets.
Not that AMD would want to sell such a chip on current sockets.
An octo-core MCM with only a single memory controller with access to memory would strangle performance.
Then there's the heat concern.
Obviously, the 130W TDP of the higher end is out of the question.
260W per socket is out of the realm of consumer cooling.
Even a 65W TDP per chip would be excessive.
It seems difficult to fit that much silicon onto a package that will fit in current sockets.
Not that AMD would want to sell such a chip on current sockets.
An octo-core MCM with only a single memory controller with access to memory would strangle performance.
Then there's the heat concern.
Obviously, the 130W TDP of the higher end is out of the question.
260W per socket is out of the realm of consumer cooling.
Even a 65W TDP per chip would be excessive.
Yeah, AMD doesn't seem to have enough all-working cores to avoid "tri-core" CPUs or ability to put out anything over 2.4GHz in volume, but slapping together two ~290mm^2, 600M transistor cores should be both simple and inexpensive.
Similar threads
