Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
econo, oh please, please, please don't give them an excuse to start up the QS witchhunt again!!
wbmw, it was relevant. Now it's beating a dead horse. To a pulp.
Nothing new. Already been discussed. Drop it already.
Newisys 4300 page
Paul
http://www.newisys.com/products/4300.html
Keith, several things are interesting here. For one the MP 2800+ is at 2133MHz while the XP 2800+(Barton) is at 2083MHz. Look at the compared specs below:
Frequency--Voltage--Max Therm Pwr--Typ Therm Pwr--Max Die Temp
MP2800
2133MHz-----1.60V--------60W--------------47.2W------------90C
XP2800 Barton
2083MHz-----1.65V--------68.3W------------53.7W------------85C
XP3000 Barton
2167MHz-----1.65V--------74.3W------------58.4W------------85C
When you consider the frequency is between the two XPs it's even more impressive. Part of it's due to the lower voltage. Interesting that the Max temp is back up to 90 degrees, too.
from these docs:
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26426.PDF
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/26237.pdf
Keith, 'nuff said - my lips are sealed on this subject. :)
chipguy, your average "Joe Sixpack" that buys his PC is the majority of the consumer market, and he has no clue about the ins and outs of PC performance. He compares some basic numbers - frequency(or the number presented, such as QS), disk size, maybe memory size. Frankly, you give the average PC consumer way too much credit. Heck, half the darn salespersons are clueless fools when it comes to the reality of PC performance! The folks who frequent these boards are in no way representative of the buying public!
Paul
yb, what I meant was the metal disk that was actually an integral part of the chip package, on which the chip info was printed. The new version doesn't have this. It still needs a heat sink, though. ;)
Paul
Keith, one more step! :)
NaS, what a ludicrous set of "arguments"!
Good news re: competitiveness of XP-M. Be sure to note the difference in battery sizes when comparing the XP-M's and Centrino's battery life. It also says that PowerNow! increased the battery life by 50% over running with all power saving features turned off.
Paul
http://www.eweek.com/article2/0,3959,1054658,00.asp
yb, informally, it's generally referred to as nForce2 ver. 2. I think it's a third revision, although on the actual package it'll say A1(as in "0309A1" shown below). xbitlabs has some info in this review. Of course, if you can wait, you may want to hold off for boards with the nForce2 Ultra 400 version of the chipset that's coming soon.
Paul
PS. Also note the lack of the metal circle on top of the package.
http://www.xbitlabs.com/articles/mainboards/display/abit-nf7.html
wbmw, the problem is that AMD has to deal with the constraints of reality. Intel is the big dog, and you know darn well they'll never allow a TPI effort to succeed, because it would not reflect well on them. I can't necessarily blame them for that, they have to do what's good for them. I don't think AMD has the influence to pull off TPI by themselves(it is not a trivial undertaking), and they're the most powerful of the alternative manufacturers. So, they have to do what they realistically can accomplish. It's not the best but it beats the old alternative by a long shot. So, let's move on to other subjects!
Paul
wbmw, OK, so make it a PIII 1.26 or 1.4GHz vs. P4 1.6GHz.
Paul
http://www.intel.com/design/pentiumiii/prodbref/index2.htm?iid=ipp_dlc_procp3p+prodb_512kb&
Have we wasted enough bandwidth on this QS vs. MHz bit yet? Let's agree to disagree(not you and I, Keith; I mean QS suppporters vs. MHz supporters) and move on. It's here, and it's staying in one form or another, and it's worked OK. They haven't pulled a Cyrix, and that's what was most important to avoid. I'm dropping the subject now.
Paul
Boxx nForce3 "workstation" pics at AMDZone LAN party. Notice the wimpy little fan on the Opteron HS, it's one of the skinny 10-15mm thick ones. Also notice the manky old green bogus HS on the nVidia "chipset". Must run cool, too. I notice they're not using the vacuum cleaner GeForceFX 5800 Ultra, can't tell what card it is. It does use the extra HD power connector, though. Interesting that Asus chose the P8 prefix for the board designator, instead of A8.
directory of jpg files(Chris Tom hasn't fully posted on it yet)
http://burmese.home.texas.net/Images/2003-05-03/LansEnd/
PS. interesting Khepri boot screen image, too. re: HT links.
Tenchu, how is comparing Celeron to Banias any less valid than comparing Banias to XP-M(totally different architectures) or Celeron to XP(which Intel has tried to market it against)?
dougSF30, LOL!!
Tenchu, in addition, how do you judge the slugly performance of any Celeron based laptop(regardless of GHz chosen) compared to the same Banias unit? A little pot calling the kettle black here?
Tenchu, not that it accounts for it all, but ATI's chipsets so far stink. Eventually, either nVidia or someone else will produce one that performs comparably to Intel's, then we'll have a better picture of how much is CPU and how much is chipset.
Paul
wbmw, actually I was instituting a 9-month countdown to IBM's Opteron offering. You're not saying you found an Opteron system at their site, are you?!
Paul
Interesting K9 architecture rumors. FWIW, probably not much.
Paul
http://www.digit-life.com/news.html?1615#1615
wbmw, I can live with 9mos. and counting! ;)
EP, the momentum is building - for both.
Joe, I'm conflicted re: Dell. If AMD were as big as Intel, or at least considerably bigger, it'd be a no-brainer. But I think that with the products AMD has coming, there's going to be plenty of demand, at higher ASPs than they'll get from Dell, to use up their capacity. Of course, I suppose there's always IBMs fab, but would that be cost-effective? Maybe reestablish the UMC relationship to service Dell alone? In any case, I think there are points to be made for both sides, I'd just be real nervous making Dell a customer, which would mean THE customer, and not a very generous one at that.
Paul
fyodor, 1MB based on Tbred's cache density would be exactly 54.16mm^2. It's 13.54mm^2 for 256KB. From this doc:
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/23794.pdf
internal document page 4. Adobe window page 14 of 30.
wbmw, it can certainly give you a very good ballpark if you're familiar with the heatsink in operation, and you can get a very good idea whether the heat sink is warmer or cooler than with another processor, and therefore whether it's dissipating more or less heat. There are at least several heat sinks I know of that work with both P4 and Opteron that would allow a good comparison. Absolute precision, no, but a very good idea of relative powers.
Paul
wbmw, point taken - I can live with that take on it.
re: "Intel enthusiasts"
Funny, it strikes me that whenever AMD is struggling whether it's deep financial d**-d**, process issues or intro delays, the "Intel enthusiasts" here are very reasonable, offering good info and constructive criticism, fostering a civil exchange. But when AMD does something right, or when it looks like they have a killer product, the FUD/flame wars begin. Unfortunately, some "AMD enthusiasts"(myself sometimes included <:/ ), forget this and allow stuff to be stirred up. Fortunately, most come around pretty quickly. Also fortunately, there are some "Intel enthusiasts", wbmw for one, who tend to stay out of that mess and are a big plus to the board, not pulling punches, but willing to call a spade a spade, whether it sheds positive light on AMD or on Intel.
Paul
wbmw, your classifications seem a little hefty to me, but it strikes me that, since there is no hard and fast designation that applies across OEMs, this whole line of arguments is a little absurd. Both sides.
Paul
wbmw, actually Johan used CPUBurn to maximize power consumption, and estimated from that. Whether that would be closer to an absolute max power rating or a TDP, I don't know.
Paul
http://www.aceshardware.com/read.jsp?id=55000268
SemiconEng, funny how you apparently chose not to read the end of my post.
Paul
SemiconEng, I don't recall if I posted it here or elsewhere, but I commented (yesterday?) on this already. I figure either they decided that since this is the most important intro in their history, that they decided to bump up the voltage on the review systems to make sure to the nth degree that there wouldn't be any stability issues of any kind. I also noted that it may have been counterproductive, since it makes people wonder if there is a headroom issue. Who knows? In any case, if the Inquirer story is true, we should know soon enough if it's a red herring.
Paul
XP40000 - you think that would account for the cache of the XP being twice as dense as Opteron's? Why?
chipguy, it's not all that impressive. If you scale it up, Tbred's L2 cache is 54mm^2 per 1MB(actual is 13.54mm^2 per 256KB). Opteron's is another story. I'll agree there's something odd about it, as I've already posted below.
Paul
D, at least JB did at times offer very useful info and technical help; SZ -- not!
I don't necessarily feel the need to boot him, but the "Ignore" feature works quite nicely. :)
PS. He's the only one I've used it on.
Spokeshave, I agree, this board's been excellent! I'm amazed this hasn't happened before now. We've got a great crew here, and though things may get a little heated now and then, there is generally a refreshing degree of honesty and wanting to get to the truth(yeah, with a little posturing, evangelizing and FUD now and then).
CJ, what subzero can't seem to get is that the estimate was for, as you noted, ClawHammer with 256K L2 cache. But we can't have facts interfering with subzero's agenda, can we? Unfortunately, this seems to be the first real instance of this kind of cr*p going on on this board since spokes and Keith took over. Or, fortunately, I guess I should say.
Paul
fyodor, you seem to have a point about the L2 cache. According to this pdf, http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/23794.pdf
XP's L2 cache is 13.54mm^2 per 256K, which would give us 54.16mm^2 for 1024K. Since this is clearly far less than the @50%(@95mm^2) of the die image that Opteron's L2 cache takes up, one has to wonder. Could it be they're having enough trouble with cache defects that they had to add enough to be able to disable some and still have the 1Meg? 95mm^2 vs 54mm^2 is quite a disparity. Any thoughts?
Paul