1. low demand

2. 0.13u lowered costs

3. yields have improved (different spin on #2)

4. Preperations for new speed grade

Good logical response Semi.

Ok gang, pick one or more answer and let us know which ones and why...

It has been widely reported that Computer sales are on the rebound. Now I'm sure that SOME may believe (or more accurately.... Hope) that all of a sudden Centrino is going to do an about face, and go into the crapper, but based on reported turnaround by motherboard makers, and reported improvement in overall sales, and estimates of Mobile gains in general vs. desktops, and the spread of WiFi to even more locations, that doesn't make sense, now does it. Clue enough for AMD Fans? I doubt it.

And why do I have a feeling, if the situation was reversed, and it was AMD mobile gains being reported, then none of you would have an issue accepting, and cheering for, the "Data that I hold onto so dearly" Hummmmmm?

If sales go up then INTC has to be able to produce to meet demand. The item they questioned was availability not demand. I didn't say it but I am asking that those that reply on the subject at least keep on the point of discussion and not skew the discussion with facts that aren't relevant.

Just don't fly off the handle and cram market data down someone’s throat when they voice an idle thought. Maybe just maybe even an erroneous premise could lead to a valid discussion of fact if you don't take it as a personal threat to one side or the other.

I don't run that store, ask them.... Besides, what was being discussed was overall Centrino sales, and as long as Centrino is IMO certainly pulling it's weight, by taking market share from AMD, according to Market DATA, that's what really should matter to Intel and AMD investors, not some transparent attempt to deflect that fact, with some B.S. about why a speed grade, is or is not sold at a particular vendor.

The data you hold onto so dearly is confirmation of what happened in the past. The clues the people on this board are searching for are indications of current and/or future performance. I might trust a market share report but I don't trust market share projections and thus I keep reading looking for that next clue...

I was going to jump on it but Keith beat me to it.

Whatever the problem is with the x1000 it's bios/mb/periprials not the centrino that's to blame.

"So you think Intel and OEMs agreeing to skip an 800 MHz Xeon
intermediate frequency Xeon in favor of a jump from 700 to 900
MHz isn't true because a year later production of the 900 MHz
had to be suspended temporarily to solve a problem? Apparently
causality and timelines don't matter much in your mental model
of the functioning of the universe."

You can look at it with rose colored glasses if you want. Intel didn't have any real competition in that sector so they can say anything they want about their ability to satisfy the market. To me spotty shipment and occasional recalls of product is a sign of a long term trend. Especially in light of the similarity between the xeon issues of old and itanium issues that followed. They may be unrelated in most aspects but the coverage of the issues have the same tone...

Looks to me like just another of the Intel Benchmark Specials that had to be recalled.

No end users reported any flaws. I wonder if any end users ever had one?

http://taipeitimes.com/chnews/2001/08/30/story/0000100844

"Intel Vice President Abhi Talwalkar said the company has resumed
sales of 900MHz Pentium III Xeon chips for server computers after
halting shipments in April because of a flaw. Last month, the
company said only a small number of chips were affected."

So they halted the shipments in april and didn't resume shipments until Sometime in late July or early August. Thats 3-5 months (depending on if you count april and august) that they weren't shipping them plus the time they hemmed and hawed about the 800 mhz version. It sure looks like you are digging up info that supports the claim Dan3 made about Intel having a hard time producing reliable parts above 700 mhz that spring and summer.

"Yes I do, quite well. Too bad it doesn't apply here for three
very big reasons:

1) Intel drove the move to 32 bit x86 chips so everyone knew that was the direction the market would go.
2) The 386 offered substantially better peformance for16 bit apps than the 286.
3) Intel drove a large and successful marketing campaign to deprecate the 286 in the eyes of users (the red circle
around and slash across "286")

AMD will not be able to duplicate any of these three factors
let alone all three."

1. True. The market leader (Intel) drove the trasition to the 386.

2. False. The 286 offered better performance at the time of the introduction of the 386, just as the 808x offered better performance around the time of the 286 introduction. The key to remember is that Intel wasn't the only one to make x86 parts during those times.

3. True. But isn't this point 1 restated?

"F.e. it's pity to see that the "slowest" Athlon 2000 is usually around $55 on pricewatch. So AMD is probably selling it for $45. And many, many of those chips are going to China, where they are remarked and then resold back as Athlons 2400. So the market is ready to pay more, just AMD is not able to collect those money."

The US market buys these as well and overclocks them or uses as is in the DYI segment. Just because its cheap doesn't mean the enthusiasts won't buy it.

can you stop and let the little people know what AC stands for?

"Alternating current although that is a bit of a misnomer."

snipped a bit...

"When a chip is tested for speed capability that is called AC
testing because the relevent spec(s) were in the AC section
of the data sheet."

Thanks, that makes more sense to me now.

ok, good enough, I just wanted to know what the abbreviation meant. I really didn't expect it to mean alternating current.

and now that the two of you agree that 10% + 20% = 30% can you stop and let the little people know what AC stands for?

"FYI, a little birdie in IAG told me last year that Intel won't
introduce a new speed grade unless it is yielding at least 15%."

Don't you mean binning not yielding?

I'm assuming if there are sellable speeds below the new one that the chip should be yeilding way higher than 15%.

"I thought that EPIC already had an AMD64 port of UT. Maybe it was just a tech demo."

My take is that there is already a port of UT but some unamed product (as in a new game that marketing hasn't even assigned a name to) is in the works.

There is nothing new about a game designer saying he has a new game in the works every 2-3 years...

Actually you can tell windows to put all the temp and swap files on another drive. Of course you have to do so in at least 4 different places (counting IE settings) and the number increases if you add NT/W2K/WXP profiles into the mess.

You name the microsoft OS and I can show you where to change the settings. I'm not going to say it's fun or easy but it is possible and not even arcane for anyone with basic dos/windows/regedit/tweakui knowledge.

"Not a Short, I'm sorry, that was fully incorrect:

Unfortunately such arrangements don't tend to hold up to the test of time. Companies like 3com are probably around to the 3rd or 4th loop on reusing mac addresses. Of course you aren't likely to use that ancient 8bit or 16 bit nic on a modern network but the mac address it has is most definately not unique if it's manufacturer is still making NICs today.

MAC addresses do have two parts, the company identifier and the serial number. Each is 3 bytes (six hexadecimal digits) long, making a 12 digit number in total. For instance, here is a US Robotics number:

00-C0-49-B4-01-00

The first part, 00C049, identifies US Robotics. The second part, B40100, is a serial number.

Now, what does US Robotics (or 3Com or anyone else) do when they run out of numbers in the serial number?

Simple. They request another block and IEEE obliges. MAC numbers do not get reused, 3Com (for instance) has several company ID blocks.

This scheme has been working now for more than a quarter century. Pretty impressive!"

Sorry but mac addresses do get reused wether you beleive it or not. Maybe if I took the time to look for it I could find some documentation of the fact but I haven't bothered to look for that info for at least 3 years so it'd take me a while to find it (if it is on the web at all anymore). I could give you some googled up URLs that make passing references to reused mac addresses but I expect you'd just dismiss them.

http://www.firewall.cx/mac_addresses.php

http://archives.neohapsis.com/archives/freebsd/2000-12/0146.html

I remember reading articles sometime in the 90s in Computer Shopper and PC Magazine that detailed the issue of reused MAC addresses unfortunately I don't know any easy way to find the text of those issues.

"Not a Short, I'm sorry, that was fully incorrect:

Unfortunately such arrangements don't tend to hold up to the test of time. Companies like 3com are probably around to the 3rd or 4th loop on reusing mac addresses. Of course you aren't likely to use that ancient 8bit or 16 bit nic on a modern network but the mac address it has is most definately not unique if it's manufacturer is still making NICs today.

MAC addresses do have two parts, the company identifier and the serial number. Each is 3 bytes (six hexadecimal digits) long, making a 12 digit number in total. For instance, here is a US Robotics number:

00-C0-49-B4-01-00

The first part, 00C049, identifies US Robotics. The second part, B40100, is a serial number.

Now, what does US Robotics (or 3Com or anyone else) do when they run out of numbers in the serial number?

Simple. They request another block and IEEE obliges. MAC numbers do not get reused, 3Com (for instance) has several company ID blocks.

This scheme has been working now for more than a quarter century. Pretty impressive!"[/I]

Sorry but mac addresses do get reused wether you beleive it or not. Maybe if I took the time to look for it I could find some documentation of the fact but I haven't bothered to look for that info for at least 3 years so it'd take me a while to find it (if it is on the web at all anymore). I could give you some googled up URLs that make passing references to reused mac addresses but I expect you'd just dismiss them.

"I can think of dozens of ways of differentiating which would not require different sockets. You could even put in a serial number. Do you think this is a cost? Not on an Ethernet NIC, every one which contains a world-wide unique serial number since 1976!"

Heh, do you beleive in fairy tales too? Sorry, I don't mean to be rude I just got a laugh out of that.

MAC addresses (the supposedly unique identifier in a network card) were designed much in the way phone numbers and ip addresses are (with subsections allocated to each manufacturer in this case as opposed to area codes or subnets). Unfortunately such arrangements don't tend to hold up to the test of time. Companies like 3com are probably around to the 3rd or 4th loop on reusing mac addresses. Of course you aren't likely to use that ancient 8bit or 16 bit nic on a modern network but the mac address it has is most definately not unique if it's manufacturer is still making NICs today.

"re: 1.7a and 2.0a P4 fans -- Yikes, is that the amperage? You'll need a fan to cool the fan!"

Yes, that is amperage. The fans are about 2" thick and may possibly be thermally controlled. I know they definately have a thermal sensor the sticks up in front of the intake side (monitoring the temp between the processor and fan).

Heck I'd have put Niceguy767 on ignore a long time ago if I were a paying member on SI. As it is I read the messages without looking at who wrote them most of the time and I still can spot his posts and skip them with great regularity.

Why shouldn't someone point out to him that he is just wasting bandwidth with his style/frequency of posts on SI?

ahh, yes, I see it now. Unfortunately the by/to lines are reversed between here and there so I sometimes get confused on who is reponding to who.

mostly off topic (biblical references on SI) and a P4 story.

http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=19160892

Maybe someone should tell Niceguy767 and Epinephrine that Samson never fought Goliath? As in the pairing should be (Samson and Delilah) and (David and Goliath).


Oh well, while I'm here let me also mention my latest run in with P4s that overheat. Dell Precision 350 P4 3.x GHz 533 FSB. Has a plastic shroud over a heat sink that would otherwise be passive. The shroud contains a really thick 80 mm fan. The rev A00 shroud/fan combo featured a 1.7a fan. The rev A01 shroud/fan combo has a 2.0a fan with the same size and slightly different blades. The 1.7a fan wasn't cooling the P4 consistantly (bios event log shows fan out of range and cpu temp out of range) PC shuts down on its own and Dell has to ship the new part to you. The fun part is that the PC is useless until they do and the last time I called them the part was on backorder for a week.

Obviously this is more Dell's problem than Intel's but I'm sure the people at Dell aren't happy about it. Did Dell cut corner's or did Intel spec the cooling requirements too low? I've replaced several of these lately and the old fan seems to spin ok on it's own, if it's below the min RPM it can't be by much.

agreed on the driver issue but your statement spoke of compatability not support.

Besides do you deny that W2K is supported by the hardware industry at this point?

Also do I have to remind you yet again that you can run W2K instead of or in addition to a 64 bit OS if you need the support that is already in place for legacy devices?

"wbmw - Porting to AMD64 FAQ

So it's not quite so "fully compatible" as we've been lead to believe..."

only if you haven't been paying attention. The 16 bit apps and x87 limitations have been known for months at the least. The can't mix 32 and 64 in same process has been know for years.

"Only 64-bit drivers can be used with 64-bit operating systems." This one has been around a while. You have to be sure you don't confuse claims about mixing 32 bit and 64 bit applications with claims of mixing 32 bit apps with a 64 bit OS and drivers.

"I thought Jerry Sanders created AMD. He might take exception to your comment. "

You might want to read some history books. IBM forced INTC to share the processor market with AMD as a second source for the original IBM PC. Essentially IBM was the orginal driving force to makeing AMD go somewhere. INTC just did the other half later by giving AMD reason to stop being nothing but a copycat design/manufacturing house and forcing them to develop new technology.

I can only say as a consumer I'm glad I can get fast CPUs at low costs. Thanks IBM and INTC.

I should probably mention that the Dell configuration page offers HT on or HT off as part of the ordering process but lists a disclaimer that the on setting only works with Windows XP.

I found that quite odd as I thought HT would have benefits in NT/W2K as well, maybe even in 98se.

http://configure.us.dell.com/dellstore/config.aspx?cs=555&oc=GX270SDTPAD&m_1=24G8D&c=us&...

Hyper-Threading is available only on the processors that have 800 Front Side Bus and may increase the performance of some applications. This feature is ONLY supported in Windows XP and can be enabled or disabled in the system BIOS at any time. The below option presets Hyper-Threading to "ON". Consult the Learn More Processor tool for more information.

You know I just noticed today that the Pentium M 1.4 GHz cuts back to 600 Mhz in battery mode. That seems a little extreem to me.

I mean 2 years ago I was using a C600 with a 900 mhz processor that if I remember correctly cut back to 4 or 5 hundred. Even if my memory is slightly off the lowest speed on the Pentium M seems low to me now.

*This is relative. Most laptops consider 6-cell as standard, rather than high-capacity. The small size of the Gateway 200XL simply prohibits large size batteries, unfortunately.


I was comparing the stadard and high capacity battery on a Latitued CSx earlierer today and subjectively the high capacity one felt close to 3 times heavier (probably around 2.5x the weight of the other one). It's definately more noticable on the thin and lights than on a C600 were there probably isn't a noticable difference in weight between the standard and high capacity batteries.

I have a GW store in my town so maybe I'll stop in one of these days...

You've taken the case apart?

I'll have to admit that I haven't seen the gateway builds. Are they still healthy (gateway not the their pcs)?

Keep in mind most dell laptops have a CPU fan and a "case" fan (or maybe I should say two case fans an in and an out). One or both of these will be a 3 way switched item (off, low, high). In the designs with a "passive" cpu setup the "case" fan is ducted straight to the CPU and only has 2 levels mid and high. Just because the fan isn't screwed onto the HS doesn't make it a passive design in my book.

So with all the dell designs in mind you may have 10-20 fan combinations (between types of fans, number of fans, and the combinations of speeds.

You generally won't hear any of the fans while docked unless it goes to 100% cpu util. Undocked is a different story of course but I'm sure your system is running at least one fan at low/mid speed at all times.

thank's, that install fixed some graphic artifacts in a video I had that played fine on another PC.

Windows Media Player and avi files are a hit and miss thing. There are tons of codecs that may be a factor and not all of them are compatable with each other (especially if someone has installed any version of the divx codecs and/or any of the older unofficial DVD ripping codecs).

It's also the case that the newer versions of WMP are specifically designed to not work with some of the older codecs (presumably in an effort to keep people from watching movies ripped from DVDs).

"You just managed to confirm that the Xeon is a 32 bit machne"

When was that ever in question?

"People believe (emphasis mine) that INTEL is doing a lot of things to deny competitors access to markets, and that certainly is not a rumour whatsoever." - I think that was Keith, forgive me if I'm wrong

"You refute your position in your own position statement. Absent any proof, people believe. If this isn't rumor, I surely do not know what is." - Greg s

You are just parsing his statement differently than he intended. Look at it like this:

"People believe ..., and that certainly is not a rumor whatsoever."

Assuming I parsed it correctly he is saying that the Inquirer is reporting on the way INTC is perceived which is worthy of a news story as it affects us all.

The only way I can see to "cherry pick" during manufacturing based on your explination (which I find resonable) is if you assume the parts that bin lower won't sell at all. Then you get the highest bin parts getting shipped and the lowest binning parts getting destroyed (hopefully before they get to packaging or assembly or whatever you call it).

Now at the distributor/retail level you could test for speed assuming the manufacturer downbinned parts for marketing reasons. At that point someone could cherry pick. Of course that only matters to the retailer/end user.

I think his comments are based on concern about AMD's ability to manufacture sufficient parts to meet market demand above the threshold already set by thier previous products and previous products of their nearest competitor (INTC). Wether or not he understands manufacturing processes is immaterial to his concerns about AMD's ability to make a profit.

The 'last-minute' diet of the P4 is what scares me most about it's future. If they put back in the features they cut the P4 could gain some IPC and then the old comparisons are shot.

If AMD can't get some traction soon INTC will eat 'em for breakfast.

"I'm reminded of an old OS/2 joke, "I can recalculate my spreadsheet, type up a document, download a file, play a game, AND format a floppy, all at the same time!" (For some odd reason, OS/2 fanatics always bragged about formatting floppies in the background.)"

It's not that odd a boast. I always get pissed when I need to move a bunch of files and format a floppy on a W2K or NT4 box and explorer decides to get slow as molasses or cpu usage goes to 100%.

If you plan to do work in other apps (browse the web, type in word, sort in excel, etc) it isn't that bad but if you try to do anything in windows explorer or my computer it just isn't pretty.

I can't say that I know from experience that OS/2 handles it better but I'd brag about it if Windows stopped that bad behaviour whether MSFT fixed it or it was offloaded to a hardware process.

"Buying a motherboard now for an upgrade to a CPU later is always a losing proposition. It just never works. A number of people are going to discover that.

People always want to make a future-proof hardware purchase. It just never works out."

It works for me all the time. The trick is to buy a midrange board and a low end processor then upgrade to the highest processor that board will allow when the price of the higher processor = the price paid for the lower processor. Often this gives you the higher processor and lower processor for less than half of the price of the higher processor at the time of the first build.

It also generally means that you are buying ram from an older generation (like SDRAM instead of DDR or DDR instead of DDR II) and see a significant price savings there as well.

Occasionally this doesn't pan out as well as you hope (like when the Athlon stalled for a while around 1 GHz) but then you sometimes luck out as well (like when the socket 7 board that you had a 300-400 mhz K6-2 in gets a super cheap upgrade to a K6-III+ that runs at 600 MHz and adds another layer of cache.

"Well, Tim Sweeney is talking about the engine content-authoring from the side of the game developer, not the game player."

And you couldn't tell that from the discussion?

Go back and reread http://www.investorshub.com/boards/read_msg.asp?message_id=1188661 with an open mind and a fresh memory.

"I recall a very recent interview with a big game developer (was he from UT or Doom?)" should have been enough to key the core concept for you as much as it was bandied about here. I bet if I dug them up you'd be at the heart of a chain of messages discussing the importance of Sweeny's statement.

http://money.cnn.com/2003/07/09/commentary/bidask/bidask/index.htm

The headline reads "Intel: out of options?" with a sub of "Microsoft's decision could put pressure on the chip maker."

But it's just another rehash of the old expensing options debate so I don't know how relevant it really is...

"NaS, I'm saying we have until the end of 2003 to release a benchmark that is named after the year 2003. Once 2004 comes, *then* it's overdue."

I'm saying if the 2002 version was released in January of 2002 then the 2003 version is already 18 months after that, and waiting 'till the end of year would make it 23-24 months later. If the benchmarks are a yearly thing (which they used to be) that is my idea of overdue.

Now considering Sysmark 2000, 2001, and 2002 all used Office 2000 you might say there isn't much need to update them yearly. But then again I thought you were the big proponent of using the absolute latest and greatest software in benchmark suites?

Personaly I wouldn't mind it if they slowed it down and just did a 2004, 2006, 2008 kind of cycle but the anouncement from last year that there would be a 2003 version makes the slow down unexpected.