Intel man says INQUIRER, you're a mule

Intel says Prescott to get new stepping

As I sit here in my cube at Intel, my response is -- yes, entirely.

My mother likes to say "you can lead a horse to water, but you can't make it drink. Especially when the horse is a mule." Usually she is referring to my father. After having written to you several times over the years to help you out with inaccuracies in your stories -- inaccuracies that make you look foolish -- it is clear from your replies that you also are firmly in the 'mule' camp. I will make no further efforts to help you.

David J. Ayers

[David Ayers "help" came yesterday when he wrote us an emal from his "Intel cubicle" describing our sources as BS. We replied to the emal saying that it wasn't entirely BS, and we had a reason for that. That's why we wrote yesterday saying we were puzzled by Intel August roadmaps that incorrectly flagged the first Prescott at 3.40GHz as not being 90 nanometers. In fact, we took the trouble to ask Intel about this when we visited its stand at ECTS yesterday, hence the additional information contained in the story linked above. Sheesh. Some people are just so rude. We suppose it's part of the "confrontative culture" that's part of the nuts and bolts of Chipzilla's success. Ed.]

US Republican Party outsources fund raising to India
You whingeing pommies

Adamson Rust chooses to write:

"We do hope and trust here at the INQUIRER that the irony of underpaid people in Harayana helping robots to call possibly out of work Americans because of a widespread policy of corporate outsourcing is not lost on our readers."

I don't know about that, but the irony of a technology magazine bemoaning the cost savings enabled by technology seems to be quite lost on the writer and the editor.

Whinge on.

Nan DU

Just a sheer rant about Real

Hi Mike,

This is my personal rant about www.real.com site experience, their player and file format. Feel free to use it any way you like.

I went to the damned site to download latest RealPlayer (free version) just to be able to watch some clips. First I had to enable JavaScript (which is something I extremely dislike) and then I was all stuffed with cookies like a parrot. Then it asked me to install the ActivX control (applet) to be able to pause/resume download. Pause/resume my ass! Resumable downloads are essential for sites that are offering large files and are usually implemented on server side without any need for installing applets on client computers. After all, every literate computer user have some sort of download manager -- GetRight, FlashGet, ReGet, etc...

Why they decided to take different approach? Simple, to be able to push advertisement on you! While downloading, substantial amount of bandwidth is used to display ads in their "download manager". Remember that you had to enable JavaScript and cookies? That means that every ad shown could store its cookie on your computer. So after exposing you to a security risk by asking you to enable JavaScript they endanger your privacy online by throwing ads with cookies on you and last but not least they waste substantial amount of your valuable dearly paid bandwidth.

Not to mention that RealPlayer by default puts itself into the system tray thus taking chunk of your system resources constantly. How about taking over all audio/video file formats in addition to their own? I had a hard time to restore some custom settings for avi and mpeg files that I added before I installed their cocky player.

Oh, I forgot to say that their format sucks! Picture quality is low, sound quality is poor and all that regardless of a bitrate. Player is a slow starter and it crashes sometimes since it can't handle corrupted files very well. What I would expect from them if I were optimistic enough to expect anything from that kind of company/corporation is to release some "RealPlayerReallyReallyBasic" version of their product which would enable us to play just their native audio/video file and streaming format and nothing else. Or even better just a codec pack to enable other players to play them. All that should be available for download through the use of regular download managers without JavaScript, cookies and ads.

My 2 cents -- I recommend every person on this planet to stay away from Real Networks, their player and file format. But I guess no one would take those two cents for Real if it weren't for the rant :)

Regards,

Igor

EU Microsoft punishment should fit the Bill

Hi,

You say microsoft should improve security etc..and that people should move toward linux. I think Linux needs some works before it can be on desktops though.

My dad, who knows close to nothing about computers can install windows on his machine and have run it just fine. The same can't be said about Linux, even I run into tons of problems with Linux, most of which have to do about security, permissions etc..which makes Linux more secure but at the expense of being user-friendly.

Until Linux offer the ease of use that windows offer, it will never ever replace windows on the desktops.

Another thing, while I agree that windows has lot of security flaws, tons of the problems with virus, and exploits that appear are because people don't patch and don't do everything to make it secure in the first place.

The Linux community isn't made of, 99% of people that are computer illeterate. Hackers don't put their time in trying to break linux because they realize that the people using it know what they are doing, and will protect themselves, and react fast to possible problems. Also if I am to create a virus, or try to hack into a computer why bother with the Linux community when the chaos will be much less then on windows if only for the fact that there are tons more windows out there.

Hackers want to be in the news and they won't be by breaking into 10 people computers, but they will when they break in 10000.

Linux is nice, but it is highly overated. Windows is nice too, but it could be better. No OS is good enough right now to be perfect. They both need work, they both have room for improvement.

I will do my best to not rant. Your suggestion to punish Microsoft by forcing it to give a grant to the Linux community is absurd. There are two major flaws with your argument that just don't stand up with a walker.

1. Using Microsoft software IS a choice.

Microsoft; contrary to popular belief, does not force every personal computer user to use their software. Consumers have a choice of what software to run as well as a choice as to what to do with that software once they've purchased it. Because of this, people should “vote” for the best software with their wallet. If Linux is a better solution for your needs; then by all means, go install Linux on your computer and give Microsoft a big juicy raspberry while you do it.

You should not blame Microsoft for writing, distributing, and ultimately profiting from their buggy software. You should blame the consumers who have consistently given Microsoft the message that they love what they bought by buying it again and again for every system they've ever owned. That was their choice and Microsoft is getting the message that people like what they are getting because people keep buying it.

2. Who decides where this silly "grant" money goes?

Give me a break. No company would stand by silently if a court ruling decided random company B gets free money. Jeez, everyone and their momma would be filing and complaining to get a piece of that pie. Why would Linux be so special as to be the only company to get “grant” money? What about all of the other companies which could use the money to encourage Microsoft to improve their wares? What about all the colleges and universities which are developing the next best thing? What about … Well, you get the idea.

Truth be told; you cannot sue a baker because his pies taste bad. Not only does it not make sense to sue the baker for selling nasty pies; but how the heck does it make sense to force the baker to donate money to the bakery across the street as a punishment? What about the bakery down the block? Here's a suggestion; just start buying your pies from the bakery across the street in the first place.

Now just swallow your humble pie and be happy.

Tom Shelley
Router Solutions Incorporated

Dear Mr. Magee,

I write to you because of your understanding of all things soft.

Yesterday when I went to the Microsoft Update Site I was notified that a new Critical Update 823559 (referable to Microsoft Security Bulletin MS03-023) was available for download. I downloaded it and went on my way. This morning I checked back at the Update Site and was notified that, once again, I lacked 83559. Thinking that it had been somehow deleted I downloaded it once more and restarted for good measure. Just to be sure I visited Update and, after finding that I again lacked 823559 I looked up my installation history. Sure enough, it showed that I had installed the thing twice. I reinstalled for the third time and tried the Update process to check. Guess what! I was informed that I needed Critical Update 823559!

My question to you is. Am I in some sort of cyber time warp condemned to repeat my tasks over and over like Sisyphus? There is clearly too much Update Deja in my Vu.

Yours faithfully,

Robert M. Schwartz

Good morning,

One of the tools government has to help internal market against foreign competition is to impose tariffs. Offering Microsoft products at a lower price only reinforces its monopoly. If the US government imposed a tax on monopoly products, it could redistribute wealth as grants to education institutions where software innovation has good chances of taking place.

Windows XP at $99.00 may be overpriced for what it is worth, but does not constitute a serious barrier. $500.00 would make me think twice.

Pierre Chartier

Hi,

I don't think your idea for the EU to give a "grant" to the linux community is very well thought out. Who would you give it to? What conditions would you attach? Not a very market based approach and it wouldnt help consumers at all because linux is already more secure then windows and neither MS nor consumers care. A far better approach would be to require MS to open the windows source code and allow competitors to recompile their own editions of Windows for resale to the public. I like having a product as easy to use as windows, but it would sure be nice to have it a bit leaner and more secure....

Cheers,

Damo

Power Consumption

Hello!

As I am very interested in power consumption of modern processors and I know something about CPU architecture (I was and still I am a computer magazine editor), I have some reasons why I believe Prescott will be very power hungry even when manufactured on 90nm process.

1. Thermal Design Power is typical power, not maximum. While Northwood 3.2 GHz has TDP od 82W, it's maximum power consumption is 96W. TDP is in fact power consumption while running typical programs (eg. not utilizing all Micro-OPs possible). HyperThreading increases TDP because it improves Micro-OPs utilisation, but it doesn't increase max. power, because that is achieved by utilising all available Micro-OPs - and HyperThreading only improves Micro-OPs utilization, it can not create any new OPs.

2. For CPUs with HyperThreading, TDP to max. power ratio is about 85%. For CPUs without it, it's much lower.

3. Prescott processor will be able to execute 4 Micro-OPs per clock cycle - this was discowered by Chip Architect web page by comparing Trace cache sizes. That is the most of all desktop processors ever - Pentium was able to do 2 instructions a clock cycle, Pentium Pro / II / III 1 complex + 2 simple, Athlon 3 complex, Pentium 4 3 decoded in trace cache. And now we have 4 instructions (Micro-OPs) per cycle. Sure it won't result in 25% power cunsumption because not all power comes from execution units. But it can result in 10 to 15% power increase.

4. Pentium 4 processors use controled decrease of core voltage with increasing amps. In case of Willamette and Northwood, this is -0.0019A per 1V. While 3.2 GHz Northwood has VID (Voltage Identification) signals configured to 1.55V, it's max. power formula is 67.4A * 1.42194V = 95.8W. See that 1.42V? That's the voltage while at max. amps. VR Down 10 guidelines suggest Prescott would have lower decrease of voltage with rising amps. That however implies that with those high amps, voltage would not be so much lover that that of Northwood - and because wattage is voltage times amps...

5. Intel's 90nm process creates 50nm transistor gates, while 0.13um process made 60nm and 0.18um did 130nm. Transition from Willamette to Northwood had very noticeable effect on transistor size and hence on their frequency potential and voltage abilities. 90nm process won't give such a high boost. It's a trend - 0.35um processors used voltage of 2.8V, 0.25um did 2.0V, 0.18um did 1.75V, 0.13um did 1.525V. Transistor voltage is nowadays not decreasing so rapidly as it did few years ago. That means Prescott could have relatively high VID settings.

6. Prescott is more complex than Northwood. It can have somewhat 100 milion transistors, supports new instructions, has larger L1 data cache and possibly also larger Trace cache (that is so power hungry Pentium M doesn't use it).

Now do some mathematics. Prescott FMB 1.0 TDP was 89W. Divide it by 85% and you have around 100W max. power. At 78Amps maximum, that is a voltage of 1.3V at max. amps. So we can presume VID settings would be for 1.4V or even little more (depends on voltage decrease with increasing amps). That is high but corresponds to note 5 I wrote. Now we have FMB 1.5 TDP number of 103W. Do the same divide by 85% and we have 120W. That is terrible and one would guess something like this is not possible. But can you tell me where would those 91Amps go if it is not possible? Even with higher TDP to max. power ratio, it has to be very high - at 90% it is 114W, at 95% (is that even possible?) it is 110W. By the way, did you know Nocona (a Prescott Xeon variant) is to use 900 gram heatsink?

Do you still think first Prescotts will be on 0.13um? There is no reason, Intel has 90nm production capacity on D1C development fab - this fab was the first to manufacture 0.13um Intel processors and it is equiped with most up-to-date technology. They develop process technologies here and also manufacture first batches of processors until they copy exactly to production fabs. Production shouldn't be a problem.

So what can Intel do to lower Prescott power consumption? There is interesting thing to note - Celeron processors have the same power requirements as Pentium chips. And because the difference is only in cache size, does that mean Intel just disconnects data paths while leaving power ones intact? There may be a reason to do so - disconnect 30 milion transistors and you have electrically very different CPU. Now, remember that article at Chip Architect - Prescott and Yamhill project. For sure they doubled ALU unit count to be able to emulate 64 bit ALU units by doing half of required 64 bit math on each pair. Just imagine what would happen if they just disconnect this second pair but leave power supply intact. Running at 6 to 8 GHz. Maybe they are now working on disconnecting it properly.

Regards,

Petr