Opteron prices way out of line

SUBJECT: XEON price attack

Thank God! The Opteron prices were way out of line with their price-to-performance ratio. Add on top of the outrageously priced chip the cheapest motherboard is $533!

AMD is going to have to shit or get off the pot.

David

"But for MP Xeons, Itanics, it's buttock clenching time"

This article causes me great concern for Intel. I am afraid that if they are to clench only one buttock they may come to irreparable asymetrical harm. Or do they alternate between left and right?

Russ Stemler

Subject: IE Bug

The IE Bug has actually been know about for over 1 1/2 years, just never reported By M$. A friend of mine has a web site link on his Page that is

"CLick here if you want to Crash Internet Explorer"

The Link goes to a page that does exactly that. He put the page up just under 2 years ago. We tested the page with IE 5.5 and 6 Both flavors crash.

Its kind of funny the person that told my friend is a MSCE for a large M$ supporting firm with ties to the GM and they reported the bug back then and MS hasn't fixed it yet

William Zimmerlee

Your correspondent should know that your latest example isn't a valid HTML comment. A correct HTML comment containing the text would be:

<!-- <input type crash> -->

which doesn't cause a crash. The example <!-- ><input type code> will cause the problem since IE appears to interpret <!-- > as a fully closed comment (although the comment close tag is supposed to be '-->', IE is very lenient, as are most browsers).

My own investigations have shown that the problem is that the type attribute isn't assigned a value but is mentioned - the 'crash' can be omitted completely: <input type>is sufficient.

The other piece of misinformation is that the problem is in shlwapi.dll. The line of code that actually causes the access violation is in the StrCmpICW() routine in shlwapi.dll, it's true - but the problem is caused by a NULL pointer passed as the first parameter to the routine. This routine was called by a private routine in MSHTML.DLL, the HTML parser, which is where the problem lies.

Technically, according to the HTML spec, an attribute which is present but not assigned a value should be treated as if the attribute had been assigned a value of its own name. Hence <input type> should be treated as if it were <input type="type">.

It's possible that other unassigned attributes which should have values will cause problems like this, however I have so far been unable to find any.

Mike Dimmick

Arron replies

<!-- <input type --> kills IE and is a valid comment.

for amusement, for those who don't believe just how poor Outlook is (the guys at work refused to believe it would crash just receiving an email until I demonstrated it) I've put a demo up on my site ( http://www.ivor.it/ieboom.html)

Anyway... thats enough of this IE crashing stuff, it does get boring after a while. I'm sure there'll be another more interesting exploit along any day now.

Cheers,
Ivor

Subject: Tests show Microsoft product 66-95% faster

Let them comment, and comment they will. People will get their backs up, but calmer minds will learn from this report and Microsoft's days as 'king of the file server', if valid, will be numbered.

More importantly, those benefiting from the improvements to Windows 2003 and file sharing should take the time to thank Samba and the open source community. Microsoft stated that they focused heavily on improving the performance of file sharing because it was embarrassing that Samba was doing a better job than Microsoft on their own turf. Without Samba's excellent work, MS would more than like have ignored file sharing performance assuming they could do no better.

It's amazing how much competition can improve services for everyone.

Rodd

Sorry but this does significantly compromise their basic validity.

First of all it would appear that Microsoft chose the systems on which the testing lab carried out the comparison , and the Red Hat system seems to have been running a 2 year old Kernel. The latest Linux stable kernel (2.4.19 ?) is now lacking about a years worth of SMP performance work on the development series (2.5.x) while the Linux development series is probably not yet stable enough in its product cycle for a fair comparison.

Secondly I would have thought that it is common knowledge that what Microsoft pays for is by definition not independent. The testing firm claims to have investigated performance issues concerning the Samba configuration, but was there any open invitation to the Samba development community to recommend configuration options for the setup, or to debug problems with this ? Presumably Microsoft's leading system developers were involved in the choice and configuration of the Windows 2003 version.

Thirdly the test involved using a network file-sharing protocol which is of Microsoft's design, which Microsoft has chosen not to publish full documentation for. The fact that previous versions of Samba have outperformed previous versions of Windows is testament to the reverse engineering skills of the Samba developers, but when the choice of playing field is so clearly sloping in one direction this is hardly comparing like with like. How about a comparison involving a Microsoft system chosen and setup by Linux NFS developers serving NFS clients in competition with Linux ?

Sincerely,
Richard Kay

I'm honestly not surprised about the VeriTest Win2K3 benchmarks--not because I think MS has better developers than Linux, but simply because I never expect either side to sit still. Prior to Win2K3, Windows has taken second place to Linux performance-wise pretty much since the release of the 2.4 kernel, and we all knew MS couldn't bear to let that continue.

Perhaps the VeriTest benchmark is flawed. I'm sure I could turn up a few fallacies if I wanted to, but I'm just not interested. You gain a proper perspective on things when you see benchmarks like this.

If these marks are any indication, the apparent enormous lead Win2K3 has will probably vanish after about two months (when Linux 2.6 is released). MSFT's two months in the sun just aren't worth bickering about.

"Anonymous Hero"

There were three things that concerned me about the benchmark, and which were not covered in letters posted today:

(1) - The benchmark measured "peak" performance, rather than sustained performance.

It's very easy to cheat and give a higher peak performance. One example is to buffer data from disk for a while, then spit it out over the network all at once. Given a 1-in-4 duty cycle of this type, a system can exhibit a peak score double that of it's rival, but actually have sustained performance of half it's rival.

This kind of benchmarketing is far from limited to computers - most loudspeakers are sold based on "music power", "peak power", "peak maximum power output", etc etc, rather than the more stringent RMS power rating (no relation to Mr. Stallman). To qualify for an RMS rating, a speaker must sustain a continuous power input without overheating, for a number of hours. Power, incidentally, is like MHz - different speakers can produce very different volumes of sound, when given the same power input.

(2) - Only one set of benchmark results was shown.

This is a far cry from the battery of tests that "hardware review" sites throw at machines almost every week. As anyone who reads those sites knows, different benchmarks can throw the balance between two rivalling products one way or the other.

This raises a great deal of suspicion about whether Microsoft (the benchmark sponsor) or the benchmarking company carefully searched for a benchmark that showed Windows 2003 in the best light.

Something else to think about - why were Pentium III Xeons used, rather than the newer Pentium 4 Xeons that have been available for ages now? I can forgive them for not using AMD, because the Athlon-MP isn't available in enterprise-sized configurations, and the Opteron has only just been released. I'm no fan of the Pentium-4, but the decision not to use it still stinks to my mind.

No discussion of the hardware choices was presented. No rationale for the test selection. The whole paper is big on low-level details, but lacks the big picture.

(3) - The disk setup did not even approximate a typical enterprise fileserver.

Please. Tell me who, when administrating a mission-critical fileserver that they've bought a very expensive support package for, would proceed to use RAID-0 on it?

For the uninitiated, RAID-0 offers maximum disk performance and no redundancy at all - if one disk fails, all data on the array is lost. Backups are all very well, but a server running RAID-0 will be out of action until the replacement disk is fitted and the backup restored onto the array.

Most sysadmins would choose RAID-1, RAID-10, or RAID-5 - all of which offer redundancy - depending on their performance needs. Of those choices, RAID-10 probably offers the best performance, and can be used with the set of four or six disks available to the servers tested. RAID-5 is the most popular choice, however.

I'd also like to further some comments already posted in letters:

(4) - An old version of Linux was tested against a not-even-released-yet version of Windows.

A quick look at redhat.com shows that "Red Hat Linux Advanced Server" is not on sale. Instead, they're selling "Red Hat Enterprise Linux AS" as a completely equivalent product, so at the very least, the product description in the paper is noticeably out of date. Further, although RH Enterprise is only released on an annual basis, there are quarterly updates - there is no indication in the report that these were applied to the system under test.

As others have already pointed out, using old versions of software - particularly the kernel and the SAMBA tools - can have significant performance as well as security implications. As it turns out, they did in fact install a later version of SAMBA on the Linux Advanced Server system, because the original version was not reliable enough to complete the test. (Don't ask me how that one got past Red Hat quality control.)

(5) - Extended-3 filesystem, journalling modes, ReiserFS...

There is no indication that the test included an evaluation of different filesystems and options that Linux offers. Even though they conducted a review of the SAMBA options (concluding that settings very close to the defaults were appropriate - but they didn't say which options they HAD changed?), they simply chose the default for the filesystem itself, and used only the default mount options for it.

Any competent Linux enthusiast or sysadmin knows about the different ext3 journalling options, even if he doesn't use them. In summary, the journal can be used for metadata alone (aka the inode tables, and the equivalent of the DOS FAT), or also for data. If the journal is restricted to metadata, you can also choose whether data is flushed to the disk before metadata is committed to the journal, or not. There is a tradeoff between performance and power-loss data security, between the three available options.

Of course, there is the convenient and simple "noatime" option, as mentioned, which prevents the need for a write to the metadata every time a file is opened. The equivalent option was explicitly selected on the Windows server, so why not on Linux?

There is a blatant error in the documentation of one of the enhancements made to the NTFS filesystem. The command "CHKDSK /X /L:65536" was described as "increasing the size of the NTFS log [journal in Linux parlance] to 64KB". In fact, this increases the journal size to 64MB, as can be verified by running "CHKDSK /?" to view the option help. (And didn't Microsoft deprecate CHKDSK in favour of SCANDISK, back in the DOS era?)

Obviously the guy writing the paper is not the same guy as performed the test - or else, the guy performing the test doesn't really know what he's doing, and is just following a "performance tuning recipe". Or both. It's hard to tell from over here, so I'll let you make your own mind up.

It's interesting to note that no such journal-size modification was attempted for the Ext3 filesystem. I can tell you from personal experience that it makes a difference, especially if the journal is being used for data as well as for metadata.

On that note, I should point out that the default block size for Ext3 is 4KB for large volumes, not 1KB as stated by one reader. I agree that this is much smaller than the block size selected for NTFS. It does, however, match the memory page size used by Linux, but I don't know how much effect this has. It should be noted that the mke2fs manpage states that 4KB is the largest block size supported.

No mention of ReiserFS, which is included as standard by most recent Linux distros, and appears to have been stable for quite some time, was made at all. Some say that ReiserFS makes a big difference to filesystem performance when using many small files, or directories containing many files. I have been unable to verify these claims using the limited hardware and tools available to me (bonnie++ reports broadly lower or similar performance when the working fileset exceeds physical RAM, on my particular hardware), but I don't think these people even tried.

I hope this sheds a little light on the weird and wonderful world of Microsoft-sponsored benchmarketing. The aforementioned hardware-review sites would have gone out of their way to avoid precisely the types of problems I just highlighted. :-)

Disclaimer: I use Windows 2000 for gaming, Linux for everything serious, and I'm sending this e-mail using MacOS X. I'm not affiliated with any involved companies.

Have a nice day.
Jonathan Morton

Subject: Robert's Ignorance ( yesterday's letters)

Even though I long ago swore off participating in Apple vs. MS online flamewars, I am driven to write today by the particularly bothersome way that Robert tries to sound cool by making disparaging remarks about Apple that are not based on fact. Robert should have checked Apple's most recent 10-K before writing, as therein he could have read that Apple had in fact nicked MS's idea pretty well over the past 5 years, "turning a profit" in every one of them but 2001.

See page 18.

Further, while Cupertino has had its share of well-hyped flops (G4 Cube, for one), it's hardly fair to say that all Apple does is go from one failure to the next. Even though I can't stand to look at one now, was the first generation iMac really a "soon forgotten failure"? The iPod certainly received a lot of hype after its introduction, and it really hasn't turned out to be a Segway. People of all OS faiths still covet them.

Finally, in Steve's last sentence we again see that he hasn't done his homework. Yes, the stores have probably been a waste of AAPL's money. But if he read the news, he might have learned that the "fantastic direct sales website" has sold 1 million songs in its first week.

Yes, yes, this implies nothing regarding the long-term success of this latest venture, but don't call it a failure already.

People are quick to point out that Apple users live in a Reality Distortion Field, but I think we see here that, just as there is both matter and anti-matter, there is an anti-Apple RDF as well. Some people refuse to admit that Apple can do *anything* right, and never admit that it has *some* good ideas and even some successes. I don't really understand why. Does Apple's continuing financial solvency threaten their manhood?

Not a zealot,
John

Subject: Carly F delivers anniversary message to HP troups

Having been an "underling" of Carly for a brief period of time, I tend to keep an eye on her, just can't help it. It was a couple hundred million dollars earlier in her life, when she was Wonderwoman at Lucent at the time... What is that about rats and sinking ships?

Anyway, even in my conservative estimation, the announcement was a rather expensive one, so I hope it was worth it. It took me about 10 minutes to read it. Maybe I'm a bit slow, but adding some allowance for the usual discussion HP folks might be involved in afterwards, I think each HP staff member, too, spent at least that much time on it. According to the HP web site, they have 141,000 employees world wide and I'm sure they all heard or read it. Counting wages and usual overheads, this must equate to about $1 million in costs (unless they did it in unpaid overtime). Add the 1$ million it cost in Carly's valuable time and it's a hell of an expensive announcement.

Regards,
Sandor

Subject: US government subsidises Business Software Alliance

"HEAVILY SUBSIDED TRADE organization the Business Software Alliance (BSA) has received $200,000 funding from the US government to promote a software piracy scheme aimed at children."

This statement is just one hell-of-a head scratcher? This reveals more of the mind of the writer than the US gov. This seems to be a scheme to create a scheme and make the US gov accomplices in trickery and misrepresentation. SCHEME...that's the word! Perhaps the BSA, out of the goodness of their hearts, should have taken on this devious propaganda by themselves? It seems realistic enough for at least one person at the Inq to believe so. And maybe the editor of one of the most popular web sites in the dot com world would also contribute a fraction of his wealth to such a cause he too very much believes in? So I guess the US gov is not only conniving, but really stupid. The BSA represents some of the most reputable software makers in the world, but they're not selfless and genuine in conviction to promote the truth without more pay. I guess the US gov should learn from the Inq's reputaion to judge honest representation with no scheming and exploitation of cause for self gain, and really caring for the worlds children. And I guess we can all agree... no more "drugged up weasels!"

Steve

Subject: OpAserv

http://www.fireav.com/virusinfo/library/opaserv.htm says that this worm program claims to come from the BSA. But I expect you knew that.

I'm ambivalent about copyright overall; I think present-day law qoes too far. And I was raised Christian, with the evangelism ethic: if you got good-quality informational assets, you should share them freely... But I don't have a particular objection to U.S. government part-funding one particular copyright education programme of the BSA's as well as subsidising the whole BSA enterprise.

Robert

Subject: A bill passed in Indiana in 1897 apparently decided to plump on the value of Pi as 3.2, although various other numbers are suggested.

If you would read the article, you will see that the bill passed only one chamber (the House), but not the other, an action which (in 49 states in this country) by itself merely generates fodder for the shredder.

Your item implies, in the typical tittilating Inq half-story, that it was made law.

Ah, tabloids. (But, aren't they entertaining?)

Oh boy I get to flame!
Mike Heald might like to reconsider his interpretation of the Video Transfer question in his article.

Anyone with enough l33t IT skillz to write for The Inq' should have the brains to realize that if you defrag your drive BEFORE a large file transfer, the transfer will be completed more quickly and with less drive activity as the heads don't have to travel all over hell and back scattering snippets of your porn movie all over your hard drive.
Please buy Mr. Heald a pint or two and explain this to him.

Thank You,
Pietro Boggio

Subject: Please for the love of God!!

I'm begging you please change your picture in the side bar today. That woman's incredible smile is just way too distracting...

How am I supposed to read your articles whith her smiling like that?

Ken

Mike replies

Raquel is a fine lass. She used to work for a distributor, and if you click on the link, you'll see she was a Miss UK finalist in 1989. Admittedly a long time ago. The pic is far more recent. She'll disappear tomorrow.

Haven't seen her for around 5 years, was hoping someone would say hey Raquel, you're INQ pic of the day.

Mike

Subject: Man to claim RIAA collusion at DMCA hearings

This article seems to ignore that Microsoft has been able to beat numerous antitrust battles, although I wish the author success!.

Joe

Subject: Using old PCs "makes CEOs look stupid"

This is totally fictional and any resemblance to reality is not intended ;^)

Secure Union Corporation, a US Government Security Consultant Firm, warns that using Microsoft PCs is costly and dangerous.

The corporation has warned their client, the entire US government, that the pervasive use of Microsoft PC's with their proprietary architecture and security infirmities makes them vulnerable attack by hackers aged 8 to 80 from any country, manipulation by an unnamed giant in the PC software industry and lawsuits in areas ranging from lost identities to PTSD and mouser's wrist.

Buying a Microsoft PC, SUC warned, leaves the buyer trapped in a situation of continuously upgrading software to fix bugs and security holes while simultaneously introducing new bugs and security holes. In addition users are left fealing like they have lost half of their expertise on how to do anything at all on a PC. "It's like building a sand castle at the water's edge when the tide is coming in." commented their chief consultant: N. Gin Eyah.

"It is extremely expensive to constantly retrain employees to use new versions of the software that they use every day. Why? Well, simply stated, although these products may share the same name, they are actually completely different software packages with different interfaces, commands, features and ofcourse bugs."

Mr. Eyah, who authored the report warned government executives that they run the risk of looking increasingly too stupid to live and terminally uncool to the point that at the next change of administration, they may be viewed as political liabilities. Baaaaa.

Analyists at SUC advise that puchasing Linux PCs at Mall Wart's may yield better stability, not to mention major cost savings, but warned that issues of integration, management and security might be better addressed by even more robust solutions with greater up front costs.

Mr Eyah noted that "its not so much the make and model of PC that yields productivity and efficiency; it's how you use the tool that determines the efficacy of the user and the organization. Of course if the tool breaks periodically and is unusable often enough, one should consider buying an industrial strength tool that is of demonstrable higher quality, security and reliability. It may cost more up front, but in the long run, it gets a lot more done."

Email address supplied

Subject: SATA Maxtors Kinda Available

Another interesting problem with Maxtor and SATA occurs when Maxtor ATA133 drives are used with SATA-PATA converters/bridges. These adapters (such as the one supplied by ASUS) often use Marvell bridge chips which wont work in ATA133 mode. Unless your BIOS allows you to force ATA 100 you can't use these drives on your nice shiny SATA controller with it's oh so slim leads....

Cheers,
Russ

Subject: The Legend of Zelda: The Wind Waker

While i agree that the game (Legend of Zelda: Wind Waker) is truly an Excellent game, as a fan of the prior games in the series i have to admit to being a bit dissapointed with the graphics in particular. In all the prior games, the design of the characters and items, as well as the landscape, all created a certain atmosphere that has drasticly changed with the latest game. The game play may be similar and the characters are the same but the game just doesnt have the "feel" that the previous games had. I really cant even find the words to adequitely describe that feeling but i can assert that it is missing from the newest installment. I attribute that difference of feel to the design of the characters and surrounding landscape. This more than anything else is why the game has drawn so much controversy.

Brandon Johnson

Subject: " Crusade starts for Pentium Ms in desktops"

Well done, Oh Nobel Servant of the masses !!

I notice you didn't specify the PRICE that you want Intel to charge for said desktop Pentium Ms.

Is that omission on purpose - so you can then BLAST Intel for whatever price they might charge should they succumb to your demands?

Paul R. Engel

count me in, i can see it now, letters of marque, flying the jolly roger off the sears tower, oops wrong channel;

make a great itx mini boxen, great utility, lots of potential, use it for just about anything.

don

Subject: Atari rises again

The Frogs pinching Atari, its just not on.

To me Atari will always be remembered as a purveyor of home consoles and arcade machines(do arcades still exist?)

Much missed but very much dead. Its name will mean nothing to anyone under 30 and is only nostalgic to those over 30.

Dead but not forgotten, but still dead as a parrot.

Michael Batt

µ