A grain of salt with your c|net, Apple still shines.

by Alan Graham

I'm currently working on a book about the Virginia Tech Apple Supercomputer Cluster, so when I saw a recent c|net commentary on the cluster project entitled, "A grain of salt with your Apple," it really got my goat. Three things really bother me about poorly written commentary like this:

1. Commentary is fine, but base it on facts and be sure to get your facts straight.

2. If you don't know what your you're talking about, you shouldn't publish your opinion as if it were news.

3. If you are going to publish commentary, at least have the guts to put a name(s) to it.

The first thing in the piece to catch my eye was this statement:

"But when it comes to cluster supercomputers, an important technology expected to become the foundation of utility computing, there are hidden costs aplenty. It cost $5.2 million to buy the Virginia Tech gear, but that figure doesn't include what the school says were "hundreds of volunteered hours of Virginia Tech faculty, staff and students to help set up the 19.25 tons of computers, routers and other equipment."

Without actually doing any fact checking, the implication here is that if the University had strictly paid for all of the labor incurred when building this project, the costs would have significantly increased. Therefore there are enormous amounts of hidden labor costs that affect the bottom line.

First of all, the $5.2 million also included construction costs (actually construction costs for power and cooling upgrades, were an additional $2 million - however, those are not platform specific, so whatever vendor they selected, those costs would be the same), not just the "gear." And while there was a large body of volunteers working on the project, their role consisted primarily in removing 1,100 G5 computers from the box, seating a special network card, powering the machine to test for DOA, and helping to place the machines on the specially constructed racks. The cabling and setup of routers and other equipment was primarily done by University Staff and representatives from hardware manufacturers. In fact, while I was there documenting the project, I saw the University staff (who are on salary) on their hands and knees running miles of cable. I'd also like to point out that the reason such a large body of volunteers was required was not because they wanted to exploit a free workforce, but because the University was on a very tight deadline in order to make the Supercomputer list.

When you work out the details (and you're very generous), you might add an additional $400,000 in hidden labor costs, but this doesn't really diminish the accomplishment. Even at $5.6 million it would still have been the best price performance leader on the TOP500 List in history!

The next thing to catch my eye:

"In an academic environment, there are plenty of graduate students on hand to figure out the best arrangement of processors, memory and network gear for a given task. Students also can translate software written for other computers to Apple's systems, which with a single machine now on the top 500 list are far from prevalent in the supercomputing arena."

Sorry, but this information is not just wrong, but completely disrespectful to the highly skilled project leaders like
Dr. Srinidhi Varadarajan, an assistant professor of computer science at Virginia Tech, and Jason Lockhart, director of the College of Engineering's High Performance Computing and Technology Innovation, who initiated the project. Come on folks, you don't let grad students design critical systems. Dr. Varadarajan did most of the coding, along with the Mellanox team that wrote the InfiniBand driver. And while the Virginia Tech cluster is the first Mac on the list, it is in good company as one of the many Unix systems on the same list. And I'd also like to point out that all of the software used on the project was open source...yes all.

"So the Apple project at Virginia Tech may be a wonderful educational project, but commercial customers who have less interest in experimentation are more likely to pay specialists at Linux Networx, RLX Technologies, IBM, Sun Microsystems, Dell or Hewlett-Packard to plan the plumbing, package the software and plug in the cables. And those companies aren't going to rely on Macs."

Experimentation? With a quickly approaching deadline, and millions of University dollars on the line, there was no room for experimentation. This had to work and had to work the first time, mistakes or miscalculations would be devastating. The folks at Virginia Tech are specialists. The Mac cluster isn't their only cluster and Dr. Varadarajan is a respected expert in his field. Why this assumption that the experts from Apple, Mellanox, Virginia Tech and others didn't know what they were doing and only folks from Dell (of all people) are the only experts who can build these solutions?

The c|net piece was nothing but spurious speculation and Mac bashing from a publisher who has a history of negative press about the Apple platform. Shame on anyone who assumes that this solution cannot be duplicated again with less money than competitive solutions, and shame on c|net for implying that this was all a worthless exercise and corporate clients would be smart to seek their solutions from a more reliable hardware company.

If corporate clients were smart they'd pick up the phone and call Virginia Tech and Apple computer who have the third fastest supercomputer on the Top 500 list, and ignore c|net, which to my knowledge, didn't make the list again this year.


2004-01-15 07:44:13
good work!
Thanks for setting the record straight, and good luck on your book. It seems that what Dr. Varadarajan and his crew at Virginia Tech was quite an amazing accomplishment, and is sure to make a fascinating story.

Is it just my imagination, or has c|net had a decidedly anti-Apple bias as of late?

2004-01-15 09:01:18
They are Just Jealous
The fact is you could not have built it with other computing platforms for a similar price and that bugs C|Net.

IBM is listed as one of the companies people will go to, the 970 is a IBM processor what are they griping about?

The G5 also has a great IO system which is why it does as well as it does -- Apple has eliminated all bottlenecks.

2004-01-15 09:13:22
c/net and beyond
Reading c/net is akin to watching Fox "news". Facts are irrelevant and honesty is a stranger. I thank you for the specifics exposing the c/net cluster distortion and hope that others will dare to call a lie a lie wherever it may be found.
2004-01-15 09:40:57
c/net and beyond
you lefty socialists are so pathetic.
2004-01-15 09:46:35
I think it's funny...
Between this and the iPod+HP announcements, the Apple FUD crowd is really losing its cool. Their arguments are getting more rash and like foaming rants, rather than the old "pfft, Apple, who cares?", and they're starting to sound like a bunch of people who are angry for no good reason.

Five years ago, it was easy to say Apple didn't matter, and most folks would believe you. Now, these people are finding themselves in a position where they have to *prove* Apple's a bad choice, and it's making them mad. The G5 supercomputer article is laughable, because anyone who can do math knows this the G5 supercluster provides a really, really good cost to performance ratio. Even Dell or Gateway aren't dumb enough to spout the old "hidden costs" argument on this because they know it will only make them look desperate. The only people who will really get taken into this article are the "choir" - people who already agreed with him before he wrote the article.

In a way, I think articles like this are a good thing. Articles like this mean that everyone's talking about Apple now and that they are quickly regaining a powerful presence in the industry. Those who don't want this to happen (maybe because they make a living writing about Windows annoyances =) are trying hard to protect their investment in Windows, but what these articles show is that they're feeling increasingly threatened by Apple's presence. Writing these articles thus confirms that Apple does indeed matter. =)

In the end it is c|net which will be hurt by this, not Apple.

2004-01-15 09:52:24
Finally, someone who has more respect than the whole C|Net combined showed what a pop-trash site C|Net is. Unfortunately, you can't shame someone who is shameless.
2004-01-15 09:55:49
c/net and beyond
umm, you are really confused.
Fox expresses the facts and it very much cares about honest. I believe you met the c/net is like cnn.
2004-01-15 10:02:59
Yep, that C/Net commentary was odious...
...but it must have made any potential supercomputer buyers laugh. Anyone serious enough to consider such a purpose, would KNOW that any published figures often does not include the full cost, as each installation is unique with unique requirements. And, anyone worth his salt would be able to ballpark his labor costs in a new york minute. The real question was, who was this "commentary" targeted to? Couldn't be serious supercomputer buyers. Could it just be for casual readers to reassure them that the status quo rules? Could it just be to reassure their advertisers, the same companies he cites as supercomputer experts? Could it just be FUD to drive website hits? Inquiring minds need to know!
2004-01-15 10:42:16
Let me be the first to say
Thank You!.
2004-01-15 10:57:34
c/net and beyond
Personally, I agree with the first poster. To believe that both c|net & Fox *don't* have their own agenda shows a lack of observation..."MY news source doesn't have a slant, but all the others do!" is kind of childish. Read (or watch) widely, and consider opposing views. You'll see that no one is unbiased. Well, unless they agree with you. :)
2004-01-15 11:06:00
c/net and beyond
Exactly, it just depends on which direction they spin: Counterclockwise (left) = CNN, Clockwise (right) = Fox.

Remember, "Right" is a direction, not a statement of correctness.

2004-01-15 11:24:54
Great Work and Thanks
CNet, like most of the IT "news" sites, is HEAVILY dependent upon Wintel Cartel advertising to pay the light bill. Can you ever go to their site and NOT see a Microsoft/Dell/HP ad? What do you expect from a Wintel bought and paid for site? You get what you pay for.
2004-01-15 11:46:12
Normally I wouldn't even bother...
remarking on the cnet piece. I mean look, they just grabbed a few "facts" or perceived "facts" from the Internet and pieced together a commentary piece that had very little basis in reality. For example, they have no idea about the months and months of design that went into the solution. Or the months and months of going from vendor to vendor to find the most affordable, best performing solution. Apple wasn't the first or the fifth solution they looked at...more of which will be in my book. They just make it sound like they slapped this thing together with a bunch of grad students.

I was there when the construction crew, electricians, welders, HVAC, plumbers, and other misc. workers were pulling double shifts (on the clock and on the payroll) to get this project done on time.

This was a professional installation from the start.

2004-01-15 12:35:22
c/net and beyond
Very good!!!! Discredit the speaker when you cannot discredit the speaker's argument.
2004-01-15 12:59:42
Those in a glass house shouldn't throw stones
Anyone care to verifiably prove that c|net has a history of negative press about the Apple platform? The author felt secure in his journalism (and I use the term loosely) to attack c|net about their facts and then doesn't bother to back this assertion with his own. Ahh hypocrisy... it's alive and well.

Too bad his points are mostly valid until that bit of tripe.

2004-01-15 13:12:56
Those in a glass house shouldn't throw stones
Obviously you don't read C|NET or ZDNET news very often if ever at all. It's as obvious as the nose on a person's face with the way they cover any "news" story relating to anything by Apple.
2004-01-15 13:27:13
Those in a glass house shouldn't throw stones
How about this for starters:


2004-01-15 13:37:57
Those in a glass house shouldn't throw stones
He's attacking C|Net for labeling an opinion piece as news. The same mistake couldn't be for a weblog entry.
2004-01-15 13:43:13
c/net and beyond
2004-01-15 13:44:37
Re: I think it's funny...
"First they ignore you ... then they laugh at you ... then they fight you ... then you win."

[Gandhi, I think]

I know, rather too high-falutin' for this topic, but I always think of that quote when I see stuff like this.

2004-01-15 13:52:39
Don't be so hard on C|net
It takes a lot of effort to make all that stuff up. They do a first class job and fully deserve the MS sponsoring.

(Great read, thanks!)

2004-01-15 13:59:17
Thanks for this article!
Thank god that there is still good people around :)
I will buy that book when it is ready.
Keep up the good work :)
2004-01-15 14:04:46
Those in a glass house shouldn't throw stones
i worked for c|nyet after they bought out zdnet. it was sad watching them whittle away at the skilled cadre of mac reporters until all that was left were third rate writers willing to push the editorial anti-mac agenda... all the mac writers who were at c|nyet with any self-respect have moved on to other jobs.

the worst of the bunch has to be charlie "shill for whoever pays me" cooper, who gave up objectivity for lent many years ago.

if you want your tech news with a pro-wintel bias, ain't none more biased than c|nyet.

2004-01-15 14:36:00
Those in a glass house shouldn't throw stones
Your reply does a disservice to the word"ignorant."
2004-01-15 14:41:04
Those in a glass house shouldn't throw stones
If you post commentary without a specific name, it reflects the view of the editorial staff...which sets the tone for the publication. So yes...I think it is fair to say they are biased, when as a publication they posted an editorial of this nature. If it was the view of one or two people...it would have mentioned their names.
2004-01-15 15:01:24
Thanks for setting everybody straight
It's about time somebody covered this in more detail. Too bad for the MS pundits at CNet. Cowards that don't take credit for their own lies.
2004-01-15 15:12:32
Those in a glass house shouldn't throw stones
well, lets see.....

1. There was the "5 Reasons Why You Shouldn't Buy an iPod" headline. I defy you to find the "5 Reasons You Shouldn't Use Windows" equivalent.
2. Then there was the "Apple refuses to fix Security Holes in Jaguar" headline based on the FUD from the slack-ass security outfit that was sucking on MS consulting dollars and had fired its CTO because he co-authored a security study critical of MS.

.....that was 30 seconds of thought....I am sure there are plenty more in addition to the BS on the VA tech piece.

2004-01-15 15:13:44
Don't be so hard on C|net
Ah hell, I tell better fishin' stories that are better than the C|net flub.
2004-01-15 15:24:37
c/net and beyond
Don't try to get flamed.

I mean that was such a non-statement.

Go away.

2004-01-15 15:30:14
Can you say "I hate Apple"
2004-01-15 15:39:30
Can you say "I hate Apple"
I read the c|net story when it first came out and thought to myself, self, that is just crap. Not a lick of it true, none if it other than the fact that the cluster is a that college and the computers were from Apple. How many Cluster project go off with exempt IT people who make less then $30K a year and all they do is haul equipment around all day and night, and as I saw in the pictures they got pizza....Did he not figure that in. Pizza alone for 2200 CPUs and hundreds of able bodied folks would run another million at least, 5 times a day for a few months. What a moron!

I stop really reading c|net a while ago. They don't like Apple, and for that matter have an odd stance against Linux too.

C|net seems to want to write bad news stuffs to just attention. Apple has faired well in the past 7 years and the new stuff is doing better. Our company is looking at a couple of Apple products right now and will be replacing some Windows/Novell servers that do nothing but cause trouble. Just as a side not I am a long time AS/400 Admin and go to bed at night and sleep well knowing that the IBM PowerPC is running my data center, and now my iBook.

IBM as cool as they are have never understood Windows or what to do with it. DELL peddles Windows like "Cure All Snake Oil" and HP kind of keeps quiet about the whole thing.

Call Apple if you need a solution. I think they need to increase the way they approach businesses and start selling into companies here and now. They are ready for the big time, and I look forward to it.


2004-01-15 15:41:59
Can you say "I hate Apple"
I look forward to the book. That should be a good read.


2004-01-15 15:44:56
It's just really irritating...
It's not ``if you don't know what your talking about'', it's ``if you don't know what you're talking about''.
2004-01-15 15:47:25
Apple doesn't always help itself
I've jsut spen two days at a journalist briefing on Apple servers, Xserv in particular. I couldn't have got the information I know have about hands-on use of Apple servers any other way because Apple is still way behind other vendors in supplying journalists with review kit to get experience with and it doesn't do pre-release code or briefings to get people up to speed in advance.

Apple also wants to suggest that it's very easy to write cluster/distributing computing code and tells us things like 'this application was written in 5 days by an intern'. That's easy to turn into a perception about grad students doing the work.

But I'm a little confused; first you say there's all this expertise in the team and then you say grad students don't have the experience. Don't most university reserach teams include a lot of grad students working on the projects they put together?

You obviously have lots more information about how things are working 'on the ground' but here's a journalist perspective. Virginia Tech is a bit of a poster boy for Apple at the moment; discussing various key server technologies, again and again Virginia Tech was the example Apple rolled out. As well as the cluster setup, they talked about the hours Apple and Virginia Tech folk spent crawling around fixing cables in the maths lab; they talked about the free swap they're doing for G5/Xserves. I'm not sure how many companies would get that much hands-on and financial support for setting up systems.

I'm not defending c|net; I don't work there, I don;t know what research they did or what their position is (although I know the 5 reasons piece on iPod wasn't iPod bashing, ore a look at the alternatives for folks who wanted comparisons). I'm just giving a journalist perspective on some of the issues on reporting on Apple

Mary Branscombe

2004-01-15 16:09:31
Apple doesn't always help itself
Do journalists learn about grammar and spelling these days?
2004-01-15 16:24:50
Apple doesn't always help itself
i have a website named after myself. does that make me a journalist too? or just give me "journalist perspective (tm)"

"5 reasons not to buy an iPod" is hardly an unbiased comparison of alternatives, from the title on down, except if you look at it with "journalist perspective (tm)."

the again, on her site, mary cops to working for AOL and PCPlus... must be where she learned spelling, grammar and objectivity, the hallmarks of anyone imbued with "journalist perspective (tm)!

2004-01-15 16:24:58
Apple doesn't always help itself
i have a website named after myself. does that make me a journalist too? or just give me "journalist perspective (tm)"

"5 reasons not to buy an iPod" is hardly an unbiased comparison of alternatives, from the title on down, except if you look at it with "journalist perspective (tm)."

the again, on her site, mary cops to working for AOL and PCPlus... must be where she learned spelling, grammar and objectivity, the hallmarks of anyone imbued with "journalist perspective (tm)!"

2004-01-15 17:27:12
Those in a glass house shouldn't throw stones
> I defy you to find the "5 Reasons You Shouldn't Use Windows" equivalent.

Just to add to the invective, I myself have such a list:

5. ILoveYou
4. Nimda
3. CodeRed
2. Melissa
1. MSBlast

2004-01-15 17:37:36
Those in a glass house shouldn't throw stones
How 'bout the fact that IAN Fried who routinely writes about Apple primarily writes his pieces based on press releases and rumor sites and has never spoken to anyone at Apple?

How 'bout the fact that many months ago he changed his byline to INA Fried so that email sent to him pointing out his inaccuracies would be conveniently dumped to an inactive mail address?

2004-01-15 18:47:24
good work!
What an enjoyable romp the prior postings were. First, thank you for the thought and work on the orginal piece.
Now, thanks to Mary who thinks she is a journalist - I are one two.
From that point on very enjoyable. Finally, to the nincompoop that thinks Fox is honest and their "we report you decide is legit" you have the TV and the government you deserve.
I'll come back and good luck on your book. As Paul O'neill (sp?) said on 60 minutes with reference to his own just published book, (paraphrase) "I wrote it because I believe there is still a market for the truth."
I believe that too. Thanks again.
2004-01-15 19:35:12
Looking forward to your book
I work in the same field of study which uses the current fastest supercomputer in the world, the Earth Simulator. We envy our colleagues who get to work with them, but for a small institution like ours, we need a different solution.

We are seriously considering a Virginia -Tech like PC cluster solusion. It would be a joy to read about the challenges and requirements needed for such project. Can't wait to read it.

2004-01-15 21:05:31
Don't be so hard on C|net
Did I tell you about the one that got away?
2004-01-15 21:35:06
I sent this to the editors at C|Net
Dear cNet,
I am writing you to call attention to some misinformation being packaged as authoritative commentary in an unsigned article from your website. Full of fluff and unsubstantiated assertions that fly in the face of reality, the piece is called _A grain of salt with your Apple_. I hope that you will forward the following URL (along with this message) to whomever wrote the erroneous and
obviously agenda-driven (or at best motivated by prejudice) article. I would
be intrigued to hear his response.

The response to the irresponsible "journalism" is here:

Your relationship with Intel notwithstanding, it is your professional ethical
responsibility to at least fact-check your news and commentary.

Now, on to MY take on why the X is significant:
When it was posted that the X cluster had exceeded 10 TFlops, I did some simple
math in order to compare the G5's floating point performance versus Intel's
offerings per GHz. The figures I used are were published by the maintainer of
the Top500 list and available here:

The figures I used were compiled from the highest representativeof each processor in the list, and the thousandths were rounded up if the
following digit were 5 or greater.

The equation used is
(1 chip/GHz per chip) * (Gflops/No. of chips) = Gflops/GHz

Here is what I found:
G5: 2.336 Gflops/GHz
Xeon: 1.284 Gflops/GHz
Itanium2: 2.972 Gflops/GHz
Itanium: 2.648 Gflops/GHz
Pentium 4: 1.197 Gflops/GHz
AMD Opteron: 1.268 Gflops/GHz

As you can see, the Itaniums lead the pack, followed by the G5. Nothing else
bears mention.

But let's look at pricing:
A 900 MHz Itanium2 single processor system costs a few hundred more than a G5
flagship model. Let's compare the performance, using the extrapolated data
from Top500:

HP entry level Itanium:
1 chip * .9 GHz * 2.972 Gflops/GHz=2.67 Gflop/s
Apple Flagship G5:
2 chip * 2 GHz * 2.336 Gflops/GHz=9.344 Gflop/s

So, for less money, you get almost 3.5 times the floating point performance.
And this isn't even taking into account the lack of a fully G5-optimized
compiler or the power of the SIMD unit for vector computations.

So even on the level of an individual workstation purchase, Intel has no answer
for the G5 in terms of price/performance, no matter what spin journalists and
columnists and marketers might try to put on this hard, unbiased, scientific
data. Furthermore, there is no answer as yet for the G5 as a supercomputing
and serving platform, especially when you consider OS X Server is sold in an
unlimited license version.

The Terascale X should only get faster, too, as they tweak it. An even
scarier thought: as you know, they have released G5 xServes specifically
designed for clustering. I imagine a comparable system to the X made of these
would be yet faster than this hand-made phenomenon.

I hope that your writers in the future dispense with the rhetoric, or at least
do more scrupulous fact-checking before they publish, in keeping with the
standards of their profession.


2004-01-15 22:38:34
It's just really irritating...
Hey idiot, skip down to the middle of the article where it says:

"When you work out the details (and you're very generous), you might add an additional $400,000 in hidden labor costs, but this doesn't really diminish the accomplishment."

That's right, it says: "you're very generous".

What does this imply? Is the author stupid? No. In all likelyhood M$ Word was at it again, autocorrecting things where it shouldn't.

2004-01-16 01:22:44
C/Net Crap
I think C/Net writes crap like this about the Mac, knowing that they’ll get lots of page hits from Mac users, and hence more ad hits. They aren’t interested in getting their facts straight, that wouldn’t generate as many ad hits. This is one Mac user who has stopped taking the bait, I don't read C/Net articles any more. That's one of the best ways to hurt them, simply ignore them.
2004-01-16 05:56:56
Please be Smart
If you're going to publish criticism of someone for not getting their facts straight, please get your facts straight. The excerpt below has the word "your" when the author clearly means "you're".

2. If you don't know what your talking about, you shouldn't publish your opinion as if it were news.

2004-01-16 06:04:37
Please be Smart
First of all...it is a grammatical/spelling error, not a fact. And second, this is a blog not a news site...on occasion these things slip through (but I'm sure you never miss these things in your writing). If you are going to criticize, pick something of substance.
2004-01-16 08:38:47
XServe costs
"they have released G5 xServes specifically
designed for clustering"

Has anyone figured out how much a new cluster would cost if it was built using the new XServe rather than G5 towers? I would guess that the machines themselves would be more expensive but you could use standard racks rather than custom built ones, and they should take up quite a bit less space.

2004-01-16 08:51:34
Yep, that C/Net commentary was odious...
That's a good point. The audience definitely isn't university or research corporation CTOs, who would be more sophisticated in their analysis and were surely floored by System X.
2004-01-16 10:10:40
Those in a glass house shouldn't throw stones
Well, actually, Ian is in the process of becoming Ina... not to ditch email, but to outwardly portray his/her true inner gender...