PDA

View Full Version : CNet Commentary: A grain of salt with your Apple


MacBytes
Jan 11, 2004, 02:22 AM
Category: Opinion/Interviews
Link: CNet Commentary: A grain of salt with your Apple (http://www.macbytes.com/link.php?sid=20040111032248)

Posted on MacBytes.com (http://www.macbytes.com)

Approved by arn

iJon
Jan 11, 2004, 02:30 AM
wow, why am i not completly surprised that cnet wrote this.

iJon

ITR 81
Jan 11, 2004, 03:12 AM
Where is the authors name?

In most companies they have their own tech teams assemble computer setups.
The gov. and mil. both use current employees and put them into teams to tackle new installations. Do they get paid extra?? No. They are paid via their current salary. This is the same for most IT companies. If a company is building a supercomputer they usually want to go the cheapest route. Right now Apple under cuts all of them. Most companies want a UNIX or something close to UNIX like BSD not Linux or Windows.

Whoever wrote the article isn't too bright because I doubt it would take more then a $7 million to hire a team to program and install the SC cluster. $7 million is still cheaper then any of the top 5 SC's.

Also now with the new Xserves and Xraids the deal is just getting better and smaller.

Balin64
Jan 11, 2004, 03:48 AM
This "article" is so off-keel I am wondering if it's a fake/hoax. If this is not coming from a myopic Windows fan the what is?

It is a university: of course students will help.

Software did not need to be re-written by faculty: the inter-connectivity company ported it to Mac OS X.

Perhaps most business will stay with the PC vendors the article mentions. Is that a surprise? Like their employees, most companies are parsimonious, cheap-assed pennie pinchers.

Bottom line: This writer is pissed off Virginia Tech did what they did: with Macs.

I am very, very close to taking CNET off my News Toolbar. What is this ******?

Xero
Jan 11, 2004, 03:49 AM
and what was the point of that article?

...another meaningless Cnet article bashing apple, hurray! :rolleyes:

backspinner
Jan 11, 2004, 05:22 AM
Right. If you need a supercomputer, you "normally" just ask some supplier to do the plumbing and you're set. Software comes out of the box, no need for your own work to get things useful :rolleyes:

How many supercomputer setups would the anonymous author have seen himself?

goku
Jan 11, 2004, 05:54 AM
I see the guys point.

1 First generation consumer hardware from Apple proves faster than generation x hardware from Sun, Intel, HP including custom hardware.

2. Mac OS X 10.2.8 which is Second revision to an OS released in 2000 allows it to be compative for Super Computer space with the likes of Irix, Sun OS, Linux, Cray OS, AIX and others *nix which are at least 7 years OS X senior.

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

3. Who says? VT looked at this and found that Mac worked for them others may do the same. The US Navy chose Apple (for a project http://www.cbronline.com/todaysnews/fbfb09645989096d80256d7c0018cf6d
(though running Linux) NASA uses Macs at JPL. Also for Large Scale Super Computers like this the big market is Education, Defense and National labs.

Most companies would not need the scale of computing power that VT was after (outside Oil and Mining that need it to compute geological data). The reasons for clusters may vary some are for HA like Windows Cluster in which you can only have two nodes. Some to solve varying degrees of complex problems like a brokerage firm doing finical forecast to medial doing DNA sequencing, others to just share load like web server cluster or render farm not all clusters are Super Computers or need to be.

Bottom Line is that the Apple gives everyone one more Hardware/Software platform to choice from at a good price and with the Xserver G5 the real hidden cost of rack space, power and cooling are addressed.

However I take it choice is bad thing if it includes Apple, funny few years ago they said the same thing about Linux in the iT rags.

AppleMatt
Jan 11, 2004, 06:06 AM
I think if it saves them $200m the chances are pretty high they will pay people to install it for them.

Do you need any business or economic qualifications to work at cnet?

AppleMatt

Sabenth
Jan 11, 2004, 06:44 AM
I dont see the point of this artical at all in fact its a bit embarassing to even read really. 2 -300 million dollars vs say 5 -8 million dollars hmm. i dont know this cnet thta well but so far i aint impressed.

yamabushi
Jan 11, 2004, 06:47 AM
CNet has had a lot of biased articles but this one takes the cake.

FlashXL
Jan 11, 2004, 07:07 AM
This must be the guy who got conned into paying the pizza bill for the students.

Jerry Spoon
Jan 11, 2004, 07:50 AM
I know cNet writes some biased articles against apple, but I don't know if I've ever read one that was so lacking in facts, research, and common sense.
Disappointing that people will read this and believe it.

edesignuk
Jan 11, 2004, 07:51 AM
Wow, I just wasted a minute of my life reading this crap :rolleyes:

duce
Jan 11, 2004, 07:54 AM
Had Virginia Tech chosen another vendor they would still have used these same resources to assemble their system. Sour grapes at CNet once more. I'm sure this guy will conclude as did MS that the iPod/HP deal will limit the user experience.

billyboy
Jan 11, 2004, 07:56 AM
Whether you know anything about supercomputers or not, that article was a piece of junk - the content had nothing to do with the conclusion. It reads like "There is this company with a record as long as your arm for hi tech innovation which works. Now it offers a groundbreaking and economically viable solution to anyone in the market for a supercomputer, but hey, it is a crap option because that company is Apple which is not one of the current industry big boys I prefer to talk up in my paper.

mainstreetmark
Jan 11, 2004, 10:42 AM
Originally posted by billyboy
Whether you know anything about supercomputers or not, that article was a piece of junk - the content had nothing to do with the conclusion. It reads like "There is this company with a record as long as your arm for hi tech innovation which works. Now it offers a groundbreaking and economically viable solution to anyone in the market for a supercomputer, but hey, it is a crap option because that company is Apple which is not one of the current industry big boys I prefer to talk up in my paper.

....and unlike other super computers, this one didn't walk in the front door and plug itself into an outlet and a network. Someone actually had to physically set it up!

nagromme
Jan 11, 2004, 11:08 AM
So with today's components and prices, and volunteer labor (which lots if institutions would also have)... could other HW vendors provide the power of Big Mac for the same low price? VA Tech tried desperately before learning of the G5 in the nick of time. The answer is 'no' and the article is absurd. The "valid points" they make alter the price comparisons only slightly--the G5's massive cost advantage is indisputable. And they don't scrutinize the costs of the other projects, only Apple's--great writing.

Not to mention... these Macs were so easy to set up that the volunteer labor did NOT need to be skilled. It would have been cheap even if it were paid.

There was something unexpected and experimental about Big Mac. That's something Apple and VA Tech are both very open about--and it equals risk that most companies wouldn't choose... yet. But they'll lose a lot of money--or speed--by not taking VA Tech's Mac leap.

And does the article mention the things that are now differtent, making things even BETTER for someone bulding a Mac supercluster?

1. XServe G5 and Cluster Node--a BIG advantage over VT. ECC RAM and high density/redundancy. (And a new RAID.)

2. XGrid -- looking promising already

3. VT put together a "how to" kit to tell others what to do.

4. VT's Deja Vu fault tolerance software has been ported to OS X and is now available.

5. The G5 (and Panther) has been out longer now and is proven: less early-adopter fear.

6. Apple and IBM are working on optimized compilers and dev tools that get even more speed from the G5.

Hmmm.... I wonder why the article didn't compare Big Mac to the Dell supercluster project in Texas? The one with TODAY'S generation of hardware--and probably student volunteers--that falls flat in price/performance compared to the G5 cluster.

Don't give CNet hits. Don't support their advertisers. We know Mac and we know how poorly they report on Mac topics. What about the topics we don't even know how poorly they are handling? I'll get my news elsewhere.

schmooze
Jan 11, 2004, 11:32 AM
Apparently they welcome comments at the "Get up to Speed" Department of CNET from which this article spawned.

Have some feedback? Tell us what you think of the GUTS sections by sending e-mail togetuptospeed@cnet.com.


I say this article demands a little "feedback"... don't you?

Mr.Hey
Jan 11, 2004, 12:04 PM
I think even with those "hidden costs" exposed!!!111 (;) ) it would still cost far FAR less then what you would have to normally pay for an x86 SC system. He(?) should have discussed VT SC lack of ECC (but it should be stated that its been added to the Xserver); and from what I read its far FAR (:D) more important for super computing than these "hidden costs".

John Baughman
Jan 11, 2004, 12:45 PM
I say this article demands a little "feedback"... don't you?
The togetuptospeed address bounced back for me. Try...

Send-letters-to-news@cnet.com

I would be curious to read what others send. Here is my lettter to the editor....

Whoever wrote "Commentary: A grain of salt with your Apple" was obviously too embarrassed to reveal his/her identity.

Does the article suggest that the difference between the cost of Virginia Tech's $5 million Mac Cluster and the cheapest site in the top 10 list at $40 million is $35 million paid to "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"? Man I'm in the wrong business.

And... Isn't part of the point that it doesn't take those "specialists" to put the Mac cluster together?

Did the editor even read this commentary before publishing it?

shadowband
Jan 11, 2004, 01:34 PM
The author is grasping for something negative to say about a complete success, and this clearly comes out in the article. Think about it - this was a very risky project with an incredibly tight schedule, and vtech pulled it off. Hardware installation in only three weeks and hitting the number three spot when the goal was just to make it onto the top ten list - it's hard to pour cold water on a success story like this.

FosterKanig
Jan 11, 2004, 01:56 PM
I think sometimes that Mac users are a little too paranoid about "people" being against them, but this article was so over the top in its venom towards what VT achieved it is scary. Is the 5.2 a true number of what it cost? I think that they aren't including a lot of the room setup. But do others do that? Or just include the cost of the computer? Even if you were to make the G5/VT setup figure 9-10 million (including pizza) it it still an incredible accomplishment. Cnet may be pulled from my bookmarks before too long.

gerardrj
Jan 11, 2004, 04:01 PM
I think I've read that the $5.2 included the systems, networks, software, racks and the backup generators, but I could be mistaken.

In any case, the cost claimed for the VA project was $5.2mil. The next fastest system claims to have cost $200mil+.
So the VA team had "...hundreds of volunteered hours of Virginia Tech faculty, staff and students...". Lets say that "hundreds" means 900. Lets say that you were to hire people to do that assembly instead of using volunteers. It's rather unskilled work, you could show someone how to do it in half an hour. So lets say you'd pay $30 per hour to such workers, an insanely hight figure.
900 hours * $30/hour = $27,000. So fine, the VA super computer project now costs $5.227 mil. Whoo hoo! ALERT THE MEDIA, the VA super computer is not not affordable any more!
You would need to hire about 550 people to each work 12 hours on the project to spend about $200,000 on labor. That's a LOT of labor, and still an insignificant cost increase for the overal price tag.

mrsebastian
Jan 11, 2004, 09:34 PM
would cnet and the author of this completely useless article, please feel free to go f@#k yourselves? even if you paid a team one million dollars to set this thing up, you'd still have a cheap supercomputer.

Sir_Giggles
Jan 12, 2004, 09:05 AM
It's so sad that CNet has to resort to these types of uneducated, biased articles to maintain some sort of PC userbase.

Still, I'm sure some PC users will be taken in by the article.

In doing so, it has lost all credibility for people using Macs or the few knowledgeable PC users out there.

So sad, adios CNet. :(