Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
maverick13 said:
owned who? With what links? Anybode with some basic knowledge can come here and claim what they will with no references.
How convenient...

That's exactly what Tortoise is doing. No credentials, no references, but we're supposed to believe this anonymous loudmouth over Virginia Tech (who spent $5,000,000 on their system), Computerworld (who gave it an award for top scientific computing solution) and other magazines who also gave it an award.....

Riiiiiiggghhht..
 
jragosta said:
That's exactly what Tortoise is doing. No credentials, no references, but we're supposed to believe this anonymous loudmouth over Virginia Tech (who spent $5,000,000 on their system), Computerworld (who gave it an award for top scientific computing solution) and other magazines who also gave it an award.....

Riiiiiiggghhht..

Virginia Tech isn't exactly a top tier school for engineering... and look, they forgot ECC... HAHAHA... man, that still cracks me up... gg newbs!!
 
Another supercomputer G5 cluster

Another supercomputer PowerPC 970FX cluster that is being built:
IBM PowerPC 970FX and BladeCenter servers to power Spanish supercluster
Recently, IBM and the new Spanish Center for Supercomputing signed a deal to build a large supercluster. This cluster will consist of 2282 JS20 blades. Each blade will have two next-generation PowerPC 970FX processors. The total cluster will have a theoretical peak capacity of over 40 trillion floating-point operations per second.
More

http://www-306.ibm.com/chips/products/powerpc/newsletter/jun2004/index.html

40 Tera flops and this with G5s ,not custom 440s ,sweet

PS: The super computer cluster will run Linux 2.6
 
yuphorix said:
Virginia Tech isn't exactly a top tier school for engineering... and look, they forgot ECC... HAHAHA... man, that still cracks me up... gg newbs!!

Ummm.... no they didn't forget it. The Professor and his colleges hav worked far harder to gain what they know and he/they likely know far more then you, so show some respect.

They very likely knew that the Xserve was coming out in the not so distant future (Apple worked with them on this thing you know) but they had a deadline to meet in order to get the project funded. So they built one with PM G5s in short term to prove what can be done and to get them on the list so they could get the funding locked in.

Now they are working on building the real work horse cluster, one with ECC, lower power and cooling needs overall and far smaller space needed. They will also likely add nodes to the cluster.

Regardless the cluster can work without ECC but additional verification has to be done which reduces the overall throughput of the system.
 
Money Management

Stella said:
Still no G5 PB soon..
:-(

What do I do.. buy G4 PB or hang on to G3 iBook until G5 PB.. getting a G4 iBook seems not much of an upgrade, since its still the same machine.

It depends upon your money management abilities.
- A G4 IBook is probably faster then your current machine, if you don't need superdrive, and it's less expensive.
- If you can afford a 1.5 Powerbook that too would hold you over.
( Remember EBay is your friend. We need to populate the used computer market to help more people switch into an Apple. )

I couldn't take the wait any longer, I've got the 1.5 17" PB.
I'm now waiting for the G6 PB. ;)
 
shawnce said:
Ummm.... no they didn't forget it. The Professor and his colleges hav worked far harder to gain what they know and he/they likely know far more then you, so show some respect.

They very likely knew that the Xserve was coming out in the not so distant future (Apple worked with them on this thing you know) but they had a deadline to meet in order to get the project funded. So they built one with PM G5s in short term to prove what can be done and to get them on the list so they could get the funding locked in.

Now they are working on building the real work horse cluster, one with ECC, lower power and cooling needs overall and far smaller space needed. They will also likely add nodes to the cluster.

Regardless the cluster can work without ECC but additional verification has to be done which reduces the overall throughput of the system.

No... you don't understand... without ECC, computating and processing large amounts of data will cause frequent errors. I'd also like to know what level of RAID they setup, if any, to prevent even more system errors. Of course the cluster can work, but like the article said, it's unstable...

AND, if they know the xserve was coming out, why not wait? Their system hasn't been up for much time, atleast not enough to do extensive research (no articles have been published about anything they've done)... why not wait a couple of months? Obviously they don't see $5million as a large amount of money, why not consider this a publicity act? There are reasons why certain schools rank high and certain don't... Whens the last time Virginia Tech was in the top10? When's the last time they contributed to the research of computer engineering and computer science? Personally, I'm only interested in instituations that can give something to the community, other than $5mil to buy some boxes to show off to real research instituations.
 
After reading this thread ( well, 70% of it ) I have come to realize that about 50% of the posters to these forums are too stupid to own any sort of computer.

Either that, or they're all 15 years old. About half of you seriously need to get girlfriends and lives. Go outside and play. Walk. Go Do Something Else.

:rolleyes:
 
AMD Flamers with a Bug up their A##

maverick13 said:
http://www.apple.com/education/science/profiles/vatech/architecture.html

Lockhart says several features of the new Power Mac G5 computers were especially attractive for Virginia Tech. “AMD’s Opteron can only execute two double-precision floating-point instructions per clock cycle, which kind of limits system performance,” he notes. The G5 processor has two floating-point units. Thus, unlike the Opteron, it can perform fused multiply-adds, giving four operations per clock cycle and achieving a theoretical limit of 8 gigaflops from a 2 GHz processor. All this is in addition to the G5 processor’s Velocity Engine floating point and integer units.

Adds Varadarajan, “When we were evaluating machines and platforms, I realized that the PowerPC 970 processor would be ideal for us. Its fused multiply-add operation gives it a floating-point performance equal to — if not better than — Intel’s Itanium2 solution. Pretty
quickly, I knew that the Power Mac G5 machines would help us reach the goals we had in mind for our supercomputer.”

Also an interesting link, http://forgetcomputers.com/~jdroz/pages/09.html


• The first module of IBM's Blue Gene/L supercomputer placed 73rd with a sustained performance of 1.4 teraflops. When the full Blue Gene/L is completed in 2005, a system with 128 times as many processors as the current prototype, IBM expects performance of 360 teraflops, enough to top the list by a wide margin.
http://news.com.com/New+blood+joins+supercomputer+ranking/2100-1006_3-5107944.html?part=msnbc-cnet
http://www.research.ibm.com/bluegene/
That's PowerPC 440 -based.

Good Point.
I've got to laugh, these AMD Flamers must think Apple is stealing market share from them. HaHaHa....

The real problems AMD has to deal with is...
1- Intel Marketing money
2- Intel strong arming vendors to not sell quality AMD systems, low end comsumer systems with cheap motherboards and graphic cards( HP ) only hurts.
3- AMD Needs a decent OS Vendor: Microsoft has really screwed them by continuing to drag it's feet on that Windows 64 bit edition, which will only be shipped when INTEL is ready to ship 64 bit systems.
 
MikeBike said:
Good Point.
I've got to laugh, these AMD Flamers must think Apple is stealing market share from them. HaHaHa....

The real problems AMD has to deal with is...
1- Intel Marketing money
2- Intel strong arming vendors to not sell quality AMD systems, low end comsumer systems with cheap motherboards and graphic cards( HP ) only hurts.
3- AMD Needs a decent OS Vendor: Microsoft has really screwed them by continuing to drag it's feet on that Windows 64 bit edition, which will only be shipped when INTEL is ready to ship 64 bit systems.

AMD does have the perfect OS... it's called Linux... AMD's are used frequently as server chips, and Linux is the perfect OS to run a server with...
 
Amd should be Loving SUN

MikeBike said:
Good Point.
I've got to laugh, these AMD Flamers must think Apple is stealing market share from them. HaHaHa....

The real problems AMD has to deal with is...
1- Intel Marketing money
2- Intel strong arming vendors to not sell quality AMD systems, low end comsumer systems with cheap motherboards and graphic cards( HP ) only hurts.
3- AMD Needs a decent OS Vendor: Microsoft has really screwed them by continuing to drag it's feet on that Windows 64 bit edition, which will only be shipped when INTEL is ready to ship 64 bit systems.

Sun is one company that's selling a decent OS on Opterons.
Their move pushed IBM and Dell into the Opteron action.
 
jragosta said:
So?

Apple made an announcement based on information available to them at the time. In 2003, they believed they would meet the target.

Things turned out to be more difficult than they thought and they're not at 3.0. No one has a crystal ball. They assume that when they make an announcement (particularly in the computer field) that people will be bright enough to realize that it's their best estimate at the time and could change.

During the 130->90 transition, IBM increased the G5 clock speed by 25%. Intel increased the P4 clock speed by 6%.

It turned out to be a more difficult transition than ANYONE expected.

Get over it.

There was never anything for me to get over, thus my use of the term, plain and simple.

Why are people so damn quick to flame others?
 
yuphorix said:
AMD does have the perfect OS... it's called Linux... AMD's are used frequently as server chips, and Linux is the perfect OS to run a server with...

Some would argue FreeBSD is the perfect server OS. :rolleyes:
Linux is perfect overall UNIX-clone though,being using it since '95 ;)
 
yuphorix said:
Virginia Tech isn't exactly a top tier school for engineering... and look, they forgot ECC... HAHAHA... man, that still cracks me up... gg newbs!!

Showing simply that you haven't followed what when on.

They got the original G5, built the system and came in #3 on the all-time list.

Then, apparently due to some arrangment with Apple, traded them in on xServe systems which have ECC.

That allowed them to test things, get the system up and running, and come in #3 on the list - and even after considering the time to swap out G5s for xServes, it was still done far faster than any comparable system would have been.
 
maverick13 said:
Another supercomputer PowerPC 970FX cluster that is being built:
IBM PowerPC 970FX and BladeCenter servers to power Spanish supercluster
Recently, IBM and the new Spanish Center for Supercomputing signed a deal to build a large supercluster. This cluster will consist of 2282 JS20 blades. Each blade will have two next-generation PowerPC 970FX processors. The total cluster will have a theoretical peak capacity of over 40 trillion floating-point operations per second.
More

http://www-306.ibm.com/chips/products/powerpc/newsletter/jun2004/index.html

40 Tera flops and this with G5s ,not custom 440s ,sweet

PS: The super computer cluster will run Linux 2.6

Oh, no.

Tortoise will insist that they don't know what they're doing, either.
 
yuphorix said:
No... you don't understand... without ECC, computating and processing large amounts of data will cause frequent errors. I'd also like to know what level of RAID they setup, if any, to prevent even more system errors. Of course the cluster can work, but like the article said, it's unstable...

Just WTF are you talking about?

How the heck is a RAID system supposed to prevent system memory errors (which is what the ECC would do)? The two concepts are entirely unrelated.

And as far as the old cluster working, they simply had to insert some SOFTWARE error checking since the hardware wasn't there. And it was STILL #3 in the world.
 
Does PCI-X matter? Please advise.

I'm pondering the new G5s and weighing the 1.8 and 2.0 boxes. I'm frankly leaning toward saving $500 on the 1.8 Ghz machine. That way I see it, I will be upgrading RAM and the hard drive myself later for far less money. The only key difference, other than a modest FSB and processing speed bump, is the lack of PCI-X on the low end unit. With DIMM densities increasing all the time, I'm not too worried about having the extra slots on the larger board.

You should know that I am still milking a six-year-old Beige G3 266 Mhz with bigger hard drive, PCI video card, Firewire/USB card and lots more RAM. When I bought the machine, I was running Lightwave, Adobe Aftereffects and other demanding apps, and that Mac was the state of the art. But my career has changed, and more recently I've just been running Quicken, web surfing and some light-weight web design.

The sad news is that I just bought my wife a Dell 4600 for her home-based business. Total was $1,000 including 17" LCD monitor. I hate Windows XP and the thing is a pain in the butt, but compatibility woes forced me to join the dark side.

On the other hand, my daughter loves her iBook and is now on her second iPod. Now it's time for me to buy myself a new Mac. Because my demands have waned over the years, my chief concern is the longevity of my investment that Apple represents. That Dell will be useless in three years, based on the Windows machines that I'm forced to use at work, while my six-year-old Mac is just now becoming frustrating to use.

My instinct is to pay more for the 2.0 G5, because in four years I will be shopping for upgrades that take advantage of the 133 Mhz PCI-X slot. Then again, I suspect that Rev. C of the G5s will jump to PCI Express. While I am certain the 33 Mhz PCI slots will be downright ancient in no time, I wonder if PCI-X is destined for history's dustbin, with no vendors supporting the thing in just a few years.

Any strong opinions or fortune tellers out there? Thanks.

P.S. -- Sorry about the cross post, but I got no feedback on the other thread.
 
rdowns said:
There was never anything for me to get over, thus my use of the term, plain and simple.

Why are people so damn quick to flame others?

I don't know. Why are you attacking Jobs on the basis of an honest miscalculation?
 
metamotion said:
I'm pondering the new G5s and weighing the 1.8 and 2.0 boxes. I'm frankly leaning toward saving $500 on the 1.8 Ghz machine. That way I see it, I will be upgrading RAM and the hard drive myself later for far less money. The only key difference, other than a modest FSB and processing speed bump, is the lack of PCI-X on the low end unit. With DIMM densities increasing all the time, I'm not too worried about having the extra slots on the larger board.

You should know that I am still milking a six-year-old Beige G3 266 Mhz with bigger hard drive, PCI video card, Firewire/USB card and lots more RAM. When I bought the machine, I was running Lightwave, Adobe Aftereffects and other demanding apps, and that Mac was the state of the art. But my career has changed, and more recently I've just been running Quicken, web surfing and some light-weight web design.

The sad news is that I just bought my wife a Dell 4600 for her home-based business. Total was $1,000 including 17" LCD monitor. I hate Windows XP and the thing is a pain in the butt, but compatibility woes forced me to join the dark side.

On the other hand, my daughter loves her iBook and is now on her second iPod. Now it's time for me to buy myself a new Mac. Because my demands have waned over the years, my chief concern is the longevity of my investment that Apple represents. That Dell will be useless in three years, based on the Windows machines that I'm forced to use at work, while my six-year-old Mac is just now becoming frustrating to use.

My instinct is to pay more for the 2.0 G5, because in four years I will be shopping for upgrades that take advantage of the 133 Mhz PCI-X slot. Then again, I suspect that Rev. C of the G5s will jump to PCI Express. While I am certain the 33 Mhz PCI slots will be downright ancient in no time, I wonder if PCI-X is destined for history's dustbin, with no vendors supporting the thing in just a few years.

Any strong opinions or fortune tellers out there? Thanks.

P.S. -- Sorry about the cross post, but I got no feedback on the other thread.

There are alot of PCI-X products out there, but they tend to be really high end or server products, like high end SCSI RAID controllers, HiDef Video editing suites, Fibre Channel controllers (apart from the the Apple one), gigabit Ethernet controllers. You don't see many consumer level PCI-X devices (thats why some of the more ignorant people on this forum claim that PCI-X is a crap standard). PCI-X is not a relatively new standard, its been around for years co-existing with normal PCI slots, but on the PC front, only servers/workstations tend to have PCI-X slots. Consumer level devices will probably skip PCI-X and headstraight to PCI Express.
 
please check your "facts"

MikeBike said:
Sun is one company that's selling a decent OS on Opterons.
Their move pushed IBM and Dell into the Opteron action.

http://www.sun.com/servers/entry/v20z/

The Sun Fire V20z server also offers exceptional flexibility and choice through support of multiple operating systems, including 32-bit Solaris 9 4/04 x86 Platform Edition, 32/64-bit Red Hat Enterprise Linux 3, and 64-bit SuSE Linux Enterprise Server 8. The Solaris Operating System (x86 Platform Edition) in 64-bit will be available soon.


Gee, Microsoft has 32-bit operating systems for Opteron, with 64-bit soon (public preview is freely downloadable today). Sounds the same to me.

And, by the way, Dell is not selling any systems with Opteron chips. IBM has had them for quite some time, as well as HP.

You are just inventing "facts" here, you are quite off the mark.
 
tortoise said:
...Media marketing copy does not constitute a refutation of some pretty basic and well established engineering fact....

...What kind of twit waves content-free marketing copy around as "proof" that all the scientists and engineers are wrong? You don't even have a clue what typical supercomputing loads look like or how they interact with system architectures.

Nobody who works on large clusters thinks the PPCs are good for anything but mult-add codes, and that is a tiny segment of the market. Please show me all the supercomputing cluster folks that are asserting that the G5 is good for STREAM-bound codes, or even non-mult-add floating point codes. Any references? Links?

Go back to playing with Photoshop and leave the supercomputing stuff to the guys that actually do it. You are not technically competent to be making such assertions, and that you use content-free marketing copy to support your position is embarrasing.


I don't know where to start with this bozo

"What kind of twit waves content-free marketing copy around as "proof" that all the scientists and engineers are wrong? You don't even have a clue what typical supercomputing loads look like or how they interact with system architectures."

Since 1988, I have continually heard this kind of crap from "experts," first, to justify staying with DOS, then to justify no user-friendly UNIX interface, and most recently to refute the idea that powerful computing can be accessible to people who can't write C++ (e.g., OS-X).

If we bought this kind of cretin's omniscience, we would stilil be working in raw code on *8088s."

I once had the occasion to meet Grace Hopper (you hard-core coding geeks spin your beenie propellers and you might remember she INVENTED COBOL, the first step toward a real-language programming language). She met huge resistence; in fact, she might have been stymied completely had the Navy not needed to automate.

"Go back to playing with Photoshop and leave the supercomputing stuff to the guys that actually do it. You are not technically competent to be making such assertions, and that you use content-free marketing copy to support your position is embarrasing."

Contrary to your inflated ego, computers are not about programmers; they are about enabling people to DO something. I am a retired technical writer. Yes, I have used Photoshop, Canvas, Corel, FrameMaker, and the whole spectrum of design, publishing, and web apps. What was I doing, documenting the work that elitist, inarticulate, self-absorbed boobs like this produced--because they were unable to explain their products to the people who BUY them!

Be doggone glad that somebody is out here "playing" with Photoshop, et al. Without them you'd have to depend on your personality for a living!
 
VA Tech cluster has been offline for months

jragosta said:
That allowed them to test things, get the system up and running, and come in #3 on the list - and even after considering the time to swap out G5s for xServes, it was still done far faster than any comparable system would have been.


Your timeline is way off

http://www.tcf.vt.edu/

Current System Status: Offline for XServe G5 Upgrade


Note this article from February: http://maccentral.macworld.com/news/2004/02/11/macmall/

MacMall sells pieces of Virginia Tech G5 supercomputer
By Peter Cohen pcohen@maccentral.com
February 11, 2004 6:40 pm ET

Virginia Tech's Power Mac G5-based supercomputer has already passed into legend among Mac fans, and now you can own a piece thanks to Mac catalog reseller MacMall. They're advertising refurbished Power Mac G5s that until recently powered Virginia Tech's Terascale Computing Facility -- the supercomputing cluster known as System X, sometimes affectionately referred to as Big Mac instead.

"Own a piece of history," announces an ad that appeared on MacMall's homepage recently, accompanied by Virginia Tech's stylized logo. "Buy a Power Mac G5 2GHz [Dual Processor] refurbished by Apple from Virginia Tech's Supercomputer!"

so, the PowerMac systems were removed, reboxed, shipped back to Apple, refurbished, and back on the market by mid February. It's now mid-June, and the cluster is still offline!

Your "record time" is mainly a "record amount of downtime" for a large cluster.


According to a story in ThinkSecret (http://www.thinksecret.com/news/virginiatech3.html)

"As a result of the extended downtime, Virginia Tech is expected to temporarily drop off of Linpack's TOP500 list of the world's fastest supercomputers, where the inexpensively-made cluster sits at third place"


All-in-all, it sounds like the cluster is being run by the School of Computer System Mis-Management, not by people who understand how to build and run a super cluster!


In fact, the VATech's website is pretty empty - none of the old pages of photos and plans are still around, although there's a page with what should be links to them....
 
AidenShaw said:
Your timeline is way off

http://www.tcf.vt.edu/

Current System Status: Offline for XServe G5 Upgrade


Note this article from February: http://maccentral.macworld.com/news/2004/02/11/macmall/



so, the PowerMac systems were removed, reboxed, shipped back to Apple, refurbished, and back on the market by mid February. It's now mid-June, and the cluster is still offline!

Your "record time" is mainly a "record amount of downtime" for a large cluster.


According to a story in ThinkSecret (http://www.thinksecret.com/news/virginiatech3.html)




All-in-all, it sounds like the cluster is being run by the School of Computer System Mis-Management, not by people who understand how to build and run a super cluster!


In fact, the VATech's website is pretty empty - none of the old pages of photos and plans are still around, although there's a page with what should be links to them....

You seem really intent on proving how clueless you are on these matters.

Why not look up how long it took to get the rest of the top 10 working before shooting off your mouth?

Even with the packaging, sending the original computers back to Apple, setting up new ones, reinstalling everything, it will STILL be less than half hte time it took to get systems of comparable performance running.
 
Thunder, 20 TFLOP, 5 months - VA Tech, 10 TFLOP, 11 months and still not running

jragosta said:
You seem really intent on proving how clueless you are on these matters.

Why not look up how long it took to get the rest of the top 10 working before shooting off your mouth?

Even with the packaging, sending the original computers back to Apple, setting up new ones, reinstalling everything, it will STILL be less than half hte time it took to get systems of comparable performance running.

Look at the headline story on Dr. Varadarajan's company's website: http://www.californiadigital.com/thunder.shtml

Despite the technical sophistication of Thunder and the incorporation of new technologies, California Digital deployed Thunder in five months, speeding delivery of computing solutions to support Lawrence Livermore's national security and science programs in fields such as materials science, structural mechanics, electromagnetics, atmospheric science, seismology, biology, and inertial confinement fusion.

"Thunder represents the next generation of Linux cluster for scientific simulation," remarked Mark Seager, Livermore's Assistant Department Head for Advanced Technology. "Our applications are seeing a 50% to 400% speed up over our Xeon base clusters."

So, VA Tech started last July, got some numbers to publish for a conference (see "publicity stunt" in preceding posts), couldn't get the system stable enough for production use, tore it apart and sold it, waited months for replacement Xserver systems, still offline and working on the upgrade today....

And you have the cojones to say that *I* am clueless!
 
AidenShaw said:
Look at the headline story on Dr. Varadarajan's company's website: http://www.californiadigital.com/thunder.shtml



So, VA Tech started last July, got some numbers to publish for a conference (see "publicity stunt" in preceding posts), couldn't get the system stable enough for production use, tore it apart and sold it, waited months for replacement Xserver systems, still offline and working on the upgrade today....

And you have the cojones to say that *I* am clueless!

Except for one thing. My statement was that this was record time for assembling a supercomputer - and you disagreed.

You were wrong - as usual. In less than 2 years, VA Tech has built TWO supercomputers from stock G5s and xServes. Check how long it took to build the rest of the top 10.

Hint - it took roughly twice as long to build ONE computer for each of the rest of the top ten.

Your assertion that VA Tech was slow is just plain wrong. As usual.
 
AidenShaw said:
Look at the headline story on Dr. Varadarajan's company's website: http://www.californiadigital.com/thunder.shtml



So, VA Tech started last July, got some numbers to publish for a conference (see "publicity stunt" in preceding posts), couldn't get the system stable enough for production use, tore it apart and sold it, waited months for replacement Xserver systems, still offline and working on the upgrade today....

And you have the cojones to say that *I* am clueless!

Oh, as for the Thunder system? That's 5 months _to deploy_ it. Add in the time to order it, obtain the components, build it and THEN start deployment and you're talking many times the amount of time they claimed.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.