Slashdot Mirror

If your app requires lots of vector crunching, the SV1 is one hellofa machine that'll keep you more than happy. The specs (mentioned above) are staggering... up to 1 TB of RAM, up to 1229 CPUs, air and/or water cooled.

However, it's not alone. There are some other pretty mighty machines out there. The NEC SX-5 has faster RAM and more powerful vector CPUs than the SV1, but does not scale as large. The SGI Origin 3000 series is not vector, but rather a of (somewhat) traditional CPU design. It's available with up to 512 CPUs and 1 TB of RAM. Unlike both the SV1 and SX-5, the Origin can be ordered with graphics (which turns it into an Onyx).

Then, there's the upcoming Cray SV2, which will be a combination of massive parallel & vector processing. Up to several thousand CPUs and a staggering RAM thruput of 250 GB/sec per bank!! (The Origin 3000 mentioned above has a total system bandwidth of 716 GB/sec.... but that's the entire machine. The SV2 will have more than that with just three banks of RAM alone).

Some of these machines are single image systems (in the case of the Origin 3000, SX-5 and >33 CPU SV1)... meaning they are one single machine, not a cluster. Most run very specific OSes made just for their hardware, with the possible exception of the Origin. SGI's big Origin and Onyx 3000 machines run IRIX 6.5, the same OS that runs on a $150 e-bay special SGI Indy workstation. Kinda cool. The compilers and math libraries are also heavily tuned and generally come with lots of example code and performance tips. When my university purchased a 96 CPU Origin 2000 a few years ago, SGI included a *box* of binders and CDs from some past performance computing seminars they had held. Our university still holds a support contract for the Origin, and thus we're still getting significant compiler and library updates.

Sort of belittles dual bank PC2600 DDR-SDRAM (2x 2.6 Gigabyte/sec = 5.2 Gigabyte/sec) and Myrinet (1 Gigabit/sec = 125 Megabyte/sec interconnect), doesn't it.

Of course... a 16 node x86 cluster doesn't cost $500K - $50M either...

If your app requires lots of vector crunching, the SV1 is one hellofa machine that'll keep you more than happy. The specs (mentioned above) are staggering... up to 1 TB of RAM, up to 1229 CPUs, air and/or water cooled.

However, it's not alone. There are some other pretty mighty machines out there. The NEC SX-5 has faster RAM and more powerful vector CPUs than the SV1, but does not scale as large. The SGI Origin 3000 series is not vector, but rather a of (somewhat) traditional CPU design. It's available with up to 512 CPUs and 1 TB of RAM. Unlike both the SV1 and SX-5, the Origin can be ordered with graphics (which turns it into an Onyx).

Then, there's the upcoming Cray SV2, which will be a combination of massive parallel & vector processing. Up to several thousand CPUs and a staggering RAM thruput of 250 GB/sec per bank!! (The Origin 3000 mentioned above has a total system bandwidth of 716 GB/sec.... but that's the entire machine. The SV2 will have more than that with just three banks of RAM alone).

Some of these machines are single image systems (in the case of the Origin 3000, SX-5 and >33 CPU SV1)... meaning they are one single machine, not a cluster. Most run very specific OSes made just for their hardware, with the possible exception of the Origin. SGI's big Origin and Onyx 3000 machines run IRIX 6.5, the same OS that runs on a $150 e-bay special SGI Indy workstation. Kinda cool. The compilers and math libraries are also heavily tuned and generally come with lots of example code and performance tips. When my university purchased a 96 CPU Origin 2000 a few years ago, SGI included a *box* of binders and CDs from some past performance computing seminars they had held. Our university still holds a support contract for the Origin, and thus we're still getting significant compiler and library updates.

Sort of belittles dual bank PC2600 DDR-SDRAM (2x 2.6 Gigabyte/sec = 5.2 Gigabyte/sec) and Myrinet (1 Gigabit/sec = 125 Megabyte/sec interconnect), doesn't it.

Of course... a 16 node x86 cluster doesn't cost $500K - $50M either...

If your app requires lots of vector crunching, the SV1 is one hellofa machine that'll keep you more than happy. The specs (mentioned above) are staggering... up to 1 TB of RAM, up to 1229 CPUs, air and/or water cooled.

However, it's not alone. There are some other pretty mighty machines out there. The NEC SX-5 has faster RAM and more powerful vector CPUs than the SV1, but does not scale as large. The SGI Origin 3000 series is not vector, but rather a of (somewhat) traditional CPU design. It's available with up to 512 CPUs and 1 TB of RAM. Unlike both the SV1 and SX-5, the Origin can be ordered with graphics (which turns it into an Onyx).

Then, there's the upcoming Cray SV2, which will be a combination of massive parallel & vector processing. Up to several thousand CPUs and a staggering RAM thruput of 250 GB/sec per bank!! (The Origin 3000 mentioned above has a total system bandwidth of 716 GB/sec.... but that's the entire machine. The SV2 will have more than that with just three banks of RAM alone).

Some of these machines are single image systems (in the case of the Origin 3000, SX-5 and >33 CPU SV1)... meaning they are one single machine, not a cluster. Most run very specific OSes made just for their hardware, with the possible exception of the Origin. SGI's big Origin and Onyx 3000 machines run IRIX 6.5, the same OS that runs on a $150 e-bay special SGI Indy workstation. Kinda cool. The compilers and math libraries are also heavily tuned and generally come with lots of example code and performance tips. When my university purchased a 96 CPU Origin 2000 a few years ago, SGI included a *box* of binders and CDs from some past performance computing seminars they had held. Our university still holds a support contract for the Origin, and thus we're still getting significant compiler and library updates.

Sort of belittles dual bank PC2600 DDR-SDRAM (2x 2.6 Gigabyte/sec = 5.2 Gigabyte/sec) and Myrinet (1 Gigabit/sec = 125 Megabyte/sec interconnect), doesn't it.

Of course... a 16 node x86 cluster doesn't cost $500K - $50M either...

Heh, I can just see a good trivia question...

Who is the "Cray woman on the upper right-hand side of most cray.com pages?"

http://www.cray.com/products/index.html

Cray, Inc. is much more alive than their former owner, SGI...

Lots of new products and they're even making a profit.

http://www.cray.com/products/systems.

Nice varitey of systems, from their own SV1/SV1ex/SV2 machines, to Linux clusters, to maspar Alphas, to NEC vector-based machines, and more.

Cray, Inc. is much more alive than their former owner, SGI...

Lots of new products and they're even making a profit.

http://www.cray.com/products/systems.

Nice varitey of systems, from their own SV1/SV1ex/SV2 machines, to Linux clusters, to maspar Alphas, to NEC vector-based machines, and more.

And yes, this (the one in the article) is the same Cray that built all the vector machines. Check out the gallery if you don't believe me.

Okay, I'm a NEC/HNSX Supercomputers employee, on the verge of becoming a Cray employee (because of the agreement they signed), but I'm not speaking for anyone else but me here, of course. :-)

I don't know why people bother with such a news. Sun's gonna provide the I/O processor for a not-so-high-end supercomputer. And?

A few weeks ago, there was a real bombshell: Cray would drop the anti-dumping legal action, re-opening the US market to japanese supercomputers. Cray will even become the sole reseller of the NEC SX Series in North America!

If you go take a look at www.cray.com, you'll see that this agreement with Sun occupies a single line in their news listing, while the NEC agreement is a big framed box that occupies about half of my screen here.

For some time now, american supercomputer customers were petitioning to get japanese machines, because it been a long time the american machines had been up to any good. Instead, we hear about the SV2, which will barely surpass the few years old SX-5 processing power, with less memory throughput than the SX-5.

I won't deal with the "no need for big clunky vector supercomputers, we have clusters". I believe a whole lot into clusters, but they're freakin' hard to program, and some things just won't be as fast (hey, the SX-5 CPU has a 256 bytes wide memory path! that's not bits, that's bytes! what can you do with your puny gigabit ethernet cluster interconnections?).

Look at these bandwidth benchmark scores. The closest thing to a cluster, the Origin machines, are literally crushed to bits by the SX-5. And they're doing twice as good than the SV1.

As for using old big iron machines for stuff like fridges and so on, there was a cool thing at one of our customer site, at the University of Stuttgart: a Cray coffee table. :-)

Nothing beats talking about supercomputer technology while drinking some orange juice on top of a Cray machine. NOTHING.

Or you could look at their employment page here: http://www.cray.com/company/employment/openings/ea gan/index.html

Read up on MTA, it's cool. Supports 128 active threads per processor.

Not just IBM, but SGI, CRAY, Compaq, VA and Penguin are all hardware vendors putting serious resources to making linux better.

Bear in mind that there have been at least four entities called "Cray". All but one would be perfectly at home building clustered micros.

1. The original Cray, pioneer of vector supercomputers. Probably the company you're thinking of.
2. The same company after it branched out into microprocessor-based high-performance computing, system integration, and consulting. This entity not only made MIPS, SPARC, and Alpha-based systems, they resold Sun and SGI workstations.
3. A loosely-defined entity never completely assimilated by SGI. The only parts of Cray SGI really wanted was Craylink (as much to keep it away from Sun as for themselves) and maybe the compiler software. The rest they more or less ran into the ground.
4. Tera Computer, which bought the Cray name from SGI, together with the Alpha-based and vector supercomputer lines. SGI had already sold the SPARC-based line to Sun and kept the MIPS-based line for itself.

__________________

No, this is actually from Tera/Cray. Check out their website.

Where exactly do you read Cray will build Beowulf clusters?

"Starting mid-2001, Cray will offer world's first production-orientated clusters, based on Alpha Linux and scalable to 1000s of processors."

we start to see Redhat Ready Crays? Shouldn't be too hard, since the non-vector Cray T3E uses modified Alphas.

The latest biggest Cray machines have all been Alpha clusters probably running some clusterable DU variant.

I think you mean the T3E, and while it's based on Alphas, it's about as far from a cluster as you can get and still have a parallel machine. It also runs UNICOS/mk (a microkernel version of UNICOS), not Digital/Tru64/name-of-the-week Unix.

--Troy

The Cray SV1ex multi-streaming vector processors are 50 percent faster at 7.2 billion calculations per second (gigaflops) each, sustained memory bandwidth effectively doubles to 40 gigabytes per second, cache latency improves by 50 percent, and maximum memory size jumps four-fold to 128 gigabytes--all without affecting industry-leading Cray SV1 reliability (MTTI) averaging more than one year between interrupts.

Cray SV1ex systems will be available with:
* 8 to 32 processors, 32 gigabytes (GB) of main memory, and 32 to 96 GB of SSD memory. Groups of four 1.8-gigaflop processors can be run as a 7.2-gigaflop multi-streaming processor.
* A powerful suite of clustering tools that allow multiple Cray SV1ex nodes to be combined to form terascale superclusters[tm] of up to 1,024 processors (1.8 teraflops).
* The company's fifth-generation CMOS architecture (0.12 micron copper), SDRAM DIMM and FPGA (field-programmable gate array) technologies.

I assure you, Vector computing in the US is not dead.

Maybe you've never heard of a company called SGI. They're down right now, but not necessarily out. Their new Orign 3000 server line uses a cc:NUMA architecure. They hope to run Linux on these someday, but Intel has been jerking them around with Itanium delays.

I suspect that Cray is also using cc:NUMA technology (which they got from SGI) in their next computer, the SV2.

Cray does a good job of a psuedo-distributed operating environment. Depends on if you count a processor+memory node as independent.

Check it out.

As an aside for those who don't care to read whitepapers and didn't already know this:
cc = Cache Coherent. The memory interface logic keeps track of what cache has what data cached so the data doesn't get corrupted.
NUMA = Non-Uniform Memory Access. There's no one point of access to core memory. Each node or board (or processor, let's just call them system blocks) has its own memory inteface, and if other system blocks need some data from another system block's memory, it must ask for it. This removes bottlenecks to relatively slow (insert-favorite-flavor-of-)DRAM. This allows for a greater theoretical system-bandwidth peak through parallelization of slow storage pipes. Writing 16 bytes, 1 byte at a time will always be 16 times slower than writing 16 bytes to 16 banks of DRAM all at once.

UI should be lightweight, hell who wants a long wait for a command you entered to be acknowledged.

funny animated effects hiding your menu items, one
button mice, transparency , stupid noises and
"shit head" the paper click. possibly the worst UI you could create(short of the S.I.R.D UI that Cray were working on)

point and click, simple boxes with a bit of text and maybe a few modeless popup fine,

just hot keys even better,

command line, do exactly what you tell it to when you tell it, hell now were talking UI.

hell ain't we going backwards here,

right the ultimate UI,

two pieces of bread, some butter and a toaster.
And you can even smell it.

The Cray 1 was freon cooled, Cray 2 was fluorinert, C90 was air (I think), XMP/YMP were air...

Then the T90s, introduced in ~1994, used a big ol' fluorinert tank. Highend, high density Cray boxen (T90) use this cooling, smaller machines (J90, SV1) are air cooled in temperature controlled environments, and some, like the T3E can be water cooled or air cooled, depending on the configuration.

The Cray 1 was freon cooled, Cray 2 was fluorinert, C90 was air (I think), XMP/YMP were air...

Then the T90s, introduced in ~1994, used a big ol' fluorinert tank. Highend, high density Cray boxen (T90) use this cooling, smaller machines (J90, SV1) are air cooled in temperature controlled environments, and some, like the T3E can be water cooled or air cooled, depending on the configuration.

The Cray 1 was freon cooled, Cray 2 was fluorinert, C90 was air (I think), XMP/YMP were air...

Then the T90s, introduced in ~1994, used a big ol' fluorinert tank. Highend, high density Cray boxen (T90) use this cooling, smaller machines (J90, SV1) are air cooled in temperature controlled environments, and some, like the T3E can be water cooled or air cooled, depending on the configuration.

The Cray 1 was freon cooled, Cray 2 was fluorinert, C90 was air (I think), XMP/YMP were air...

Then the T90s, introduced in ~1994, used a big ol' fluorinert tank. Highend, high density Cray boxen (T90) use this cooling, smaller machines (J90, SV1) are air cooled in temperature controlled environments, and some, like the T3E can be water cooled or air cooled, depending on the configuration.

The Cray 1 was freon cooled, Cray 2 was fluorinert, C90 was air (I think), XMP/YMP were air...

Then the T90s, introduced in ~1994, used a big ol' fluorinert tank. Highend, high density Cray boxen (T90) use this cooling, smaller machines (J90, SV1) are air cooled in temperature controlled environments, and some, like the T3E can be water cooled or air cooled, depending on the configuration.

The first Cray super computer to use Flourinert was the Cray 2, circa 1988. Instead of using a liquid nitrogen intercooler system, the flourinert was circulated to a refrigaration unit mounted outside the building. This made it one of the quietest supercomputers ever. Also one of the prettiest! (Looked like a giant Xmas tree bubble light.)

For more pictures and info, go to Cray.com's gallery. In the picture, the C shaped tower in the center is the computer. The two red and white boxes at the rear are pumps for the flourinert, and the clear tube at rear center is a recirculation system.

Who are you? Where are we going? And why are we in this handbasket?

It also listed refridgeration with air conditioning. I think they're refering to a whole range of devices, not just air conditioners. Like food storage and transport, liquid cooled supercomputers, and cold fusion. Don't these things positively effect your daily lives? Okay, so maybe not those things, but the freezing and refrideration of food has had a great impact on the development of society. Frozen food is much more resiliant to disease and easier to transport larger distances. Cheaper and More available food directly relates to an increase in population. Although debatable whether an increase in human population is good, one cannot doubt that refridgeration has significantly contributed to the development of society.

So, Cray was torn apart by SGI and Sun. Sun jerks over their portion and comes up with the Enterprise 10000. SGI flounders a while with their chunk and ends up only holding a name to sell.

While I'm glad SGI is embracing Linux, in light of what they did to Cray, do we want the Linux name associated with this company? I would be happier if Sun truly supported Linux (instead of just paying lip service, community source... gack).

In SGI's defense, $100 million is a pretty good price for a trademark and a domain name.

http://www.cray.com
A Cray T3E, which is the fastest massively parallel Cray, runs at up to 1.2 TeraFLOPS, with 2048 DEC Alphas. I believe that there is an upgrade to get it to run 600 MHz Alphas instead of 450's, which would increase the max speed. I remember reading that they put together a machine that did over 1 TeraFLOPS on real code. So basically, the answer to "how fast is a Cray" is "Very". :)

Forest Godfrey