Slashdot Mirror


Cray XD1 Now Available

cyngus writes "Cray announced the availability of their XD1 systems. Each XD1 chassis has up to 12 AMD Operton processors. Up to 12 chassis can be clustered together in a rack. The XD1 uses Cray RapidArray Interconnect technology, based on HyperTransport, for high bandwidth and low latency communications between processors and chassises. The XD1 also has a handful of other technologies aimed at the HPC market, including Xilinx FPGAs, communications accelerators, etc."

20 of 200 comments (clear)

  1. Re:long time no news... by cyngus · · Score: 5, Interesting

    SGI does not own CRAY. They did buy them back in 1996. SGI sold its Cray unit in 2000 to Tera Computer.

  2. Re:long time no news... by Anonymous Coward · · Score: 1, Interesting

    well I think since the advent of home computer beowulf clusters it has take the awe out of huge supercomputers. Before it wasn't technically possible to scale a system to such high performance, but now with a bit of cash you can go buy OTS G%s and make a top500 computer. It is taking a bit more to inspire dreams with simply performance specs.

  3. Does the XD1 give the illusion of shared memory? by Anonymous Coward · · Score: 3, Interesting

    I've heard conflicting reports on this - reading Cray's own literature, you see them say:

    "Tightly coupled to the AMD Opterons and switching fabric, [the RapidArray Communications Processors] handle memory to memory copies, global memory management, and system wide process synchronization, freeing..."

    (Emphasis mine)

    Does this mean the HT links give the OS the view of a single-system for each chassis? (Or rack, even?) Ie, can I utilize a single processor out of those 12 in a chassis, and access 96GB of RAM with that one process WITHOUT using MPI or rDMA?

  4. Re:Not the top end by LoudMusic · · Score: 2, Interesting

    It would take sixty racks of these to best the Earth Simulator's theoretical peak; more than 60% more processors.

    Still, if they need someone to, uh, test one...


    Interesting numbers. Also to note, NEC's Earth Simulator is now nearly three years old - the Cray XD1 is made with modern AMDs.

    I guess there's no getting around it. For the time being our really fast computers will just be fucking huge. Oh, and call NEC if you want a big computer. (:

    --
    No sig for you. YOU GET NO SIG!
  5. Cray doesn't do Clusters? by Edward+Ka-Spel · · Score: 3, Interesting

    I thought Cray was trying to convince the world that Clusters were not as good as true supercomputers, but this looks like a glorified cluster. In looking under the hood it appears to be just a collection of 2-way SMP Opterons with a superfast proprietary network backbone.

    And it's running Linux, if that matters to you

    1. Re:Cray doesn't do Clusters? by ari_j · · Score: 3, Interesting

      This is nothing compared to the 1024-way massively parallel computers Cray has built in the past. That's why they don't like clusters - to them, a cluster is an inferior solution in the light of massively parallel systems and, on the other end, vector supercomputers. What can you do with a cluster that you can't with one of these?

      Given the financial status of Cray, embracing clusters is just a common sense move, not necessarily an ideological one.

  6. Re:Not the top end by Anonymous Coward · · Score: 1, Interesting

    I guess there's no getting around it. For the time being our really fast computers will just be fucking huge. Oh, and call NEC if you want a big computer. (:

    Actually:

    Blue Gene

    Let's see what happens come Supercomputing '04... good things can come in small sizes. ;)

  7. Hell, yeah! by DarkMan · · Score: 4, Interesting

    For my apps, I do iterative matrix calculations. However, one of the required data tables scales as n^2.3 (ish) of the system size. These can be precalculated, or calculated on demand. Typical size for a small run is 4-6 GB. I've filled a 40 GB array with data tables before.

    Thus, the part that impacts runtimes the most is either the on disc lookup, which is still faster than direct calculation, which we've also had to do.

    I looked into FPGA's a while back. Some back of envelope calculations show that a single FPGA should be able to calculated the data table on demand, and it'll be faster than reading from disc.

    (Turns out, that to actually get a usable solution for a basic PC would need to hack up the whole tool chain. FPGA cards for a PC are all designed for DSP, rather than numerics).

    So, with an FPGA and a CPU, I could elminated the slowest part of the job, and scale up to, what, a 1GB working matrix, which is about 8 time larger than the biggest job I've ever run, which hogged a T3E1200 for 6 hours.

    So, in short, gimme an FPGA and some reasonable tool chain, and I will be able to about half runtimes, and, more importantly, scale up to 10 times larger calculations. 5 time larger calculations is the most I've ever been asked about.

    Time to brush up on my VHDL, I think.

  8. Re:Not the top end by Rasta+Prefect · · Score: 2, Interesting
    It would take sixty racks of these to best the Earth Simulator's theoretical peak; more than 60% more processors.

    Earth Simulator uses vector processors. If you want a comparable Cray system, you should be looking at the X1 which is also a vector processor. Incidentally, the X1's silicon runs so hot they use evaporative florinert cooling instead of a straight liquid - the florinert is heated to just under the evaporation point and sprayed onto the processor so that the phase change will remove more heat than just immersion.

    --
    Why?
  9. Re:right.. but what happened to Tera?! by fitten · · Score: 2, Interesting

    Tera's MTA isn't exactly like HyperThreading. HyperThreading looks at (currently two) threads and sees which instructions from each stream it can schedule each clock. MTA was more like round-robin scheduling of threads on a per-clock basis. At each clock N, it scheduled an instruction from thread N, on clock N+1, it scheduled an instruction from thread N+1. In other words, if your process had only one thread, and the MTA processor ran at 1GHz and had 128 "threads", then your process ran as if it were on a single threaded 8MHz CPU.

  10. Let's See... is my G5 math right? by Wingsy · · Score: 2, Interesting

    12 Opterons deliver 58* GFlops (where * = peak). The Army's recent G5 cluster (1566*2 G5 processors running at 2GHz) deliver 25* TFlops. 58 divided by 12 yields 4.8* GFlops per chip for an Opteron, and 25000 divided by 3132 yields 8* GFlops per chip for the G5. What's wrong with this math? I didn't think the G5 had numbers THAT much better than an Opteron. And with G5s hitting 2.5GHz today the numbers would be much worse (or better, depending on your point of view).

    --
    If I didn't have absolutely NOTHING to do, I wouldn't be here.
  11. What happened to RedStorm? by telemonster · · Score: 2, Interesting

    Cray not-too-long-ago had major announcements with the RedStorm project. I believe that system is supposed to be a single image 10,000 CPU AMD based rig. There are some oddities friends have pointed out, like the OS is based on IRIX I believe...

    Yea check this out:

    Cray Unicos/mp"

    Actually that references the X1, which is not based on PeeCee stuff, but actually a 8 core MPM.

    Sad thing is, even with Red Storm I think IBM will remain on top as their contract calls for 130,000 of their powerPCs on one system?

    It would be nice to see Cray on top, with something other than a commoditiy processors. I realize the T3D and T3E were both Alpha based systems.

    PS, I still have a J932se 32 proc Vector Cray ( for sale ) if anyone wants a Cray for home. $4500, real deal 3 cabinet Cray from 97', most likely used for gov't nuclear energy something-or-other. Located in Southeastern Virginia.

    --
    Southeastern Virginia REPRESENT!
  12. Re:Interesting specs and density by 14erCleaner · · Score: 2, Interesting
    However, part of me is a bit saddened by seeing the Cray name attached to X86s.

    Actually, in the year between crash of Cray Computer (in March 1995) and his death in an auto accident, Seymour Cray started a new company, SRC Computers, which still exists, and makes a parallel Pentium-based computer (which also incorporates custom hardware processing elements). I believe that this product is the same thing he was working on from the start of that company in 1996.

    --
    Have you read my blog lately?
  13. Re:Sheesh...what happened to Cray? by Anonymous Coward · · Score: 1, Interesting

    The Cray-1 and X-MP machines had soft padded versions and the Y-MP had hard plastic versions. Most of the really loud stuff was in the motor generator room off to the side. The ambient noise in the typical larger computer room was louder than anything coming from the Cray. The soft padded versions weren't the most comfortable things to sleep on, but they did in a pinch very late at night.

  14. I love Cray by Superfreaker · · Score: 3, Interesting

    After searching everywhere for the legendary "Wang Computer" tshirt, I decided to fall abck on teh second geekiest computer company to get a shirt from, Cray. I couldn't find a shirt through the normal outlets (eBay/ThinkGeek), so I called them directly. The woman that answered was glad to help and shipped out, not a tshirt, but a very nice collared shirt that makes it look like I work for Cray! I wer it to all the conventions and I become cool(er).

    *queue calls to Cray*

    1. Re:I love Cray by fingusernames · · Score: 2, Interesting

      Cray was always big on the memorabilia. I worked there as an intern for two years in the mid-90s. I still have many t-shirts (including our custom "World's Fastest Interns" version), some very cool posters, a ceramic Y-MP model, obligatory mugs, a cool 1-800-BUG-CRAY coaster, so on. I would call them up every so often after leaving there, and have them send me rolls of new posters and such.

      That was a unique experience... I had a security pass to the machine room, full of Cray C-90, Y-MP, X-MP, Cray-1 & Cray-2, lots of others. Awesome environment.

      Larry

  15. Re:Where's the source code by Troy+Baer · · Score: 2, Interesting
    Actually, as far as I can tell, their interconnect is IB at the hardware level, but without a PCI bus between the HCA and the memory controller. That drops their latency by a couple microseconds, and it means they don't need PCI Express to get full bandwidth out of the network. The claim is that they use their own software stack in place of the VAPI stuff, and that they don't use MVAPICH from Ohio State like most IB sites do. I haven't had a chance to look at our XD1 (we've had 3 chassis' worth for almost a month) to see if that's true or not.

    Their story as far as storage is kinda lame; I think they're praying Lustre won't suck.

    --
    "My life's work has been to prompt others... and be forgotten." --Cyrano de Bergerac
  16. Re:Not the top end by flaming-opus · · Score: 2, Interesting

    ISR (Isothermal Systems Research, Inc.) and cray have cross licensed patents on this technology. I don't know if ISR plans on productizing this or not.

    I imagine that this is extremely expensive stuff to do. Since a cray can charge $40,000 per processor for the X1, they can get away with a $700 cooler. Not so easy on a PC.

  17. Re:Not the top end by bobbozzo · · Score: 2, Interesting

    You shouldn't need phase-change to cool a PC.

    Just circulate the stuff and run it through a radiator.

    BTW, HP was recently researching cooling chips with inkjet nozzles spraying a coolant which evaporates easily onto the CPU.

    --
    Nothing to see here; Move along.
  18. Re:Still waiting for... by Fnord · · Score: 2, Interesting

    You think its a joke, but I used to work there and they do refer to employees of the company as Crayons.