Slashdot Mirror
96 Processors Under Your Desktop
Roland Piquepaille writes "A small Santa Clara-based company, Orion Multisystems, today unveils a new concept in computing, 'cluster workstations.' In October, you'll be able to choose between a 12-processor unit for less than $10,000 or a 96-processor system for less than $100,000. These new systems are powered by Efficeon processor from Transmeta and are running Fedora Linux version 2.6.6. Apparently, this new company has friends in the industry. You already can read articles in CNET News.com ("A renaissance for the workstation?"), the New York Times ("A PC That Packs Real Power, and All Just for Me," free registration, permanent link) and the Wall Street Journal ("Orion Sees Gold in Moribund Workstations," paid registration). The company is targeting engineers, life scientists and movie animators. It's too early to know if the company can be successful, but I would certainly have to get one of these systems under my desk. In this overview, I've picked the essential details from the three stories mentioned above."
Any ideas?
Great... by october my brand-new machine will be hopelessly out-of-date. I knew it would happen, but had no idea they'd usurp me by 94 processors.
I thought '96 processors under your desktop! That would be the Pentium at 133MHz!
Seriously, why 96? Why not 64 or 128?
-
If you keep throwing chairs, one day you'll break windows....
Apparently, this new company has friends in the industry.
Apparently Slashdot is one of them
Can I run Doom 3 on it in maximum resolution mode?
Check out the best P2P sharing website: MEDIACHEST.COM
why not get a huge server where more users can benefit from the processing powers? and what kind of videocard does this baby pack? that must give some great doom3 performance :)
http://www.virtualconcepts.nl/
It'd be cool to have 12 high-end AMD processors instead of relatively slow Transmeta CPUs in this workstation. But I guess their total disspated heat will melt computer case :(
Did I miss something? I thought this was 2004, not 1998.
One thing perplexes me:
Chips on the same board communicate using Gigabit Ethernet, while board-to-board communication takes place on 10 Gigabit Ethernet.
Wouldn't same board communication be more frequent, hence needing the faster connection?
Better yet, why not 10GBe for both?
I dont want that thing running anywhere close to my crotch. .. course this is slashdot so i'm sure most wont care :)
I really admire this guy; although the ventures he took part in haven't gone anywhere financially, they were pretty cool. Transmeta, OQO, and now this! Go Colin Hunter!
With the power requirements on this thing, the case will be half PSU. I can see the warning on the case now 'Do not place in carpeted areas.' I bet the electro-static discharge would make you sterile faster than the speed of rubbing socks.
ASCII pr0n. Coming to a Lunar Lander near you!
I'd like to see what the employees do with them after hours...
"Gotta work late today, honey. Oh, sorry, is this the Pizza place? Could you please hold one sec?"
"Gotta work late again today, honey..."
WARNING: Smartphones have side effects--most of them undocumented.
This time we don't have to imagine what a cluster would be like. It's already in the box!
The comment has already been made. Let's move it along people. Nothing to see here.
That'll be $9999, please.
Man.... I truly hope that they don't make it Longhorn compat at some point. That would definitely be scary. It would at least fit the requirements of the dual/quad procs.
-- Friends don't let friends buy Nokia.
Fedora currently is either Core 1 or Core 2. 2.6.6 is a kernel version number.
Kernel version != Distribution
Saying "Fedora 2.6.6" is like saying a car is a "Ford 2.4 liter".
96 processors running Fedora? I want one!
Actually, I would be willing to bet that the university I'm at could use a few of these things. After all, we've got undergrads doing BLAST database work, just to teach them about it. Having been through that hell myself, it'd be a lot easier if you didn't have to have a cluster to do the work by computer. For those who don't know, BLAST is a genetic sequencing database that allows for comparison with an extracted gene (retrived through polymerase chain reaction) with a known, sequenced gene in their database.
Haec merda tauri est. Ceterum censeo Carthaginem esse delendam.
....but I would certainly have to get one of these systems
under my desk. seriously,.. where do u intend to put ur foot ?
fifteen jugglers, five believers
The pricing seems quite steep. 800/cpu for 12 configuration, 1000+/cpu for the 96 cpu configuration. I can see why they have friends in the industry the prospect of selling 10 to 100 times the equipment per seat must have marketing departments salivating.
If your'e going to spend that kind of money though theres alllready solutions that will provide that level of processing cheaper.
There is also the utilization isssue, programming tasks hardly require 96 processors except on compile and link. You don't need 96 processors to wait for a keystroke. The same holds true in applications. You don't need cpu's waiting for a user to decide what to do.
If this does well over the long term, I suppose I'll be looking at an n-cluster computing system in ten or fifteen years, when the technology percolates out to the masses.
Canthros
If you're having lag problems, I feel bad for you son.
I got 96 processors and you got one!"
OK, apparently, someone took the "Imagine a beowulf cluster of these" a bit too seriously.
Way to go!
...in 3, 2, 1...
How many of these units do you think they're going to sell? These things don't design themselves and the company has to cover these costs. Of course getting the pricing right will dictate sales.
SCSI, SATA or what?
Leandro Guimarães Faria Corcete DUTRA
DA, DBA, SysAdmin, Data Modeller
GNU Project, Debian GNU/Lin
No matter what the actual news is about you will always have more then enough posters on /. that freely share their expertise with us and tell us again and again that Linux is _not_ ready for the desktop.
Putting 96 CPUs under hood simply will not change this fact!!!11!!11one!
now thats a system i'd like to install gentoo on :)
-- Karma: beyond good and evil - mostly affected by posting political
I've got well over a hundred in the box under my desk. Unfortunately, it's just that. A box of over 100 CPUs, mostly Pentiums/Pentium IIs.
If so, queue megadeth's "99 ways to die" ha ha!
Seriously, though, I bet Strongbad can check the hell out of his email with that!
stuff |
So what's that in theoretical peak MIPS? Or even bogomips?
tasks(723) drafts(105) languages(484) examples(29106)
The interconnect is tcp/ip over ethernet, just think of the the overhead it generates on the poor transmeta CPU. Actually, quite strange given that each efficieon has its own Hyper Transport Bus they could have been much clever about it and on the way also use Single System Image OS - maybe Mosix. As I see it there is not much usability for the staff, unless you like noice and heat beneath your table...
Now we just need to see a company come up with a complete LAN solution where every machine acts as a SETI@Home type client in a cluster, giving SMB's a supercomputer that works while it's various workstations comprising it are idling.
So what's the benefit? How does this compare to a dual opeteron or something similar? If a dual opteron can do something like 12G flops for $2.5K, what can this thing do with 12 processors at $10K...?
Anybody have a clue how the performance of an Efficeon processor stacks up against Xenon's, Opterons, & etc-
Having a 12 p box is great but if its performance is a bit less or the same as a 4k$ 2p Opteron box what's the "real world" advantage?
This kinda stuff (if it were cheaper) could pave the way for half-decent mmo's, assuming net speeds could hurry up and level out (I can't stand seeing people in the world with 56k's and others with 20mb net).
Warning ! Warning ! Warning ! Warning ! Warning !
Attention, a public service announcement follows : do not read the "overview" touted by Mr. Piquepaille. This person constantly spams Slashdot, trying to get traffic to his site on Radio Userland (which I'm not linking to, for obvious reasons). Do NOT go to his overview, you're only giving traffic to a spammer. See these recurring complaints, for instance. Not to mention he steals the images he puts on his blog and sometimes also spews bullshit for lack of knowing better. This must stop. In any way, do not fall for the spam, and do not provide him any more traffic. Please also warn fellow readers when you see one of his self-serving posts.
And now, a personal message (warning : verbal abuse in foreign language follows) : Roland, tu nous les brises. Va te pendre, hé Ducon !
[disclaimer : I'm not commenting on whether the subject is interesting, or not. But the kind of astroturfing the submitter engages in regularly is just wrong]
Warning ! Warning ! Warning ! Warning ! Warning !
Xenu brings order!
This wouldn't be as reliable as having 96/48/24/12 computers with 1/2/4/8 processors each, which would be important for things like movie animation.
And besides for movies, we already know to just fit as many Opterons in a rack as possible. What advantage does this have (except for cost)?
But can you imagine the licensing fees for using Windows on this many processors? I bet it wouldn't be too cheap. :-D
How many stories a week does this frigging guy get into Slashdot?
His business plan:
1) Sell Ad Space on "News" Website
2) Shovel In Content From Online Articles
3) Submit To Slashdot Daily
4) Tout "Slashdot Coverage" To Advertises
5) Profit!
And looking at his site, it works fine and dandy indeed.
Then again, is he just doing a service to us?
...If you've got kernel problems, I feel for you son,
I've got 96 problems but my chips ain't one....
and Windows ain't on one. :D ...
Sorry.
That means, not so many applications will benefit from it. In fact, I'd say that 90% of desktop apps will run better on a 2 GHz single-cpu computer than on this one. In fact, a 2 or 4 CPU (non-cluster) computer would offer better performance for a larger number of apps, than any cluster. Software that takes advantage of clustered configurations has to be specifically written for them.
Oracle RAC is one, but I can't think of any other popular title that would, expecially not for the desktop.
Sigged!
Imagine a Beo.. uhh. Never mind.
With so many GigE connections inside that box, I guess Sun's "the network is the computer" finally came through.
On the other hand, this is also "the computer is the cluster" at the same time.
Am the only one who saw "A small Santa Claus-based company" I guess after seeing 96 processors under your desktop, that's what first came to mind.
The only 2.4L engine Ford manufactures is, to my knowledge, the diesel engine used in the Ford Transit van, but the Transit isn't sold in the US. So the engine may not do a spectacular job identifying the vehicle, but it could also be a lot worse.
Is there a 2.4 liter Ford engine that I'm not aware of?
Now if you'll excuse me, I'm off to download the internet on the CD Disk with my American Online after I finish installing a new RAM Memory.
100k for 96 processors? Figure you can get a barebones system with 256 MB ram for around $250. That's $24k for the boxes, a 96-port switch, and some good clustering software.
Where's the rest of the cost coming from?
I mean it's cool, but if I had $100k that absolutely needed to be spent, I'd get a Viper or something instead of a big server.
<ducks>
Wer mit Ungeheuern kämpft, mag zusehn, dass er nicht dabei zum Ungeheuer wird. --Nietzsche
Orion Sees Gold in Moribund Workstations.
------------------
OT
If you have a gmail account, you might want to read this.
Good ol' WSJ, that hardcore tech rag. Next they'll tell me the brains march off little soldiers to various parts of Computer Land to give orders and bring back messages.
free online diet tracking.
Maybe this is what I have been waiting for?
Nice little subliminal message.
~lake
First, I see that Transmeta are finally making 1.5GHz Efficeons, which is a good sign, they looked to be stuck at 1GHz for so long.
This merely looks like 12 computers on a single motherboard with a GigE switch connecting them together. Each computer is highly optimised of course, just a processor, memory, support chipset (GigE, IDE).
I do have to wonder how it compares with something similar made with Opterons or Pentium-Ms. Opteron has the advantage of being able to do SMP so the per-system processing power would be much higher, each board could have 4 low power 2GHz Opterons which will probably be close to the 12 Efficeons in terms of computing power and power consumption.
But still, this is a cool system. I wonder how fast it can do a kernel compile?
I was just curious- what would the average desktop user do with 12 processors? Now they will be able to run much more spyware, I would look forward to all of the support calls. "I don't get it- I have 30 things running in the background and my computer is slow. It worked fine yesterday."
Crack: It's not just for breakfast anymore
Everyone misses the point of these. A supercomputer under your desk is pretty pointless to most folk. No neon lights, so the boss won't want one to show off. Real workers most likely value decent single thread performance more than many CPUs running many threads miserably.
For the tasks that it could do well, performance will be stunted by miserable disk performance. You can fit 96 CPUs under your desk, but they all share a 4500RPM notebook HD. Heh, yeah. Does Fedora Core have a countdown timer for measuring time till disk swap? Does it have an hours digit? Two of them?
Seriously though, this box will be stunningly good for two things. First is testing code for the next gen of gaming consoles. The more important one is that it will be really good at simulating the performance of the 'next big thing' in desktop CPUs, the massively multicore stuff.
Names like Conroe, Tukwila and AMDAskedMeNotToSay all come to mind. So do PS3, XBox2 and Nintendo Somethingorother. What do you develop on to test your code that needs to synchronise 8 threads across 5 CPUs? How do you test that? Simulations only go so far.
These boxes are a great way to test far future algorithms on far future platforms using safe x86/iAMD64 instructions. Would you rather buy shitloads of PS3 dev stations to write platform neutral algorithms, or a cheap box your devs already understand.
All this said, I don't see a huge market for the box, but it sounds nifty. And it is black. Black is goth, and goth is ummm, it is black though.
-Charlie
If they were Intel or AMD processors, I'd be interested.
Imagine a beowulf cluster of.... ok, ok, I'll shut up.
"If a boss demands loyalty, give him integrity. But if he demands integrity, give him loyalty." (John Boyd, 1927-1997)
If I remember correct, the transmeta uses code-morphing to emulate a X86. Would it be possible to remove/alter this step and create a compile option on GCC which makes code run "natively"? would this lead to noticeable improvements??
10 ?"Hello World" life was simple then
Apparantly, we also need a "bitch slap" mod...
.
Here, I'll start things off for everybody:
.
In Soviet Russia....bitch slaps YOU!
.
Thank you, and have a pleasant evening....
.
A goal is a dream with a deadline
Would make a great platform for either a Web Server or database (or both). Imagine trying to slashdot a Web site running on one of these puppies. Would also be handy for doing load testing, and large scale compiles (re-build the universe for a large software project).
[Insert pithy quote here]
Can the editors PLEASE STOP POSTING PRESS RELEASE COPY?!?!
If you're going to post a story announcing a product or discovery, at least link to a weblog or site that actually has a little commentary on the subject, or the original site itself.
Roland "Fuckyfacey" Piquepaille is neither of these.
Thanks.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
By itself it's going to look seriously off-topic.
I thought I had attached this to one of the many "imagine a Beowulf cluster" comments.
strange....
A goal is a dream with a deadline
# of them that are advertisements, not news: 3.
slashdot is dead.
I'm surprised there are already over 100 posts and noone has asked about the fedora version...
The article says "Fedora version 2.6.6". I take that means Fedora Core 2, with a 2.6.6 kernel, but it is rather imprecise... At that price, why not RHEL?
> . . . These things don't design themselves . . .
They may not design themselves, but perhaps they'll design the next generation.
Trying to scale up that many processors in a single small chassis seems like the wrong direction to go. Especially since the power/heat requirements make a relatively slow processor necessary.
Wouldn't someone be better off using a good Gig-E network and some processing nodes with fast CPUs to do the crunching?
Apple's Xgrid is a good example of this.. Use a G5 workstation, and process on some Xserves.
I guess the main downside to Xgrid is that it needs application support, where SMP is divided up in the OS.
Check out this article regarding throttling with the new TM6800 processor:
5 _efficeon/040405_efficeon.htm
http://www.vanshardware.com/reviews/2004/04/04040
V
Do you have anything interesting to say about this interesting looking workstation?
Friends don't help friends install M$ junk.
Or lack there of. I was imagining one of those SGI deskside Onyx servers when I read the post but these are just wide full towers that pack 96 processors. Quite nice. I am sure the movie industry is all over these babys. The 12 unit is around the size of a Sun pizza box.
How do the Transmeta CPU's do in fp computations? That is obviously the metric to note. I wonder how long it would take to render a movie? Is the USB USB 2? No firewire though.
Get quad opteron. It should get about better performance for same price as the other but without need for clustering, for the small system. For bigger system, you could use myrinet and dual opterons. Oh what the heck. It costs 600-800$ per processor to build a rack of Athlon64 based cluster with Gb ethernet. So this effineon based cluster would be beaten with system costing less than their solution. So it is beaten from two different solutions.
A) Getting single image opteron system if communicationlatencies are important.
B) Getting cluster of AMD64 if price/performance was important.
ONLY thing they bring is density of A for system type of B while costing more than A.
Emacs is good operating system, but it has one flaw: Its text editor could be better.
There have been many posts about underutilization of these computers when the (scientific or artistic) users are not at their desks rendering the next CGI movie or modeling their semiconductor quantum dots.
For these to be a useful expendature, they need to integrate into the main batch job processing system at night or when the user is out to lunch. Obviously, the local user's processes would be highest priority, but it should certainly not be strictly limited to a single-user "workstation". That would be a major waste of money. When you spend this much money on a set of CPUs, you just don't let them sit idle -- especially when they're designed to be a cluster and you've got cluster work to be done.
I could easily see a day when Dreamworks does away with their batch processing cluster and uses clusters of their employee's workstations to render batch jobs at night or when the employee is tinkering with something simple.
a beowulf cluster of theese.
Sorry, I coldn't help it.
The base model of this thing will peak at 36 GFLOPS for about $10,000. A single G5 Xserve will get 30 GFLOPS for $3,000. Start looking at putting together several of these for less than you can get one of these mini-cluster machines and get higher performance. What am I missing?
(Please no flame-bait posts. I am not trying to start a PPC vs. [name of favorite processor] here. I am seriously curious why this is a good deal?}
Imagine a beowulf cluster of these.
*smack* *pound* *boot-to-the-head*
*groan*
how about this for a submission template
check out this overview for more details, guess who comes up first, can we say formula ? he could at least tr7y a bit harder, then a hack like him probably isn't clever enough to actually work for a living
not convinced ? how about
this overview, can we say no creativity ? (plagarism aside)
perhaps he should get a real job, i wonder how he did in his exams ? do you think he wrote his own thesis or just ripped off someone else ?
trolling takes many forms but plagarism pays a lot more
In only a couple of months' time, you'll be able to put 8 Opteron cores (4 dual core CPUs) in a desktop-size case - and this is a rather reliable information. It is also very likely that similarly sized boxes with 8 CPU sockets (and thus possibly 16 cores) will appear next year: infoworld.com article.
"I love my job, but I hate talking to people like you" (Freddie Mercury)
he says:
They may not design themselves, but perhaps they'll design the next generation.
"I spare not a single unit of thought on these cybernetic simpletons!" he boomed. "I speak of none but the computer that is to come after me!"
Fook was losing patience. He pushed his notebook aside and muttered, "I think this is getting needlessly messianic."
music lover since 1969
http://www.tomshardware.com/hardnews/20040830_0920 16.html
It says:
"According to Hunter, the Efficeon architecture allows Orion to reach a performance of one flop per Watt - more than would be possible with any competing processor."
I'm familiar with megaflops and gigaflops and teraflops and petaflops, but what is so magical about "one floating point operation per watt"? Is this just a misquote, or does it mean something?
.. pa-ra-bo-la, pa-ra-bo-la, 2 pi R, 2 pi R, where's your latus rectum, where's your latus rectum, 2 pi R
$699 X 96 + taxes !
Darl McBride
getSexySig();
... for each of those CPUs!
in other news: Orion sits down w/ID Software for preliminary planning of reference platform for Doom 4
Being able to have one of these in a standard office that doesn't need a server room setup is the major factor, as I see it. It means that the company saves on racks, rackspace, the cost of electricity (which is going up as you may have noticed by the price of oil being over $40 per barrel), administrator costs (his salary alone will be worth it) and air conditioning for the server room.
Think of a small CG effects movie company (say around 5 to 10 employees): They want to be able to render their CG movie frames as quickly as possible in order to save money, but don't have the capital to buy and maintain an expensive server room setup along with the requisite admin type to look after it. These machines provide and effective alternative: 1 or 2 of the 12 CPU machines for the guys who do the rendering of small snippets and 1 96 CPU unit as an office renderer for resource expensive renderers.
VNC feels about the same regardless if you're using 100BaseT, GigE, or 10GigE. You still have the overhead and latency of ethernet, TCP/IP, and VNC itself. Upgrading from 100BaseT to 10GigE will greatly increase the amount of thruput possible, but it will do very little to reduce the total lag.
This isn't really a show stopper, unless your GUI applications are tied directly to the 96 CPUs. In that case, they really outta add at least one x16 PCI-E slot to that beast for graphics.
Otherwise, just use a $250 PC as your main desktop GUI box and make use of Linux's wonderful networking abilities to tap into the power of that 96 CPU beast sitting beside it (and hopefully connected via a sweet GigE or 10GigE crossover cable).
I know people that still use an SGI Indy as their desktop workstation, but almost every one of their CPU hungry apps runs on a server in some other room. With X forwarding and NFS mounts it all blends together.
OMG you nailed it! Haha fast computer + Doom 3 system requirements = real fucking funny HAHAHAHAHA!!!!!!!
Here's what really perplexes me:
What they've basiclly done is make a cluster-in-a-box; each board contains 12 self-contained single CPU circuits and a gigabit ethernet switch circuit (with 10GigE uplink). Multiple boards are connected together via their 10GigE uplinks.
If there were aiming high with 96 CPUs in one box, why didn't they really try to impress potential users by making each board a *single* PC?! That is, rather than 12 individual PC nodes on each board (each running its own copy of the OS), why not make each board a single 12-CPU PC? That way you'd still have 96 CPUs, but with much greater thruput and much faster shared memory between every 12 CPUs. You'd still have 8 total PC nodes in the box, so reliaiblity would still be reasonable.
When I think "multiple CPU workstation" I tend to think that at least SOME of the CPUs will be directly connected in an SMP manner for faster shared memory, much easier process migration, and the need for only one copy of the OS to reside in memory. Gigabit ethernet is a nice technology, but a CPU interconnect it is not.
I have some side questions:
How upgradable is this monster? With 96 individual nodes, I would assume it will have at least 96 SO-DIMM slots. 192 slots would be nice as it would allow me to upgrade the RAM without throwing away the original sticks.
How does this thing bootstrap? Am I correct in assuming that at least one node has a hard drive from which it boots, then serves itself up as a netboot server to the other (diskless) nodes?
This brings me back to one of my earlier points: with 96 nodes you have 96 copies of the OS to load up. That may be cool in a geeky way, but I would really much rather just deal with 8 nodes, each a 12-CPU SMP PC. Heck, you could even afford to have one real hard drive for each of those 8 nodes.
Maybe I just don't "get it".
Okay, here's what I don't get. We are going with 96 procs that run at what speed? Knowing Transmeta, (partners with VIA right), why would anyone whant a cluster of those procs? Yeah, they may be somewhat fast, but the FP calculation on these things stink (I've heard). Now why not go for this system http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&cate gory=11223&item=5716808314&rd=1 (ebay). A 128 proc SGI origin with R10000 procs. for $14,000 or how about http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&cate gory=11223&item=5714399696&rd=1(ebay). Same thing with 4GB RAM. These things are full SMP mahcines (which means they are not a cluster). Look it up the SPEC benchmarks site, and you will find that one 320Mhz will run the same speed as many low end P4s! and 128 of those at that!
Now if only by budget would support a beowulf cluster of these machines.....;)
Man is the lowest-cost, 150-pound, nonlinear, all-purpose computer system which can be mass-produced by unskilled labor.
Can I get the Greeg Destiny (density) model at 240 nodes in one rack?
MASSIVELY PARALLEL SYSTEMS ARE THE FUTURE. LEARN YOUR PARALLEL PROCESSING THEORY NOW, CS STUDENTS!!!!
If you think about it, it's the natural evolutionary direction to take. Single processor systems will be come a thing of the past. Think about the dual core Intel chips in the news recently. If you want a good job, learn your parallel programming theory!
Here's a 12-node cluster made of VIA EPIA V8000's with 800MHz C3 processors. Idle power consumption is ~140W.
"The machine runs FreeBSD 4.8, and MPICH 1.2.5.2. After working with his machine and running some basic tests, Glen's cluster looks to be equivalent to at least 4 (maybe 6) 2.4Ghz Pentium IV boxes in parallel on a similar network - achieving a performance of around 3.6 GFLP."
In the end, why put this thing under your desk? Just leave it in the server room and enjoy the quiet.
I got 96 processors but DNF ain't done.
this is not gonna end up like the ill-fated Project Orion. That too was a supposed to be a a new concept, but the sucker never got a chance to take off...
Well, maybe it's better for us all... That was pretty cool, though.
There are 32-bit processors, which is probably not what certain scientists and others who need lots of memory want.
At first I thought that this was a joke, you know... a scam or something - but I guess that's wrong - these machines are real. I suppose they might have some interesting uses.
Probably the greatest drawback is that they aren't 64 bit.
Is there a market for this? If there is, it shouldn't be hard to get the price down by a factor of 2 or 3, at which point it starts to look reasonable.
It's amusing that they used Transmeta CPUs. Transmeta's most useful feature is that their CPUs use very little power when doing nothing. That may say something about the CPU utilization these guys get in their clusters.
May be they are using a GiE switch with a 10G bit uplink port? It is a matter of using what's available.
That makes sense.
Another issue is power: 10 GiE takes more than 1 GiE. Silicon area goes up, too.
If you don't need 10G of comm per processor, why pay a penalty - and another in the hub - to get it. And how much computing are you going to get done if your processor is spending a bunch of time wrapping 10G of comm? You'd need hardware TCP/IP accelleration, which means MORE silicon and power, in a situation where heating budgets are your limit. 1G per processor sounds good to me. (Should let you have even interprocessor comm intensive computations proceeding with reasonable efficiency.)
But a single 10G link offboard, using a relatively cheap commodity switch chip, makes sense, too. Your cluster becomes nearly symmetrical, rather than having a significant intra-board bandwidth advantage making it lumpy and making your algorithm partitioning problem more difficult.
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
DES in 96 seconds,
Any key, any time, any place.
My 96 CPUs are hungry,
AES a little longer,
Linger while I cook the keys.
Blowfish? Blown!
Twofish? Battered, fried and eaten.
And elipticals make smile,
Albeit for a very short while.
We just installed another 32 dual 2.7 GHz Xeons in racks with 200 gig drives with the cost of approximately 1 K per processor which is roughly what these clusters are going for. In terms of sheer number crunching - it ain't gonna cut it. 1.5 GFLOPs per processor as opposed to 5 GFLOPs per processor. This is probably an optimistic figure - given that the Efficeon translates from X86 code to it's own native code before execution and the translation performance is likely variable with real world apps and probably hard to optimize code for. There is also the issue of reliability - what happens when one of the chips goes down?
However, the fact that most applications can take advantage of the processors with no tweaking, the low power consumption (no computer room necessary...) and the high interprocessor bandwidth might make it appealing to casual users with memory intensive applications and $100K kicking around...
...high-CPU-count SMP is hard.
Why do you think the market of huge single-system-image thousands-of-CPUs supercomputers has, essentially, only one vendor? (Silicon Graphics, a.k.a. SGI) We're not counting here superclusters and things like that, where IBM, HP and so on are happy to sell you a bunchload of systems. We're only discussing pure single-image SMP.
The SGI supercomputers are single-system-image machines. This Transmeta-based thing is essentially just a cluster.
Clusters are easy to design and build. Single-image machines are hard. Unfortunately, there are entire classes of mathematical algorithms that cannot be solved efficiently except on single-image SMP machines.
Imagine a beowulf cluster of these!
Okay I would love to have one of these just like I was drooling over SGI Octanes some years ago. I'm curious about graphics support on the motherboard. Can this thing drive a high-res video wall like one rich person was mentioned on slashdot some time ago? I haven't used that many cards on linux yet but I understand the number of agp ports matters. Perhaps another tower standing next to it with a bunch of graphics cards and some more of that 10Gbps fabric? Thinking private planetaria and snowcrash..
Now we have something serious, if true. You should have made that part of your public service announcement to begin with. Because you are so familiar with this, it should not be so hard to provide links.
Friends don't help friends install M$ junk.
I wonder what kind of fps these things will get running Doom 3 will all the goodies turned on. Our would it require 12 or 96 Radeon x800 Platinum Edition video cards for that?
These Efficieon processors better be pretty reliable, or this system better be able to fall back to SMP with a lower number of proccesors or downtime from dead CPUs would be a big problem with a 96 CPU one. If an Efficieon lasts 10 years on average (an overestimate, I'm sure), that means one will break every month.
They're not gonna make a dent in the market without a FOSS AutoCAD or Microstation clone..
WTF is plagerism? What is M$NBC? Who is Micro$oft?
Twitter, why don't you buy a damn D-I-C-T-E-N-A-R-Y already?
it seems so. its as if they cant pack any more power into a single core design so they either goes for multicore or multichip designs. it will be interesting to see this beast fly, and would anyone like to crank up doom3 on it;)
;)
hell, with sli comeing back in force frome nvidia and now this i realydo wonder what the gameing rig 5 years from now will look like
comment first, facts later. http://chem.tufts.edu/AnswersInScience/RelativityofWrong.htm
Heh, I use the same argument for my support of DSL over Cable for broadband hookups.
I prefer to call him Trolland Picknose.
and bitch-x aint on one.
Thes funny/odd thing about the 12 processor machine is it produces less gflops than a dual G5 so the only reason that you would eb spending that much for a machine is what? the extra ram?
Fedora eh? I wonder how fast till SCO Comes knocking.
I have already built a "cluster in a box" for my own computer research by taking a Dell PowerEdge 1655 blade server (12 processors) and running Redhat 9 with the Oscar clustering software. All in a single 4u box. At around $20k it isn't that much more expensive than the "coming soon" machines mentioned in this article and comes with 1.4 GHz P-III instead of the not exactly commonplace chips mentioned in the Orion machine. Only downside was the machine is just way too loud to use on my desktop (my original plan). Had to stuff it into a spare room as the fans really make a lot of noise. Still, the beauty of 'ssh -X' means I essentially have a 12 processor desktop machine for code development and running applications.
http://shit.slashdot.org/article.pl?sid=04/08/30/1 112254&threshold=-1&tid=137&tid=126&tid=1&tid= 218
Now I can see once again.
I also really hate the name Piquepaille.
It looks completely fake and made up. Very trollish, actually. Like Pan T. Hose, Seth Finkelstein, Bowie J Poag, etc. If I had that name, every kid at school would make fun of me and think of rude titles that would be far easier to say.
If I knew him in school, I'd probably do the same thing if I didn't feel sorry him. Since he's such a fuck face, I like to call him "Fuckyfacey", which probably rhymes.
I used to see the word "Roland" and my nipples would get hard at the thought of awesome music instruments I could never afford.
Now I see "Roland" and think: "Fuckyfacey". I just get angry now, not aroused. v_v
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
# 2004-06-17 18:00:22 INDUCE Act introducted to ban filesharing. (Your Rights Online,The Internet) (rejected)
# 2004-04-27 22:25:03 Microsoft media player now runs on Linux (Index,Linux) (rejected)
# 2004-03-09 17:23:00 Sveasoft Charging for Firmware (Linux Based) (Index,Linux) (rejected)
# 2004-01-26 18:03:00 Howard Dean calls for Internet ID checks (Index,Privacy) (rejected)
# 2004-01-11 05:10:53 Best Wireless Card for Linux? (Ask Slashdot,Linux) (rejected)
# 2003-11-20 06:33:25 RMS fires lead Hurd Maintainer Thomas Bushnell (Index,GNU is Not Unix) (rejected)
# 2003-10-21 04:23:28 As U.S. high-tech wages slide, fewer jobs may head (Index,Technology) (rejected)
# 2003-10-06 02:56:10 Intel CEO admits: jobs aren't coming back to US (Index,Technology) (rejected)
Why would I want to pay >$10K for a newbee box?
"...choose between a 12-processor unit for less than $10,000 or a..."
Regards,
-Ocelot Wreak
"I figure you're here 'cause you need some whacko who's willing to stick his finger in the fan. So who are we helping?
I didn't see anybody else link to Green Density. Look at the Compute Density (Mflops/sf) of Green Destiny+ vs ASCI Q from back in 2002. It makes a very strong case for using a ton of weaker more effcient processors. I don't think I'd want a bunch of p4 or athlons sitting under my desk making the room super hot.
Of course "Which would you rather use to plow a field, two strong oxen or 1024 chickens" - Seymore Cray (or something along those lines)
Seems like a new product just waiting to be made. A PCI express card with a bunch of CPUs on it. A plug in cluster :) :)
Of course you have could have the what is better a Cluster of G5s vs Intel.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
96 processors ought to be enough for anybody
I'm guessing it could use Seagate Savvio Drives - 10k RPM, SATA or FC, 36 or 72 GB, in a 2.5" drive designed for servers.
I have no evidence, but they are perfect for this type of application.
HIV Crosses Species Barrier... into Muppets
No, this is not funny any more. All of you idiots who think it still is just shut up. And you idiots with mod points, this is off-topic, but please mod the parent down instead. Thank You.
Colin Hunter did not co-found OQO.
http://www.oqo.com/company/team/
``In 2000, Jory Bell co-founded OQO with Jonathan Betts-LaCroix.``
http://biz.yahoo.com/ic/106/106342.html
``OQO believes computer users can have their cake and eat it, too. The company has developed a handheld computer with the functionality of a full-size PC. Powered by processors from Transmeta, its devices run Microsoft's Windows XP and feature LCD touch screens. OQO's computers can also be connected to a cradle and docked to a keyboard or monitor. Apple Computer veteran and OQO CEO Jory Bell, CTO Jonathan Betts-Lacroix, and members of the team that developed Apple's popular Titanium notebook founded OQO in 2000. The company has received funding from a venture capital firm formed by former Transmeta employees Colin Hunter and Edmund Kelly.``
come on, loosen the tie somewhat... after all, serious people do not look up slashdot, ok?;-)
"If a boss demands loyalty, give him integrity. But if he demands integrity, give him loyalty." (John Boyd, 1927-1997)
This machine went up for sale on ebay today. Go to ebay and search for "mini itx cluster" and give it a look.
$2500, almost 7 days and still no bids...
I used to get high on life, but I developed a tolerance. Now I need something stronger.