Beta
×

Welcome to the Slashdot Beta site -- learn more here. Use the link in the footer or click here to return to the Classic version of Slashdot.

Thank you!

Before you choose to head back to the Classic look of the site, we'd appreciate it if you share your thoughts on the Beta; your feedback is what drives our ongoing development.

Beta is different and we value you taking the time to try it out. Please take a look at the changes we've made in Beta and  learn more about it. Thanks for reading, and for making the site better!

IBM Touts Supercomputers for Enterprise

samzenpus posted more than 6 years ago | from the always-use-a-big-hammer dept.

Supercomputing 94

Stony Stevenson writes "IBM has announced an initiative to offer smaller versions of its high-performance computers to enterprise customers. The first new machine is a QS22 BladeCenter server powered by a Cell processor. Developed to power gaming systems, the Cell chip has also garnered interest from the supercomputing community owing to its ability to handle large amounts of floating point calculations. IBM hopes that the chips, which currently power climate modeling and other traditional supercomputing tasks, will also appeal to customers ranging from financial analysis firms to animation studios."

Sorry! There are no comments related to the filter you selected.

Oblig.. (4, Funny)

mrbluze (1034940) | more than 6 years ago | (#23413600)

But is it Vista capable?

Re:Oblig.. (1)

poopdeville (841677) | more than 6 years ago | (#23413622)

Only if you max out the RAM.

Re:Oblig.. (2, Funny)

mrbluze (1034940) | more than 6 years ago | (#23413634)

owing to its ability to handle large amounts of floating point calculations
Oh, so you can't run Excel on it.

Re:Oblig.. (2, Funny)

ozmanjusri (601766) | more than 6 years ago | (#23413704)

Oh, so you can't run Excel on it.

You can run 65,535 or 100,000 instances of it, whichever comes first.

Re:Oblig.. (2, Funny)

mrbluze (1034940) | more than 6 years ago | (#23413708)

You can run 65,535 or 100,000 instances of it, whichever comes first.
And then clean up the mess afterwards... ewwww!

Re:Oblig.. (1)

rubycodez (864176) | more than 6 years ago | (#23417034)

but without the expected Pentium floating point bugs our accountants would get confused

Re:Oblig.. (1, Funny)

RuBLed (995686) | more than 6 years ago | (#23413752)

And reversed the polarity...

Re:Oblig.. (1)

dwater (72834) | more than 6 years ago | (#23413770)

I wish people would stop using my name in such disgusting ways.

That's a Welsh pass time, and I'm English!

Re:Oblig.. (0)

Alwin Henseler (640539) | more than 6 years ago | (#23413776)

But is it Vista capable?
Well duh.. unless Vista gets ported to the Cell architecture, no (emulation aside).

Can't believe IBM doesn't mention that in the specsheet. Features: * Does NOT run Vista

Re:Oblig.. (1, Informative)

AKAImBatman (238306) | more than 6 years ago | (#23413898)

But is it Vista capable?

Wrong question. The correct question is: Will it run Java?

No? Then What The !@#$ have you been selling me WebFear^W^W^W^WSphere for?

BTW, for all you Eclipse lovers? NEVER install Rational Application Developer (RAD). IBM managed to take a half-decent product, add tons of suck to it, and make sure it was the SLOWEST application in the history of mankind. You need at least 4GB to run it at a decent speed. Which is pretty sad when you consider that Windows only goes up to 3.5... ;-)

Re:Oblig.. (1)

AKAImBatman (238306) | more than 6 years ago | (#23413904)

Err.. s/^W/^H/g

Sorry, too distracted by my new Mac. :-P

Re:Oblig.. (1)

Hal_Porter (817932) | more than 6 years ago | (#23414402)

Don't you have to escape the ^s?

i.e.
s/\^W/\^H/g

Java? Who cares? (5, Informative)

jpmorgan (517966) | more than 6 years ago | (#23414260)

Completely wrong. This is industrial supercomputing, not random web applications. The applications IBM is targetting Cell at are things like seismic analysis for oil companies, and this requires highly tuned implementations and specialised algorithms. This is expensive. It takes a lot of developer hours, and those developers are top-rung, very highly paid, because this stuff is very hard. And it costs a lot in hardware. Unfortunately for IBM, while Cell is fast, it's not fast enough to justify the cost for most companies. And it has a lot of competition from NVIDIA's Tesla platform, AMD/ATI's FireStream, and plain old clusters.

Re:Java? Who cares? (2, Insightful)

AKAImBatman (238306) | more than 6 years ago | (#23414328)

It's only completely wrong if we assume the story is a non-story. The markets you're talking about are traditional super-computer markets. The summary at least (too tired to read the full story today) is suggesting that IBM is targeting "Enterprise" customers. Which, by the classic definition of who those customers are, means that a lot of folks are going to be asking, "Where's the 0xDEADBE... Err.... Java?!?!"

Re:Java? Who cares? (2, Insightful)

Fred_A (10934) | more than 6 years ago | (#23415626)

Completely wrong. This is industrial supercomputing, not random web applications.
Unless they know something we don't... Could it be that Duke Nukem Forever is about to be released ?

Re:Java? Who cares? (1)

littlewink (996298) | more than 6 years ago | (#23419338)

It takes a lot of developer hours, and those developers are top-rung, very highly paid, because this stuff is very hard.


I've worked in that industry for over 20 years and I assure you that the developers are the same run-of-the-mill klutzes as are found in any shop in the USA. Most are definitely not Google material.

OTOH the software libraries are thoroughly picked over, having been developed over the last 50 years and scrutinized by many eyes. So it's pretty good code, but nothing like NASA's.

Re:Oblig.. (1)

Ambidisastrous (964023) | more than 6 years ago | (#23420798)

Of course it will. Java is huge in high-performance computing, oddly enough. The old-timers like using Fortran, C and C++, but only because they associate that with performance psychologically. Java's thread support wins on distributed-memory systems with large numbers of processors -- not as much as languages actually designed for parallelism, but better than C++ by a long shot.

It hits "good enough" on enough of these points -- familiarity, memory safety, concurrency support, comprehensive library, ubiquitous VM that lets hotshots run a better language (e.g. Scala or Clojure) on top of it -- that Java's probably the best bet for enterprise supercomputing.

Re:Oblig.. (1)

tallguywithglasseson (944783) | more than 6 years ago | (#23422744)

NEVER install Rational Application Developer (RAD). IBM managed to take a half-decent product, add tons of suck to it, and make sure it was the SLOWEST application in the history of mankind. You need at least 4GB to run it at a decent speed. Which is pretty sad when you consider that Windows only goes up to 3.5...
Not to mention it takes like 10 CDs and 5 hours to install...

Re:Oblig.. (1)

Durdenator (1288094) | more than 6 years ago | (#23413998)

Its just about as vista capable as your x86 box. Meaning it will crash every 3 seconds.

Re:Oblig.. (2, Informative)

DAldredge (2353) | more than 6 years ago | (#23414100)

What am I doing wrong because, hard as I try, I am unable to get my Vista box to crash. Please let me know what I have to do to get it to crash.

Re:Oblig.. (2, Informative)

Durdenator (1288094) | more than 6 years ago | (#23414122)

Well last poll showed 29% of vista crashes where from nvidia drivers, Try that.

Re:Oblig.. (1)

Hal_Porter (817932) | more than 6 years ago | (#23414460)

I have a Asus G1S Vista laptop with an 8600M GT and it's rock solid. Try the "Bioshock edition" drivers. That seems to be a good vintage.

Re:Oblig.. (2, Funny)

mrbluze (1034940) | more than 6 years ago | (#23414412)

What am I doing wrong because, hard as I try, I am unable to get my Vista box to crash. Please let me know what I have to do to get it to crash.
It obviously takes quite a number of instructions before Vista will crash. Perhaps your processor is too slow and you should upgrade to the above mentioned supercomputer.

Have you tried turning it on? (1)

Rexifer (81021) | more than 6 years ago | (#23415806)

You can't crash the car if you don't turn the key and back out of the driveway... so to speak.

You're SO lucky. (1)

DrYak (748999) | more than 6 years ago | (#23416974)

I am unable to get my Vista box to crash
You are *incredibly* lucky.
My brother has a fairly recent computer (AMD Phenom, ATI 3870X2, AMD 790 : All hardware has Vista drivers). Antivirus/FireWall/Updates. He's not exactly the kind to install every possible widget and explorer-bar.

Vista will systematically crash every couple of days, sometimes even twice a day.

Whenever we try googling the latest error message du jour to get some online help, AFTER THE FEW FIRST LETTERS in the search box of Firefox, GOOGLE ALREADY SUGGESTS the exact message DOWN TO THE ERROR CODES and other popular keywords such as "Vista" "blue screen" etc. (This gives you an idea how much such requests are popular to the point that google can auto suggest them based on the first few letters).

And then you hit "enter" and get something like "results 1-10 out of 35'678". (Among which, hundreds of pages of users complaining of the exact same bluescreen in Vista on Forum).

So either you are indeed incredibly lucky because there are an incredible number of Vista users experiencing massive amounts of bluescreens.
Or, there's indeed only a couple user to ever experience bluescreens in Vista, but they massively spam the whole internet about it. Somewhat I think that this explanation is less likely.

First, open the box... (1)

mkcmkc (197982) | more than 6 years ago | (#23428810)

and insert the enclosed CD-ROM into a PC... :-)

Re:Oblig.. (2, Funny)

Anonymous Coward | more than 6 years ago | (#23414140)

I think the proper obligatory question should be, "Can it run LCARS?"

Re:Oblig.. And I was going to say, (1)

davidsyes (765062) | more than 6 years ago | (#23421298)

"Make it so, Number One!"

The trend towards commodity hardware continues... (4, Interesting)

HockeyPuck (141947) | more than 6 years ago | (#23413682)

I've seen more than my share of traditionally big iron applicatoins (databases, data warehousing etc..) being moved off of specialized hardware (ibm p595s, sun e15k, HPSuperdomes etc..) being moved onto (or attempted to) commodity hardware. Management hears,

"We can replace that $2m server with 10racks of servers each $1k and after 3yrs we just throw them away and replace them with the latest and greatest x86 based hardware with 2x performance still $1k/server?
Now IBM wants to push highly specialized blades. Somewhere someone's saying, "How many x86 servers can we get for one Cell blade?"

Personally, I'm sick of managing farms of physical servers, and with the introduction of VMWare, I'm now managing 3x the number of machines (albeit virtual machines). Have an FTP server? Run that in it's own image. Also have a syslog server? Yet another virtual machine. I really hope this sells well. Maybe I can now play PS3 games in the datacenter.

Re:The trend towards commodity hardware continues. (4, Insightful)

explosivejared (1186049) | more than 6 years ago | (#23413884)

If specialized hardware returns to vogue, then there problems will crop up with the new specialized hardware. Dude face it, if you are a sysadmin, God will provide you with your share of things to complain about. It is the natural working order of things.

Re:The trend towards commodity hardware continues. (5, Insightful)

icebike (68054) | more than 6 years ago | (#23413896)

> Personally, I'm sick of managing farms of
> physical servers, and with the introduction of
> VMWare, I'm now managing 3x the number of
> machines (albeit virtual machines).

All those Virtual machines to do the same thing with 4 times the resources as one well configured Linux box. Tsk Tsk.

Oh, but don't you LOOK busy.....

Re:The trend towards commodity hardware continues. (3, Insightful)

Kadin2048 (468275) | more than 6 years ago | (#23414254)

> Have an FTP server? Run that in it's own image. Also have a syslog server? Yet another virtual machine.

Which as you have no doubt discovered, are sort of a PITA to administer because they're all in separate VMs. I suspect the next big thing in commodity server virtualization will be nice management interfaces and protocols that break down some of the management walls between VMs, while still leaving the more important parts of the virtual environment intact. And being able to change the hardware assigned to a VM on the fly will probably become more common, too. I'd give it 5 or 6 more years, and VMWare will probably have managed to reinvent the LPAR.

Gotta love how this stuff goes around in cycles. Anything cool today in microcomputers was probably boring people to tears 10 or 15 years ago on large systems. (Cf. multitasking, multiple users, parallel processing, network-oriented filesystems, virtualization, hypervisors...)

Re:The trend towards commodity hardware continues. (1, Interesting)

Anonymous Coward | more than 6 years ago | (#23414350)

I agree, VM has more development coming out. Lately they seem to be on the bleeding edge. I think the VM Ranger product will go by the wayside soon, as it will be incorportated into virtual center.

If you have looked at the new Virtual Center, you can now patch all types of virtual machines using the VM console, even while they are off. They are obviously reading the files of the virtuals to see what patch levels they are at, etc. It is only a small step from there to see what services are running and use VC to manage those. That will make the virtual server creep that most vm admins experience much more managable.

Re:The trend towards commodity hardware continues. (1)

makapuf (412290) | more than 6 years ago | (#23416080)

Anything cool today in microcomputers was probably boring people to tears 10 or 15 years ago on large systems.


OK then what are the interesting technologies currently happening in large systems ?

Re:The trend towards commodity hardware continues. (2, Informative)

Kamokazi (1080091) | more than 6 years ago | (#23414560)

Totally agree on virtualization...over-hyped BS.

Don't get me wrong, it definately had its uses and we do use it some where I work, but it's not the be-all end-all that VMWare and co. would like you to believe. Basically if you're not using it for:

a) Test environments

or

b) Services + apps that don't play well with others

then it is probably a waste of resources. Like you said, running a virtual machine for just an FTP server is absurd...."Let's use 50x the ram and hard drive space necessary for an FTP server!"

Re:The trend towards commodity hardware continues. (0)

Anonymous Coward | more than 6 years ago | (#23414636)

Your doing it wrong. No. I mean that.

ex. 1 Print servers. Zebra drivers dont play nice with other drivers. But a properly configured windows print server only needs 384mb of ram to run, and uses very little processor. A 10 gig OS drive, single proc, 384mb print server works perfectly. We use a handful of those for a 7000 user Hospital system. Much more effective than trying to use multiple physical machines.

ex 2. I run several MS SQL servers, all of which are no bigger than 1 proc, 2gb of ram virtuals. I have over 1000 connections to some of them. We run them inside a blade environment, and it provides us the uptime of a cluster(actually better) with much less hardware. I can add disk space on the fly in seconds, without ever touching a disk drive or raid controller. Technically, 1 proc/2Gigs is way below spec. But you have to learn to ignore the specs presented and translate it into the virtual world. When you do that, you do not waste resources.

VM can be very effective if it is properly administered. If it is not well thought out, you end up over provisioning very quickly.

Re:The trend towards commodity hardware continues. (1)

RiotingPacifist (1228016) | more than 6 years ago | (#23415460)

ex. 1 Print servers. Zebra drivers dont play nice with other drivers.

b) Services + apps that don't play well with others

ex 2. I run several MS SQL servers, all of which are no bigger than 1 proc, 2gb of ram virtuals.
A well configured box could host them with n*384 less overhead. I dont know about windows but under linux you can ad drivers to lvm and expand partitions.

But you have to learn to ignore the specs presented and translate it into the virtual world. When you do that, you do not waste resources. You still have the overhead of running windows for each SQL server, for SQL that may not be significant (its still 10% of your spec) & the overhead of unusable ram, HDD space, processing time.

Re:The trend towards commodity hardware continues. (1)

Kamokazi (1080091) | more than 6 years ago | (#23442542)

1) That fits my example of 'stuff that doesn't play nice'. (And I work with some Zebra drivers myself, shipping and some internal labeling, they are a bitch...I use 3rd party if I can, Seagull, etc.)

2) I'm no SQL guru by any means, but why can't you just run multiple databases in a single SQL instance? I guess it could be some proprietary crap with licensing I suppose, but I would think for mose things you can consolidate them. I know we have some DBs that share a single instance.

And my main point was that it's just over-hyped, not useless.

Re:The trend towards commodity hardware continues. (1, Insightful)

Anonymous Coward | more than 6 years ago | (#23414576)

Have an FTP server? Run that in it's own image. Also have a syslog server? Yet another virtual machine.


Ya know thats a pci requirement, right?

Re:The trend towards commodity hardware continues. (1)

Ed Avis (5917) | more than 6 years ago | (#23415210)

Have an FTP server? Run that in it's own image.
Why?

Re:The trend towards commodity hardware continues. (2, Interesting)

njcoder (657816) | more than 6 years ago | (#23416618)

I've seen more than my share of traditionally big iron applicatoins (databases, data warehousing etc..) being moved off of specialized hardware (ibm p595s, sun e15k, HPSuperdomes etc..) being moved onto (or attempted to) commodity hardware.
I'm honestly curious. How well does this really work out for databases and data warehousing?

One of the benefits I understand from going with one big-ass server is that the memory to pipelines between cpu and memory are much better than the ethernet/myrinet/infiband/whatever connections between cluster nodes.

Depending how you're going to be doing the clustering, you're either going to have some type of cluster fs or you're going to be using a shared NAS rather than SAN for the data. This is also going to be slower than local disks or a dedicated iSCSI connection.

From my understanding, the clustering technologies for databases aren't very good when you have a lot of writes to the db. Data warehousing is different, but I would think that something like an e15k or M9000 would be better for really big databases.

Things like 3d rendering and certain types of data analysis and modeling can be clustered easier. For these companies that need this type of service, it's probably better and cheaper to use a utility computing provider like Sun's Grid. Why pay the electric/cooling/sqft costs for running a supercomputer when you're only running reports quarterly/annually or if you're rendering on a per job basis that gets billed to the client. Easier and probably cheaper to bill for a service than to try and factor in your overhead for you're own rendering engine.

Add to that, you're now maintaining 10 racks of servers vs the equivalent of 1. Assume 40 nodes per rack with 2u for switches. That's about 40-80 amps of power per rack. Times 10 racks that's a lot of power, as well as heat that will work the AC harder. If I remember correctly an e15k is going to be somewhere around 50-80amps. All the extra power consumption and cooling is going to add up. Not to mention the space, cost of wiring up and testing all those nodes.

With 400 servers in 10 racks, even at 1k per server, between the racks and the labor to set it all up you got to be in the $750K-$1M ballpark. Even if you only really need 200 servers to get similar performance to an M9000 that's still not a huge savings compared to the operating costs, setup costs and simplicity.

In the long run does it really make sense?

I remember reading a comment about how PayPal is setup. They have a large number of linux servers as their front and middle tiers. In the back are 3 big Sun boxes that handle the databases. That to me seems like the right type of setup.

As for virtualization. A lot of people are just plain doing it wrong.

Flamage (4, Funny)

explosivejared (1186049) | more than 6 years ago | (#23413686)

Developed to power gaming systems, the Cell chip...

That's just what we need to introduce to enterprise computing, the flame wars and invectives of the console world.

Admin #1 cell totally rockzorz!!!11!!!1 u n00bs using virtualization are in the past
Admin #2 IBM SUCKS!!! YOU KNOW IT!!! YOUR WHOLE BUSINESS IS TOTALLY GAY FOR THEM!!!!

Re:Flamage (2, Interesting)

zappepcs (820751) | more than 6 years ago | (#23413876)

Actually, the Cell processor is a rocking piece of hardware. I'd like to see something like it added to motherboards as an optional coprocessor arrangement. Yes, I realize that the code/compilers would have to be redone, but I think that this is something that would make a huge difference in performance with little actual hardware/cost increase. It's a thought anyway.

Re: Flamage du jour? (2, Insightful)

shanen (462549) | more than 6 years ago | (#23414002)

A coprocessor for what? You do realize that each Cell already has 8 SPEs attached to it, and it's designed to be tightly paired with another Cell? It already delivers kind of ridiculous bang for the buck, so I'm curious what you're asking for.

(Usual disclaimer: I'm in the IBM food chain, but I don't speak for anyone and I don't know anything.)

Re: Flamage du jour? (3, Interesting)

zappepcs (820751) | more than 6 years ago | (#23414174)

Even though this might make me sound a bit off, we already have co-processors for video, network, etc. Why not go a bit further and specialize the hardware just a bit more. Let the Cell do all the real work and sandbox the user on the x86 cpu in a way that allows the user to be rather free in operation while the real work is done on the Cell processor in protected manner. That "should" be enough processing power to isolate the user completely from the tasks of the computer itself. The idea would be to sort-of create a mainframe/client environment where it would be nearly impossible for the user to accidentally introduce viri to the system.

The UI and i/o shelled through the x86 system. There are examples of this in some smaller embedded systems where system memory is separate from user memory etc. The details of this seem sketchy as I have not worked them out to any degree that would make the proposal sound workable thus far. I do know of examples where techniques like this are used to protect the 'system' while 'user agents' do what they want without the intrusion of security software at every turn. When the system is turned off, the user space is cleared. The protected system space is always protected.

Yes, that leaves room for infections on the Cell side to act like root kits as there is always some spot that is vulnerable, but it does offer a much more bullet resistant setup. The effects are not too different from working from a live CD all the time. Reboot and all is clean again, but with a more permanent and less inconvenient process. If you run some version of Linux/Unix on the client side, and strictly control the communications to the Cell side it becomes a much tighter box to try to squeeze a virus into. It may provide opportunity for the Cell side to monitor processes in the client/UI side meaning that keyloggers and such wou9ld become a thing of the past. In general, I mean to add horsepower by splitting system tasks from UI tasks and add a much stronger sandbox for the client to operate in, rather than continue lumping all the work on one cpu and letting security run in the same sandbox as the questionable software.

It's an idea... obviously I do not design motherboards or OSes for a living (IANACSPHD ??)

Re: Flamage du jour? (1)

zappepcs (820751) | more than 6 years ago | (#23414268)

arghh, I forgot one idea/thought. Within a workgroup in a business, say you have 15 workstations/desktops, and all of them running with the Cell/x86 setup. The Cell sides are all networked (on a separate network) as a grid and each x86/UI portion as clients on that grid. Suddenly you have HUGE computing potential with a common UI for each workstation that seems transparent and is scalable by adding more workstations.

Additionally, with SaaS and GoogleApps et al, you can see the value of the mainframe/client approach. It's basically the mainframe sitting inside a thick client box. Separating user and system spaces would fall along the same lines... or so it occurs to me.

Re: Flamage du jour? (2, Insightful)

shanen (462549) | more than 6 years ago | (#23414852)

Sorry, but I think you're off on a pretty wild tangent as to what the Cell processor is really about... However, I don't feel that I'm in a position to do more than to refer you to the publicly available documentation. There's quite a bit available about the architecture of the Cell and the interfaces to external processors and resources.

In some ways it might help to think about the Cell in comparison with a CDC 6600--but they probably stopped production on those before the average /. reader was born. Seems to be a relatively young crowd around here.

Re:Flamage (2, Informative)

maglor_83 (856254) | more than 6 years ago | (#23413974)

Admin #2 IBM SUCKS!!! YOU KNOW IT!!! YOUR WHOLE BUSINESS IS TOTALLY GAY FOR THEM!!!!
Gamer #1 But the 360, PS2 and Wii all use IBM chips.

Grrr lameness filter says I'm yelling.

Re:Flamage (1)

maglor_83 (856254) | more than 6 years ago | (#23414586)

And obviously that's meant to be PS3, not PS2.

Let me guess (3, Funny)

iminplaya (723125) | more than 6 years ago | (#23413690)

LCARS

no I'm not yelling, you dillhole

Re:Let me guess (3, Funny)

quanticle (843097) | more than 6 years ago | (#23413858)

Heh. Like with you, the first thing that popped into my head upon reading the headline was, "Wait, IBM make supercomputers for starships now? Since when?"

Re:Let me guess (1)

Bat Country (829565) | more than 6 years ago | (#23423814)

The first thing I thought of is "why the hell does an aircraft carrier need a supercomputer?"

Scotty was heard to say... (4, Funny)

Anonymous Coward | more than 6 years ago | (#23413882)

SCOTTY: Computer. Computer! Hello computer!


SAM PALMISANO: Uh, just use the keyboard.


SCOTTY: Keyboard... how quaint.

All I want to know is.. (1)

Durdenator (1288094) | more than 6 years ago | (#23413962)

Does it make toast too?

Yes, It Does Run Linux (4, Interesting)

Doc Ruby (173196) | more than 6 years ago | (#23413996)

From IBM's detailed press release [ibm.com] :

the QS22 boasts an open environment, utilizing the flexibility of Red Hat Enterprise Linux as the primary operating system and the open development environment of Eclipse.


That means that a PS3 running Linux [psubuntu.com] , even with its ridiculously low 512MB RAM, can be used as a $500 development platform for these CellBE BladeServers.

And, in turn, some QS22 SW might be usable on the PS3, if it can be ported to use the tiny RAM. Or if someone hooks an i-RAM bank to the SATA port as swap/ramdisk, using perhaps iSCSI over its Gb-e for storage.

Re:Yes, It Does Run Linux (0)

Anonymous Coward | more than 6 years ago | (#23419622)

The PlayStation 3 has 256 MB of XDR main memory and 256 MB of GDDR3 video memory for the RSX.

Re:Yes, It Does Run Linux (1)

Doc Ruby (173196) | more than 6 years ago | (#23419986)

There are hacks to use the VRAM as malloc()'able RAM for the CellBE.

And there's the i-RAM I mentioned.

Re:Yes, It Does Run Linux (1)

CronoCloud (590650) | more than 6 years ago | (#23423986)

That means that a PS3 running Linux, even with its ridiculously low 512MB RAM, can be used as a $500 development platform for these CellBE BladeServers.


Yes, Terrasoft promotes the PS3 as a cheap testbed/devstation for Cell blades, been doing so practically since the PS3 launch. They even sell individual PS3's with Linux pre-installed, as well as clusters.

If IBM their ServRAID controller right... (1)

Ang31us (1132361) | more than 6 years ago | (#23414050)

I would not be putting out fires every few weeks...do people who know hardware still pay an IBM premium for the same parts everyone else sells? I'm finding that their RAID QA leaves a lot to be desired.

IBM pricing (1, Insightful)

Anonymous Coward | more than 6 years ago | (#23414114)

IBM hopes that the chips, which currently power climate modelling and other traditional supercomputing tasks, will also appeal to customers ranging from financial analysis firms to animation studios.


Speaking as someone who architects, implements and maintains systems for animation studios, the sorts of dollars IBM ask for *anything* means they'll never appeal to animation studios.

There's only one Pixar and there's only one ILM. The other million animation studios out there don't have budgets even close to these guys, particularly considering the turn around on hardware (today's super cluster is tomorrow's pile of junk). Renderfarms will be staying with cheaper vendors (which also means white box for most) for some time to come yet.

If IBM want to appeal to that lot, they're going to have to knock a few zeros off the end of their quotes.

Re:IBM pricing (2, Interesting)

BiggerIsBetter (682164) | more than 6 years ago | (#23414166)

There's only one Pixar and there's only one ILM. The other million animation studios out there don't have budgets even close to these guys, particularly considering the turn around on hardware (today's super cluster is tomorrow's pile of junk). Renderfarms will be staying with cheaper vendors (which also means white box for most) for some time to come yet.
I understand your argument, but there's an upside. Maybe someone will put energy into Cell processor renderfarm software, so all the fiscally-challenged shops can buy a small rack of PS3s and go at it? IBM or not, the Cell is still a damn quick CPU for serious number crunching (say raytracing, etc).

Re:IBM pricing (-1, Troll)

Anonymous Coward | more than 6 years ago | (#23414306)

I am a senior manager and I control a lot of money. I have been waiting for some kind of expensive toy to purchase, so I can argue for it, make it happen, and advance my career by adding one major item to my "has done this" list. This QS22 IBM thing might be it, what do you think? I am too busy to follow the industry, so I count on you college kids to tell me if you think this would work.

Re:IBM pricing (1)

njcoder (657816) | more than 6 years ago | (#23417116)

Why don't small animation studios ditch the render farm and use a utility computing service? Sun's Grid [network.com] digipede [digipede.net] 3Tera [3tera.com]

Which Enterprise? (1)

rossdee (243626) | more than 6 years ago | (#23414120)

[ ] CVAN65
[ ] NX1
[ ] NCC1701
[ ] NCC1701 A
[ ] NCC1701 B
[ ] NCC1701 C
[ ] NCC1701 D
[ ] NCC1701 E

Re:Which Enterprise? (5, Funny)

compro01 (777531) | more than 6 years ago | (#23414250)

N-C-C one seven oh one. no bloody A, B, C, or D.

Re:Which Enterprise? (1)

spacefiddle (620205) | more than 6 years ago | (#23417026)

i am so not a ST fanboi and mostly don't like the kidgloves treatment they gave themes they could have explored so much better (in TNG, anyway... commence flamage), but that was one of the best episodes of anything ever. mr. doohan was one of the most underutilized actors yet seen :(.

Re:Which Enterprise? (1)

MiKM (752717) | more than 6 years ago | (#23414558)

You forgot "Voyager".

Re:Which Enterprise? (4, Funny)

RuBLed (995686) | more than 6 years ago | (#23414598)

You make it sound like it's a bad thing...

Re:Which Enterprise? (0)

Anonymous Coward | more than 6 years ago | (#23416714)

I was always under the impression that "Voyager" (NCC-74656) was Voyager... not Enterprise (NCC-1701). Last I checked 1701 != 74656... except for, maybe, the rare case of an extremely small value of 74656.

M5? (1)

SmokeSerpent (106200) | more than 6 years ago | (#23414278)

I don't know if that's a good idea...

This is a Failure (4, Informative)

InvaderXimian (609659) | more than 6 years ago | (#23414374)

First off, I should admit that the Cell is quite a good processor, however it has shortcomings. Many shortcomings.

I'll begin first by listing the positives:
It has very good single precision floating point performance
Very high bandwidth on chip (25 GB/s)
With the e[nhanced]D[ouble]P[recision] addition to it IBM is adding for Los Alamos' Road Runner, it should be even better.
IBM Developerworks is a useful resource for programmers
It's a processor that gives the academic community a chance to publish more papers

The negatives:
Far too expensive
Floating point performance, single or otherwise, is useless for most enterprise work (no point running a database on a Cell)
The things that makes the Cell fast and unique are the SPEs (SIMD processors) are useless for servers*(I'll elaborate below)
Developing software to make use of the SPEs is time consuming and difficult. Orders of magnitude more so if your algorithm isn't suited for being split across SPEs
Computer Science or Engineering students who are doing research in this field have to read redundant papers that reword the IBM Cell manual (plagarism?) for 70% of it and what they actually did with the Cell for about 10%. 20% is of course left for citations.
You can't actually play games on a Cell. The PS3 games use them for physics engines, sound or such; the video is done on a video card.
Worst of all, YOU HAVE TO GET DATA FROM MAIN MEMORY TO SPU'S CACHE YOURSELF! (also known as Local Store, 256KB)

Now to talk about the SPEs... They are what makes the Cell tick. If the Cell didn't have them, it would just be an old Mac processor. (IBM Power 4 was it? I've trying to avoid research papers on the Cell) The graduate students (a year away from a PhD) that I was observing who were doing developing a scientific application on the Cell.. when we summed their year in development, they essentially told me it was a pain in the ass. Why? Because you have to get data from main memory to local store. Imagine if you had to get data from memory to your x86 processors L2 or L1 cache. (No, you don't actually want that. Trust me, you don't.) Scientists don't want to develop applications on this platform. Researchers like it because they can publish papers talking about it. This is why Los Alamos' Road Runner might be a flop. Bye bye to $100m of US tax payer money.

Back to the discussion with the grad students... I then came up with the idea that the Cell would be perfectly suited to an Asterix server. Why? Because sometimes Asterix needs to convert from one format to another when audio codecs differ. This is a perfect SIMD application, except we have a problem. While the Cell does have great on chip bandwidth, it has very poor Power5 processor (PPU) to SPU latency*. There is a research paper out there that puts PPU initiated memory transfers to SPU latency at about 4 microseconds. Compare this to SPU initiated transfer to SPU latency of about a tenth of a microsecond. Huge difference. You can't really avoid this unless you want to develop something horrendous that avoid the PPU but that's unlikely. The PPU would run the actual server and then it would need to notify the SPU of data, either by sending message (slow) and the SPU getting the data itself (2*slow=2slow) or sending the data and a message (slower).

It will be about 5 years until good software development tools are written for this architecture that will address most of the issues. This is assuming that IBM doesn't ditch it, which seemed likely before this announcement.

Oh, by the way, the QS22 doesn't support a hard drive which means you need fast NFS. The cost of ownership is more than just the cost of single blade.

Re:This is a Failure (1, Interesting)

Anonymous Coward | more than 6 years ago | (#23415560)

OK, so you have observed people facing problems getting performance out of the Cell.

These people are not alone, but it looks like they are not the most "advanced" when it comes to programming the Cell.

More and more games developers are starting to "appreciate" the Cell.

No, the Cell is not meant to be to take on the X86 market. So running a MySQL DB on the Cell is not a good idea.

IBM are targetting specific markets/customers.

One project where you see the tremendous power of the Cell is with the Folding project.

Perhaps you should talk with the people at Stanford, instead of just complaining about the Cell.

And why would IBM ditch the Cell, when Toshiba is even preparing a laptop with the Cell as a copro?

Re:This is a Failure (1)

HockeyPuck (141947) | more than 6 years ago | (#23415974)

Oh, by the way, the QS22 doesn't support a hard drive which means you need fast NFS. The cost of ownership is more than just the cost of single blade.
Ever heard of SAN boot? 4Gb or 8Gb access to SAN disk isn't fast enough for you? Give yourself some time and you'll see FCoE (single 10GigEthernet interface carrying both IP and FibreChannel on a single interface) http://en.wikipedia.org/wiki/FCoE [wikipedia.org] and there are already shipping products http://www.cisco.com/en/US/products/ps9670/index.html [cisco.com]

Re:This is a Failure (2, Insightful)

pohl (872) | more than 6 years ago | (#23416450)

Your post could have been much shorter if you wrote it this way: "This will fail because there are tradeoffs for using it. I know this because I'm in college and other college kids said it was hard to use. I tried to come up with an idea, but there are those darned tradeoffs again. No tool that has ever succeeded has had tradeoffs or was hard to use."

Nonsense (2, Insightful)

Funk_dat69 (215898) | more than 6 years ago | (#23417388)

What a rambling bunch of nonsense. While the 'negatives' you post are true, they in no way make the design a failure.

The chip was not designed to replace any type of general purpose processor in the Enterprise space. Cell is a large parallel FP number cruncher. So, no, databases or file servers and other solution that basically involves moves data around between storage and the network, is not a good fit.

Real time Video/Audio transcoding and distribution on the other hand, if needed in an Enterprise shop somewhere, would be a perfect fit, if done correctly.

The PPU is *not* a POWER5, and while DMA transfers do have inherent latency, it is not really a problem at all. Especially, if you schedule your DMAs correctly, you can almost completely recover the latency times. Message sending from PPU and SPU has no latency (their on the same chip!).

All your bitching about difficulty in programming, are grounded in fact, but compare these techniques to programming in Cuda or writing fake Opengl programs in order to take advantage of a GPUs floating point power and you'll realize that while Cell is difficult, at least it will run straight C - even on the SPUs - and you can optimize from there however much you want.

You seem misinformed.

Re:This is a Failure (1)

grazier (118138) | more than 6 years ago | (#23418058)

...
Oh, by the way, the QS22 doesn't support a hard drive which means you need fast NFS. The cost of ownership is more than just the cost of single blade.
This is misleading. It is always more than the cost of a single blade when dealing with bladecenters; Chassis, network modules, SAN modules, redundant power, etc... You only need one server, buy a rack mount or desktop workstation. Bladecenters are for when you need many that stay up as long as possible. 7-14 is where it shines.

A FiberChannel card connected to external storage unit will work fine. Bladecenters were designed for complete redundancy and internal HDs (regardless of RAID config) are crude forms of redundancy in the enterprise these days. Far better to have multiple paths to a sharable storage unit that can have the LUNs automatically shifted to another blade in event of failure AND can be mirrored to another unit somewhere else in case of catastrophe.

Fibre Channel. (0)

Anonymous Coward | more than 6 years ago | (#23429150)

Sorry...

Re:This is a Failure (1)

rockmuelle (575982) | more than 6 years ago | (#23419052)

As one of those "graduate students" who got some good publications and a thesis out of the Cell, I can tell you that the processor is definitely good for that. :)

I want to take issue with one of your negatives, however. The programmable local store is one of the best features of the Cell and actually simplifies programming for performance significantly. Once you get used to moving data and code from main memory to local store (admittedly a foreign concept for most programmers weened on Java/Ruby/PHP/etc), you wonder why you ever relied on the cache in the first place.

With the local store, there is no guessing when your data is in L1 or playing prefetching tricks (which is where most development time is spent on performance critical code). You load the data you need and it's there. No surprises. The model is actually very similar to working on a client/server system (think servlet loading data from database), but in this case the client is a core on your processor and the server is your main memory.

On your point about the PPE: it wasn't designed to be used heavily by application code. The PPE was designed to run essential OS services. All application code was supposed to be developed for the SPEs. Unfortunately, IBM never really publicized this point and a lot of developers got burned trying to use the PPE to dispatch work to SPEs.

-Chris

Re:This is a Failure (0)

Anonymous Coward | more than 6 years ago | (#23423962)

Floating point performance, single or otherwise, is useless for most enterprise work (no point running a database on a Cell)

Not all enterprises are created equal. A company which does image processing and render farm work, large scale video archive conversion from digital security camera formats, or audio signal processing is constantly starved for powerful floating point systems.

Something like this would be an affordable compressor and archiver rig for a large callcenter which wanted to log its calls for legal purposes too.

Fast wavelet or fractal compression of tens of thousands of minutes of audio or video per day sounds good to me.

As far as the engineering problems with programming are concerned, if it works now (albeit less efficiently than it could) and your business can grow into it as better versions of your software become available, does it really matter how much of a nightmare it is to develop for? The end user (the business) only cares if it works fast, costs little, and lasts a long time.

Cell's Problem (0)

Anonymous Coward | more than 6 years ago | (#23414376)

GFlops:

CPUs > Cell > GPUs

Flexibility/Ease of programming:

CPUs Cell GPUs

Seriously, I'm not really sure who's supposed to be Cell's target audience, but it's not anyone concerned with return on investment.

Re:Cell's Problem (0)

Anonymous Coward | more than 6 years ago | (#23420842)

I think that should be GFlops: CPUs Less than Cell Less than GPU's

Really, editors. (1, Troll)

tzjanii (1170411) | more than 6 years ago | (#23414484)

Seriously, editors, this is Slashfuckingdot. If there is a single place on the whole internet where you need to be more careful about throwing around the capitalized word 'Enterprise', point me there, please.

For future reference, this also goes for the terms 'Serenity','Falcon', and 'Nebuchadnezzar.'

Oh darn (2)

Haoie (1277294) | more than 6 years ago | (#23415030)

I came in here thinking this was about Star Trek, it was even tagged as such.

Oh well.

Not many people will be able to afford such high end PCs. I myself can't, that's for sure.

And here I got my hopes up... (2)

Glenn Rubenstein (1288436) | more than 6 years ago | (#23415196)

My first thought was "Star Trek" story. That's what happens when I read the phrase "The Enterprise."

To quote Patton Oswalt, "My geekiness is getting in the way of my nerdiness."

Of course it's for enterprise... (2, Insightful)

800DeadCCs (996359) | more than 6 years ago | (#23416006)

Of course it's for enterprise...
because you and I can't afford a $10k blade let alone the rack slot farm to put it in.
http://www-03.ibm.com/systems/bladecenter/hardware/servers/qs22/index.html [ibm.com]
(browse and buy)

The updates to it, DDR2 and better floating point, those are nice.
Now if they would make a small system, somewhere between a mini-mac and a shuttle pc in size,
DVD/RW, 2 Gb lan ports, some usb, basic video, 2 or 4 gigs ram (or empty memory slots we could fill ourselves) and a place for a SATA drive
not using it as a co-pro, but as the main CPU, with a linux install disk and the SDK,

compare my imagined specs with the PS3 crippled cell (7 out of 8 working, and 1 of them against you for system security).
If it were in the same price range, that'd be nice, maybe a way to get more people to experiment/develop for it.

What's so great about Cell? (1)

DoofusOfDeath (636671) | more than 6 years ago | (#23416102)

When I was doing some scientific computations on it, I found that on a per-core, per-clock-cycle basis, it performed 32-bit floating calculations about as fast as an Intel Core2 E4400 processor.

Back when x86 chips had at most two cores, that made the Cell look pretty good. But...

- Intel chips now have 4 cores, with 8 on the horizon.

- It was a PITA to program the Cell, because it's not a
    shared-memory architecture. The cores need to communicate
    via message passing. And in the current generation,
    the software running on most of the cores (the "SPEs")
    could take up 256KB for object code AND working data.

- The Cell's 64-bit floating point performance sucked ass
    compared to Intel's.

So in my experience, it just wasn't that worthwhile a processor. Does anyone know what the big deal is?

Re:What's so great about Cell? (1)

Funk_dat69 (215898) | more than 6 years ago | (#23417498)

Um, you do realize the SPU have 128-bit registers, right? Which means you can do 4 32-bit floats at a time.

I'm not sure what kind of 'scientific' computations you were doing, but it sounds like you were not using the SPUs properly.

Re:What's so great about Cell? (1)

DoofusOfDeath (636671) | more than 6 years ago | (#23417612)

Um, you do realize the SPU have 128-bit registers, right? Which means you can do 4 32-bit floats at a time.

I'm not sure what kind of 'scientific' computations you were doing, but it sounds like you were not using the SPUs properly.

We were doing acoustic ray tracing. The problem we had was that any branching that occurred quickly led to different rays needing different *kinds* of operations. So it really wasn't practical or efficient to keep on tracing four or even two rays concurrently using the SIMD operations.

But even then, what about x86's SIMD operations? I imagine that a Core2's registers are only 64 bits wide, compared to Cell's 128-bit registers. But I wouldn't expect that to be a compelling reason to use Cell.

Re:What's so great about Cell? (1)

Ant P. (974313) | more than 6 years ago | (#23425650)

Does anyone know what the big deal is?
It's mostly hype. The Cell is the Power-architecture equivalent of a Pentium 4 Extreme. Big numbers on paper, but not much use outside of a pissing contest (or applications like a game console where you can drown out the sound of the required cooling system).

aer (0)

Anonymous Coward | more than 6 years ago | (#23416988)

I came in here thinking this was about Star Trek, it was even tagged as such.è±èf [hk-beauty-centre.com]

Replies (1)

InvaderXimian (609659) | more than 6 years ago | (#23418214)

Have any of you actually worked on a Cell (blade) before? Sounds like you've only read about it. To HockeyPuck: Nowhere did I say that SAN boots are slow, but I did mention it'll increase the cost of owning a blade. Don't forget the cost of the BladeCenter as well. Funk_dat69: The design is not a failure. IBM belief in thinking that it would succeed in the enterprise is however, a huge failure. The "college kids" I've worked with would be the same skilled programmers developing code for the oil industry or certain financial applications. They've spent quite a while working on techniques to make it easier, but apparently it's still a pain in the ass. Scientific/acedemic computing is the Cell main industry, outside of gaming. The tradeoffs you need to make for the Cell are far too many to make it successful. The PPU is a Power4, not Power5. I did say it was a Power4. You can't schedule the DMA all that well for on-demand loads such as given by servers. Also, the Power4 is a fairly weak processor so it can't schedule for all 8 SPUs independently without a noticable performance penalty. While GPGPU is a pain as well, it's on a different level than the Cell. We're focusing on the Cell here. I know Cuda and friends aren't great. grazier: Just cleaning up some confusion. Not every on /. is a blade expert.

I'm sure... (0)

Anonymous Coward | more than 6 years ago | (#23419924)

...Spock will be excited to hear that!
Check for New Comments
Slashdot Login

Need an Account?

Forgot your password?