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!

Linux May Need a Rewrite Beyond 48 Cores

CmdrTaco posted more than 3 years ago | from the it's-all-stacking-blocks dept.

Operating Systems 462

An anonymous reader writes "There is interesting new research coming out of MIT which suggests current operating systems are struggling with the addition of more cores to the CPU. It appears that the problem, which affects the available memory in a chip when multiple cores are working on the same chunks of data, is getting worse and may be hitting a peak somewhere in the neighborhood of 48 cores, when entirely new operating systems will be needed, the report says. Luckily, we aren't anywhere near 48 cores and there is some time left to come up with a new Linux (Windows?)."

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

Original Source and Actual Paper (5, Informative)

eldavojohn (898314) | more than 3 years ago | (#33748820)

It appears that the problem, that affect the available memory in a chip when multiple cores are working on the same chunks of data, is getting worse and may be hitting a peak somewhere in the neighborhood of 48 cores, when entirely new operating systems will be needed, the report says.

Seriously? You picked that over my submission?

I submitted this earlier this morning I guess my submission was lacking [slashdot.org] . But if you're interested in the original MIT article [mit.edu] and the actual paper [mit.edu] (PDF):

eldavojohn writes "Multicore (think tens or hundreds of cores) will come at a price for current operating systems. A team at MIT found that as they approached 48 cores their operating system slowed down [mit.edu] . After activating more and more cores in their simulation, a sort of memory leak occurred whereby data had to remain in memory as long as a core might need it in its calculations. But the good news is that in their paper [mit.edu] (PDF), they showed that for at least several years Linux should be able to keep up with chip enhancements in the multicore realm. To handle multiple cores, Linux keeps a counter of which cores are working on the data. As a core starts to work on a piece of data, Linux increments the number. When the core is done, Linux decrements the number. As the core count approached 48, the amount of actual work decreased and Linux spent more time managing counters. But the team found that 'Slightly rewriting the Linux code so that each core kept a local count, which was only occasionally synchronized with those of the other cores, greatly improved the system's overall performance.' The researchers caution that as the number of cores skyrockets [slashdot.org] , operating systems will have to be completely redesigned [slashdot.org] to handle managing these cores and SMP [wikipedia.org] . After reviewing the paper, one researcher is confident Linux will remain viable for five to eight years without need for a major redesign."

I don't know, guess I picked a bad title or something?

Luckily we aren't anywhere near 48 cores and there is some time left to come up with a new Linux (Windows?).

Again, seriously? What does "(Windows?)" even mean? As you pass a certain number of cores, modern operating systems will need to be redesigned to handle extreme SMP. It's going to differ from OS to OS but we won't know about Windows until somebody takes the time to test it.

Re:Original Source and Actual Paper (0)

Anonymous Coward | more than 3 years ago | (#33748850)

U mad?

Re:Original Source and Actual Paper (1)

bindoeve (1912478) | more than 3 years ago | (#33749050)

Understandable. This is nothing but inferior compared to his submission.

Re:Original Source and Actual Paper (0, Informative)

Anonymous Coward | more than 3 years ago | (#33748904)

I’d just like to interject for a moment. What you’re refering to as Linux, is in fact, GNU/LInux, or as I’ve recently taken to calling it, GNU plus Linux. Linux is not an operating system unto itself, but rather another free component of a fully functioning GNU system made useful by the GNU corelibs, shell utilities and vital system components comprising a full OS as defined by POSIX.

Many computer users run a modified version of the GNU system every day, without realizing it. Through a peculiar turn of events, the version of GNU which is widely used today is often called “Linux”, and many of its users are not aware that it is basically the GNU system, developed by the GNU Project.

There really is a Linux, and these people are using it, but it is just a part of the system they use. Linux is the kernel: the program in the system that allocates the machine’s resources to the other programs that you run. The kernel is an essential part of an operating system, but useless by itself; it can only function in the context of a complete operating system. Linux is normally used in combination with the GNU operating system: the whole system is basically GNU with Linux added, or GNU/Linux. All the so-called “Linux” distributions are really distributions of GNU/Linux.

So this blog needs to be renamed to the GNU/Linux Hater's Blog. Have a nice day.

Re:Original Source and Actual Paper (3, Funny)

RCGodward (1235102) | more than 3 years ago | (#33749292)

Don't bother checking the box, RMS, we know who it is.

Re:Original Source and Actual Paper (2, Funny)

spazdor (902907) | more than 3 years ago | (#33749434)

Fuck, dude, we hurd you the first time. and "GNU Plus Linux" is terrible marketing.

Re:Original Source and Actual Paper (4, Funny)

VorpalRodent (964940) | more than 3 years ago | (#33748932)

What does "(Windows?)" even mean?

I read that as saying "Windows is the new Linux!". Clearly the submitter is trying to incite violence in the Slashdot community.

Re:Original Source and Actual Paper (4, Interesting)

Dragoniz3r (992309) | more than 3 years ago | (#33748940)

Oh look, CmdrTaco published yet another story with a poorly-written, hypersensationalist summary! Par for the course.

Re:Original Source and Actual Paper (4, Interesting)

NevarMore (248971) | more than 3 years ago | (#33749236)

The thing is eldavojohn practically *is* an editor for /. , just check out his submission page. Despite having such a high UID he's got a solid reputation, a good writing style, and offers good commentary on a wide variety of topics.

Re:Original Source and Actual Paper (1, Insightful)

davev2.0 (1873518) | more than 3 years ago | (#33749476)

Good summaries do not offer commentary. Save the commentary for the comments.

Re:Original Source and Actual Paper (5, Interesting)

klingens (147173) | more than 3 years ago | (#33748962)

Yes it is lacking: it's too long for a /. "story". Editors want small, easily digested soundbites, not articles with actual information.

Re:Original Source and Actual Paper (5, Informative)

eudaemon (320983) | more than 3 years ago | (#33748964)

I just laughed at the "we aren't anywhere near 48 cores" comment - there are already commercial products with more than 48 cores now. I mean even a crappy old T5220 pretends to have 64 CPUs due to the 8 CPU, 8 thread design.

Re:Original Source and Actual Paper (3, Informative)

WinterSolstice (223271) | more than 3 years ago | (#33749116)

Got a pile of AIX servers here like that:
http://www-03.ibm.com/systems/power/hardware/780/index.html [ibm.com]

I was kind of wondering about the "modern operating systems" comment... I think he meant "desktop operating systems".
Many of the big OS vendors (IBM, DEC (now HP), CRAY, etc) are well beyond this point. Even OS/2 could scale to 1024 processors if I recall correctly.

Re:Original Source and Actual Paper (3, Informative)

Anonymous Coward | more than 3 years ago | (#33749376)

OS/2's SMP support is a joke. I'm sure that somewhere in that tangle is a comment like "up to 1024 processors". But it's as relevant as a sticker on a Ford Cortina warning not to exceed the speed of sound.

Officially the SMP version of OS/2 "Warp Server" supported 64 processors. In practice anything other than an embarrassingly parallel task would see rapidly diminishing returns after just a couple of CPUs. The stuff that this article is moaning about, that Linux doesn't do well enough on 48 CPUs? OS/2 doesn't even attempt it, the official docs just say to "avoid" such things. This test case on 48 CPUs on OS/2 would just leave the OS constantly thrashing trying to move pages from one CPU to another, and no work being done.

Now maybe if OS/2 had been a huge success, and IBM were now the dominant OS vendor on the desktop, there'd be a 1024 CPU version of OS/2 today. But in our reality, where OS/2 support was gradually abandoned and handed over to an underfunded little independent outfit, it sucks on SMP.

Re:Original Source and Actual Paper (2, Informative)

Skal Tura (595728) | more than 3 years ago | (#33749144)

nevermind quite an standard server, a dual xeon 6core HT... total reported CPUs is 24, and it's quite a lot used and nothing special.

Re:Original Source and Actual Paper (4, Insightful)

Perl-Pusher (555592) | more than 3 years ago | (#33749182)

Core !=CPU

Re:Original Source and Actual Paper (3, Interesting)

TheRaven64 (641858) | more than 3 years ago | (#33749234)

And it's worth noting that the most common application for that kind of machine is to partition it and run several different operating systems on it. Solaris has already had some major redesign work for scaling that well. For example, the networking stack is partitioned both horizontally and vertically. Separate connections are independent except at the very bottom of the stack (and sometimes even then, if they go via different NICs), and each layer in the stack communicates with the ones above it via message passing and runs in a separate thread.

However, it sounds like this paper is focussing on a very specific issue: process accounting. To fairly schedule processes, you need to work out how much time they have spent running already, relative to others. I'm a bit surprised that Linux actually works as they seem to be describing, since their 'change' was to make it work in the same way as pretty much every other SMP-aware scheduler that I've come across; schedule processes on cores independently and periodically migrate processes off overloaded cores and onto spare ones.

There are lots of potential bottlenecks. The one I was expecting to hear about was cache contention. In a monolithic kernel, there are some data structures that must be shared among each core and every tim you do an update on one core you must flush the caches on all of them, which can start to hurt performance when you have lots of concurrent updates. A few important data structures in the Linux kernel were rewritten in the last year to ensure that unrelated portions of them ended up in different cache lines, to help reduce this.

Even then, it's not a problem that's easy to solve at the software level. Hardware transactional memory would go a long way towards helping us scale to 128+ processors, but the only chip I know of to implement it (Sun's Rock) was cancelled before it made it into production.

Re:Original Source and Actual Paper (2, Informative)

Anonymous Coward | more than 3 years ago | (#33748966)

I don't know, guess I picked a bad title or something?

No. Your summary was too long.

Seriously, the purpose of a summary is not to include every last fact and detail mentioned in the article; it's to give the reader enough information to decide whether reading the full article is worth it. Don't try to put everything in there.

Re:Original Source and Actual Paper (2, Informative)

Dahamma (304068) | more than 3 years ago | (#33749282)

the purpose of a summary is not to include every last fact and detail mentioned in the article; it's to give the reader enough information to decide whether reading the full article is worth it.

If you think a summary can actually help get a /. reader to RTFA, you must be new here...

Re:Original Source and Actual Paper (4, Informative)

BeardedChimp (1416531) | more than 3 years ago | (#33749300)

The purpose of an editor is to edit any submissions to make them ready for print.

If the summary was too long, the editor should have got off his arse rather than wait for the summary that fits the word count to come along.

Re:Original Source and Actual Paper (3, Insightful)

Wonko the Sane (25252) | more than 3 years ago | (#33749414)

Your summary was too long.

Yes, but the submission that got accepted has a bullshit headline.

Of course "Linux May Need to Continue Making Incremental Changes Like It Has Been Doing For The Last Several Years To Scale Beyond 48 Cores" doesn't draw in as many clicks.

Re:Original Source and Actual Paper (1)

TheLink (130905) | more than 3 years ago | (#33749418)

Yes his summary was a bit long.

But the purpose of many slashdot summaries seem to be to generate more comments about errors in the summary, or due to misunderstanding of the summary, or the summary just being crap. A bit like trolling for hits ;).

Re:Original Source and Actual Paper (2, Informative)

X0563511 (793323) | more than 3 years ago | (#33749422)

I've seen longer stories about lamer things get published...

Re:Original Source and Actual Paper (1)

characterZer0 (138196) | more than 3 years ago | (#33749054)

(Windows?)

I thought he was implying that we will also need to come up with a new Windows.

Re:Original Source and Actual Paper (3, Interesting)

Skal Tura (595728) | more than 3 years ago | (#33749100)

Scare piece.

Your submission wasn't scaring enough. From your submission, it seems that it's not that big of a deal and rather easy solution. This submission makes it sound like linux kernel needs a complete rewrite ground-up, as in starting from scratch.
Plus yours was a bit long and lots of details.

Re:Original Source and Actual Paper (0, Flamebait)

poetmatt (793785) | more than 3 years ago | (#33749138)

toot your own horn much?

way to link your own article, which I will avoid now.

Re:Original Source and Actual Paper (2, Insightful)

Anonymous Coward | more than 3 years ago | (#33749408)

Wow, really just wow you sir are the cream of the crop! /sarcasm

the OP has a very valid point, i come to read about technology news on slashdot not scare pieces with little or no information or value, his post was far superior in every respect and yet got passed over for this garbage post. And you devalue his point further by not even giving him the time of day, way to go asshole.

Re:Original Source and Actual Paper (1)

jpmorgan (517966) | more than 3 years ago | (#33749148)

Well, Windows historically did have problem scaling beyond a fairly small number of processors. So with Windows 7, Microsoft replaced the original NT system executive with a new system called MinWin. Microsoft claims MinWin efficiently scales to 256 cores: http://tech.slashdot.org/tech/08/11/02/0130253.shtml [slashdot.org]

Re:Original Source and Actual Paper (1, Redundant)

wastedlife (1319259) | more than 3 years ago | (#33749394)

You've been misinformed, the NT executive is still alive and kicking [wikipedia.org] :

MinWin is not, in and of itself a kernel, but rather a set of components that includes both the Windows NT Executive and several other components that Russinovich has described as "Cutler's NT".[16]

It's all still NT, Windows 7 is just NT version 6.1. I guess "6.1" doesn't have the same ring to it as a whole number. Will Windows 8 be NT 6.2, or will they move the version up to NT 7.0?

Re:Original Source and Actual Paper (1, Informative)

Anonymous Coward | more than 3 years ago | (#33749214)

The effect is otherwise known as "Amdahls Law", well documented by Gene Amdahl in 1967. Is this news at all?

Re:Original Source and Actual Paper (1)

poet (8021) | more than 3 years ago | (#33749268)

Of course they picked it over yours. Yours is intelligently written and has an expectation that people will understand what you are talking about.

Unfortunately, this is Slashdot.

Re:Original Source and Actual Paper (2, Insightful)

interkin3tic (1469267) | more than 3 years ago | (#33749276)

I don't know, guess I picked a bad title or something?

Slashdot: dramatically overstated news for nerds... since that seems to be the evolution of news services for some reason?

I'm working on a submission: Fox news just had a bit about the internet, I'm assuming that their headline is something like "WILL USING OBAMANET 'IPv6' KILL YOU AND MAKE YOUR CHILDREN TERRORISTS?"

Linux needs a rewrite anyway. (-1, Troll)

Anonymous Coward | more than 3 years ago | (#33748844)

Shit sucks.

Re:Linux needs a rewrite anyway. (1)

arkane1234 (457605) | more than 3 years ago | (#33749216)

What part?
Be specific, since I've been using it since '95 and it's only gotten better.
Granted, it's a little more bloated than back then, but hey you get that when you have sixteen billion subsystems.

Re:Linux needs a rewrite anyway. (0)

Anonymous Coward | more than 3 years ago | (#33749288)

The part that requires the user to grow a neckbeard and masturbate to lolicon.

Barrelfish (1, Informative)

Anonymous Coward | more than 3 years ago | (#33748858)

This is exactly why people are doing research on Barrelfish (http://www.barrelfish.org/).

Re:Barrelfish (1)

ciderbrew (1860166) | more than 3 years ago | (#33748894)

The more I see photos like that, the more I think computers make you go bald.

Linux already runs on thousands of cores (2, Insightful)

Chirs (87576) | more than 3 years ago | (#33748882)

SGI has some awfully big single-system-image linux boxes.

I saw a comment on the kernel mailing list about someone running into problems with 16 terabytes of RAM.

Re:Linux already runs on thousands of cores (4, Interesting)

Gaygirlie (1657131) | more than 3 years ago | (#33748936)

It's not the case of not being able to do such, but instead about where there are performance regressions. Of course it's possible to run Linux on multiple hundreds of cores, but it seems that after 48 cores there is a performance regression and thus all those cores don't benefit as much as they could. That is the issue here.

Re:Linux already runs on thousands of cores (4, Informative)

DrgnDancer (137700) | more than 3 years ago | (#33749274)

I thought this as well, but after more carefully reading the article, I *think* I see what the problem is. It's not really a problem with large numbers of cores in a system, so much as a problem with large numbers of cores on a chip. Since the multicore chips share caches (level 2 cache is shared, level 1 cache isn't IIRC, but I could be wrong) it's actually cache memory where the issue lies. I've worked on single system image SGI systems with 512 cores, but those systems were actually 256 dual core chips. That works fine, and assuming well written SMP code performance scales as you'd expect with number of cores.

Re:Linux already runs on thousands of cores (2, Interesting)

Gaygirlie (1657131) | more than 3 years ago | (#33749390)

Since the multicore chips share caches (level 2 cache is shared, level 1 cache isn't IIRC, but I could be wrong) it's actually cache memory where the issue lies.

That's what I thought too, but after thinking it a bit more I'd dare to claim it's both a hardware and software issue. Too small cache of course does cause issues like the researchers noticed but it's mostly because the method how memory accesses and cache is handled in software that makes it such a big issue. Rethinking the approach how kernel handles such could very well minimize the impact even in cases where there is not all that much cache available.

Of course, I'm not an expert in SMP or multi-core systems so I could have verily misunderstood it.

Re:Linux already runs on thousands of cores (3, Interesting)

TheRaven64 (641858) | more than 3 years ago | (#33749324)

SGI has some awfully big single-system-image linux boxes.

Not really. SGI has big NUMA machines, with a single Linux kernel per node (typically under 8 processors), some support for process / thread migration between nodes, and a very clever memory controller for automatically handle accessing and caching remote RAM. Each kernel instance is only responsible for a few processes. They also have a lot of middleware on top of the kernel that handles process distribution among nodes.

It's an interesting design, and the SGI guys have given a lot of public talks about their systems so it's easy to find out more, but it is definitely not an example of Linux scaling to large multicore systems.

Error in their math (5, Funny)

El_Muerte_TDS (592157) | more than 3 years ago | (#33748898)

They have an one-off error in their math, it's actually 9 times a 6 core CPU. So, at 42 cores a rewrite is needed.

Re:Error in their math (1, Funny)

Anonymous Coward | more than 3 years ago | (#33749284)

Dude, nobody makes jokes in base 13!

not anywhere near 48 cores? (0)

Anonymous Coward | more than 3 years ago | (#33748900)

Not anywhere near 48 cores? Stick 4 AMD Magny Cours Opterons (12 cores each) in a quad socket motherboard and you will have 48 cores. Not that uncommon.

Not close yet? (1)

BWJones (18351) | more than 3 years ago | (#33748902)

Dunno... I am typing this on a system with 12 cores and 24 virtual cores. And the GPU has somewhere around 1600 cores... Other systems I've worked with have hundreds to thousands of cores so I think we are pretty close...

Seriously though, these issues have been known for a while but will have to trickle down to desktop OSs to deal with caching and shared memory.

Re:Not close yet? (1)

blueg3 (192743) | more than 3 years ago | (#33749278)

And the GPU has somewhere around 1600 cores...

It doesn't. Its resources also aren't managed by your operating system, so how Linux behaves in multicore environments is irrelevant to your GPU's operation.

Enough (2, Funny)

wooferhound (546132) | more than 3 years ago | (#33748928)

640 cores ought to be enough for anybody . . .

Re:Enough (1)

Ukab the Great (87152) | more than 3 years ago | (#33749164)

All you need is a single monolithic chip that's 640 times bigger than a regular core.

Re:Enough (1, Funny)

Anonymous Coward | more than 3 years ago | (#33749190)

Obligatory xkcd [xkcd.com]

Re:Enough (0)

Anonymous Coward | more than 3 years ago | (#33749358)

lol i dont see why people cant be happy with a quad it gets the job done if you need more than 4 cores you should just shoot yourself

What are they talking about (4, Insightful)

pclminion (145572) | more than 3 years ago | (#33748948)

Can somebody please explain what the fuck they are actually talking about? They've dumbed down the terminology to the point I have no idea what they are saying. Is this some kind of cache-related issue? Inefficient bouncing of processes between cores? What?

Re:What are they talking about (1, Funny)

Anonymous Coward | more than 3 years ago | (#33749110)

... from the original MIT article it sounds like there is lock contention on shared memory reference counts. ... but I'm making that up.

Simple solution (0)

Haxamanish (1564673) | more than 3 years ago | (#33748970)

Just write a little AWK script to replace evey occurence of "48' in the source code by, say, 256 or 1024.

Only Linux? (3, Interesting)

Ltap (1572175) | more than 3 years ago | (#33748990)

It looks like TFS was written by a Windows fanboy; why mention Linux specifically when it is a general problem? Why try to half-assedly imply that Windows is more advanced than Linux?

Re:Only Linux? (0, Redundant)

jpmorgan (517966) | more than 3 years ago | (#33749070)

Probably because Microsoft rewrote the NT kernel for Windows 7, to eliminate the kinds of problems this study discovered:

http://www.zdnet.com/blog/microsoft/windows-7-to-scale-to-256-processors/1687 [zdnet.com]

Re:Only Linux? (1, Redundant)

Jorl17 (1716772) | more than 3 years ago | (#33749372)

No, their rewrite is also subject to to this issue. Go publicize Windows somewhere else.

Re:Only Linux? (0)

Anonymous Coward | more than 3 years ago | (#33749458)

And yet SGI Altrix UV line goes up to a total of 2048 cores per system, and runs Linux.

The only reasonable explanation I can find for this is that Linux treats multiple CPUs and multiple cores on the same CPU differently. Windows may be just as affected.

Re:Only Linux? (2, Informative)

wastedlife (1319259) | more than 3 years ago | (#33749464)

They did not "rewrite the kernel" for 7. They updated the code, just like every other piece of software normally does when it moves from version to version. Rewriting the kernel implies that they tore it down and started over, which is most certainly not true. Vista/2008 is NT version 6.0, 7/2008 R2 is NT version 6.1, not a rewrite.

Re:Only Linux? (3, Insightful)

Attila Dimedici (1036002) | more than 3 years ago | (#33749178)

Having read eldavojohn's post that summarizes the article, it appears that the reason to pick out Linux specifically is because that is the OS that the writers of the paper actually tested. Since Windows uses a different system for keeping track of what various cores are doing it is likely that Windows will run into this problem at a different number of cores. However, until someone conducts a similar test using Windows we will not know if that number is more or less than 48.

Re:Only Linux? (1)

KarmaMB84 (743001) | more than 3 years ago | (#33749342)

As a post already pointed out, Microsoft modified the NT kernel to scale to 256 cores with Windows 7 and Windows Server 2008 R2.

based on a 1970s OS and language (1, Troll)

peter303 (12292) | more than 3 years ago | (#33748992)

UNIX and C were great in their days. But perhaps not in the meg-core era.

Re:based on a 1970s OS and language (2, Insightful)

Anonymous Coward | more than 3 years ago | (#33749112)

UNIX and C were great in their days. But perhaps not in the meg-core era.

So, what is better in your opinion? Java? Or maybe even ruby? Oh yes, that would be great. Run-time OS reflection through kernel drivers implemented as ruby modules.

Too bad CPU's don't come with built-in ruby interpreters.

Re:based on a 1970s OS and language (0)

Anonymous Coward | more than 3 years ago | (#33749290)

Hah! that's a good one, thanks for making my day! Hey can you tell me where you've bought the low id?

Re:based on a 1970s OS and language (4, Insightful)

geekoid (135745) | more than 3 years ago | (#33749294)

Hahaha. Oh arrogances from ignorance, how I loath you.

64 cores (2, Interesting)

hansamurai (907719) | more than 3 years ago | (#33749000)

At my last job we had a bunch of Sun T5120s which housed 64 cores. So yeah, we are "anywhere near 48".

Re:64 cores (0)

Anonymous Coward | more than 3 years ago | (#33749254)

No, it had 64 Virtual CPU's. The T5120's have 8 cores and 8 threads per core so Solaris reports 64 vCPUs. Only one thread can run at a time on a core (the rest are parked waiting for a page fault to be serviced...maybe other reasons too).

Re:64 cores (1)

Splab (574204) | more than 3 years ago | (#33749266)

They aren't talking about physical CPUs, they are talking about cores within any given CPU, which I doubt you where having 64 of. While this may seem like a moot point, when scheduling it is very important to keep track of what's going where since you are usually at least sharing L3 cache on a single die.

Question is, what to do... (0)

Anonymous Coward | more than 3 years ago | (#33749004)

...with the other 46 cores we are not using. Most people still do one thing at a time and only need a couple of cores at best. Three if you throw in Windows anti-malware software. :)

Re:Question is, what to do... (0)

Anonymous Coward | more than 3 years ago | (#33749056)

Well let's see... The user's Facebook Firefox tab takes up one core. There's probably a keylogger running on another core. And various other forms of malware are probably slamming the other 40+ cores.

Re:Question is, what to do... (1)

MachineShedFred (621896) | more than 3 years ago | (#33749352)

Or, if you have an application that isn't written for a system from the 1990's, it will spawn multiple threads and use all the cores to do that one task much faster. You know, things like video compression which is nothing but massive amounts of math - nobody [apple.com] does [live.com] that [pinnaclesys.com] on a consumer [adobe.com] level [apple.com] at all.

Jaguar? (2, Insightful)

MrFurious5150 (1189479) | more than 3 years ago | (#33749008)

Cray [wikipedia.org] seems to have addressed this problem, yes?

Re:Jaguar? (0)

Anonymous Coward | more than 3 years ago | (#33749224)

No, because the Jaguar isn't a "monolithic" computer like a personal PC. It's a series of nodes - "Each XT5 compute node contains dual hex-core AMD Opteron 2435 (Istanbul) processors".

Re:Jaguar? (0)

Anonymous Coward | more than 3 years ago | (#33749360)

Did you even read the wikipedia article you posted? Jaguar contains 26,520 nodes. Each of those only have up to twelve cores apiece.

48 cores? (4, Funny)

drunkennewfiemidget (712572) | more than 3 years ago | (#33749036)

I'm still waiting for Windows to work well on ONE.

too bad, (1)

Major Downtime (1840554) | more than 3 years ago | (#33749094)

i was hoping to see Crysis 2 running on Linux

seeing as Linux does 10240 cores already, WTF? (4, Interesting)

r00t (33219) | more than 3 years ago | (#33749102)

No kidding. SGI's Altix is a huge box full of multi-core IA-64 processors. 512 to 2048 cores is more normal, but they were reaching 10240 last I checked. This is SMP (NUMA of course), not a cluster. I won't say things work just lovely at that level, but it does run.

48 cores is nothing.

Re:seeing as Linux does 10240 cores already, WTF? (0)

Anonymous Coward | more than 3 years ago | (#33749338)

They mean cores per socket.

Re:seeing as Linux does 10240 cores already, WTF? (1)

varmittang (849469) | more than 3 years ago | (#33749406)

But that is over multiple processors in the whole machine. They are talking about a single processor having 48 cores, not the whole machine over multiple processors.

Priorities (0, Troll)

jeff4747 (256583) | more than 3 years ago | (#33749170)

Perhaps they should worry about getting Flash to work without stuttering before they worry 'bout 48 cores.

...unless the plan is to use 48 cores to make Flash work.

Re:Priorities (1)

silas_moeckel (234313) | more than 3 years ago | (#33749470)

Somebody cares about flash video working? Only people I can think of is Adobe flash is there closed DRM ridden POS it's in there interests to make it work. If they start using open video streaming protocols they would not have a problem.

48 isn't far off at all. (0)

Anonymous Coward | more than 3 years ago | (#33749176)

My new HP DL-385 G7 has a 12 core AMD processor. A four fold increase is not far off.

Sun E10Ks were at 72 cores over a decade ago (0)

Anonymous Coward | more than 3 years ago | (#33749186)

And guess what? With near linear scaling.

These [oracle.com] have 512.

These [oracle.com] have 256.

Appears to be a Linux problem.

Windows is good (0)

Anonymous Coward | more than 3 years ago | (#33749194)

http://channel9.msdn.com/shows/Going+Deep/Mark-Russinovich-Inside-Windows-7/

Windows 7 can scale to 256 processors.

Who uses that (2, Funny)

MSDos-486 (779223) | more than 3 years ago | (#33749210)

http://xkcd.com/619/

Obligatory xkcd reference (2, Interesting)

zill (1690130) | more than 3 years ago | (#33749232)

Do they have support for smooth full-screen flash video yet? [xkcd.com]

My Ubuntu 10.04 system still can't play embedded youtube videos. At least Adobe provided a work-around by adding a "play on youtube" option in the right click context menu.

Why? (1)

geekoid (135745) | more than 3 years ago | (#33749270)

Why aren't we even close to 48 cores? There have been chip with 16 or more cores. Why are we still mulling around 6 cores?

I suspect it's a fab issue. In that so may ships on a die with such small lines means a lot more flawed chips and bad wafers.

Of course, Who will rewrite Linux? We could never recaptures it's unique origins.

We passed that point (0)

Anonymous Coward | more than 3 years ago | (#33749310)

I seem to recall seeing operating systems running on more than 48 cores. In fact, doesn't Linux power some of the giant super computers with 64+ cores?

OpenIndiana?? (0, Troll)

andersenep (944566) | more than 3 years ago | (#33749320)

Why write a new Linux when Solaris already does such a fine job scaling to large numbers of cores/threads? OpenIndiana is just getting off the ground, but it's open source, free, and works now.

48 Cores in 1U (2, Informative)

kybur (1002682) | more than 3 years ago | (#33749350)

I'm not affiliated with Supermicro in any way, but they have four 1U serverboards designed for the 12 core opterons, so that's 48 cores in a 1U server. I'm guessing that Supermicro is not the only vendor of quad opteron boards supporting the latest chips. There are most likely quite a few of these in use by real people. Anyone want to speak up?

I know from personal experience that the socket F opterons performed very poorly in an 8 way configuration compared to the previous generation (socket 940 gen). I ran multiple tests on dual core chips (885s, I think), back in 2006 or 7 where I'd get nearly double the performance in going from a quad configuration to an 8 way configuration, but with the socket F breed of chips, there was no performance boost at all, it was like the clock speed was being cut in half and all the threads took twice as long to complete. I saw this behavior again and again, and the motherboard manufacturer that I was testing the chips with told me that it was an issue with the chips themselves. I think this is the reason why 8-way opteron systems are very rare now.

that's crazy (2, Funny)

Punto (100573) | more than 3 years ago | (#33749354)

Nobody's every going to need more than 640 cores

how is this news? (3, Insightful)

dirtyhippie (259852) | more than 3 years ago | (#33749382)

We've known about this problem for ... well, as long as we've had more than one core - actually as long as we've had SMP... You increase the number of cores/CPUs, you decrease available memory thruput per core, which was already the bottleneck anyway. Am I missing something here?

Infinite (1)

Chameleon Man (1304729) | more than 3 years ago | (#33749392)

It's amazing how we live in a world involving an infinite, non-discrete numeric system yet the computers we construct are always bound by some finite, discrete limitation.

Yeah, on Windows, 47 for viruses (0, Troll)

cjonslashdot (904508) | more than 3 years ago | (#33749420)

If a Windows machine had 48 cores, 47 of them would be running viruses, spyware, and anti-virus/anti-spyware software and one would be running the user's applications.

Large linux systems today have 3072 processors (1)

kroyd (29866) | more than 3 years ago | (#33749424)

(or more, probably)

http://lkml.org/lkml/2010/7/22/252 [lkml.org] is a fun post on the Linux-Kernel list about missing caching of ACPI tables leading to 20 minute boot times. I get that problem every day! (I wish :P)

It is a pretty safe bet that you don't have to worry about Linux and more than 48 cores, as it is the OS of choice for a lot of the top supercomputers and OS research in general. Of course, applications which can take advantage of such systems is another problem, but that is hardly a Linux problem.

Hundreds of cores on today's Linux (1)

Florian Weimer (88405) | more than 3 years ago | (#33749426)

One SGI Altix [sgi.com] version comes with 2,048 cores running a single image.

The benchmarks in the paper are a bit suspicious because they avoid disk I/O. tmpfs is used instead, which may skew results significantly. Surprisingly, they do not describe the architecture of the test machine, but perhaps I've missed that. They suggest that a workload which does not spend much time in the kernel cannot have scaling issues caused by the kernel, which seems rather dubious to me.

I don't understand... (2, Informative)

Anonymous Coward | more than 3 years ago | (#33749462)

I'm trying to understand the point of this article..Do we really need a new paper to say that centralized memory bandwidth is at some point a limiting problem in an SMP environment? Isn't this why we have NUMA?

If you want to go after linux internals like the BKL more power to you but that horse left the stable a long long time ago as well.

You could talk about the software problem in dealing with decentralized memory access, synchronization, scalable algorithms...etc but this is all likely something needing to be addressed in application space rather than at the kernel where this paper seems to focus.

There are no shortage of huge single system image linux systems with thousands of processor cores and not a single one of them use SMP architecture. They are all NUMA based (decentralized memory access).

other kernels (1)

TheSHAD0W (258774) | more than 3 years ago | (#33749480)

Are there other open-source OSes which are better suited to more parallelism? The Hurd, perhaps?

Load More Comments
Slashdot Login

Need an Account?

Forgot your password?