Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
The Internet

Transatlantic Gigabit Gaming.. err, Research 117

Posted by timothy
from the can-beat-up-your-network dept.
YourHero writes: "Science Technology And Research Light-Illuminated Gigabit High-Performance Transit. All hail acronyms. U Illinois at Chicago has hooked up to SURFnet (Netherlands) at 2.5Gbps, with plans to go to 10Gbps and hook up Canada, Asia and other parts of Europe. StarLight as its called makes a monster gaming ... err. I mean 'real-time, multi-site virtual reality.' Looks like they've been racking up killer ping times for a few weeks now.
This discussion has been archived. No new comments can be posted.

Transatlantic Gigabit Gaming.. err, Research

Comments Filter:
  • I dunno about you but I don't like killer pings. And why would you time them? "Woo Hoo! I can send a ping 'o' death to Amsterdam in 1 nanosecond!"
  • by ekrout (139379) on Thursday December 06, 2001 @01:38PM (#2666064) Journal
    STARLIGHPT (pronounced "starlipped"?) just doesn't have a nice ring to it...

    ;-)
  • STARLIGHT (Score:2, Insightful)

    by jargoone (166102)
    Science Technology And Research Light-Illuminated Gigabit High-Performance Transit. All hail acronyms.

    Yeah. More like "All hail naming a project to fit a stupid acronym".
  • Now they need to hook this up to the NCSA's Cave, and get some low-ping, immersive Quake deathmatch going. *drool*
  • I always wondered what a 1 Mbps mp3 stream would sound like....
    -Ben
  • Wait a minute... I thought this was for research.
  • Slashdot (Score:3, Funny)

    by fliplap (113705) on Thursday December 06, 2001 @01:41PM (#2666098) Homepage Journal
    All that and they still can't survive the slashdot effect?
  • by Dixie_Flatline (5077) <vincent.jan.gohNO@SPAMgmail.com> on Thursday December 06, 2001 @01:44PM (#2666129) Homepage
    That no matter how big the pipe is, you can't reduce ping times past a certain limit, right? You've got the whole 'speed of light' thing tripping you up. There is a certain latency that can't be defeated, no matter how many gigabits your pipe is.
    • Nonsense. Just broadcast over the Sub-Etha Net. Updates through the whole galaxy (and multiple unknown dimensions) instantaneously.
      • It's not that easy. although M. Jacques Toussaint Benoit (de l'Herault) and M. Biat-Chretein did pioneering research into snail-telegraphs (c.f. Parsian newspaper Presse, 10/27/1850), most of the modern research into the union of galvanic, magnetic and symaphetic fluids as applied to the Etherial fluid was done as recently as 1996 by Alexis Hastings (I presume that her works were later published in the Paradigma, but a transcript of her work is available in Horizon, WW4012). Although I know that the cupric sulfate crystals are necessary to raise the imbalance to necessary levels, the dosage requirements (so as to prevent toxicity to the snails) is not recorded.

        Regardless, I am sure that more research is necessary before this is useful as a reliable medium for communication, let alone the transmission of IP Datagrams.

        --
        Evan

    • hmmmm well perhaps then we should begin work on a quantum entagled ethernet.

      ZERO PING..... whoo yeah!!!
      • Well, actually, if the Quantum Physics laws still hold, it is not possible to transmit information in FTL speeds. The phenomenon of qubit teleportation exists, but to achieve it you have to transmit a bit of information through classical means, and so you get bitten by the light speed limit.

        For more information, see Quantum Computation and Quantum Information, by Michael A. Nielsen, Isaac L. Chuang

        Carlos

        • actualy Quantom entanglement is quite interesting in that 2 particles can communicate instantly.....when you observe a quantom entangled particle, the other particle, regardless of distance has the exact same measurments and therefor communication occured faster than light (in a spooky quantom sort of way.)
    • by Anonymous Coward
      A ping of 1 milli-second is ALOT slower
      than the speed of light , the routing
      a switching and packet opening and
      re-encapsulation is your time killer .

      optical networking is the key to beating
      this . it routes the light based on
      frquency spectrum to a MAN of a cetain city
      without having to open and rebuild the packets.

      AKA assign a certain spectrum to a certain city,
      and all DWDM optical routing sends that light
      frequency to that one city , there it is
      opened , and local transport over Sonet/ATM
      arranged .

      Alot of companies have yet to do this, but
      this model is on its way to most if not all.

      onsiterepair@yahoo.com
      • A ping of 1 milli-second is ALOT slower than the speed of light , the routing a switching and packet opening and re-encapsulation is your time killer.

        Hmm... let's do some math: Say we have 10 000 km from europe to the USA and light travels with about 330 000 km/s ->

        10 000 km / 330 000 km/s = 0.030s = 30ms

        And that is just one way... take it twice and you have your ping.

        Maybe you could make a hole through the earth to shorten the way ;)

        Routers etc. should not add much to the latency... maybe 1 or 2ms, not more.

        But seriously... it will never happen to do "real realtime" worldwide applications or mega-fast-fat-networks with low latency. It is time to invent "light2" - man, light is so slow :)

    • Unless you use tachyons, then your packets would arrive before you send them.
      • I think this would only be from the perspective of the tachyons...that is, they would arrive younger than when you sent them, but not before.

        Now, if you held the tachyons in place and accelerated the earth to faster than lightspeed and returned to the tachyons, then you might be there before you left.

        In fact, if we get back early enough I'm gonna save that cat from the box.
    • That no matter how big the pipe is, you can't reduce ping times past a certain limit, right? You've got the whole 'speed of light' thing tripping you up. There is a certain latency that can't be defeated, no matter how many gigabits your pipe is.
      Most high-performance networking applications (besides Quake or Counter-Strike :-p) are bandwidth-limited, not latency-limited. It's alright that it takes a bit 67 milliseconds to get from Chicago to Amsterdam as long as you can push a gigabyte per second down the pipe.
      • And of course I agree. But the news post of course made reference to gaming and ping times. I'm just pointing out that it doesn't matter that they made a bigger pipe...pings don't necessarily drop.
        • True true. And now you know (to first-order) the minimum latency you could ever expect to get from Chicago to Amsterdam. That is, until we start laying fiber through the Earth's core rather than along the surface... ;-)
    • The propagation delay created by the gates in the circuitry of the switches and routers affects how long it takes a signal to get from one point to another. Newer transistor technologies allow these to run a bit faster which is why gigabit/s is possible. Also, if a system is buffering/compressing/error checking any large quantity of the data before it is sent again this creates yet another delay.
    • I'd like to point out to you people that modded me down after being modded up, that I was the first one to post about the light-speed delay. Have you read the moderator guidelines? Favour modding up instead of modding down. You're wasting your points.
    • Well ya of course that's a big set back, not! have you ever got on say a dual-t3 at 3:45am and pinged texas over mci routers? you get a under 25ms ping consistantly, if everything was 100gig/sec you'd get 50ms everywhere in north-america even at peak times.. and I'm sure the 5sec all accross the east coast would make alot of ppl smile! :)
  • ping times? (Score:5, Informative)

    by alienmole (15522) on Thursday December 06, 2001 @01:46PM (#2666151)
    Looks like they've been racking up killer ping times for a few weeks now.

    Someone correct me if I'm wrong, but ping times on high speed links (with no particularly slow routers or switches in the way) are primarily limited by the speed of light and the distance travelled.

    So no matter how fast the bandwidth of this connection, if it's between the US and Europe, the ping times aren't going to be a whole heck of a lot better than they were before - the distance is the main limiting factor, and it's a pretty "hard" limit too, according to my old pal, Einstein.

    • Re:ping times? (Score:5, Informative)

      by cnkeller (181482) <cnkeller AT gmail DOT com> on Thursday December 06, 2001 @01:57PM (#2666258) Homepage
      Someone correct me if I'm wrong, but ping times on high speed links (with no particularly slow routers or switches in the way) are primarily limited by the speed of light and the distance travelled.

      You're basically right on. This [stanford.edu] is my favorite essay about latency.

    • So the next performance gain will be running cable straight through earth, rather than along the arc of the surface, to reduce the physical distane.

      It'd be worth it.

      Really.
      • Re:ping times? (Score:2, Interesting)

        by haystor (102186)
        The force of gravity effectively propogates instantly, thereby reducing ping times to the speed at which you can create and destroy matter.
        • The force of gravity effectively propogates instantly

          Gravity propagates at the speed of light. If you knocked the Sun away with a big hammer, the Earth would continue to orbit around it for 8 minutes, before noticing something strange.
          • If gravity were to propogate at merely the speed of light the force exerted on the earth by the sun would no longer be perpendicular to path of earth around the sun. This would result in the earth spiraling outward away from the sun.

            Think about it. An object moving in a circle at a constant rate requires a force to be constantly acting on it perpendicular to the path its is currently traveling, not perendicular to the path it was traveling 8 minutes ago.
    • In the best world there would be pure light transport. But today the routers/switches used inside the networks are using ASICs and PNPs (Programable Network Processors) so there are alot of light-to/from-electronic conversion taken place at the input and output of the routers/switches. Pure optical switches based on MEMS mirror technology or bubble technology are on the way but we are'nt there yet for production use. Imagine to have a 10 Gigabit Ethernet optical network card in your computer with fullblow out on the pure-optical MAN.
    • Yes, and ping times aren't the most important factor. Admittedly the Medical Faculty might not be the most important bandwidth-hog, but I recall that they're always talking about 'remote medicine'. Well, here I am sitting on my SURFNet connection, @10Mbps to my desktop, from a 100Mbit switch (GBit backplane), with a 38MBit link to the main Rotterdam SURFNet connection.

      Not that I ever have speed problems here, but still, I bit more spreading of the bandwidth might be nice. They're upgrading the network here, so I should have 100Mbps to my desktop soon, with cables rated at 1Gbit.
    • Will we ever see good all world online gaming for games like q3?

      Let's assume that there is a central server located in city x and two gamers in city y and z respectively. The information about y's doings to get to z tunneled through the server have to travel (in worst case) a distance of 40 Mm. That means a theoretical latency of about 200 ms. With a peer-to-peer network we could get a maximun latency of 100 ms instead. In the practical case this is at much higher.

      For the sake of online gaming I propose that we should pack all the people of the world in for example Europe. The theoretical ping time between worst case connections will be dramatically reduced.
  • by Mark4ST (249650) on Thursday December 06, 2001 @01:47PM (#2666163) Homepage
    ...plans to go to 10Gbps and hook up Canada, Asia and other parts of Europe

    Since when did Canada and Asia become "parts of Europe"? I'm always the last to know this stuff. *sigh*

  • Difference in pings? (Score:4, Interesting)

    by FortKnox (169099) on Thursday December 06, 2001 @01:47PM (#2666169) Homepage Journal
    Honestly, is there THAT big of a difference in 80 ping and 5 ping (as far as games are concerned)?

    I haven't been to a LAN party in ages, because all my gamer friends and I have broadband, and its just as fun to play over the internet than on a LAN (and we don't have to lug our machines around).

    Having a ping lower than 80 isn't all that with the games out today. Now its all about video cards, memory, and CPU power. Broadband is making people happy and satisfied (for the time being).
    • Depends on the game (Score:5, Informative)

      by Ted V (67691) on Thursday December 06, 2001 @01:58PM (#2666278) Homepage
      It honestly depends on the game. I can't notice any difference between 0ms and 25ms of ping, but you can feel a little bit between 50ms and 0ms in an online game. I did work with the Unlagged Q3 [planetquake.com] mod as part of unlagging my own mod, Art of War [planetquake.com]. It turns out that 50ms of time is about half of a player bounding box in Q3. So lets say a player is moving from left to right and you have a 50ms ping. If you are aimed just a little bit to the left of center, but still lined up with their bounding box, you will miss with 50ms ping and hit with 0ms ping. That's pretty important. With unlagging, I notice a huge improvement in all my hitscan weapons, even in the sub 100 ping range.

      That said, Packet Loss is far worse than a high ping, and high pings don't mean as much in slower paced games like RTS games.

      -Ted

      • by Anonymous Coward
        The same is noticable in Half-life since they released their new netcode (probably about a year or so ago now), with the lag compensation enabled hitscan weapons do far better, even at low pings. Before they added that, I noticed that my chances of using hitscan weapons against other players dropped significantly once my ping rose above 50ms.

        Of course, the problem with lag compensation is that it causes weird side-effects, like people hitting others around corners and without facing the proper direction, and it only gets worse as the two players involved have higher pings.

        Even worse, still, is when a router hiccoughs and your ping spikes to 80 from 20, or when your cable modem service provider decides they can squeeze three more people onto your node, and one of those people decides they need to run Napster 24/7.
        • by grammar fascist (239789) on Thursday December 06, 2001 @04:15PM (#2667145) Homepage
          I'm the author of Unlagged Quake 3 [planetquake.com], so naturally I'm biased. For those of you who don't know, it's a server-side hit-scan lag compensation mod, which means there's no extra cheating and nothing to download to play. It works by remembering every player's position - and when instant-hit weapon hit tests need to be done, it tests with the players in the positions they were in at the time of attack as opposed to the time the command was actually received.

          Of course, the problem with lag compensation is that it causes weird side-effects, like people hitting others around corners and without facing the proper direction, and it only gets worse as the two players involved have higher pings.

          You're right, but those side-effects are much less noticeable than people like to make them out to be. The reason these "problems" sometimes seem so big is that people generally sit around and theorize about the idea rather than test it. Another issue is that there are always, always, always inconsistencies when you try to make anything real-time with latency involved. The game designer has to choose which inconsistencies he thinks he and his audience can deal with.

          So let's take a look at lag compensation vs. no lag compensation in the inconsistency category to see how they stack up. First, lag compensation:

          • You sometimes, but rarely, notice that you get shot when you think you're hidden
          • You sometimes, but even more rarely, notice that the person who shot you isn't currently aiming at you (this also happens with no lag compensation, but to a lesser extent)
          • HPB's sometimes, and still rarely, seem to make impossible shots - like they seem to rail you right after you hit a bouncepad

          And now, without lag compensation:

          • HPB's can line up a rail shot perfectly and miss
          • Game balance is skewed by the fact that four of the weapons are nearly useless to anyone pinging over 100


          I stand by my assertion that the inconsistencies in hit-scan lag compensation are very rare. I've been testing it for weeks now, and I've only noticed them a few times - like three. Just about everyone else who's tried it and given feedback has said about the same thing - and usually, they love it.

          The only people who really hate it seem to be those who consider their broadband edge to be more important than fairer play. ("Fairer" because the low-ping player will always have the advantage.) They'll avoid it like the plague.

          Even worse, still, is when a router hiccoughs and your ping spikes to 80 from 20...

          Hit-scan lag compensation takes care of that, too, so it isn't just for HPB's.

          My point is that people should try it out before they let the supposed problems dissuade them from it. I've just finished a server listing page [planetquake.com] that updates every five minutes. Check it out. Just remember not to aim ahead of your target with the railgun.
      • by iconian (222724)
        You are absolutely right when you say you notice a difference between 80 ms and 25 ms lag. From the human visual system perspective, 80 ms is a LONG time. Ask any psychologist and they'll tell you that many low level perceptual processes occurs under 20 ms.

        Here's an indirect piece of evidence: Do you notice the difference between a motion video clip played at 15 frames/sec versus 30 frames/sec? At 15 frames/sec, one frame is displayed every ~67 ms. At 30 frames/sec, one frame is displayed every ~33 ms. The fact that most people are able to see a big difference between 15 and 30 fps suggests that most people will also notice a difference between playing a first person shooter with 30 ms lag vs 80 ms lag.

        I would say when you have at least get below 15 ms to not notice any difference.
    • oh your so outta the loop...

      First of all as reported several times, only 5% of the people connected to the internet have broadband, thus telling us one thing... we sure aren't close to having a fast broadband network.. now of course that percentile is higher in canada and the us.. however I myself am still stuck on 56k, and being in Ontario, right close to London, I must say that since London was the second testing city in canada to get cable modems, over 5years ago, I am glad to see that within 2months I'll finally have cable.

      Going on.. 80ping and 5ping, yes they make a remarkable difference, not to mention have you tried gaming between 8-12pm EST lately? most broadband connections show wonderful 300ms delays all over the place, and there is few services out there that provide good, solid, consistant pings anywhere in north-america. As well, I dunno what kind of lan parties you goto but, 80ping vs. 5 ping and a lan vs. internet is a huge diff., lan parties are extremely fun, you should check out a big exciting, fun lan party, it would sway your opinion no doubt about it. (checkout the url I have listed above)

      Games out today? even the newest games still are ping dependant, sure I must admit their getting better, but I've played alot of games, and nothing has ever been "good" or to an "acceptable level" for 56k play... the best I've experienced is Q2, with tweaked settings, and FPS caps enabled.. mind you I've tried plenty of games online, and I will be very happy when cable arrives, well providing it works too. :)

      One more quick note, broadband is having massive problems right now in north-america, I mean @HOME is horrible and has been for along time, ppl are losing service and getting re-routed to things that are even worse.. when you have a problem with your connection you rarely get it fixed in a timely fashion, I've heard things such as people with old TV sets cause noise on the lines for cable modem users, and until they replace those TV sets, then they're isn't much anyone really does to fix the problem, besides setting up new lines etc.. still causes major problems.. and there's so many other factors, the problems are endless it seems... take for example Videotron cable in quebec, they had consistant pings from 10-60 up and down the east coastline for the last year, however in the past 3months they've had nothing but horrid 900ms delays from the hours of say 5:00pm to 12:30am EST.. which just goes to show ya, broadband suppliers want money, not to offer a good, gaming connection. :o

  • by imrdkl (302224) on Thursday December 06, 2001 @01:51PM (#2666208) Homepage Journal
    They're taking proposals testbed projects. See the iGRID project page under the startap URL.
    From the iGRID page:

    How would you use a 2.5Gb (10Gb? 40Gb?) global testbed? How will you change your application codes? How will you expand the complexity of the problem you are solving? How will you take advantage of the fact that the networks are now faster than the computers driving them?

    A slashdotting should show them if it works.

    • Er, I was actually trying to be serious here. Slashdotting could be a real profit-generating, and instantly-deployable network testing tool!

      ;)

      Seriously, could any of the projects I hear mentioned here on /. be considered as a contender for the testbed, provided the bandwidth/latency will improve the result? Gene searching, ET searching, or any of the other distributed problems?

      Get a piece of I2 while it's open.

  • The main problem with Gnutella and its derivatives is the fact that it relies on aggregate network bandwidth to perform searches. The Gnutella derivatives such as Limewire and Bearshare all try to get around this by keeping central search databases, but, of course, that makes them vulnerable to the same argument that killed Napster, and is starting to kill Kazaa/Morpheus/Fasttrack.

    Having a huge backbone like this could make Gnutella work, at least as far as the backbone goes.
  • Great for Virtual reality with those high intensity feedback sensors.

    You know someone is going to want to be first on the holodeck with pr0n, etc.

  • by MongooseCN (139203) on Thursday December 06, 2001 @02:00PM (#2666295) Homepage
    To quote a famouse obvious scientist. Sure it may be 1000X faster than a regular network but that just means we'll soon have machines putting out 1000x more data or have a 1000x more machines on a network, etc. No matter how much bandwidth there is it will always be maxed out.
    • And perhaps more importantly...who's paying for all this bandwidth? It's already nearly impossible to get free webspace without a draconian bandwidth cap. Are we to assume bandwidth costs under this network are going to be 1000x less?
  • by Anonymous Coward
    High Bandwidth != low ping time

    It can be a pipeline with 100Gbps and yet have a very high latency.
  • ah hem Try 480gbps! (Score:4, Interesting)

    by lonney (539537) <lonney@gmail.com> on Thursday December 06, 2001 @02:07PM (#2666355) Homepage
    The Southern Cross Cable [ http://www.southerncrosscables.com/ ]that connects us in New Zealand and Australia to the US is running at 120gbps with plans to boost it to 480...enough to transfer a 3km-high stack of typed documents or eight full-length motion pictures every second.
  • He translated for us:

    "The upgraded connection between SURFnet, the advanced research and higher education network in the Netherlands, and Abilene, an Internet2® backbone network serving over 200 universities and research centers in the United States, will enable applications such as TV-quality videoconferencing, MPEG2 video streaming, data mining and remote collaboration between researchers, teachers and students in the US and the Netherlands. "

    Should be:

    "The upgraded connection between SURFnet, the advanced research and higher education network in the Netherlands, and Abilene, an Internet2® backbone network serving over 200 universities and research centers in the United States, will enable applications such as TV-quality Live-Sex videoconferencing, MPEG2 video streaming of DVD porn, pleasure mining and remote collaboration (also known as Live Sex with force feedback devices, or LS-FFD) between researchers, teachers and *especially* students in the US and the Netherlands. Sometimes between teachers and students but only where extreme secrecy and power-abuse is employed."
  • I am connected to the net through surfnet, but I can't say I notice any differences. According to the surfnet website the link is done through TeleGlobe, and my traceroute shows teleglobe hops. Unfortunately my ping to America (specifically www.internet2.edu and www.uic.edu ) are still over 100ms ... so much for low pings.

    Since this link was established half a month ago, it can't be routers that need to ajust their tables. Too bad, no high-speed pr0n^H^H^H^Hresearch material for me.
    • I have a colo in a nearby building with a 1Gbit link to SURFnet, and my pings to www.uic.edu go through uu.net, Alter.net and bbnplanet. So much for us Internet1'ers...

      All my pings to "the other side" (YOUR side, iow) are about 100ms ... but the first hop on US soil is about 17ms ...

      Dave
  • by Evro (18923)
    "Science Technology And Research Light-Illuminated Gigabit High-Performance Transit."
    Yes, I truly love the brand new "STARLIGHPT" . Nothing rolls off the tongue quite like the word Starlighpt. Yessir!

    :-D
  • I remember reading something, and I just can't find it - that you could effectively communicate at a speed much greater than light. (Theoretically, 0 ms to the other end of the Universe!)

    Rather than send a beam of light to the destination, as we're doing now, you take an existing beam of light and change an interference pattern within the beam.

    Then you measure the interference pattern. I honestly didn't fully understand the full implications of it, involving quantum theory and all, but the upshot is that the message could, in theory, be transmitted at many times the speed of light this way.

    Anybody know of a link? (Damn!)
  • has anyone else had horrible disconnects and lag from EE online?

fortune: cpu time/usefulness ratio too high -- core dumped.

Working...