I still have my Roland MT-32, and would love to use it for those old DOS games that support it. Can anybody suggest how I'd go about doing that in DOSBox?
The original Tron was pretty poor *as a movie*, but nevertheless I enjoyed it, and continue to enjoy it today. What it really came down to, though, was that the movie was *extremely* metaphorical, and that those metaphors made sense in the context of computing at the time. (And in fact those metaphors hold up today, which is what makes the movie so much fun to watch today.)
The new Tron: Legacy didn't actually try to play with metaphors. It used the old ones from time to time, and it threw in some Unix and open-source allusions here and there (inappropriately, I might add), but other than that it just spent its effort on making things look pretty.
As a side-note: my reading of the movie is that the central theme of the movie is that open-source is good, but that the GPL is bad. And Clu is Richard Stallman. My review (which doesn't go into that in detail, admittedly) is here.
Usenet will never die until the last news server goes down; it'll just fade away.
Even "fading away" is a little pessimistic. Usenet still has too many benefits for real discussion - consistent interface, a wide variety of tools, killfiles, newsrcs, universal access through the flood-and-fill protocol, spam fighting, the wide variety of cultural forces that Usenet introduced - and the world is slowly coming around to accept them in other protocols. Even if another article were never transmitted via NNTP/UUCP, the lessons of Usenet will be taught to the next protocols - or, if not, then the lessons will be re-learned after they are poorly implemented a few more times.
Me, I hope that a smaller (read: binary-free) Usenet might lead to a resurgence of popularity, as people realize that they can easily pull down a full feed of the text groups to their private machines and share them to the world, just like any web server in the world. It's a *little* quixotic, sure, but not insanely so.
Google Groups probably does more with usenet than anyone else.
But they still don't know what they're doing with it, sadly.
Re:The average user does not know about usenet
on
RIAA Sues Usenet.com
·
· Score: 2, Informative
(Do they still bar any proposals for the creation of binaries groups under rec? Do any comp.binaries.* groups survive today?)
Most of comp.binaries.* is gone. We're not looking to make more binaries groups in the Big-8. You can see the current creation policies here, and the list of "discouraged" proposal types is here.
There is a reason why Usenet was forgotten: it was the birthplace of spam.
Naah, that didn't have much to do with it. Spam was actually effectively defeated on Usenet. The problem is that nobody tells anybody else that it exists anymore, and so the number of posters have gone down... I personally think that the difficulty of making a new group caused problems too, and I hope that the new system may help.
Text-based Usenet is a useful service. More people should use it; it does a better job of allowing discussion than most web forums out there, and there's little threat of centralized control over the discussion.
I've been tempted to make comp.internet.services.news.slashdot in the Big-8...
It bothers me that the only time that Usenet is brought up anymore on sites like this is in regards to the binary content on Usenet. Yes, alt.binaries.* exists, and it's 99.9% of Usenet's traffic; but the text-based groups still exist, and they're doing pretty well lately, especially with our new Big-8newsgroup creation system. But it's difficult-to-impossible to publicize this through discussion fora such as Slashdot, which aren't interested in the old, less-catchy systems of yesteryear...
If anybody is any good at publicity and wants to help with Usenet, give me a mail.
Umm... what do you think scales better than NAMD, exactly?
GigE lets us scale to 32-64 nodes reasonably, with the ApoA benchmark. Using Myrinet or Infiniband, you can bring that up to a thousand or so processors.
Also of note: we've done a series of cluster-building workshops specifically focusing on the software and hardware required to run these kinds of simulations. Copies of the presentations and tutorials are linked off of that page.
And if you want to see how we designed our clusters, I've got full specifications up here.
Let's say you've got a cluster of 100 tightly-coupled nodes running a parallel job, each with something eating 1% of their CPU time. What happens when one of those nodes goes into the 1%? The other nodes have to wait for that node to finish up...and they're not computing in the mean time. You haven't wasted 1% of one node, you've wasted 1% of *50* nodes.
But wait! What about the other 49 nodes? They're going to have that 1% taken off of them, too. These times aren't going to be synchronized, much as you'd love for them to be. Instead, on average, you're going to lose 1-(1-0.01)^50 of your compute time waiting on that 1%. That's 40% of your cycles.
What, 1% is unrealistic? Let's go to.1% then. With that and 50 nodes, you're wasting 5% of your cycles. This adds up.
The moral of the story: don't run any background daemons on compute nodes. Or run less tightly coupled software, of course...
Not using HP? Well, none of us can stand HP-UX, and their guys screwed up on our campus a while ago and essentially stopped supporting us. Fine, we said, and cheerfully went to Sun. Same goes for SGI. IBM we still talk to, but their prices are even worse...
I will take a look at the Quadros some time; it couldn't hurt, after all. But I'm still dubious.
Never buy a Sun with anything less than the best video card available for that platform, if you're going to use it at the console. Me, I was always more than content with my Ultra 10 with Elite3Dm6 graphics. Never caused me any problems, never felt slow...
I suspect we push our systems harder, but I'm not positive so I won't push the case... At any rate, our Suns are essentially rock solid. Why not Quadros instead of GeForces? Price, the GeForces have always seemed quicker to market with new features to me, and the knowledge that it's the drivers that seem to have caused us most of our problems so far, not the cards themselves. We don't have the budget to buy the quantities of SB2000s w/XVR1000s that we'd like; we just have to de-rack old cluster machines most of the time. Still, even the Linux boxes offer us weeks of uptime if we don't push the graphics. They're stable systems. I wouldn't settle for anything less. And as for OpenGL drivers with Sun, I will admit that part of my biases come from the fact that we've got Sun engineers more than happy to help us fix any problems we come across rapidly. At the same time, our developers have been literally begging nVidia and other companies to fix OpenGL and driver bugs for years, only to be rebuffed with "we don't have the manpower" constantly. I've seen what they go through... And oh yes, stereo. Can't forget stereo. Linux doesn't do it, Suns do...and we need it, at least some of the time. Yes, it's a matter of the cards, but still...
Even if the hardware was close, Sun can't compete driver wise with nVidia. You're kidding, right? Sun's video drivers work. nVidia's don't. Case closed, as far as I'm concerned. Before I get modded down as a troll: I run both SB2000s and a whole pile of Linux boxes with GeForce cards. The GeForce drivers aren't *awful*, but they definitely fall over when you push their OpenGL. New versions of the drivers have consistently caused just as many problems as they've fixed. My users are used to having X crash once every couple of days on the Linux boxes, and occasionally gripe that they miss the stability their old Sun Ultra 10s gave them. The annoying part is that nVidia has the best Linux drivers on the market. We can barely get ATI's to work in our environment at all, and even then just on test systems; and don't get me started on 3DLabs...
In case you're interested, I've also got a page up of Neal Stephenson's short work, fiction and non-fiction. BTW, this book is the first book of three in Baroque Cycle, and they'll be released at six month intervals. So says HarperCollins.
Slashdot Mirror
Wallet is a Kerberos-based secret management tool. It works well for me.
How do ProtoMechs fit into this picture?
$40? I believe that I am in. Thank you!
I still have my Roland MT-32, and would love to use it for those old DOS games that support it. Can anybody suggest how I'd go about doing that in DOSBox?
(Posted while logged in this time.)
The original Tron was pretty poor *as a movie*, but nevertheless I enjoyed it, and continue to enjoy it today. What it really came down to, though, was that the movie was *extremely* metaphorical, and that those metaphors made sense in the context of computing at the time. (And in fact those metaphors hold up today, which is what makes the movie so much fun to watch today.)
The new Tron: Legacy didn't actually try to play with metaphors. It used the old ones from time to time, and it threw in some Unix and open-source allusions here and there (inappropriately, I might add), but other than that it just spent its effort on making things look pretty.
As a side-note: my reading of the movie is that the central theme of the movie is that open-source is good, but that the GPL is bad. And Clu is Richard Stallman. My review (which doesn't go into that in detail, admittedly) is here.
Usenet will never die until the last news server goes down; it'll just fade away.
Even "fading away" is a little pessimistic. Usenet still has too many benefits for real discussion - consistent interface, a wide variety of tools, killfiles, newsrcs, universal access through the flood-and-fill protocol, spam fighting, the wide variety of cultural forces that Usenet introduced - and the world is slowly coming around to accept them in other protocols. Even if another article were never transmitted via NNTP/UUCP, the lessons of Usenet will be taught to the next protocols - or, if not, then the lessons will be re-learned after they are poorly implemented a few more times.
Me, I hope that a smaller (read: binary-free) Usenet might lead to a resurgence of popularity, as people realize that they can easily pull down a full feed of the text groups to their private machines and share them to the world, just like any web server in the world. It's a *little* quixotic, sure, but not insanely so.
How long before they take on Google Groups?
Do they carry alt.binaries.*?
Google Groups probably does more with usenet than anyone else.
But they still don't know what they're doing with it, sadly.
(Do they still bar any proposals for the creation of binaries groups under rec? Do any comp.binaries.* groups survive today?)
Most of comp.binaries.* is gone. We're not looking to make more binaries groups in the Big-8. You can see the current creation policies here, and the list of "discouraged" proposal types is here.
There is a reason why Usenet was forgotten: it was the birthplace of spam.
Naah, that didn't have much to do with it. Spam was actually effectively defeated on Usenet. The problem is that nobody tells anybody else that it exists anymore, and so the number of posters have gone down... I personally think that the difficulty of making a new group caused problems too, and I hope that the new system may help.
Text-based Usenet is a useful service. More people should use it; it does a better job of allowing discussion than most web forums out there, and there's little threat of centralized control over the discussion.
I've been tempted to make comp.internet.services.news.slashdot in the Big-8...
It bothers me that the only time that Usenet is brought up anymore on sites like this is in regards to the binary content on Usenet. Yes, alt.binaries.* exists, and it's 99.9% of Usenet's traffic; but the text-based groups still exist, and they're doing pretty well lately, especially with our new Big-8 newsgroup creation system. But it's difficult-to-impossible to publicize this through discussion fora such as Slashdot, which aren't interested in the old, less-catchy systems of yesteryear...
If anybody is any good at publicity and wants to help with Usenet, give me a mail.
The main research page may interest some of you. And for those that it doesn't help, perhaps you want to look at our Linux clusters instead?
The *simulation* was done at NCSA, using LRAC resources.
The *image* was created on our local workstations.
I did overstate our case somewhat, though...
Umm... what do you think scales better than NAMD, exactly?
GigE lets us scale to 32-64 nodes reasonably, with the ApoA benchmark. Using Myrinet or Infiniband, you can bring that up to a thousand or so processors.
Also of note: we've done a series of cluster-building workshops specifically focusing on the software and hardware required to run these kinds of simulations. Copies of the presentations and tutorials are linked off of that page.
And if you want to see how we designed our clusters, I've got full specifications up here.
1. The full research page for this project is here. This is a lot better than the stuff linked through Nature and such.
2. The image was actually generated by our group, and specifically Anton Arkhipov, using our software package VMD. NCSA didn't have anything to do with it.
Let's say you've got a cluster of 100 tightly-coupled nodes running a parallel job, each with something eating 1% of their CPU time. What happens when one of those nodes goes into the 1%? The other nodes have to wait for that node to finish up...and they're not computing in the mean time. You haven't wasted 1% of one node, you've wasted 1% of *50* nodes.
.1% then. With that and 50 nodes, you're wasting 5% of your cycles. This adds up.
But wait! What about the other 49 nodes? They're going to have that 1% taken off of them, too. These times aren't going to be synchronized, much as you'd love for them to be. Instead, on average, you're going to lose 1-(1-0.01)^50 of your compute time waiting on that 1%. That's 40% of your cycles.
What, 1% is unrealistic? Let's go to
The moral of the story: don't run any background daemons on compute nodes. Or run less tightly coupled software, of course...
Clustermatic is Scyld-but-free.
Oh, and for kicks, check out our cluster-building workshop.
My favorite part of this: on page two, they talk of how Cane Toads can be used to make zombies.
I appreciate Cane Toads, if you hadn't guessed.
- Tim Skirvin
Not using HP? Well, none of us can stand HP-UX, and their guys screwed up on our campus a while ago and essentially stopped supporting us. Fine, we said, and cheerfully went to Sun. Same goes for SGI. IBM we still talk to, but their prices are even worse...
I will take a look at the Quadros some time; it couldn't hurt, after all. But I'm still dubious.
Never buy a Sun with anything less than the best video card available for that platform, if you're going to use it at the console. Me, I was always more than content with my Ultra 10 with Elite3Dm6 graphics. Never caused me any problems, never felt slow...
I suspect we push our systems harder, but I'm not positive so I won't push the case... At any rate, our Suns are essentially rock solid.
Why not Quadros instead of GeForces? Price, the GeForces have always seemed quicker to market with new features to me, and the knowledge that it's the drivers that seem to have caused us most of our problems so far, not the cards themselves. We don't have the budget to buy the quantities of SB2000s w/XVR1000s that we'd like; we just have to de-rack old cluster machines most of the time. Still, even the Linux boxes offer us weeks of uptime if we don't push the graphics. They're stable systems. I wouldn't settle for anything less.
And as for OpenGL drivers with Sun, I will admit that part of my biases come from the fact that we've got Sun engineers more than happy to help us fix any problems we come across rapidly. At the same time, our developers have been literally begging nVidia and other companies to fix OpenGL and driver bugs for years, only to be rebuffed with "we don't have the manpower" constantly. I've seen what they go through...
And oh yes, stereo. Can't forget stereo. Linux doesn't do it, Suns do...and we need it, at least some of the time. Yes, it's a matter of the cards, but still...
Even if the hardware was close, Sun can't compete driver wise with nVidia.
You're kidding, right? Sun's video drivers work. nVidia's don't. Case closed, as far as I'm concerned.
Before I get modded down as a troll: I run both SB2000s and a whole pile of Linux boxes with GeForce cards. The GeForce drivers aren't *awful*, but they definitely fall over when you push their OpenGL. New versions of the drivers have consistently caused just as many problems as they've fixed. My users are used to having X crash once every couple of days on the Linux boxes, and occasionally gripe that they miss the stability their old Sun Ultra 10s gave them.
The annoying part is that nVidia has the best Linux drivers on the market. We can barely get ATI's to work in our environment at all, and even then just on test systems; and don't get me started on 3DLabs...
Found, read, and added. Good story, too. Thanks.
In case you're interested, I've also got a page up of Neal Stephenson's short work, fiction and non-fiction.
BTW, this book is the first book of three in Baroque Cycle, and they'll be released at six month intervals. So says HarperCollins.
...except that encap works with essentially everything that uses 'configure' fairly trivially, and most of the stuff that uses 'xmkmf'.
/opt is for.
If you want the MacOSX model (the one I actually use on a regular basis), that's what