Quote:

As to your second question. Have you tried to close application that hanged up? "Preemption" under that system is working a little differently than it should.

If the application is hanging, its not processing the window messages telling it to quit.

However, using the task manager to terminate the process works pretty damn fast ( unless the application is in an infinite loop of some description and it takes forever to load the task manager [grin] ).

On linux I usually already have some console open to kill such a process.

Quote:

Original post by wodinoneeye That might be a poor example if you consider a Quad Core (coming out quite soon and later will be affordable..) 4 X 2Ghz (and with much larger caches as Intel has been doing). Current games might not be particularly well designed to run efficiently on multi-cores, but future ones likely will be.

I think perhaps I didn't make my point clear. The problem is that people think multiple cores are some sort of magic that is going to make everything better. They won't. The only thing they solve is the fact that chip manufacturers can't just keep increasing the operating frequency to improve speed. It's just a different way of keeping up with Moore's Law. All that will happen is that computers get faster in the same way that they have for a couple of decades, except we're going to have a poor year or four in the meantime while people struggle with adjusting to concurrent programming. If anything, the switch to multiple cores will slow progress down in the short term.

Quote:

Original post by Kylotan Quote: Original post by wodinoneeye That might be a poor example if you consider a Quad Core (coming out quite soon and later will be affordable..) 4 X 2Ghz (and with much larger caches as Intel has been doing). Current games might not be particularly well designed to run efficiently on multi-cores, but future ones likely will be. I think perhaps I didn't make my point clear. The problem is that people think multiple cores are some sort of magic that is going to make everything better. They won't. The only thing they solve is the fact that chip manufacturers can't just keep increasing the operating frequency to improve speed. It's just a different way of keeping up with Moore's Law. All that will happen is that computers get faster in the same way that they have for a couple of decades, except we're going to have a poor year or four in the meantime while people struggle with adjusting to concurrent programming. If anything, the switch to multiple cores will slow progress down in the short term.

Finally, someone who thinks like me. Really, the whole "multi-care" marchitecture (marketing + architecture) is just the chip industry's way of saying, "Ok, we're out of ideas, you software guys go figure it out." And more cores doesn't make things easier, it just makes things harder. Many people keep forgeting that all the submodules of the game HAVE to talk together and share data. Rendering has to work with physics and AI, or else, how would it know what to render, when to render it, and how to render it. So, it's nice if you split the physics off to another core and keep the module that feeds the GPU on a seperate core, but these modules have to constantly talk to each other. There is actually alot of concurrency/latency involved in multicore. And in the end, you're falling down into your lowest common denominator. Even something as simple as an asynchronous file I/O can be troublesome because if the I/O doesn't finish when we need the data, the game still has to wait and vice versa. So, in the end, the whole multicore thing is only good for the chip industry, while the software industry gets thrown the ball called "performance."

On a side note, Moore's Law was a marketing and media thing as well. It outlines exponential growth in technology, but it has been redefined many many times over the year. Even Moore himself will tell you that the truth is far from his statement. If you look up what he originally said, he said that the amount of "transitors" on an affordable "consumer" chip will double ever 24 months. Even his first statement fell through as a proper projection of the industry, but the media loved it so much they hyped and redefined it to be a law.

Quote:

Original post by Kylotan Quote: Original post by wodinoneeye That might be a poor example if you consider a Quad Core (coming out quite soon and later will be affordable..) 4 X 2Ghz (and with much larger caches as Intel has been doing). Current games might not be particularly well designed to run efficiently on multi-cores, but future ones likely will be. I think perhaps I didn't make my point clear. The problem is that people think multiple cores are some sort of magic that is going to make everything better. They won't. The only thing they solve is the fact that chip manufacturers can't just keep increasing the operating frequency to improve speed. It's just a different way of keeping up with Moore's Law. All that will happen is that computers get faster in the same way that they have for a couple of decades, except we're going to have a poor year or four in the meantime while people struggle with adjusting to concurrent programming. If anything, the switch to multiple cores will slow progress down in the short term.

Unless they start getting results with Josephson Junction tech or Galium Arsenide they were hitting a wall and the multi-cores are the next alternative (along with huge L2 caches). Supposedly Intel has 16 core CPUs planned for the near rather than far future. Yes, everyone will have to adjust to doing much more multi-threading when more CPU power is needed but its inevitable as its a useful path no matter what new tech comes along. Lose coupling via clusters is also there, but its even more inflexible/costly when applied to problems requiring close coupled parallelizations.

Of course many application dont need more CPU, so wont be impacted. Reverting to 2+GHz on the new intel architecture isnt so much a loss of performance of one core versus the 4+GHz old architecture, so most people might not notice.
Game players can still use the old architecture in the interim and game companies
will come up to speed on using multi-cores which can greatly exceed mono-cores if used right(and on an awful lot of games the GPU still is much more the determining factor).