I agree with the basic idea in the modem analogy, but IMHO the problem with these modems is that they're tied to Windows. Openly specced softmodems do exist, I think they're fine. For one thing, they help save manufacturing resources. On the other hand, graphics cards are justified to have dedicated processors, and even with powerful CPUs there would be interconnect bottlenecks.
Sound cards don't seem to have these problems of CPU-intensive work or transfer bottlenecks. Plus, there are so many ways of processing sounds that there's little point in a specialiced DSP for a few of them.
Somehow I like having dedicated hardware, but when you look at the whole objectively, the advantages are not so obvious. For example driver issues, when you have all kinds of weird data going to the soundcard, rather than just raw PCM sound.
You probably don't have a sound chip. You likely have a DSP and a piece of software that runs as a sound device on your computer..... But the idea is that the software is basically a sound card and the chip only takes the information from a digital level and placed it in an analog output that your speakers can use. If you want a real sound chip, something like an older sound blaster, turtle beach and so on would offer real sound. I used to hate older slow computers with little memory and those types of chips (DSP). Playing sounds or music could drag that damn thing to a halt almost.
WTF? This is how things have developed, and IMHO it's good. Old soundcards did things like FM synth in hardware, now we have enough oomph for doing it in software, which is much more flexible. I mean, imagine you used soundcard hardware for playing MP3s, you'd have to buy a new card for playing Vorbis. It's also good unix philosophy to separate the DAC/ADC from DSP and other processing. Then you can focus on building simple high-quality devices, rather than crap with bazillions of features.
However, where I do feel the pain is, when Linux doesn't recognize my soundchip. That drives me bonkers, and it's still a running concern. I guess Linux users are not into music that much. I just tried booting the newest Xubuntu live CD, and my otherwise puny soundchip wasn't detected. (worked fine on the laptop, though, so it's hit and miss) I hope Novell's efforts will bring at least a small improvement in this area.
I use Linux for making music. As with any hardware/OS combination, if you intend to use Linux, you should do your homework on supported devices. That way you'll also encourage further Linux-friendly hardware development.
What's wrong with this picture? The problem is that most Linux people have a cooks-first mentality, and when a regular diner comes along with a question or any comment except for extreme praise, the standard answer translates into "Why haven't you read the cookbook yet? The answer is right there." Well, the reason they didn't read the cookbook is because they just want to eat a tasty Linux sandwich, not to become a master chef.
What's the problem? You can either learn to cook for yourself, or pay for the sandwich. There are commercial (b|d)istros with paid support if you want the latter.
Opening the source hasn't got much to do with gaming success. The so-called succesful platforms, Windows and all the game consoles, are closed. Linux gaming could be pretty good right now, with NVidia's drivers for example. Opening the source will hardly make things less geeky, more attractive, or easier for the masses.
Quite another thing is your definition of success. For me, Linux has been a major success since 1999, and I hope I'd have discovered it even earlier. I'm not very interested in games, so it's not a factor of success IMHO. It's the same issue with 'being ready for the desktop', it depends on what you do, so you should never generalize (except in this sentence;)
Why did they introduce the P4 in the first place? Why did Microsoft introduce Vista? In both cases the company marketed their main product as the best thing ever. Now they're admitting they were wrong. The problem is, why should we trust them this time, since they were obviously lying the last time?
The key to reducing CPU power consumption is to use lower-latency memories, which require smaller on-die caches for a given performance level. We could double the throughput of DDR SDRAM by simply demultiplexing the address and data busses, similar to the way SRAM functions. There's no requirement in the underlying storage structure of DRAM to require separate row and column addresses; it's just a historical artifact. Originally, before DRAM and SDRAM became popular, computers were built with SRAM because its lower latency allowed even slow CPUs to work efficiently. But DRAM promised lower cost (via fewer bus lines) and lower power consumption (bits stored in charged capacitors, rather than cross-coupled transistors), at the price of latency, and the rest, unfortunately, is history.
Well, it would be great to have a gig of SRAM as your main memory, but that wouldn't be too good for overall power consumption, not to mention price. Besides, modern processors can turn off the power from parts of the cache, in order to save more power.
On the other hand, I agree it's been a dumb decision to favour throughput at the expense of latency. DDR2 and DDR3 are unfortunately good examples of this. I guess we're back to the Pentium 4 -style marketing where megahertz is king, no matter what the performance.
"Why people think "performace" means "throughput" is something I'll never
understand. Throughput is _always_ secondary to latency, and really only
becomes interesting when it becomes a latency number (ie "I need higher
throughput in order to process these jobs in 4 hours instead of 8" -
notice how the real issue was again about _latency_)."
-- Linus Torvalds
And what are they going to do when Tuvalu goes under water? Will they discontinue.tv? All its going to take is a foot or so rise in sea level and tuvalu goes glug glug glug...
Apparently,.su is not discontinued even if the country has gone gulag gulag gulag...
Offtopic: had a friend once who thought that there were penguins in Canada. When I disabused her of that notion, she said, "there go all my Canadian thanksgiving jokes."
I'm sometimes asked if there are penguins in Finland. I'm still trying to figure out a snappy answer that involves Linus and Tux.
Would it be cheaper to turn off BOINC and donate the electric bill savings directly to nonprofit organizations?
Probably not, if the machine is running anyway. Both of my computers have CPUs that max out around 20 watts (Pentium M, VIA Nehemiah) and most of the electricity is wasted on other components. Of course, YMMV.
Dual, yes, to run user applications on one CPU and the antivirus on the other. But what desktop workload benefits from more than two CPUs?
Desktop schmesktop. I've always wondered why people call it according to its placement rather than its function (e.g. workstation). We should really start by asking why you need more than, say, a 500 Mhz P3 for 'desktop' stuff like web, email and office. Most of today's single CPU machines are underutilized anyway, which is why I keep things like BOINC running to get something useful done instead. It's relatively rare that an interactive process is CPU-bound, but for those times it often helps to have multiple CPUs, even if not fully utilized. For example transcoding between different media formats, where you have decoding and encoding processes. I don't have much hands-on experience with multiproc machines, so I can't tell about interactive latencies (where, supposedly, UI threads running on their own CPU make things more snappy).
Funny but seriously, my advice is, don't buy a multicore CPU until they get it right.
Just to add to my other post, if everyone thinks like you, then nobody will develop multiproc-aware software. IMHO a multiproc system is very useful even without concurrency in individual applications.
I agree with your crappy concurrency article (whichever way you parse that;) but I also wonder whether it has anything to do with scientific computing. In that field people have dealt with multiprocessing for decades, I myself worked at CERN in 2001 and used a sensible language (F90) and compiler to utilize a dual x86 system. I don't see what the problem is, apart from marketing which says that multiproc didn't exist until Intel invented the core;)
Slashdot Mirror
I guess for the authentic nipple experience you need an IBM Space Saver. I have one, and it's pretty much the best keyboard I know.
I agree with the basic idea in the modem analogy, but IMHO the problem with these modems is that they're tied to Windows. Openly specced softmodems do exist, I think they're fine. For one thing, they help save manufacturing resources. On the other hand, graphics cards are justified to have dedicated processors, and even with powerful CPUs there would be interconnect bottlenecks.
Sound cards don't seem to have these problems of CPU-intensive work or transfer bottlenecks. Plus, there are so many ways of processing sounds that there's little point in a specialiced DSP for a few of them.
Somehow I like having dedicated hardware, but when you look at the whole objectively, the advantages are not so obvious. For example driver issues, when you have all kinds of weird data going to the soundcard, rather than just raw PCM sound.
WTF? This is how things have developed, and IMHO it's good. Old soundcards did things like FM synth in hardware, now we have enough oomph for doing it in software, which is much more flexible. I mean, imagine you used soundcard hardware for playing MP3s, you'd have to buy a new card for playing Vorbis. It's also good unix philosophy to separate the DAC/ADC from DSP and other processing. Then you can focus on building simple high-quality devices, rather than crap with bazillions of features.
I use Linux for making music. As with any hardware/OS combination, if you intend to use Linux, you should do your homework on supported devices. That way you'll also encourage further Linux-friendly hardware development.
is the difference between helping your attorney Jack off with gay porn, and helping your attorney jack off with gay porn.
I'm confused. I though the WHO is on the other side of the battle.
Yep, it does sound like a B movie. Can I get my GNAA membership card now?
What's the problem? You can either learn to cook for yourself, or pay for the sandwich. There are commercial (b|d)istros with paid support if you want the latter.
What about the Core Duo? It's 32-bit, and should be fine for Leopard according to the article.
Opening the source hasn't got much to do with gaming success. The so-called succesful platforms, Windows and all the game consoles, are closed. Linux gaming could be pretty good right now, with NVidia's drivers for example. Opening the source will hardly make things less geeky, more attractive, or easier for the masses.
Quite another thing is your definition of success. For me, Linux has been a major success since 1999, and I hope I'd have discovered it even earlier. I'm not very interested in games, so it's not a factor of success IMHO. It's the same issue with 'being ready for the desktop', it depends on what you do, so you should never generalize (except in this sentence ;)
Games, schmames. If we have complete specs of the hardware, there are plenty of things besides graphics we can do with it.
two pictures > 2048 words:
http://www.cs.helsinki.fi/u/nikki/linux-communism.jpg
http://www.nixp.ru/pub/walls/linux-china-communism.jpg
Yeah, who cares about the hot progressive-minded blondes when you can have legal torrents!
Why did they introduce the P4 in the first place? Why did Microsoft introduce Vista? In both cases the company marketed their main product as the best thing ever. Now they're admitting they were wrong. The problem is, why should we trust them this time, since they were obviously lying the last time?
The key to reducing CPU power consumption is to use lower-latency memories, which require smaller on-die caches for a given performance level. We could double the throughput of DDR SDRAM by simply demultiplexing the address and data busses, similar to the way SRAM functions. There's no requirement in the underlying storage structure of DRAM to require separate row and column addresses; it's just a historical artifact. Originally, before DRAM and SDRAM became popular, computers were built with SRAM because its lower latency allowed even slow CPUs to work efficiently. But DRAM promised lower cost (via fewer bus lines) and lower power consumption (bits stored in charged capacitors, rather than cross-coupled transistors), at the price of latency, and the rest, unfortunately, is history.
Well, it would be great to have a gig of SRAM as your main memory, but that wouldn't be too good for overall power consumption, not to mention price. Besides, modern processors can turn off the power from parts of the cache, in order to save more power.
On the other hand, I agree it's been a dumb decision to favour throughput at the expense of latency. DDR2 and DDR3 are unfortunately good examples of this. I guess we're back to the Pentium 4 -style marketing where megahertz is king, no matter what the performance.
Apparently, .su is not discontinued even if the country has gone gulag gulag gulag ...
I'm sometimes asked if there are penguins in Finland. I'm still trying to figure out a snappy answer that involves Linus and Tux.
I don't see any debate there, if you can have something like (|CFS> + |SD>)/sqrt(2)
I for one welcome The Police's new electro album!
Probably not, if the machine is running anyway. Both of my computers have CPUs that max out around 20 watts (Pentium M, VIA Nehemiah) and most of the electricity is wasted on other components. Of course, YMMV.
Desktop schmesktop. I've always wondered why people call it according to its placement rather than its function (e.g. workstation). We should really start by asking why you need more than, say, a 500 Mhz P3 for 'desktop' stuff like web, email and office. Most of today's single CPU machines are underutilized anyway, which is why I keep things like BOINC running to get something useful done instead. It's relatively rare that an interactive process is CPU-bound, but for those times it often helps to have multiple CPUs, even if not fully utilized. For example transcoding between different media formats, where you have decoding and encoding processes. I don't have much hands-on experience with multiproc machines, so I can't tell about interactive latencies (where, supposedly, UI threads running on their own CPU make things more snappy).
Me too. It took several minutes to emerge openoffice-bin, due to downloading over ADSL and unpacking the binaries.
"The Kerberos Konsortium" would be even kooler.
Just to add to my other post, if everyone thinks like you, then nobody will develop multiproc-aware software. IMHO a multiproc system is very useful even without concurrency in individual applications.
I agree with your crappy concurrency article (whichever way you parse that ;) but I also wonder whether it has anything to do with scientific computing. In that field people have dealt with multiprocessing for decades, I myself worked at CERN in 2001 and used a sensible language (F90) and compiler to utilize a dual x86 system. I don't see what the problem is, apart from marketing which says that multiproc didn't exist until Intel invented the core ;)