Yeah, cause you really want the CPU to pump any kind of I/O with no buffering, byte by byte. Heck, even the memory access itself will generally be done in cache lines of 64 bytes or so.
Still, of course, a multi-threaded system won't be able to do real time for every single sample in an audio stream. Video is far easier, you only 30 FPS or so, so all you need is to do the page switch with some accuracy.
A proper server would support hotswap of internal drives, or an extremely sleek configuration where you have no internal storage at all and it's all on the SAN. A series of small, rack incompatible, external drive enclosures don't raelly cut it.
Well, the point of DRM used that way is that the user who hsa physical access to the machine and some genius will not be the one interested in breaking it. The cracker writing a worm won't be able to solder your VGA port, reboot from a Knoppix CD or anything like it. Making code signing prevalent should make the life easy for those great numbers never writing any code on their own. Those of us who do will have to live a bit less secure, or perform an iris scan for every "make"...
There has been some fuss around the problems of phishing in this area. There are, for example, cyrillic letters that are more or less identical to the Latin ones in appearance. If you register www.slashdît.org, most people won't realize that it's not identical to www.slashdot.org and risk entering their precious/. login data to a malicious site, opening for identity theft.
It's quite similar to registering a domain name with typos and still hope that people enter their login data, but it's MUCH harder to realize that this is going on when you can't realize it by just reading the domain name with your eyes, no matter how closely you look at the letters.
It will just be sad for those users relying on IDN. That may not be U.S. users, but it WILL disturb some Swedish sites, and I assume it's far worse for Japanese and Chinese users, for example. There may be other, older, domain name schemes for those users still used that I'm not aware of, though, but IDN has been seen as the way forward for quite some time.
It's not a patch anymore than turning of Javascript is a patch for several IE vulnerabilities. It might be argued that this workaround does less in the area of destroying the "experience" for normal surfers, but as I noted, I think that depends much on your nationality/language.
What might help is the work to emulate the MSHTML COM interfaces through Gecko to actually get useful. That doesn't require any genius. It just requires lots of tedious work. It should be somewhat similar to ripping out COMCTL32.dll in a Win32 system; more or less user mode only, no extremely strange dependencies, but it's quite a lot of functionality to duplicate with a very specific interface to get everything working. It should be no surprise that, last I heard, the ones pursuing any such addon to Gecko was the WINE guys.
It's not like seamless interaction with remoting, including SOAP web services, was one of the selling points of C#, although that's not a main use in practice.
MSN Messenger is highly popular in some regions/demographics (although I don't use it).
BTW, Google hadn't received that name yet and I know of no "public API" back for Backrub or Yahoo in 1997. Those were the days when a Javascript controllable HTML DOM was part of everyone's wet dreams.
Furthermore, it's quite hard to make SOAP Windows-only.
So, nice trolling, I was stupid enough to get hit.
This may just seem offtopic when zonk corrects his spelling mistake without noting it in the summary. It originally said "Neal Armstrong", which was the basis for my crude joke about confusing the two.
Doom3 doesn't provide a general purpose windowing system that any old GDI app has to be able to live happily within. Anyway, I would really consider 1024x768 a low-end resolution for Vista. For the "optimum config" stuff, which is really what they describe, I would expect 1280x1024 (maybe dual-mon) or 1600x1200. Remember that one single 1600x1200 frame is 7.680.000 bytes. At the very least, you want to maintain a double-buffering scheme of that. You will just fit 17 single frames that size onto a 128 MB card. Even if you don't go the route of making every window a texture, it doesn't seem that weird to actually have 15 frames worth of data that are used to composite your normal display. Even seemingly small things like fonts can get huge if they are bitmap-cached for many UNICODE code points.
You say you don't need that? Maybe you don't, but it is very easy to list uses for all that graphics memory that makes some sense. The font example is actually quite interesting, just ls/dir a huge directory in your desktop environment of choice with the window active vs. minimized/hidden. Time both operations. Repeat a few times so you are sure that there are no caching artefacts. Well, the text rendering is actually consuming most of the CPU time. By only refreshing n times each second and anyway placing most of that refreshing work on the graphics chip, both could (not will, but could) get equivalent. The ironical thing here is that the most highly programmable chips would actually need less memory, as more things could be represented in a more efficient way than simple bitmap textures.
Well, there is a grain of truth in this. "Units of memory" could be a non-geek obfuscation of "pointers". And noone should say that 64 bit pointers everywhere is a free thing, that it doesn't increase memory requirements or performance (mainly due to caches filling faster and increased bandwidth requirements).
In addition, I would imagine that the total mapped area of virtual memory will increase a fair bit by the thunking and duplication of code for 32-bit and 64-bit. Any libraries that are loaded into the process adress space will need to be mapped in both versions, and that will naturally increase the amount of unique code pages mapped.
Nah, plain GDI is rendered in a bitmap that's then made a texture. It's been made clear that there will be a "classic", very XP-like, mode. One reason is performance, another that some corporate environments really don't want to do anything that will confuse their users. A third reason will be that some machines will have trashy hardware where the drivers never even would load Aero, no matter what performance you would accept.
In this case, there is even the possibility of turning off Aero and running in a somewhat improved XP. Avalon apps will suffer, but they will still work.
They want the Aero rendering to be considered fast and snappy. And, oh, it will possibly be so, but only on the right hardware. If they put the official requirements too low, it would just be said that the new interface is so full of eyecandy that it can't perform.
What's really interesting here is what they manage to pull of on laptops, together with ATI and Nvidia. Will the power management for graphics chips make sense, even when 3D mode doesn't equal "battery sucking gaming mode"? The (public) slides from Microsoft even from the very first mentioning of Longhorn's 3D UI stressed this aspect. It will be interesting to see the solution. If a Mactel box will provide a sleek UI with a charge keeping the machine powered for twice as long, that'll be a very real selling point.
The backslash key is not identical in width and location on those two images. Maybe just different lots or one or the other not accurate relative to the real thing.
I wonder how fun a chess competition would be if each player would have to write a detailed spec on how they chose each move. Put another way: programming competitions with any restrictions of style easily puts much of the fun out of it.
That said, TopCoder (as mentioned elsewhere) also have the design/development competitions, where you code real commercial code. I wouldn't keep doing that without the chance of payments, though. "Algorithmic" competitions, on the other hand, are fun in their own right.
Wasn't Compaq's decision to break off their part in the Windows 2000 Alpha development, thereby stopping the product at the stage of Release Candidate 1 or something, quite significant? Windows was never very popular on Alpha, but the decline in support from upper management started long before HP got involved.
1. Recompile.
2. Rip out anything that doesn't work and doesn't have to work if you just use it for massive databases.
3. Make the press release about how you've optimized your offering for the "workloads" that still work on the system.
4. ????
5. Profit!
Yeah, you can look at the code. But doesn't it tell you something if the spec is unclear enough that you need a specific reference implementation to actually interpret it? Especially if this reference implementation wasn't written with that goal in mind, but with the goal to be a reasonable implementation for end users.
The data structures actually used to represent a word processor document are not obvious. One can go about it in several ways. Just compare HTML (overlapping tags allowed) to XHTML and you see an example of this. Now imagine you write a quick transform to make the HTML valid, duplicating any overlapping tags into two parts to create a valid tree. By doing this, you lose the connection between the two parts, so if you edit common formatting for one of them, the other won't maintain that.
That's just an example, but if you base a format on Your Way Of Doing Things, a XML based format won't be that "open" compared to a binary format. Note that it's the editing aspect that introduces many of these issues. PDF, or HTML, or SVG, or some new format only concerned about viewing is, IMHO, a better option if the intent is to make all documents available to everyone.
I just got word from my ISP that my DSL connection will "soon" be upgraded from 0.5 Mbit/s to 8. And, no, there was no reasonable way for me to get more than 0.5 for that place before. And, yes, it's located within the borders of the 4th city in size in the country, not in the real countryside or something.
Fast connections are not uncommon and I could certainly have found a place to live with one if that was my only priority. If you're just trying to get a decent place to live, weighing all things, you MAY be able to get 24+ DSL down, 1 up in some places. In a few places you may get a real RJ-45 with a fat pipe backing it up in your home, but please don't make it out as some geek utopia. It isn't. You may be worse off, but I think the difference is quite a bit smaller than you make it out to be.
If 3G penetration in Japan is as extensive as it's generally described to be, I'm more than jealous, though...
Read the post or even the article again. 1 Gb/s while begin stationary, 100 Mb/s while riding your bike in a moderate speed. One clear downside of the added use of multiplexing and adaptive technologies is that moving even within the area of the same cell will lower your bandwidth.
And, yeah, in practice I get lackluster GPRS performance many times on the road (not to mention on hi-speed trains), but that's another story. GSM and WCDMA should, theoretically, handle the within-cell scenario quite well, and I guess my problem is the frequent cell switching and bad coverage overall.
I can promise you that the analog receivers of 1985 didn't reconstruct a signal that was on par with normal definition of today and far below a HD stream.
Also, I seem to remember that one part of the multiplexing ideas for 4G was too use differently oriented antennas, dynamically adapting a signal mixing scheme to filter out the signal minus most interferences and echos, as those shouldn't be uniform for different polarizations. Therefore, saying that there simply can't be enough available bandwidth in the air isn't that relevant. We are still far from the theoretical maximums, and this kind of approach also opens the possibility of nearby transmitters sharing the same frequency with less jamming. Sure, these numbers might be optimistic, but if proper multiplexing gets into the standard, 4G will be far more interesting from a technological standpoint than 3G. Did I mention lower transmission power? (at least when not maxing the connection)
Slashdot Mirror
Still, of course, a multi-threaded system won't be able to do real time for every single sample in an audio stream. Video is far easier, you only 30 FPS or so, so all you need is to do the page switch with some accuracy.
A proper server would support hotswap of internal drives, or an extremely sleek configuration where you have no internal storage at all and it's all on the SAN. A series of small, rack incompatible, external drive enclosures don't raelly cut it.
Well, the point of DRM used that way is that the user who hsa physical access to the machine and some genius will not be the one interested in breaking it. The cracker writing a worm won't be able to solder your VGA port, reboot from a Knoppix CD or anything like it. Making code signing prevalent should make the life easy for those great numbers never writing any code on their own. Those of us who do will have to live a bit less secure, or perform an iris scan for every "make"...
It's quite similar to registering a domain name with typos and still hope that people enter their login data, but it's MUCH harder to realize that this is going on when you can't realize it by just reading the domain name with your eyes, no matter how closely you look at the letters.
It's not a patch anymore than turning of Javascript is a patch for several IE vulnerabilities. It might be argued that this workaround does less in the area of destroying the "experience" for normal surfers, but as I noted, I think that depends much on your nationality/language.
What might help is the work to emulate the MSHTML COM interfaces through Gecko to actually get useful. That doesn't require any genius. It just requires lots of tedious work. It should be somewhat similar to ripping out COMCTL32.dll in a Win32 system; more or less user mode only, no extremely strange dependencies, but it's quite a lot of functionality to duplicate with a very specific interface to get everything working. It should be no surprise that, last I heard, the ones pursuing any such addon to Gecko was the WINE guys.
It's not like seamless interaction with remoting, including SOAP web services, was one of the selling points of C#, although that's not a main use in practice.
BTW, Google hadn't received that name yet and I know of no "public API" back for Backrub or Yahoo in 1997. Those were the days when a Javascript controllable HTML DOM was part of everyone's wet dreams.
Furthermore, it's quite hard to make SOAP Windows-only.
So, nice trolling, I was stupid enough to get hit.
This may just seem offtopic when zonk corrects his spelling mistake without noting it in the summary. It originally said "Neal Armstrong", which was the basis for my crude joke about confusing the two.
You say you don't need that? Maybe you don't, but it is very easy to list uses for all that graphics memory that makes some sense. The font example is actually quite interesting, just ls/dir a huge directory in your desktop environment of choice with the window active vs. minimized/hidden. Time both operations. Repeat a few times so you are sure that there are no caching artefacts. Well, the text rendering is actually consuming most of the CPU time. By only refreshing n times each second and anyway placing most of that refreshing work on the graphics chip, both could (not will, but could) get equivalent. The ironical thing here is that the most highly programmable chips would actually need less memory, as more things could be represented in a more efficient way than simple bitmap textures.
Repeat after me: Neil Armstrong, but CowboyNeal. They are, repeat, they are NOT, one and the same.
In addition, I would imagine that the total mapped area of virtual memory will increase a fair bit by the thunking and duplication of code for 32-bit and 64-bit. Any libraries that are loaded into the process adress space will need to be mapped in both versions, and that will naturally increase the amount of unique code pages mapped.
Nah, plain GDI is rendered in a bitmap that's then made a texture. It's been made clear that there will be a "classic", very XP-like, mode. One reason is performance, another that some corporate environments really don't want to do anything that will confuse their users. A third reason will be that some machines will have trashy hardware where the drivers never even would load Aero, no matter what performance you would accept.
They want the Aero rendering to be considered fast and snappy. And, oh, it will possibly be so, but only on the right hardware. If they put the official requirements too low, it would just be said that the new interface is so full of eyecandy that it can't perform.
What's really interesting here is what they manage to pull of on laptops, together with ATI and Nvidia. Will the power management for graphics chips make sense, even when 3D mode doesn't equal "battery sucking gaming mode"? The (public) slides from Microsoft even from the very first mentioning of Longhorn's 3D UI stressed this aspect. It will be interesting to see the solution. If a Mactel box will provide a sleek UI with a charge keeping the machine powered for twice as long, that'll be a very real selling point.
The backslash key is not identical in width and location on those two images. Maybe just different lots or one or the other not accurate relative to the real thing.
That said, TopCoder (as mentioned elsewhere) also have the design/development competitions, where you code real commercial code. I wouldn't keep doing that without the chance of payments, though. "Algorithmic" competitions, on the other hand, are fun in their own right.
It's not writing games that's hard. It's writing games that actually use anything slightly close to the theoretical specs from Sony et al.
Wasn't Compaq's decision to break off their part in the Windows 2000 Alpha development, thereby stopping the product at the stage of Release Candidate 1 or something, quite significant? Windows was never very popular on Alpha, but the decline in support from upper management started long before HP got involved.
Regarding potential for being a RDBMS, I would vote for emacs. I'll leave it to you to decide if that's a good thing...
1. Recompile.
2. Rip out anything that doesn't work and doesn't have to work if you just use it for massive databases.
3. Make the press release about how you've optimized your offering for the "workloads" that still work on the system.
4. ????
5. Profit!
The data structures actually used to represent a word processor document are not obvious. One can go about it in several ways. Just compare HTML (overlapping tags allowed) to XHTML and you see an example of this. Now imagine you write a quick transform to make the HTML valid, duplicating any overlapping tags into two parts to create a valid tree. By doing this, you lose the connection between the two parts, so if you edit common formatting for one of them, the other won't maintain that.
That's just an example, but if you base a format on Your Way Of Doing Things, a XML based format won't be that "open" compared to a binary format. Note that it's the editing aspect that introduces many of these issues. PDF, or HTML, or SVG, or some new format only concerned about viewing is, IMHO, a better option if the intent is to make all documents available to everyone.
Fast connections are not uncommon and I could certainly have found a place to live with one if that was my only priority. If you're just trying to get a decent place to live, weighing all things, you MAY be able to get 24+ DSL down, 1 up in some places. In a few places you may get a real RJ-45 with a fat pipe backing it up in your home, but please don't make it out as some geek utopia. It isn't. You may be worse off, but I think the difference is quite a bit smaller than you make it out to be.
If 3G penetration in Japan is as extensive as it's generally described to be, I'm more than jealous, though...
And, yeah, in practice I get lackluster GPRS performance many times on the road (not to mention on hi-speed trains), but that's another story. GSM and WCDMA should, theoretically, handle the within-cell scenario quite well, and I guess my problem is the frequent cell switching and bad coverage overall.
Also, I seem to remember that one part of the multiplexing ideas for 4G was too use differently oriented antennas, dynamically adapting a signal mixing scheme to filter out the signal minus most interferences and echos, as those shouldn't be uniform for different polarizations. Therefore, saying that there simply can't be enough available bandwidth in the air isn't that relevant. We are still far from the theoretical maximums, and this kind of approach also opens the possibility of nearby transmitters sharing the same frequency with less jamming. Sure, these numbers might be optimistic, but if proper multiplexing gets into the standard, 4G will be far more interesting from a technological standpoint than 3G. Did I mention lower transmission power? (at least when not maxing the connection)
Femto
Pico
Nano
Micro
Milli
One
Kilo
Mega
Giga
Tera
Peta
Exa
Yota
Just adopt SI :-)