Seeing as how the entire Vivendi company's profits rose by 190% mainly on the "higher margin of the World of Warcraft business," I think Blizzard's standard response about money being a problem in the creation of dynamic events rings a little hollow.
My honors thesis at college back in 2004 was a framework that would allow you to load pixel shaders (written in CG) as 'threads' and run them in parallel on one GPU. As far as I can tell nVidia has done the same thing, but taken it a step further by translating from C (and more efficient I'm sure).
I guess I should have published that paper back then...oh well.
Why is there no railing around the semicircular pit near the Ironforge Auction House? People fall into the pit every few seconds; the lack of a railing is extremely obnoxious.
Everyone has been talking about a roll-back to fix this problem; unfortunately that will hurt Blizzard more than it helps them.
The dupe exploit is not 100% repeatable; it does depend on certain conditions being present. So on any given server, only a minority of people will have used the exploit. Of these, most will make no effort to hide their tracks (i.e. posting 10 Krol Blades), and this can easily be identified.
The remaining small fraction of players will get away with their crime, but this is OK by me. What will these players use the money for? Buying more items. From who? Everyone else on the server. So yes, one or two people will have an unfair advantage, but it is a temporary one. As long as Blizzard fixes the exploit this afternoon, the system will return to equilibrium within a week, and nobody will be the worse for it.
A rollback, on the other hand, eliminates a week's worth of work, and harms everyone on the server in a serious way.
I think the key to the setup was that it had the 'green streak' behavior on any set of intersections in that little grid, not just the main thoroughfare.
It used weight sensors under the pavement though; The radar approach sounds like it would work pretty well.
When I was in San Jose, CA a couple of years back, they had a system up where sensors in the road would pick up cars at intersections. They then used microwave antennae to broadcast the information to lights further down the road. So if you were driving along at night with nobody else on the road, you would get long strings of green lights going your way.
I just built a budget fileserver of my own - total proce $600 (including shipping). It probably could have been about $500 if I'd really shopped around a little more and looked for places with less expensive shipping (that came to $100 total). Specs:
Asus nForce 2 motherboard w/ Athlon XP 1800 4x IBM 60GB 2MB 5400RPM HDDs 3x HDD cooling units 256MB RAM Decent case, FDD, extra case fan, etc. RH9 w/ Software RAID partitioning
I don't need 7200RPM 8MB HDDs - You'd need a pretty damn high-usage server to actually require fast drives. The slower drives are more reliable, and of course cheaper. I wanted 4 drives so I could RAID them for reliability purposes, and I chose the 60GB drives because they offered the best price for the capacity I was aiming for. You could get 80 or 100 GB drives without significantly adding to the price.
The motherboard was obvious - it eliminated the lack for a graphics card, even if nForce is rather poorly supported under Linux. As for RAM, again - I'm not playing Quake on here - I rarely use more than a few percent, even with only 256MB.
All in all, the price is excellent, I get my RAID with ample (for my purposes) storage space, and speed is not an issue in any respects, so I don't need to worry about it. The case is extremely well-cooled and the system is robust in general.
It would be great if someone over at Norton or other antivirus manufacturers added this to their virus definitions. Then, when the thing tries to install, users will get a virus alert message, and SunnComm will get plenty of angry emails from customers claiming that they've released a virus-ridden disc.
I'm not so sure about that. I ran some very quick informal tests (with other apps running in the background and such) of Photoshop and some other things that are comparable to the article's tests. My single-processor, 32-bit 2gHz P4 (2+ years old now) either matched, came very close to matching, or in some cases even outperformed, the dual-processor G5. (i.e. Photoshop takes 9 seconds to start for the first time, 4.5 seconds subsequent times). That's with all the G5's beautiful fast memory and cache and 64-bit eliteness. I'm pretty certain that if there were comparable startup-time tests that I could perform on my Linux machine, the disparity would be even more glaring. Hey, even my battered 600mHz Celeron laptop with only 128 MB of RAM comes alarmingly close to the times posted in the article.
As someone who owns both Mac and PC hardware, the only conclusion I have ever been able to come to is that even outdated PC hardware beats the newest and best from Apple when it comes to things like startup times and application response times.
I had *precisely* the same experience...it took them four times as long as originally scheduled to complete the project, code had so many bugs as to be unusable; they didn't even implement some functionality. They didn't bother documenting it, so it was practically impossible to go back and fix. Then they refused to refund and even wanted us to pay them outrageous service fees for fixing what they were supposed to have done in the first place.
I agree with parts of this, but not all...In recent releases, the core of Java has improved greatly...when you write straight code and algorithms, its performance (especially with Hotspot) is very decent. This means that as a serverside language, it is a really nice choice. Add to that the existence of Tomcat and JSP, and its case is only strengthened in this regard.
I agree with you, though, that Java is a poor choice when developing client applications, especially those involving sound and graphics. I have a fair amount of experience in creating UIs and client apps, and as you mentioned, most API type routines (AWT, Swing, Java3D) are abysmally slow. Platform-independent? Not a chance. Java3D isn't even available for Mac, on Linux it suffers from OpenGL driver problems (especially on the popular NVidia cards). On the Mac, Swing has notable and serious bugs in glaringly obvious places. I've actually delayed the release of a Mac version of one of my Java apps until Apple gets their act together.
Then look at other APIs - JavaSound? Notoriously unstable and bug-fraught on anything other than Windows. Apple did not even bother implementing audio recording (added to Windows as a standard as of 1.2) on their OSX JRE untl the 1.4.1 release, where it remains very buggy. And even on Windows, the Sun engineers over a period of several years 'forgot' to implement the Port interface (arguably the one thing that would make JavaSound pretty cool) until release 1.4.2 (where it still remains buggy or nonexistent under all other operating systems)
In short, if you want to develop a nontrivial user-interface or graphics-driven client-type application that looks and performs well, Java is a dicey choice. For many other things, its speed and performance is excellent.
Slashdot Mirror
Seeing as how the entire Vivendi company's profits rose by 190% mainly on the "higher margin of the World of Warcraft business," I think Blizzard's standard response about money being a problem in the creation of dynamic events rings a little hollow.
My honors thesis at college back in 2004 was a framework that would allow you to load pixel shaders (written in CG) as 'threads' and run them in parallel on one GPU. As far as I can tell nVidia has done the same thing, but taken it a step further by translating from C (and more efficient I'm sure).
I guess I should have published that paper back then...oh well.
Why is there no railing around the semicircular pit near the Ironforge Auction House? People fall into the pit every few seconds; the lack of a railing is extremely obnoxious.
Everyone has been talking about a roll-back to fix this problem; unfortunately that will hurt Blizzard more than it helps them.
The dupe exploit is not 100% repeatable; it does depend on certain conditions being present. So on any given server, only a minority of people will have used the exploit. Of these, most will make no effort to hide their tracks (i.e. posting 10 Krol Blades), and this can easily be identified.
The remaining small fraction of players will get away with their crime, but this is OK by me. What will these players use the money for? Buying more items. From who? Everyone else on the server. So yes, one or two people will have an unfair advantage, but it is a temporary one. As long as Blizzard fixes the exploit this afternoon, the system will return to equilibrium within a week, and nobody will be the worse for it.
A rollback, on the other hand, eliminates a week's worth of work, and harms everyone on the server in a serious way.
I think the key to the setup was that it had the 'green streak' behavior on any set of intersections in that little grid, not just the main thoroughfare. It used weight sensors under the pavement though; The radar approach sounds like it would work pretty well.
When I was in San Jose, CA a couple of years back, they had a system up where sensors in the road would pick up cars at intersections. They then used microwave antennae to broadcast the information to lights further down the road. So if you were driving along at night with nobody else on the road, you would get long strings of green lights going your way.
I just built a budget fileserver of my own - total proce $600 (including shipping). It probably could have been about $500 if I'd really shopped around a little more and looked for places with less expensive shipping (that came to $100 total). Specs:
Asus nForce 2 motherboard w/ Athlon XP 1800
4x IBM 60GB 2MB 5400RPM HDDs
3x HDD cooling units
256MB RAM
Decent case, FDD, extra case fan, etc.
RH9 w/ Software RAID partitioning
I don't need 7200RPM 8MB HDDs - You'd need a pretty damn high-usage server to actually require fast drives. The slower drives are more reliable, and of course cheaper. I wanted 4 drives so I could RAID them for reliability purposes, and I chose the 60GB drives because they offered the best price for the capacity I was aiming for. You could get 80 or 100 GB drives without significantly adding to the price.
The motherboard was obvious - it eliminated the lack for a graphics card, even if nForce is rather poorly supported under Linux. As for RAM, again - I'm not playing Quake on here - I rarely use more than a few percent, even with only 256MB.
All in all, the price is excellent, I get my RAID with ample (for my purposes) storage space, and speed is not an issue in any respects, so I don't need to worry about it. The case is extremely well-cooled and the system is robust in general.
It would be great if someone over at Norton or other antivirus manufacturers added this to their virus definitions. Then, when the thing tries to install, users will get a virus alert message, and SunnComm will get plenty of angry emails from customers claiming that they've released a virus-ridden disc.
I'm not so sure about that. I ran some very quick informal tests (with other apps running in the background and such) of Photoshop and some other things that are comparable to the article's tests. My single-processor, 32-bit 2gHz P4 (2+ years old now) either matched, came very close to matching, or in some cases even outperformed, the dual-processor G5. (i.e. Photoshop takes 9 seconds to start for the first time, 4.5 seconds subsequent times). That's with all the G5's beautiful fast memory and cache and 64-bit eliteness. I'm pretty certain that if there were comparable startup-time tests that I could perform on my Linux machine, the disparity would be even more glaring. Hey, even my battered 600mHz Celeron laptop with only 128 MB of RAM comes alarmingly close to the times posted in the article. As someone who owns both Mac and PC hardware, the only conclusion I have ever been able to come to is that even outdated PC hardware beats the newest and best from Apple when it comes to things like startup times and application response times.
I had *precisely* the same experience...it took them four times as long as originally scheduled to complete the project, code had so many bugs as to be unusable; they didn't even implement some functionality. They didn't bother documenting it, so it was practically impossible to go back and fix. Then they refused to refund and even wanted us to pay them outrageous service fees for fixing what they were supposed to have done in the first place.
Hmm...I should definitely check out the OpenGL bindings. If it's fast and available on Linux too, it sounds interesting.
I agree with parts of this, but not all...In recent releases, the core of Java has improved greatly...when you write straight code and algorithms, its performance (especially with Hotspot) is very decent. This means that as a serverside language, it is a really nice choice. Add to that the existence of Tomcat and JSP, and its case is only strengthened in this regard.
I agree with you, though, that Java is a poor choice when developing client applications, especially those involving sound and graphics. I have a fair amount of experience in creating UIs and client apps, and as you mentioned, most API type routines (AWT, Swing, Java3D) are abysmally slow. Platform-independent? Not a chance. Java3D isn't even available for Mac, on Linux it suffers from OpenGL driver problems (especially on the popular NVidia cards). On the Mac, Swing has notable and serious bugs in glaringly obvious places. I've actually delayed the release of a Mac version of one of my Java apps until Apple gets their act together.
Then look at other APIs - JavaSound? Notoriously unstable and bug-fraught on anything other than Windows. Apple did not even bother implementing audio recording (added to Windows as a standard as of 1.2) on their OSX JRE untl the 1.4.1 release, where it remains very buggy. And even on Windows, the Sun engineers over a period of several years 'forgot' to implement the Port interface (arguably the one thing that would make JavaSound pretty cool) until release 1.4.2 (where it still remains buggy or nonexistent under all other operating systems)
In short, if you want to develop a nontrivial user-interface or graphics-driven client-type application that looks and performs well, Java is a dicey choice. For many other things, its speed and performance is excellent.