jb
10-23-2002, 01:14 PM
Today I wanted to bend your ears on something that's been troubling me for some time now. And its the power of FUD. FUD was a term that came up in the early IBM days were they used Fear, Uncertainty and Doubt to hinder competition. Since they were founders of the modern PC market they challenged all the new competitors by attacking them and causing FUD in the consumers minds. For a variety of reasons this failed and IBM is no longer the leader of the PC technology. However these new companies picked up and started to use FUD in their own marketing campaigns so today is common practice. And no one company is above using parts of FUD in their PR work. They all do it. The problem is, like gossip it spreads like wildfire and often with out any facts to back it up. Message boards are common places to see this effect. You have one poster that takes part of an PR releases then uses it to back up their own favorite company and spreads the word. As that word spreads it gets slightly changed from one person to the next. And before you know it no longer accurate. But people don't understand it and take it as the gospel and tell their friends, families, clan mates, ect So lets take a look to see just how bad/good ATI's drivers are?
Most people have heard that ATI's cards are good but their drivers Sux0rs! And since it takes both to run correctly a lot of people stayed away from ATI products. Now in the past (2 years ago) that was very true. ATI's drivers at the launch of the original Radon did have a lot of issues. Most of them were corrected but it took some time. During that time lots of people complained about the quality of ATI's drivers. Then came the launch of the 8500. With Smooth Vision (one of its touted features not working), a bug/cheat that provided higher FPS in Q3 with a loss of IQ and a host of other issues, ATI had well deserved their reputation for bad drivers. Even the ID god JohnyC stated that he was disappointed in ATI's driver quality. We soon began to see a change in the ATI drivers. 2 weeks after the 8500 launch we saw drivers that had fixed Q3, Smooth Vision and a host of other issues. Funny thing is that some people out there (me included) had almost no issues with their 8500. Then about every 6 weeks we saw new drivers from ATI. With each release the list of problems got smaller as well as speed gains were reached. Most people had very few issues with the 8500 about 6 months after its launch. So did ATI finally fix its drivers?
Lets look at their launch of the R9000/R9700. At this launch we have very few issues. All features were working as advertised. Most old games worked great and there was very few issues. Over all most people were pleased with the R9700. So it look like ATI has done a lot better with their drivers. But some of the newer games did not run so well. For example Mafia, The Thing and 1941 all had lock up issues on the 9700. Gack, just when it looked like ATI had started to turn there driver rap around this stuff happens. So what's the deal'o?
The deal is that its part ATI's fault and part of the game developers fault. Huh, what? What do the developers of games have to do with it? Almost everything. Game developers are usually smaller groups of people that don't have access to each and every possible hardware or software configuration. So the pick the most common platform that's reliable and use that to do all development of their game. Since 51% of the market share is controlled by nV then its pretty easy to see why these developers use nV as their primary development platform. Here during the development of games they may find conditions where a bug occurs. When this happens they can contact nV and let them know about it and/or modify their code to work around it. No matter which they chose to do most time the public never sees these issues. Now in an ideal world developers will have boards from all video card companies and enough time to test each one of them. However in the real world developers have to meet ship date on a limited budget so this type of testing is rarely done. For example Epic on UT2k3 did have a beta R9700 board to do some testing on. But it was not until after the game had shipped until they got a real R9700 to test with. See where I am going here? If developers don't have the time or resources to test other cards then bugs don't get caught. Then they show up in the public and we have to wait for either the developer to fix it in a patch or new drivers from the video card company.
Another issues is that developers can use vendor specific instructions even though there are standard instructions to use. Case in point is the shinny water issue on NWN. NWN will show shinny water surfaces on all GF3/4 cards but any one on a 8500 or R9700 will not see these effects. Of course people blamed ATI for crappy drivers. The shinny water was created by using DX8 pixel shadders which both the 8500/R9700 support. So why then was it not working? After a few emails bounced around we found out the reason. NWN is an OpenGL game. The developers used GF3/4 to do all of the development. When it came time to code the shinny water they chose to use a custom NV OpenGL call that was not available to the other cards. Funny thing is there already a standard OpenGL call that does the same thing that the developers could have used to do the same effect which would have worked on all cards that support that feature of DX8. No reason was given on why they chose not to use the standard call. At that time nV was not allowing other vendors to use their calls unless they paid nV. Not exactly in the sprit of what OpenGL stands for. So how in the world is this ATI's fault? ATI's developer relations need to work even harder at their job. From what I hear their team is good and making strides to provide the interface between game developers and ATI. The lead developer in charge of DX at MS was quoted as saying that ATI has closed the gap in their minds of driver quality so that the difference between them an nV is too small to count.
Finally some of the fault lies in the link from the keyboard to the screen (ie that is the user). For what ever reason ATI cards have more issues when they are installed on top of a system that has had an nV card in it before. Normally a clean install is the best way. However that's not an option for some people. Now should you have to wipe a system clean when installing a video card? Well for me I have found that when I change video cards that only time were I will reformat is going from one core vendor to another (ie from 3dfx to nV, from nV to ATI, from S3 to Kyro2, ect). If you changing to just a new card from the same core vendor then its not that big of a deal. Some time even running the un-install will still leave remains of that card on your system that could cause issues on the new card. A lot of users that had issues with ATI never did a clean install and so they are not starting clean. Also ATI's drivers could not have been installed correctly. People may have download the wrong one or installed it correctly. Also people could have gone in and over clock other parts of their systems which may cause a bit more un-stability. There are a ton of things that could happen with users PCs that may help to cause issues.
Bottom line is that ATI still needs to work on somethings. One of them being driver relations. If they can get more developers using their cards then more things will be fixed before they are ever released and labeled as driver bugs. ATI also need to still improve on their driver process. There is no way in heck those first drivers for the 8500 should have ever been released. Who ever was heading up the quality group at that time was sound asleep. Finally ATI needs to work more on their driver image. The bad rap they have now is not relevant to their current work. Changing that rap will take work on their part as well as helping to educate the end user. Everyone has different experiences with drivers. Just because someone's says it worked/broke their system does not mean that is the norm. It means its was there experience. I have really had very few driver issues every with any of may cards. Yes I have had some things I had to tweak and a few bugs, but in the long run all drivers seem to have worked well for me. I am not trying to say ATI drivers are great. The are really good but not perfect. At this point I see no reason why you wound not want to get an ATI card if it provided is the best fit for your needs at that time.
Most people have heard that ATI's cards are good but their drivers Sux0rs! And since it takes both to run correctly a lot of people stayed away from ATI products. Now in the past (2 years ago) that was very true. ATI's drivers at the launch of the original Radon did have a lot of issues. Most of them were corrected but it took some time. During that time lots of people complained about the quality of ATI's drivers. Then came the launch of the 8500. With Smooth Vision (one of its touted features not working), a bug/cheat that provided higher FPS in Q3 with a loss of IQ and a host of other issues, ATI had well deserved their reputation for bad drivers. Even the ID god JohnyC stated that he was disappointed in ATI's driver quality. We soon began to see a change in the ATI drivers. 2 weeks after the 8500 launch we saw drivers that had fixed Q3, Smooth Vision and a host of other issues. Funny thing is that some people out there (me included) had almost no issues with their 8500. Then about every 6 weeks we saw new drivers from ATI. With each release the list of problems got smaller as well as speed gains were reached. Most people had very few issues with the 8500 about 6 months after its launch. So did ATI finally fix its drivers?
Lets look at their launch of the R9000/R9700. At this launch we have very few issues. All features were working as advertised. Most old games worked great and there was very few issues. Over all most people were pleased with the R9700. So it look like ATI has done a lot better with their drivers. But some of the newer games did not run so well. For example Mafia, The Thing and 1941 all had lock up issues on the 9700. Gack, just when it looked like ATI had started to turn there driver rap around this stuff happens. So what's the deal'o?
The deal is that its part ATI's fault and part of the game developers fault. Huh, what? What do the developers of games have to do with it? Almost everything. Game developers are usually smaller groups of people that don't have access to each and every possible hardware or software configuration. So the pick the most common platform that's reliable and use that to do all development of their game. Since 51% of the market share is controlled by nV then its pretty easy to see why these developers use nV as their primary development platform. Here during the development of games they may find conditions where a bug occurs. When this happens they can contact nV and let them know about it and/or modify their code to work around it. No matter which they chose to do most time the public never sees these issues. Now in an ideal world developers will have boards from all video card companies and enough time to test each one of them. However in the real world developers have to meet ship date on a limited budget so this type of testing is rarely done. For example Epic on UT2k3 did have a beta R9700 board to do some testing on. But it was not until after the game had shipped until they got a real R9700 to test with. See where I am going here? If developers don't have the time or resources to test other cards then bugs don't get caught. Then they show up in the public and we have to wait for either the developer to fix it in a patch or new drivers from the video card company.
Another issues is that developers can use vendor specific instructions even though there are standard instructions to use. Case in point is the shinny water issue on NWN. NWN will show shinny water surfaces on all GF3/4 cards but any one on a 8500 or R9700 will not see these effects. Of course people blamed ATI for crappy drivers. The shinny water was created by using DX8 pixel shadders which both the 8500/R9700 support. So why then was it not working? After a few emails bounced around we found out the reason. NWN is an OpenGL game. The developers used GF3/4 to do all of the development. When it came time to code the shinny water they chose to use a custom NV OpenGL call that was not available to the other cards. Funny thing is there already a standard OpenGL call that does the same thing that the developers could have used to do the same effect which would have worked on all cards that support that feature of DX8. No reason was given on why they chose not to use the standard call. At that time nV was not allowing other vendors to use their calls unless they paid nV. Not exactly in the sprit of what OpenGL stands for. So how in the world is this ATI's fault? ATI's developer relations need to work even harder at their job. From what I hear their team is good and making strides to provide the interface between game developers and ATI. The lead developer in charge of DX at MS was quoted as saying that ATI has closed the gap in their minds of driver quality so that the difference between them an nV is too small to count.
Finally some of the fault lies in the link from the keyboard to the screen (ie that is the user). For what ever reason ATI cards have more issues when they are installed on top of a system that has had an nV card in it before. Normally a clean install is the best way. However that's not an option for some people. Now should you have to wipe a system clean when installing a video card? Well for me I have found that when I change video cards that only time were I will reformat is going from one core vendor to another (ie from 3dfx to nV, from nV to ATI, from S3 to Kyro2, ect). If you changing to just a new card from the same core vendor then its not that big of a deal. Some time even running the un-install will still leave remains of that card on your system that could cause issues on the new card. A lot of users that had issues with ATI never did a clean install and so they are not starting clean. Also ATI's drivers could not have been installed correctly. People may have download the wrong one or installed it correctly. Also people could have gone in and over clock other parts of their systems which may cause a bit more un-stability. There are a ton of things that could happen with users PCs that may help to cause issues.
Bottom line is that ATI still needs to work on somethings. One of them being driver relations. If they can get more developers using their cards then more things will be fixed before they are ever released and labeled as driver bugs. ATI also need to still improve on their driver process. There is no way in heck those first drivers for the 8500 should have ever been released. Who ever was heading up the quality group at that time was sound asleep. Finally ATI needs to work more on their driver image. The bad rap they have now is not relevant to their current work. Changing that rap will take work on their part as well as helping to educate the end user. Everyone has different experiences with drivers. Just because someone's says it worked/broke their system does not mean that is the norm. It means its was there experience. I have really had very few driver issues every with any of may cards. Yes I have had some things I had to tweak and a few bugs, but in the long run all drivers seem to have worked well for me. I am not trying to say ATI drivers are great. The are really good but not perfect. At this point I see no reason why you wound not want to get an ATI card if it provided is the best fit for your needs at that time.