Well this just asks for a comment, so I'm sure I won't be the only one.
It's easy to sit, and say you missed this differance or that, but the biggest bugbear I have about the consumer cards is the lack of proper Accum buffer. But as you rightly say about the other features, they'll certainly have that in a year anyway.
But just as the consumer cards will have all the more features in a years time, so will the workstation cards, and more than just memory too.
I look forward to curves and NURBS and a whole host of other features.
But again, (if you listen to rumours), the likes of nVidia are all ready thinking that far ahead.
3D Labs must have noticed the vast fill rate gap, (and to some degree in geometry acceleration I think ?), and presumably will be working towards solving this problem. Although I'm sure they're having fun solving the memory bandwidth issue.
All this is to the great benifit of the consumer.
One of my issues of contention is external bandwidth however. I'm not an expert in the field, but in my mind it still seams to be one of the biggest problems. I work in broadcasting, and getting video and 3D working together is highly desired. The idea of a spinning cube with 6 video textures on it is an old favourite. I think it will be a long time before consumer cards can address this issue, where as the proffesional cards can afford to look at the problems themselves. Matrox have the likes of the Movie 2 bus for their video products, and SGI add a special port to the Wildcat 4210 to help get data in and out quick. I'm sure there are other examples.
People can add all the external ports they like however, but it seams both sides are struggling with memorey bandwidth issues though, and I think that will be the presiding issue for years to come. More features will be added to counterbalance this, and draw away peoples attention. But when you start wanting real time video, and 3D textures, this is going to be a continuing problem.
Whatever happened to the likes of BitBoys and Pixel Fusion who were going have on die memory ? Is anyone else going to following in their so far seamingly unsuccessful footsteps ?
I think the only survivial for the 'workstation' manufacturors such as 3D Labs is in splitting 2 ways. Producing a card with a price/performance to match nVidia, and/or moving in the more niche markets where bandwidth and special features override any cost comparison. The features they currently offer over the likes of the GeForce 2 for most workstation needs just do not justify the price. Their cheaper cards just simply don't cut it when held up against nVidia.
Slashdot Mirror
Well this just asks for a comment, so I'm sure I won't be the only one.
It's easy to sit, and say you missed this differance or that, but the biggest bugbear I have about the consumer cards is the lack of proper Accum buffer. But as you rightly say about the other features, they'll certainly have that in a year anyway.
But just as the consumer cards will have all the more features in a years time, so will the workstation cards, and more than just memory too.
I look forward to curves and NURBS and a whole host of other features.
But again, (if you listen to rumours), the likes of nVidia are all ready thinking that far ahead.
3D Labs must have noticed the vast fill rate gap, (and to some degree in geometry acceleration I think ?), and presumably will be working towards solving this problem. Although I'm sure they're having fun solving the memory bandwidth issue.
All this is to the great benifit of the consumer.
One of my issues of contention is external bandwidth however. I'm not an expert in the field, but in my mind it still seams to be one of the biggest problems. I work in broadcasting, and getting video and 3D working together is highly desired. The idea of a spinning cube with 6 video textures on it is an old favourite. I think it will be a long time before consumer cards can address this issue, where as the proffesional cards can afford to look at the problems themselves. Matrox have the likes of the Movie 2 bus for their video products, and SGI add a special port to the Wildcat 4210 to help get data in and out quick. I'm sure there are other examples.
People can add all the external ports they like however, but it seams both sides are struggling with memorey bandwidth issues though, and I think that will be the presiding issue for years to come. More features will be added to counterbalance this, and draw away peoples attention. But when you start wanting real time video, and 3D textures, this is going to be a continuing problem.
Whatever happened to the likes of BitBoys and Pixel Fusion who were going have on die memory ? Is anyone else going to following in their so far seamingly unsuccessful footsteps ?
I think the only survivial for the 'workstation' manufacturors such as 3D Labs is in splitting 2 ways. Producing a card with a price/performance to match nVidia, and/or moving in the more niche markets where bandwidth and special features override any cost comparison. The features they currently offer over the likes of the GeForce 2 for most workstation needs just do not justify the price. Their cheaper cards just simply don't cut it when held up against nVidia.
Glenn Shoosmith