NVIDIA shows signs ... [2008 - 2017]

Status
Not open for further replies.
lololol.. Orly ?

A thoughtful reply as usual. Perhaps you could name one other company that has performant hardware and consistently provides up to date Linux driver support for both legacy as well as the very latest hardware?

There isn't any vendor even close for enterprise Linux deployments.
 
A thoughtful reply as usual. Perhaps you could name one other company that has performant hardware and consistently provides up to date Linux driver support for both legacy as well as the very latest hardware?

There isn't any vendor even close for enterprise Linux deployments.

Big deal. Truth be told, only 3D drivers are a problem in linux today. Almost all other drivers are in the mainline kernel itself. Other vendors support linux much better than they do. They push their code into kernel itself and hence don't receive much publicity.

AMD open source gu drivers are also coming along. They should be in a usable state by the end of this year.
 
Big deal. Truth be told, only 3D drivers are a problem in linux today.

Yes, they are. Although Nvidia's are pretty solid.

Almost all other drivers are in the mainline kernel itself. Other vendors support linux much better than they do. They push their code into kernel itself and hence don't receive much publicity.

Which other vendors would that be? Surely you can't be talking about Intel or ATI, IMG, Matrox or S3.. who either only have low end consumer hardware to begin with, or only support some models but not others, or have nothing.

AMD open source gu drivers are also coming along. They should be in a usable state by the end of this year.

Well, so far they've usually treated Linux as an ugly stepchild. If that is changing great. We'll see whenever it happens. Putting the specs out there and letting the community solve it is quite different from the years of steady Nvidia driver support, though.
 
Florin you have moved your goal posts from the orignal statement of :

and it is. In fact, nvidia's consistent cross platform strategy with their proprietary stuff has probably done more than anyone to cement linux' position as a serious platform for everything from rendering to finance. Tech docs would surely be helpful, but credit where credit is due.


based off that statement i can think of a few companies that have done "more" for linux, hows about these:

Oracle
IBM
Novell

i cound go on but it hink you get my point ;)
 
What does Microsoft have to do with anyone's hardware naming schemes?

Marketing and other support for OEMs. Its in their better interest to make computer hardware naming simpler because each OEM acting individually cannot bring about the change required to give the end consumer a better shot at picking an appropriate laptop/desktop. If people do not understand what they are buying its very hard to get them to understand why they would need an upgraded GPU when their capabilities are not obvious from the names they carry.
 
Marketing and other support for OEMs. Its in their better interest to make computer hardware naming simpler because each OEM acting individually cannot bring about the change required to give the end consumer a better shot at picking an appropriate laptop/desktop. If people do not understand what they are buying its very hard to get them to understand why they would need an upgraded GPU when their capabilities are not obvious from the names they carry.

Yeah, you know.. maybe it's better to forget about this idea and let them stick to naming their own products.

I'm not particularly waiting for decrees coming out of the Microsoft Kremlin stating that ATI's next card will be allowed to be called the Radeon Windows Experience Index 7.3 Ultimate.
 
I'm not particularly waiting for decrees coming out of the Microsoft Kremlin stating that ATI's next card will be allowed to be called the Radeon Windows Experience Index 7.3 Ultimate.

You know.. it actually worked out really well back in the day when a number of multimedia consortiums labeled their products according to certain performance levels, i.e. MPC1, MPC2 etc.
 
Holy guacamoli Batman!!!1

Rumor has it that this week NVIDIA will let it officially be known that XFX's European operations will cease to be an NVIDIA Approved Partner. This means that XFX will lose all underlying NVIDIA support on GPU allocation and new product launches...including Fermi. It also looks as if Visiontek will be kicked out of a MAJOR USA e-tailer, which will probably decimate Visiontek's business. We will be seeing other major changes in video card sales this year here in the USA as well. The video card manufacturer landscape will be very different in the USA this year. Expect to see two or three major brands fall by the wayside when it comes to NVIDIA and AMD GPUs.
http://www.hardocp.com/news/2010/01/09/gpu_news_from_ces
 
I wonder if it is more XFX dropping Nvidia due to lack of new products. Or Nvidia wanting to consolidate chip allocation to companies that remain faithful to Nvidia.

Regards,
SB
 
Yes, they are. Although Nvidia's are pretty solid.
Nv's drivers are generally good quality, but that can't change the fact that in oss world, oss drivers work best for both technical and maintainability reasons.

Which other vendors would that be? Surely you can't be talking about Intel or ATI, IMG, Matrox or S3.. who either only have low end consumer hardware to begin with, or only support some models but not others, or have nothing.
All non gpu hardware.
 
I wonder if it is more XFX dropping Nvidia due to lack of new products. Or Nvidia wanting to consolidate chip allocation to companies that remain faithful to Nvidia.

Regards,
SB

Given that there's no benefit to XFX to lose Nvidia as a supplier I'd go with the latter. Would still be a stupid move on their part given that XFX is very popular but I guess they figure setting an example is worth losing a good partner.
 
All non gpu hardware.
Which should tell you something...

Pretty much all hardware drivers do little more than shovel bits from piece of hardware to another. There may be some smart DMA structures to lower CPU overhead, but all in all the vast majority of the performance smarts are inside the chip, with the driver just being a thin layer around it to deploy it.

I believe there are enough performance optimizing tactics deployed in GPU driver that result a significant competitive advantage. If you're in a business where a 10% advantage over your competition can make or break a sale, it's not unreasonable to want to keep this kind of code secret.

It will be interesting to see the results of the ATI open source driver and compare its performance to a similar close source equivalent, but I wouldn't be surprised if the difference will be larger than 30%.
 
You know.. it actually worked out really well back in the day when a number of multimedia consortiums labeled their products according to certain performance levels, i.e. MPC1, MPC2 etc.

At this point since noone knows what a GPU is really, having a name which actually destribes their purpose makes a hell of a lot of sense. They aren't just GPUs anymore so you may as well call them multimedia accelerators. I think even a sticker 'Makes Youtube run better' would probably sell a GPU better than a name Nvidia GT260M.
 
Which should tell you something...

Pretty much all hardware drivers do little more than shovel bits from piece of hardware to another. There may be some smart DMA structures to lower CPU overhead, but all in all the vast majority of the performance smarts are inside the chip, with the driver just being a thin layer around it to deploy it.

I believe there are enough performance optimizing tactics deployed in GPU driver that result a significant competitive advantage. If you're in a business where a 10% advantage over your competition can make or break a sale, it's not unreasonable to want to keep this kind of code secret.

It will be interesting to see the results of the ATI open source driver and compare its performance to a similar close source equivalent, but I wouldn't be surprised if the difference will be larger than 30%.

Well, there is pretty much no shader compiler worth a damn out there today and writing one is hard, so I am not expecting the drivers to be performance competitive right out the gate. A lot of performance tuning will have to be done as well before it can be competitive.

Graphics drivers are complicated and are almost a full blown OS by themselves. The oss drivers are starting out from scratch. Once a respectable/reasonable foundation is laid, however, I am expecting the drivers to keep up with hw much better than today.
 
Why just the European side?

Not expecting enough Fermi to go around?

EVGA's behest?

I remember that quite a lot of European AIBs have been snubbed by nVidia a while ago, so it could be nV doing this so EVGA would not say, act in its own best interests perhaps (AMD/ATI).
 
I wonder if it is more XFX dropping Nvidia due to lack of new products. Or Nvidia wanting to consolidate chip allocation to companies that remain faithful to Nvidia.

Regards,
SB

Nah, I believe it has more to do with the fact XFX has serious Q&A issues going on right now that affect Nvidia based cards, board, X58 boards and even ATI based cards. They need to step it up and get back to making quality products again as it appears they suffer far more failures across all lines versus other makers.
 
Nah, I believe it has more to do with the fact XFX has serious Q&A issues going on right now that affect Nvidia based cards, board, X58 boards and even ATI based cards. They need to step it up and get back to making quality products again as it appears they suffer far more failures across all lines versus other makers.

What are you basing this off of? IIRC XFX has one of the best warranties in the biz.
 
Status
Not open for further replies.
Back
Top