News AMD hints at pro-grade graphics push

Wow, fight fight fight !

 

"Leading to stop-gap measures like the sale and leaseback of its headquarters".

Ouch, every company I've read about that has gone bust during the recent years has done the exact same thing. It makes no sense to rent your premises, that's just another overhead, we've seen it so many times that the rent goes up exceeding what they can afford to pay. So selling your premises and leasing them back is usually a short term gain and long term drain.

I'm pretty worried that they have gone down this route, it says to me that this is a company in difficulty and cannot raise enough capital, if the banks don't think they are safe enough to lend to, then that's another factor that confirms this.

 

But AMD have been doing the that for a few years now and it appears they are heading back in the right direction.

My main concern is the fact they are looking at taking another shot at "Pro-Grade" products at this very, very early stage of recovery. It could be argued that the high end of the market place nearly killed them due to such poor take up.

AMD are not Intel and shouldn't pretend to be. Intel can afford to stick it's fingers in many pies, AMD unfortunately cannot no matter how much they want to or want to be seen as a competitor to Intel.

In this instance, they are looking to take on Nvidia who with such a high market share will only push harder to keep that market share.

Stick to what you are good at AMD.

 

AMD going after workstations is a battle long lost. Its performance per watt sector where AMD does not compete anymore. Will take more than a good product to get faith back from users in this sector. AMD will need 2-3 years of good reviews with the drivers backing them up to claw back market share in this sector.

Took them 18months to fix up there Consumer GPU drivers, Businesses can not afford to wait 18months for things they can get elsewhere.

And in the business sector its not pure performance that has took nvidia to its position its good drivers and low power consumption. Niether are things AMD is Famed for.

They had no choice but to sell the headquaters as they were in real danger of been brought out or going broke if they did nothing. AMDs biggest mistake was leaving the Fab scene and relying on others for there Fab work.

 

Yeah biting and chew comes to mind...

If they can get the power usage down then no doubt they'll price low. 290x could be pretty telling as to what's going to happen with these cards. Would be weird if they did it in reverse though. IE - release the desktop card then use the same core for a workstation solution.

 

intel seems to be perfectly fine for not bug fixing hardware for a few years - and ignoring driver complaints as well....

 

Intel is the other 1% in the workstation sector they are basically a none player on the gpu side of it.

 

You are obviously too much of an nvidia fan to know that what you said is very short-sighted. Just like in the desktop graphics market, some programs work considerably better in one brand versus the other. I have a friend who does a lot of CAD engineering work and found that AMD's workstation GPUs performed considerably better than the nvidia equivalent, and for a cheaper price.

When it comes to live editing or brief renderings, either brand is suitable (though AMD would make more sense since they're cheaper). For everything else, Nvidia is probably a better choice. Nvidia specializes in GPGPUs and has for years. They're also a lot better at writing drivers, so micro-optimizations (which are important in a lot of workstation platforms) are more abundant and effective.

 

Casting fanboyism's a side, you can't deny his comment when they have 18% of the market. Just because your friend used them, doesn't warrant a sweeping statement that the other 1% and indeed the larger 81% is wrong regardless of how considerably better your friend thinks they are.

 

They've got to go for it - this is one of the few markets AMD could make money with big margins left. They used to do this in the x86 server market but their hardware just can't keep up with Intel any longer. Graphics wise however radeons are still a match for the best. The problem is this isn't an easy market for AMD to crack. Fundamentally it needs rock solid hardware, drivers and support.

The problem for AMD is they've never been much good at software - in x86 they let MS and Intel write all the software. In radeons they provide the basic driver but always want someone else to do the rest (e.g. open cl not cuda, havok/bullet not physx). For professional you've got to do better then that and AMD just isn't setup to do that (nvidia have claimed for years they have more software then hardware devs). They will need to invest heavily if they really want to win market share and not just talk about it a bit in shareholder meetings to fob them off.

 

Thats the thing it may work great in certain programs when it works but when it does not work its hopeless. Nvidia is heavily supported across most of the proffesional range of programs and that has helped them aquire a huge marketshare in business.

Thats because business can not afford to wait for driver fixes that can take so long from AMD to arive.

AMD needs to fix both software support and Drivers then let them mature in enthusaist hands. After that then business will probably look into them. This is not we launch the next fire pro and it will sell well cause it just will not happen.

Price in the high end workstation area is not really relivent, They want the best performance for the programs they use. That plus power consumed whilst doing it is the only 2 factors that will come into a decision. If nvidias top end card which costs aprox $6000 can do the job twice as quick as the one that costs $3000 then the top end card would get brought 9 times outta 10.

Consumer Workstation is a different ball game but its not as high cash flow as people are usauly after the low end of the market where the prices are cheaper ( sub $1000 / £1000 )

Took Nvidia years of dedicated Support and drivers to build up the reputation it has in business. AMD have let there reputation be tarnished by neglecting this market for so long.

Even in the consumer sector they have not really fixed Crossfire on Multi screen displays. Took Nvidia to release a program to show up AMDs frame time issues.

 

I'm not saying nvidia is a worse choice or that nvidia is undeserving of their 81%. All I'm saying is there are several tests where AMD comes out on top, and nvidia is not champion of all workstation benchmarks. My friend made his decision based on the fact that the software he intended to use worked better on AMD (such as Solidworks) before he even bought the GPU.

I would have to agree with DbD, where AMD's problem is they let other developers do all the hard work and don't spend enough time optimizing individual programs themselves. Whenever AMD comes out on top, it's because they worked together with the software devs. This really shows their hardware is great but their drivers are utter crap. If AMD had driver devs as good as nvidia, they'd likely be #1 in nearly every benchmark.

As a linux user, I have the option to go for the Catalyst drivers or the open source drivers. There are times where the open source driver performs better than catalyst, even on Windows. This also shows how the hardware is great. The way I see it, AMD is overall a poor choice if you're going for anything brand new. But, give the drivers some time to mature, and it ends up being a very worthy product.

 

Again, casting those fanboyism's aside. An 81% market share is not something you can question when the area concerned is workstations. This area is not benchtesters going for the high end gfx card as a pissing contest. It has to be fit for purpose, be able to do the job now and come in at a set budget. Do you think it's ok for someone to list his desired components for the business to pick up the bill for it to be selected on which manufacturer he likes? That decision tends to be made by historical practice and industry knowledge. If was just down to cost, then wouldn't that market share be the other way round? It's pretty much ok to assume that within that 81% market share, it has some engineers using CAD and doing well on it.

Blaming other developers for AMD failings isn't the way to go either. They are historically bad at driver support which probably adds more wait to the Nvidia dominance in this sector.

Waiting for them to get the drivers right could end up costing a company a fortune and given that by the time they get it right, if at all, the competition already has a better working product usually the next generation. So what's the point in investing thousands of $£ in a product that might not work now?

I think they'll need more than your hope you clearly have that they'll succeed.

 

The choice between AMD or nVidia, on the professional card market, is solely driven by : "What software are you using and/or what are you doing with it ?".

Example : Firepro are better with large model under 3DS max 2011, but nVidia is better for everything else in the same software




Same with Autacad, it depends on the use case ...



Full article here : http://www.xbitlabs.com/articles/graphics/display/nvidia-quadro-amd-frirepro_2.html#sect0

Another great article opposing pro and gamer cards : http://www.tomshardware.com/reviews/best-workstation-graphics-card,3493.html