NVIDIA shows signs ... [2008 - 2017]

[Silent_Buddha]

Aye, when GTX 460 launched I was left wondering why in the world they would have clocked it so conservatively. Sure there's obviously the desire not to tramp on sales of GF100 salvage chips, but considering the rather lukewarm reception of GF100 compounded by the fact that the top end model is a salvage chip and the rest are salvage chips of salvage chips. They could have easily positioned it under GTX 470 but close to 5850. GTX 465 even at current clocks was made completely irrelevant by GTX 460 so that wasn't a consideration.

However, doing so, everyone knows that AMD would have had to respond, but even then I figured it would have made some sense from a business standpoint.

But, after looking at their Q3 filing, it just makes a lot more sense in terms of deliberately avoiding triggering any sort of price war. At least until they have a more stable and competitive lineup (rather than just one chip). Added to that a price war between GF104 and Cypress would have reduced the price of 5850 and possibly 5870, making GTX 480/470 even more unattractive to the masses. And driving their price down for those would further damage their already marginal margins.

So in the end. If AMD wasn't going to push their advantage (high margins are better than squeezing the comp when wafer allocations are non-optimal), it looks to be best for Nvidia currently to maneuver around that and carefully pick their battles. At least until they are in a better position with regards to product lineup.

Regards,
SB

 

[trinibwoy]

http://www.xbitlabs.com/news/video/...Professional_Graphics_Market_Segment_JPR.html

In Q2 2010, Nvidia managed to maintain dominant positions both on the desktop and portable professional graphics market, according to Jon Peddie Research.

The market tracking company said that in the second quarter of this year 87.5% of professional graphics adapters shipped were Nvidia Quadro (up 2.2%), whereas ATI commanded only 12.5% of professional graphics accelerators with the FirePro lineup. Nonetheless, desktop ATI FirePro graphics accelerators managed to capture 11.4% of professional add-in graphics cards (up 0.9%), but the lion's share of 88.6% still belonged to Nvidia. The latter also managed to gain some share on the market of professional graphics accelerators for mobile computers.

Nvidia's achievement for Q2 2010 is quite remarkable since most of the Quadro graphics cards Nvidia shipped were based on previous generation technology...

 

[Psycho]

Isn't the "professional graphics market" a circular definition? Which is probably a big part of the reason nvidia keeps it's share here. It is NOT the total market of people using various professional 3D applications, but merely those that decides to buy quadro/firepro, instead of those going with geforce/radeon.
And I imagine that a lot of the more economical minded people that consider the FirePro alternative, will also look at the gamer cards, and thereby maybe decide to simply "leave the market" instead. A big part of the "market" doesn't really need the workstation cards, but buys them anyway as they are "professional" and they know they "just work", and the additional maybe thousand dollars doesn't matter much for a full time user of the program anyway..
I'm fully aware that for some applications workstation cards(/drivers) ARE very important and the Quadros the obvious choice. But other 3d programs works just as fine on gamer cards, or maybe their use isn't really performance intensive - and don't even mention the Quadro MX series, sold on their multi monitor capabilities for business applications...

Do we have any kind of numbers of the (trends of the) size of the "professional graphics market", compared to the size of the professional graphics market?

 

[Alexko]

Isn't the "professional graphics market" a circular definition? Which is probably a big part of the reason nvidia keeps it's share here. It is NOT the total market of people using various professional 3D applications, but merely those that decides to buy quadro/firepro, instead of those going with geforce/radeon.
And I imagine that a lot of the more economical minded people that consider the FirePro alternative, will also look at the gamer cards, and thereby maybe decide to simply "leave the market" instead. A big part of the "market" doesn't really need the workstation cards, but buys them anyway as they are "professional" and they know they "just work", and the additional maybe thousand dollars doesn't matter much for a full time user of the program anyway..
I'm fully aware that for some applications workstation cards(/drivers) ARE very important and the Quadros the obvious choice. But other 3d programs works just as fine on gamer cards, or maybe their use isn't really performance intensive - and don't even mention the Quadro MX series, sold on their multi monitor capabilities for business applications...

Do we have any kind of numbers of the (trends of the) size of the "professional graphics market", compared to the size of the professional graphics market?

I would imagine the ratio remains constant, at 1…

Seriously though, this may explain why there are so many codenames with "GL" in them related to the HD 6000 series: AMD needs to do better here. Based on hardware alone, they should be doing a LOT better. Clearly, they're doing something wrong.

 

[Florin]

I would imagine the ratio remains constant, at 1…

Seriously though, this may explain why there are so many codenames with "GL" in them related to the HD 6000 series: AMD needs to do better here. Based on hardware alone, they should be doing a LOT better. Clearly, they're doing something wrong.

They certainly are, as this recent review demonstrates, with the Quadro equivalent of a GTX465 (!) pretty much ripping the sibling of the 5870 a new one.

It might be described as ironic that the Fermi-based solution also has a TDP which is 56W lower.

 

[trinibwoy]

Isn't the "professional graphics market" a circular definition?

Not really, it's pretty straightforward. The feature set and performance of "professional" graphics cards are much better than their consumer counterparts when it comes to professional apps. The "professional graphics market" is the market for those cards, simple as that. Besides, what makes you think that the sort of people who cheap out and use consumer cards with professional apps would have a preference for Radeon? There's no evidence at all to support that.

They certainly are, as this recent review demonstrates, with the Quadro equivalent of a GTX465 (!) pretty much ripping the sibling of the 5870 a new one.

It might be described as ironic that the Fermi-based solution also has a TDP which is 56W lower.

Yeah that's a poor showing from AMD's driver team. But it goes to show the perf/mm^2 thing isn't quite that cut and dry.

 

[Alexko]

They certainly are, as this recent review demonstrates, with the Quadro equivalent of a GTX465 (!) pretty much ripping the sibling of the 5870 a new one.

It might be described as ironic that the Fermi-based solution also has a TDP which is 56W lower.

Considering the 513MHz clock, the Quadro's TDP is pretty much what you'd expect. It seems the FirePro is suffering from crappy drivers: the performance gap in Viewperf is just brutal.

Then again, the FirePro is $620 (or 34.8%) cheaper (Newegg)…

 

[Florin]

Considering the 513MHz clock, the Quadro's TDP is pretty much what you'd expect. It seems the FirePro is suffering from crappy drivers: the performance gap in Viewperf is just brutal.

Ah well, there's probably some older -a version driver with a magic hotfix and CAP update combo out there which is known to do much better.

 

[3dilettante]

Considering the 513MHz clock, the Quadro's TDP is pretty much what you'd expect. It seems the FirePro is suffering from crappy drivers: the performance gap in Viewperf is just brutal.

While I'm not discounting drivers as a major factor, isn't it also the case that a lot of those apps stress geometry performance very heavily, which Fermi should excel at?

Then again, the FirePro is $620 (or 34.8%) cheaper (Newegg)…

I believe the initial cost is considered worth it in a professional context, where productivity of an expensive employee over time can make up for the higher initial outlay.

 

[Alexko]

While I'm not discounting drivers as a major factor, isn't it also the case that a lot of those apps stress geometry performance very heavily, which Fermi should excel at?

That may play a part, but the Quadro FX 4800 is GT200-based, and still does better (sometimes much better) than the FirePro in several benchmarks. Considering the Quadro 5000 has only 352SPs clocked at 1026MHz, I think the FirePro shouldn't have too much trouble with it, even in relatively geometry-heavy applications, provided it had appropriate drivers.

I believe the initial cost is considered worth it in a professional context, where productivity of an expensive employee over time can make up for the higher initial outlay.

Hmm… I guess it makes sense, at least in some cases. But there are lots of low-end, slow professional cards that presumably still sell well. E.g. http://www.nvidia.com/object/product_quadro_fx_380_us.html (16SPs).

 

[neliz]

some OEM's only still sell things like an FX380 as the only possible graphics update for a machine. The same goes for older C2D laptops where older FX cards are the only graphics option available.

 

[Psycho]

Not really, it's pretty straightforward. The feature set and performance of "professional" graphics cards are much better than their consumer counterparts when it comes to professional apps. The "professional graphics market" is the market for those cards, simple as that. Besides, what makes you think that the sort of people who cheap out and use consumer cards with professional apps would have a preference for Radeon? There's no evidence at all to support that.

My point is that many (never said all) of these professional apps (or their use, ie some may only make smaller models) don't require or even benefit from a workstation card. So the market for graphics card for professional apps is NOT only the workstation cards. We could limit the "market" to those applications where it's important - but then you'll be selling a lot of quadros to the non-professional graphics market
Specifying the professional graphics market as those who by those 2 particular brands is what I call circular definition. What if AMD dropped the FirePros, uncrippled the Radeon drivers and pushed Radeon to the "professionals" - would the Quadro then have 100% of the "professional graphics market"?
And I ofcourse never implied they would prefer radeons (unless you don't think "gamer cards" include Geforce...).
So, to the point, Geforces (and Radeons) may be a bigger threat to the Quadro sales than the FirePros - but that won't show in these reports.

And naturally even the low end Fermis will do really well as QuadroFXs, as the apps are so focused on geometry, while basicly requiring no shader power (I think I have yet to see something that couldn't be done in SM2.0). And regarding driver optimizations it's of course going both ways with a ~90% market share.

 

[Sound_Card]

I know a few people that buy consumer cards and run the quadro drivers on them.

 

[trinibwoy]

Specifying the professional graphics market as those who by those 2 particular brands is what I call circular definition.

It's not just that. It's probably defined more by what OEM's choose to equip in their workstation products. Radeon vs FirePro and Quadro vs Geforce are just words. There's no benefit to AMD to start flogging "Radeon" to the professional market and it will probably result in the target market taking them less seriously.

But for argument's sake let's say they did that and killed the FirePro line. What would be the point? The market is obviously not price sensitive cause AMD's lower prices haven't done squat for them. And whatever is lacking in the FirePro line won't magically be fixed if they start selling them as cheap Radeons instead.

And I ofcourse never implied they would prefer radeons (unless you don't think "gamer cards" include Geforce...).

Ah, I misread. Thought you were referring to the damage to Nvidia's share of that market, not Quadro's share.

 

[dkanter]

Actually I think the workstation market is both defined by IHVs and what they put in their systems and also what ISVs will certify.

If you are running AutoCAD for business, you aren't going to monkey around with a hopped up Radeon or GeForce...since you won't get any support that way.

DK

 

[Psycho]

But as least the XBitlabs piece clearly writes Quadro and FirePro, not NVIDIA and ATI. Which lead me to believe the "market" is defined by the sales of those particular product lines and not the users/uses. And you'll both have Quadros not running any "workstation apps" (they just happen to be sold with the wanted laptops etc) and 3D professionals not using workstation cards, because their particular application/use doesn't need it.

 

[Erinyes]

Actually I think the workstation market is both defined by IHVs and what they put in their systems and also what ISVs will certify.

If you are running AutoCAD for business, you aren't going to monkey around with a hopped up Radeon or GeForce...since you won't get any support that way.

DK

I read somewhere that like 80% of Fortune 500 companies actually specify that only Nvidia Quadro cards can be used for their GPU accelerated software. Is that true?

 

[Silent_Buddha]

It wouldn't surprise me. There used to be a frequent poster on these boards who worked for a company that made software for professional use. Their software would only work with Nvidia cards because they would not validate or program for ATI cards.

Regards,
SB

 

[chavvdarrr]

It wouldn't surprise me. There used to be a frequent poster on these boards who worked for a company that made software for professional use. Their software would only work with Nvidia cards because they would not validate or program for ATI cards.

Regards,
SB

I believe pro software uses OpenGL, which was ATi weak point for years.
I still remember when in 2006 someone promised fully rewritten OpenGL driver ...

 

[Kaotik]

I believe pro software uses OpenGL, which was ATi weak point for years.
I still remember when in 2006 someone promised fully rewritten OpenGL driver ...

Didn't that in fact pop up at some point?