[Hexus]Nvidia pulls away from AMD in graphics card market share

[SirPauly]

It's not Jon Peddie or Mercury Research but curious to see:

http://store.steampowered.com/hwsurvey/directx/?sort=chg

 

[Subyman]

Hmmm... I'm sure this couldn't have anything to do with the tone of the article:



[/I]So the author of the article owns shares of Nvidia and recommends them. The website he writes for owns shares of Berkshire Hathaway and recommends them.

I see a pattern developing here.

Depends on how you look at it. He/they believe so much in their commentary that they put their money where their mouth is. On the other hand, they could be trying to bump their stock. I'm not sure how much influence they have, so the later depends on their viewership and if it could actually have any impact at all on the stock market.

 

[exar333]

$200 R9 290 is even a better deal! (It's a great card really, I have one ref. heater myself )
But did it help them one iota?

We've heard this many times before:
All AMD needs to do (insert something painful, like ridic price slash)

This one is not on AMD marketing department

No one can reinvent your business, when all you have is the choice between losing market share and money,
and losing even more market share and money, or some equally unattractive combination of the two.

I agree that the R9 290 is a GREAT deal, but according to the market share information, it is not compelling enough to sway buyers.

Marketing is MORE than pricing. Pricing helps you, but marketing transcends pricing (good or bad). Sometimes being cheap is actually counter-productive in marketing...

 

[Erenhardt]

AMD would sell more than nv if they give nvidia sticker with it.
The same with CPUs. Full AMD laptop (CPU+GPU) would sell like a hotcakes if it had intel and nv stickers on it.

Todays sticker sells the product most of the time.

 

[exar333]

This is what I see from the graph:

1 - when AMD has a clear performance advantage, they get ~40% of the market.
2 - When NV has an advantage (performance and/or power efficiency) the sky is the limit for market share

This tells me that many NV buyers will simply not purchase AMD. There IS a population of folks who just buy the best (maybe 10-20%?) and the rest are die-hard AMD buyers. Maybe look like this:

60% Nvidia die-hards
20% 'switch hitters'
20% AMD die-hards

I see a lot of recent resistance at the 60% market share where AMD cannot exceed 40% of the market, regardless of performance or efficiency advantages.

Myself - I fall in the switch buyers. I went from the 5870 to the 670, but probably would have gotten a 7970 if I had purchased a few months after the 670 release (7970 price cuts and so forth). I grabbed the 970 because I could get a cheap water block for it (old 670 block) vs. anything else.

 

[f1sherman]

I agree that the R9 290 is a GREAT deal, but according to the market share information, it is not compelling enough to sway buyers.

Marketing is MORE than pricing. Pricing helps you, but marketing transcends pricing (good or bad). Sometimes being cheap is actually counter-productive in marketing...

Ofc. its not enough, when (even with the price difference) $329 970 is simply better more complete package.

I mean that's what market says (,and I would not disagree).

Omega did more for AMD rep than all their tweets(!), countdowns and hardware giveaways.
Mantle boosted their rep. prolly even higher, at least initially. But I am not convinced that's a well spent effort.
For instance, imagine series of Omegas instead of Mantle, and plethora of features.

It's an easy living when you have a good product - only a fool would be unable to market 970/980 well against 290/X.
Hence this one is NOT on AMD marketing

 

[sontin]

It's barely beating AMD's last gen. A poor comparison imo.

It's nearly twice as fast as AMD's latest GPU Tongo while using even less power.

The reason why this is not like Conroe lies in the possibility to push the power consumption through the roof.

 

[bystander36]

I find this surprising. It seems that AMD has been improving with their GPU's, drivers and features. I would have thought they'd start to gain back shares.

 

[wand3r3r]

It's nearly twice as fast as AMD's latest GPU Tongo while using even less power.

The reason why this is not like Conroe lies in the possibility to push the power consumption through the roof.

Uh ok, cherry pick certain metrics to compare a $330-500 card with a sub $200 card to "show" a (wrong/misleading) point.

Conroe wasn't heralded for efficiency. Way to miss the point.

 

[sontin]

I didn't compare the price. Which doesnt make sense anyway.

I used Tongo because it is AMDs latest GPU architecture. And yet Maxwell offers more performance while using around 30W less power. It is like Conroe from an architecture standpoint.

 

[exar333]

I didn't compare the price. Which doesnt make sense anyway.

I used Tongo because it is AMDs latest GPU architecture. And yet Maxwell offers more performance while using around 30W less power. It is like Conroe from an architecture standpoint.

I think it's a fair comparison. Prior to the 970/980 the 750 was used to express the efficiency of Maxwell, even though it wasn't a 'performance part' (in effect, even less so than the 285...)

 

[AnandThenMan]

The bottom line is, consumers are stupid. Yes I said it, consumers buy what they think is the best not what is the best. Sometimes the two are the same, often they are not. Branding trumps actual product merits and value, this is not the fault of Nvidia the blame lays right at the feet of AMD. As a company they never grasped the critical importance of building the brand.

Nvidia has achieved the coveted status of having their loyal customers sell product for them, it's a curious phenomenon in the consumer space. People become disciples for lack of a better word and champion (as well as purchase) the same brands over and over again. It will take a truly terrible run of said product for people to get fed up and switch, and even if they do if the next product cycle turns out to be good they will switch back.

What does this have to do with marketshare? Well everything, AMD has never been rewarded solely based on product and has been punished disproportionately. I've said it before I think we are finally seeing the end of AMD as we know it, enjoy your $900 mid range GPUs. Hopefully another competitor will step in and fill the void or maybe AMD will get bought out by a company that has some marketing sense.

 

[Udgnim]

went from Nvidia 7950 GT, AMD 4870, AMD 7850, AMD R9 290

all I care about is price / performance around the $200 - $300 segment

 

[exar333]

The bottom line is, consumers are stupid. Yes I said it, consumers buy what they think is the best not what is the best. Sometimes the two are the same, often they are not. Branding trumps actual product merits and value, this is not the fault of Nvidia the blame lays right at the feet of AMD. As a company they never grasped the critical importance of building the brand.

Nvidia has achieved the coveted status of having their loyal customers sell product for them, it's a curious phenomenon in the consumer space. People become disciples for lack of a better word and champion (as well as purchase) the same brands over and over again. It will take a truly terrible run of said product for people to get fed up and switch, and even if they do if the next product cycle turns out to be good they will switch back.

What does this have to do with marketshare? Well everything, AMD has never been rewarded solely based on product and has been punished disproportionately. I've said it before I think we are finally seeing the end of AMD as we know it, enjoy your $900 mid range GPUs. Hopefully another competitor will step in and fill the void or maybe AMD will get bought out by a company that has some marketing sense.

Very good points.

I look at a few examples:

-Intel with Netburst
-Nvidia with 'FX' brand-GPUs

These were both examples where a 'popular' company made a BIG misstep and were hurt in market share more than a normal misstep would have entailed. In both cases, they were able to avoid too much market punishment and used their brand-image and marketing to get by (Intel also used some illegal stuff too...)

AMD really only made a similar misstep once (Bulldozer) and still feels the effects. I would even say Phenom wasn't in the same league because they followed with a pretty compelling Ph-II product set.

You are right though...AMD never built the brand. AMD really had just established itself and then it sent away.

 

[SteveGrabowski]

Why are people talking about the AMD price cuts on the 290s not exciting people based on stats from July-September 2014? The price cuts started late October back when the 290's were going for $400 and you'd have to be a moron to choose them over the new $330 GTX 970s. I know on the Black Friday weekend it was hard to find R9 290's, even the overpriced ones like the Vapor-X that were selling for $320.

 

[Red Hawk]

I didn't compare the price. Which doesnt make sense anyway.

I used Tongo because it is AMDs latest GPU architecture. And yet Maxwell offers more performance while using around 30W less power. It is like Conroe from an architecture standpoint.

It's AMD's latest product, but it wasn't intended to be the flagship and go for the performance crown. AMD's own 290 and 280X beat it in raw performance. Your comparison is like comparing the 290X to Nvidia's 750 Ti.

Part of what made Conroe a big deal was its price. Intel was able to undercut the price of AMD's flagship CPUs at the time in a big way while beating them in performance and power efficiency.

 

[sontin]

AMD's x2 processor were overpriced like nothing. What was the price of the FX? $1000?

On the other hand we can use the GTX970: As fast as the 290X, nearly half the power consumption and around $200 cheaper at launch. Sounds like Conroe, huh...

 

[96Firebird]

Why are people talking about the AMD price cuts on the 290s not exciting people based on stats from July-September 2014?

I guess this also begs the question: How did Nvidia gain so much marketshare, with the 900 series only being available for 12 days in Q3? Was it the initial rush of cards on release?

 

[tviceman]

It's barely beating AMD's last gen. A poor comparison imo.

Hmmm 20% faster, 15% smaller, 90% more efficient, better overclocking headroom, 1 billion less transistors, and all on the same node. If that's barely beating, then slap a wig on my head and call me Sally. Even Fermi wasn't as bad off vs. Evergreen or Cayman as Hawaii and Tonga are vs. Maxwell.

AMD is losing on every single metric except price. If your product is 85% as good as the competition's, but costs 10% more to make and you also have to sell it for 20% less the competition, that is an absolutely horrible position to be in. More expensive to make, more expensive power delivery, sell for less, which means AMD is losing on profit and money left over for R&D. And according to these results, their selling less cards AND making less money on each card. How does AMD stop this cycle? If Fiji isn't noticeably better all around in every measurable regard (in both performance metrics and cost metrics) than Maxwell, I truly believe it might be the last discrete GPU cycle we'll see out of AMD.

 

[Red Hawk]

AMD's x2 processor were overpriced like nothing. What was the price of the FX? $1000?

On the other hand we can use the GTX970: As fast as the 290X, nearly half the power consumption and around $200 cheaper at launch. Sounds like Conroe, huh...

Except for the "as fast" part.

With Conroe, even Intel's third-tier CPU was tying or beating AMD's top CPU at the time.

Hmmm 20% faster, 15% smaller, 90% more efficient, better overclocking headroom, 1 billion less transistors, and all on the same node. If that's barely beating, then slap a wig on my head and call me Sally. Even Fermi wasn't as bad off vs. Evergreen or Cayman as Hawaii and Tonga are vs. Maxwell.

AMD is losing on every single metric except price. If your product is 85% as good as the competition's, but costs 10% more to make and you also have to sell it for 20% less the competition, that is an absolutely horrible position to be in. More expensive to make, more expensive power delivery, sell for less, which means AMD is losing on profit and money left over for R&D. And according to these results, their selling less cards AND making less money on each card. How does AMD stop this cycle? If Fiji isn't noticeably better all around in every measurable regard (in both performance metrics and cost metrics) than Maxwell, I truly believe it might be the last discrete GPU cycle we'll see out of AMD.

LOL! Last discrete GPU cycle? The FUD is strong with this thread.

 

[tviceman]

LOL! Last discrete GPU cycle? The FUD is strong with this thread.

It's not FUD. It's my concern. I'm not saying AMD SUXORS or throwing flames around. AMD has more semi-custom chips on the horizon. It's what they want their bread and butter to become. It's guaranteed revenue with very little R&D overhead. They want to specialize a chip for a customer and sell that chip in high volume for 5+ years. Their margins are nowhere anywhere close to what Nvidia's margins are in discrete GPU's. If they continue to lose market share and continue to lose margins, why would they stay in that market?

 

[stahlhart]

Second, and last, warning -- keep this thread on topic.
-- stahlhart

 

[Stormflux]

One segment that AMD has absolutely no ground on is the GPGPU front. nVidia is by far and away the only player. From a CG work perspective, GPU Renderers are hitting amazing feats compared to CPU Renderers these days. There is a large industry transition happening to utilize the GPU even more.

I often see rigs containing 4+ 780Ti cards. People here argue if SLI is worth it. In these rigs they're cramming every PCI-E slot they can. With multiple rigs running concurrently. Despite LuxMark showing high favor for GCN tech, and OpenCL being pretty robust, most of the hard hitters are keeping with CUDA. AMD Drivers being not up to par and CUDA being more refined could be reasons. You could argue that this may be balanced with BitCoin but I don't know the stats for those.

In the professional market, Quadros are still preferred over FirePro cards. For the smaller guy, Geforce are still preferred over Radeons in the same market.

For me personally, if I were to go the GPGPU route, AMD isnt even an option and I've had nothing but great experiences with AMD cards. Even if AMD offers the best price/memory ratio, the support isn't there to add in performance to the equation.

 

[SolMiester]

But notice the drop-off in Q2 2014 though -- AMD is world class, fierce and very strong competitors that bring compelling choice. Just one misstep from nVidia could dramatically change this current landscape or Amd may surprise with a technology marvel.

Q2 2014 = Mining!

 

[AnandThenMan]

Just one misstep from nVidia could dramatically change this current landscape or Amd may surprise with a technology marvel.

There's no way you actually believe this. Nvidia had a massive misstep with Fermi and easily recovered, no way Nvidia makes another Fermi like blunder. Put another way, did YOU buy an AMD card when Fermi was a no show? No you didn't you waited and in fact bought a Fermi GPU IIRC.

This is exactly the reason AMD never gains any real traction, buying habits like the above.