AMD Q414 results

[Maxima1]

The vast majority is iGPU only. People play games on them.

Then they are ignorant, so how does that help your argument that the AMD option is better? You also aren't considering what games they're playing, how often, if they have a desktop, etc.

Kaveri is decent for games on a laptop, but you can find Intel + dGPU laptops that are around the same price and have better performance and battery life. It's difficult for AMD to offer great value in this area considering Intel's CPU advantage and AMD's system memory solution for graphics.

If we're talking desktop, I see no reason to go with the APUs.

 

[zentan]

Writing numerous times a thing doesn't change reality.
If the majority of the users are using integrated graphics doesn't mean the majority of them game on it necessarily or they require more powerful ones for gaming or that's there primary concern.
Anyway what about quick sync,for years amd lagged behind with their answer to quick sync.
I know 4k is not that popular or is a concern for most people..may be very few guys at present are concerned for it but AMD lagged behind when it comes to 4k decode.
Those who play some light games on facebook or flash games..extra GPU horse power means very little to them.And for most office,school,college,etc guys having a discrete card doesn't make sense if not required,both intels and amd's integrated chips are enough for them.
There's no denying AMD's integrated GPU performance is quite on a different level,but why ignore cpu part,price metrics,delays.
AMD simply needs to come in time and deliver,not like announce
a8 7600 some months ago and retail availability happens much later.
In countries like India and China amd's offering for desktop APUs are even worse.Only few Beema based budget laptops are worth considering,some of them are 10-15% cheaper than i3 4010u that's what makes them a choice.Everywhere else they are absent.Kaveri laptops are not present in those region.
They have a10 7850k,a10 7700k priced in the range on 4th gen i5s there..no way they are gonna be very good choice there in most cases.Last I checked a8 7600 wasn't available in India.
Even if we leave APUs and look towards fx line-up,they yet haven't yet launched fx6350 there,though fx 6300 is still a good choice there.Even fx6300 was very late there.Simply put they are poor executors of markets which may be due to lack of resources,feedback or other issues.
With a product portfolio like AMD have if they delay so much they have to pay the price.
AMD's processor/apu market share is not that great and that's more than enough for proof how well they are doing.

Hopefully the next releases would be lot better from amd with proper timeline.

 

[Abwx]

Kaveri is decent for games on a laptop, but you can find Intel + dGPU laptops that are around the same price and have better performance and battery life. .

This summarize the thing, AMD s APU can process all tasks decently while Intel has to rely on dGPUs to compensate for their lacks and mediocre perfs in one of the most practiced usage, games, and no, battery life with a dGPU cant be better...

 

[cbn]

Out of ~330M iGPUs sold i will take off ~120M dGPUs. So PCs/laptops with iGPUs only are more than double of all the dGPU PCs/Laptops sold.

Of those ~120M dGPUs, 50-60% are sub $100 Grapics Cards like GTX720-730-740 and AMDs R3/5 etc.

All in One

Just looking at newegg, there are 37 Celerons(BayTrail and Ivy/Haswell), 66 with Pentium (BayTrail and Ivy/Haswell), 90 with Core i3, 55 with Core i5 and 26 with Core i7.

The vast majority is iGPU only. People play games on them.

PCs
From newegg again, there are hundreds of iGPU only PCs. People play games on them.

Notebooks
1: Vast majority is iGPU only. People play games on them.
2: There are laptops with low end dGPUs that the iGPU like HD4400-4600 is faster people play games on them.

Anyone believing that High-end gamers ($200+ dGPUs) are the top of the food chain is badly mistaken. We are the minority.

AtenRA, Yes mobile devices are increasing and therefore iGPUs are increasing. But does this mean AMD should be pushing for very large igpus on desktop? (245mm2 on 28nm is trying pretty hard IMO)

 

[Maxima1]

and no, battery life with a dGPU cant be better...

You don't have to use it.

 

[Idontcare]

Out of ~330M iGPUs sold i will take off ~120M dGPUs. So PCs/laptops with iGPUs only are more than double of all the dGPU PCs/Laptops sold.

Of those ~120M dGPUs, 50-60% are sub $100 Grapics Cards like GTX720-730-740 and AMDs R3/5 etc.

All in One

Just looking at newegg, there are 37 Celerons(BayTrail and Ivy/Haswell), 66 with Pentium (BayTrail and Ivy/Haswell), 90 with Core i3, 55 with Core i5 and 26 with Core i7.

The vast majority is iGPU only. People play games on them.

PCs
From newegg again, there are hundreds of iGPU only PCs. People play games on them.

Notebooks
1: Vast majority is iGPU only. People play games on them.
2: There are laptops with low end dGPUs that the iGPU like HD4400-4600 is faster people play games on them.

Anyone believing that High-end gamers ($200+ dGPUs) are the top of the food chain is badly mistaken. We are the minority.

How did Peddie get a 52.8% CAGR for iGPU? Just looking at the data in the graph, doubling roughly 3 years, one can clearly see the CAGR is on the order of 25-26% (half of what Peddie reports in the graph).

A 50% CAGR would require the IGPU units to increase by 50% every year, something that one can easily observe is not happening in that graph.

It took 3 yrs to grow from 100m to 200m units (26% CAGR), and another 3 yrs to grow from 200m to 300m units (15% CAGR). Still better than flatlining like dGPU has, but not the >50% CAGR that the peddie graph misleads the reading into thinking is happening in the iGPU market today. (CAGR is probably 10% now)

 

[Phynaz]

Did you asked them your self ?? :whiste:

Do they know what AMD APUs are capable of ??? Does the sellers in the retail shops explain how much faster the AMD APUs are compared to the competition when it comes to the GPU performance ??

Again, casual user going to buy an OEM iGPU PC for everyday work and ligh gaming, will be way better suited with an AMD 45-65W TDP APU than any Intel counterpart at the same or lower price.

Everyone wants to play games now days, but nobody wants to spend a fortune for a PC. An AMD APU like the A8-7600 has adequate performance to do everything a casual user wants at a lower price than the Intel CPU.

Unfortunately, the vast majority of retailers are promoting Intel products due to rebates and customers end up with inferior or overpriced PCs for their needs.

It seems logic defies you.

 

[AtenRa]

How did Peddie get a 52.8% CAGR for iGPU? Just looking at the data in the graph, doubling roughly 3 years, one can clearly see the CAGR is on the order of 25-26% (half of what Peddie reports in the graph).

A 50% CAGR would require the IGPU units to increase by 50% every year, something that one can easily observe is not happening in that graph.

It took 3 yrs to grow from 100m to 200m units (26% CAGR), and another 3 yrs to grow from 200m to 300m units (15% CAGR). Still better than flatlining like dGPU has, but not the >50% CAGR that the peddie graph misleads the reading into thinking is happening in the iGPU market today. (CAGR is probably 10% now)

This is not annual, this is the CAGR from 1998 to 2010

Formula should be

((Value at end/value at start) ^ (1/years)) - 1

= ((330/2)^(1/12))-1

= 0,53

 

[mrmt]

How did Peddie get a 52.8% CAGR for iGPU? Just looking at the data in the graph, doubling roughly 3 years, one can clearly see the CAGR is on the order of 25-26% (half of what Peddie reports in the graph).

There's another missing line in that graphic. Given that the amount of CPUs sold today isn't growing nowhere near 52.8% CAGR, those extra iGPU growth must be at the expense of something already existent, and it is at the expense of chipset graphics. This graphic is relevant for chipset and MB makers only.

 

[TurtleCrusher]

18.15% of all Steam gamers use an Intel IGP as of last month. The vast majority of gamers use a lower-end mainstream GPU or APU/IGP. We're talking likely 60% or more of the market uses APU class or similar performing GPUs as primary gaming graphics cards. Any talks of dismissing the importance of integrated graphics shows how far removed from reality many of you are. The fact you'd even try to talk about business sense while overlooking this is absolute idiocy.


Another thing to note is the market-share of AMD is actually quite high for how little they are known to the common consumer. Just over 25% of the gaming market isn't anything to sneeze about, especially since their budget is a hell of a lot less than 1/4th of Intel, not having access to the latest fabs and zero marketing.

 

[TurtleCrusher]

It seems logic defies you.

People also buy landrovers despite the clear knowledge you'll have to replace it during or soon after your warranty expires. People with money still buy them like candy.

 

[Ramses]

People also buy landrovers despite the clear knowledge you'll have to replace it during or soon after your warranty expires. People with money still buy them like candy.

That is extremely true.


And there are a lot of interesting takes on Steams hardware survey. Personally I think it has great potential but it really half assed data wise.
Googling around one can find some interesting writeups on it.

Graphics cards remain scattered about the gaming landscape like the detritus blowing in the wake of a beautifully rendered explosion. Steam's most popular, for instance, is Intel's HD Graphics 4000. Thanks, laptops. Of the real 3D cards, Nvidia's GTX 660 is their most popular (2.2%), thanks to its neat balance between price and graphicsability, while AMD are slightly behind with ATI Radeon HD 7850 (1.25%). Overall, Nvidia are currently winning the eternal war for our pixels, with 51.67% of the share of Steam's survey taking userbase. Although AMD probably can't hear Nvidia's gloating over the sound of all the next-gen console cards they're making.

 

[Idontcare]

This is not annual, this is the CAGR from 1998 to 2010

Formula should be

((Value at end/value at start) ^ (1/years)) - 1

= ((330/2)^(1/12))-1

= 0,53

That is incorrect. The value at the start was zero, by definition.

If Peddie is computing CAGR from time zero in the same way you demonstrated by choosing to take the start value as "2" then no wonder he reports such a fantastically useless CAGR value.

But where is the data for the last half-decade? This is 2015, the graph ends at 2010.

A more useful CAGR valuation is one for the past 5 years, not one from 1998 to 2010.

 

[AtenRa]

That is incorrect. The value at the start was zero, by definition.

If Peddie is computing CAGR from time zero in the same way you demonstrated by choosing to take the start value as "2" then no wonder he reports such a fantastically useless CAGR value.

But where is the data for the last half-decade? This is 2015, the graph ends at 2010.

A more useful CAGR valuation is one for the past 5 years, not one from 1998 to 2010.

If you look closely at the graph, you will see that 1998 ends with 2-3M iGPUs. Not only that, but it is written in the graph that the CAGR is from 98-2010

edit:

Those are closely to current time

http://jonpeddie.com/publications/market_watch/

The one you are looking for (2013 to 2017)
http://jonpeddie.com/press-releases/details/the-cg-market-will-exceed-149-billion-by-2017/

 

[TurtleCrusher]

That is extremely true.


And there are a lot of interesting takes on Steams hardware survey. Personally I think it has great potential but it really half assed data wise.
Googling around one can find some interesting writeups on it.

In my opinion, the Steam hardware survey is pretty close to encompassing the entire PC gaming landscape in terms of hardware specifications. Not just that, but numbers updated monthly so we get a good view into what people are actually buying and new-tech adoption trends. If someone plays games on PC other than Facebook, the gross numbers are high enough that the margin of error for market speculation is likely insignificant. I trust it more than numbers from companies or outside sales forecasters. Valve might be just the one without a horse in the race.


Just my $0.02.

 

[Ramses]

In my opinion, the Steam hardware survey is pretty close to encompassing the entire PC gaming landscape in terms of hardware specifications. Not just that, but numbers updated monthly so we get a good view into what people are actually buying and new-tech adoption trends. If someone plays games on PC other than Facebook, the gross numbers are high enough that the margin of error for market speculation is likely insignificant. I trust it more than numbers from companies or outside sales forecasters. Valve might be just the one without a horse in the race.


Just my $0.02.

I agree on trusting Valve for the data, I just wish it wasn't opt-in.
That's my only beef with it really. Maybe that they were a bit more transparent with how when and what was collected.

My suspicion is that people who are savvy enough to be trained to click "no" on any optional survey/popup or anything like that from years of bad online spam experience, are the very one's that would have more interesting hardware. And I've personally never seen the question at all, I've used steam since it was new. And I think it's a ton of laptops too.
It's interesting data and I don't distrust Valve, I just wish there was more to it. With 75 Million users, if I was a game developer doing anything remotely modern 3D and looked at those results, I think I'd go write for consoles.
Which is often what seems to be happening, coincidence?


All of this continues to reinforce my dislike for the principles and philosophy behind iGPU. Eight years ago you could buy a motherboard with decent for the day GPU's integrated into the chipset, on the board. Or you could buy one with basic Windows grade stuff on the board. Surely in eight years that could have improved for those that need/want it, CPU's could not be burdened by this genpop GPU stuff, and they could not put everything including the kitchen sink on every single motherboard. It's a middle finger with a bow on it to enthusiasts imo.

 

[gammaray]

i don't understand the igpu philosophy either, it's marketing bs imho. Anyone, or any system builders can easily sell a simple mobo with a simple videocard to any customers, and it's working perfectly. how is a "leader" chip like a A7850 with a good mobo, any better than a normal mobo and say a gtx750 ?

 

[monstercameron]

i don't understand the igpu philosophy either, it's marketing bs imho. Anyone, or any system builders can easily sell a simple mobo with a simple videocard to any customers, and it's working perfectly. how is a "leader" chip like a A7850 with a good mobo, any better than a normal mobo and say a gtx750 ?

But less complexity and points of failure are more valuable imho. Why attach a nvidia gt 630 or r7 240 rather than a a10 7850k. If the apu dies it would be easier to debug and replace it.

 

[Qwertilot]

Also an overall smaller, quieter machine and cheaper to make - the consoles didn't end up that way by accident

In terms of gaming machines, the major issue with stuff like the a10 7850k is that they're just not fast enough yet. HBM and stuff might well change that for a fair chunk of the market.

 

[Ramses]

I couldn't care less really about an apu beyond what it's taking away. Ten years ago(or more I guess) it was onBOARD video, and that was fine since those of us that didn't need/want could not buy them. What now, a few xeons without? I get it and I don't expect it to change, but I think it's part of the continuing dumbing down and consumerification of my hobby and I will not smile and nod approval.

 

[witeken]

This is not annual, this is the CAGR from 1998 to 2010

Formula should be

((Value at end/value at start) ^ (1/years)) - 1

= ((330/2)^(1/12))-1

= 0,53

How does CAGR even make sense when obviously the trend is linear. And now it's flat.

CAGR makes sense whit stable exponential growth, like Intel's data center group revenue, or Moore's Law.

 

[DrMrLordX]

But less complexity and points of failure are more valuable imho. Why attach a nvidia gt 630 or r7 240 rather than a a10 7850k. If the apu dies it would be easier to debug and replace it.

There are more benefits than just that. One can apply a single cooling solution to both the CPU and iGPU. I personally find this to be quite satisfying. All the work I went through delidding/relidding, lapping, and "over-cooling" my 7700k with an nh-d14 provides me with twice the benefits (effectivley) that it would have with a standalone CPU.

Once Broadwell and DX12 become fairly ubiquitous throughout the industry, you can expect APUs to have new life in games. The gamer ecosystem will change, perhaps dramatically.

The ubiquity of GPU compute is a fishier topic. The tools for GPGPU are out there right now (stuff like C++ Amp, aparapi, etc). It's just a matter of getting people to use them. Heck, there are plenty of applications out there that aren't even making use of AVX/AVX2 that could do so.

 

[monstercameron]

There are more benefits than just that. One can apply a single cooling solution to both the CPU and iGPU. I personally find this to be quite satisfying. All the work I went through delidding/relidding, lapping, and "over-cooling" my 7700k with an nh-d14 provides me with twice the benefits (effectivley) that it would have with a standalone CPU.



Once Broadwell and DX12 become fairly ubiquitous throughout the industry, you can expect APUs to have new life in games. The gamer ecosystem will change, perhaps dramatically.



The ubiquity of GPU compute is a fishier topic. The tools for GPGPU are out there right now (stuff like C++ Amp, aparapi, etc). It's just a matter of getting people to use them. Heck, there are plenty of applications out there that aren't even making use of AVX/AVX2 that could do so.

I guess we are a rare breed, having used Kaveri and thought it a good experience. I kinda want to do a user experience experiment at a local mall to see what the layman consumer thinks. I know it is hard to find objective data from technical forums when we can be inside such a bubble.

 

[mrmt]

I guess we are a rare breed, having used Kaveri and thought it a good experience. I kinda want to do a user experience experiment at a local mall to see what the layman consumer thinks. I know it is hard to find objective data from technical forums when we can be inside such a bubble.

The problem is not convincing customers to buy AMD APUs, but convincing the supply chain it is good for them to bring AMD APUs to the customers. Given the inherent cost handicap (245mm^2 die vs Intel 100mm^2 die) I don't think there's enough money for the supply chain to bother with the APU.

 

[monstercameron]

The problem is not convincing customers to buy AMD APUs, but convincing the supply chain it is good for them to bring AMD APUs to the customers. Given the inherent cost handicap (245mm^2 die vs Intel 100mm^2 die) I don't think there's enough money for the supply chain to bother with the APU.

Ok I'll agree with you for some of that, but I'm gonna need a source on the "cost handicap" do you have a source on the pricing oems are getting for Intel and amd.