Discussion AMD failure in notebooks really was due to giving igps priority over cpu?

[Shivansps]

Ok i wanted to make this thread to discuss about this matter ever since people started chearing when AMD cut down the IGP in Renoir with some people thinking that the idea was giving "CPU higher priority because IGP priority gave nothing to AMD".

Im going to go back several years to the first notebook i ever had, the MSI U230, this was the time that in small factors your choices were a increible slow Atom CPU+useless IGP, the ION platform (Intel Atom + Nvidia chipset with a OK igp), later replaced with the ION2 platform when Intel did not allowed Nvidia to make chipsets anymore, ION2 was Intel Atom+Intel chipset+Nvidia ION2 gpu in a x1 PCI-E connection used via optimus, and finally AMD Yukon platform, with AMD Neo CPUs (it were the same CPUs but with lower clocks and voltage).
The MSI U230 as one of the best you could get in the Yukon platform, with a dual core L335 and a RS780 chipset with a HD3200 IGP, AMD was having far more CPU perf than the ATOMs, and the HD3200 was hand to hand with the IONs, because the HD3200 was a little slower than Nvidia ION GPUs(specially ION2 that had VRAM), but the CPU perf difference was way too high, so in gaming the HD3200 was better, specially with dual channel, only the ION2 was at top a few times.

Then AMD came up with i think it is the worst they ever did, the small cores. The first APU was the E-350 that was a Dual Core with Bobcat cores and a integrated HD6310 (it was a HD5450 in a IGP), but with single channel ram.
I went ahead and brought one, a HP DM1z, i was planning to give my U230 away and replace it the DM1z... You guys have no idea how dissapointed i was with the DM1z, the Bobcat cores were slower than the ones in the L335, and the IGP was held back by the single channel ram, resulting only in a very small jump gaming in performace. Only batery life were about 1 hour longer.
And thats was the last notebook i owned, by that time i was already working in a computer distrubutor with access to a lot of notebooks and general hardware.

Everything went downhill there, Intel did a far better job in keeping OEMs from using the worthless Atom off the mainstream notebooks(until Baytrail), but AMD small cores were not restricted to small factors and started to go in every type of notebook, this damaged AMD image A LOT, because performance was BAD, really, really bad, even in IGP has Intel HD3000 was outperforming the 80CU IGP in the small core APUs. The small cores are a problem even today as unsold Carrizo-L and Stoney Ridge notebooks models are still around, even with a SSD these things feel slow. And SSD arent common on those models.

And that was not the only problem i saw, the number of small core notebooks was hidding the big core AMD APU models, the big core APU models became rare, but they did a lot better in performance vs the Intel notebooks, but even big cores APUs had both CPU and GPU performance problems vs Intel, as well as using more power. At the end AMD did not had a clear GPU lead over Intel in big cores in notebooks either.

This changed only with the first Ryzen Mobile APUs, but they still had turbo and power efficiency issues that AMD fixed in Renoir.
So what i think is was never about giving the IGP priority, it was about the mistakes they did with the small cores, and they never had a clear lead on the mobile big core APUs either.

 

[A///]

This is false.

People do evaluate between Windows and Macs. In fact, I alternate between them today. Apple does not exist on an island alone.

Try again.

Dude. Have you ever seen the average Windows based laptop and your typical Macbook Air or MacBook? There is a world of a difference in quality. You can't match Mac exterior quality until you're paying 1800-3000. Say what you want about Mac people and overpaying for simple stuff, but there is a sheer quality difference in Apple's OS vs. Windows. I'm not even a Mac fan and I'll admit Apple has a leg up on the industry.

A person who wants a Mac is not going to buy a Windows laptop.

 

[A///]

My point is to refute your point below.

Your point:


My counters:

Graviton is an in-house product Amazon develops at a small scale. It is not meant for resale to third parties. They only compete on EC2 instances.

 

[GodisanAtheist]

OK, this topic went down the pisser.

Anyhow: I like to keep a nice desktop computer around to play more graphically intensive games and I keep a laptop around (my current has one of those 6/12 Intel i7 CPUs and a GTX1050) in the event that I do attend the rare lan party or if I will be traveling and want to game while traveling.

So there is that.

 

[Shmee]

Dude. Have you ever seen the average Windows based laptop and your typical Macbook Air or MacBook? There is a world of a difference in quality. You can't match Mac exterior quality until you're paying 1800-3000. Say what you want about Mac people and overpaying for simple stuff, but there is a sheer quality difference in Apple's OS vs. Windows. I'm not even a Mac fan and I'll admit Apple has a leg up on the industry.

A person who wants a Mac is not going to buy a Windows laptop.

Apple OS is better than Windows most likely, (though not for games) but then you can also put unix based OS's on just about any laptop, so I think the OS thing is irrelevant unless you really want the mac OS in particular. And then there are hackintoshes kinda sorta.

Macbooks have pretty good build quality, but then the current ones are so hard to upgrade and repair...The older ones were much better than the new ones.

 

[tcsenter]

Sorry if I duplicated someone else's sentiments, the answer is "somewhat, but...." AMD has not been competitive on power consumption/efficiency per IPC for several years now (maybe over a decade), either in the server, desktop or mobile space (heck, not even in GPUs). It just matters least (to most users) on the desktop, lesser to those biggish 17-inch laptops with dual hard drives, GPUs, none-to-quiet fans, and big 9-cell batteries. It starts mattering increasingly more as you go 'down', which is where most of the volume has been and will continue to go.

 

[Shmee]

Sorry if I duplicated someone else's sentiments, the answer is "somewhat, but...." AMD has not been competitive on power consumption/efficiency per IPC for several years now (maybe over a decade), either in the server, desktop or mobile space (heck, not even in GPUs). It just matters least (to most users) on the desktop, lesser to those biggish 17-inch laptops with dual hard drives, GPUs, none-to-quiet fans, and big 9-cell batteries. It starts mattering increasingly more as you go 'down', which is where most of the volume has been and will continue to go.

Huh, I don't think this is true. My understanding is that Zen 2 is has better IPC than Intel comet lake or the current Xeons, but also uses less power. Intel wins in some situations with comet lake due to much higher clocks. Current Zen 3 Laptop chips take very little power, and still have good IPC, intel starts coming back with Tiger lake, but those laptops are not here yet.

Now with video cards you may be right, comparing AMD vs Nvidia, at least in some cases. But overall your statement is not true unless you are mainly talking about Bulldozer based parts, but those are not the chips of recent.

 

[Gideon]

Sorry if I duplicated someone else's sentiments, the answer is "somewhat, but...." AMD has not been competitive on power consumption/efficiency per IPC for several years now (maybe over a decade), either in the server, desktop or mobile space (heck, not even in GPUs). It just matters least (to most users) on the desktop, lesser to those biggish 17-inch laptops with dual hard drives, GPUs, none-to-quiet fans, and big 9-cell batteries. It starts mattering increasingly more as you go 'down', which is where most of the volume has been and will continue to go.

This has not been true, ever since Zen 2 was released.

1. Not in Servers
2. Not on Desktop
3. Not in Laptops
(longer answers below)

Tiger Lake seems to be the first product to challenge that (in mobile).

The Stilt had a very good review on initial Zen 2 launch, here are his aggregated Power Efficiency numbers:

I could end the post here, but let's go to more detail as promised:

1. Server: Epyc 7002 review

Power draw (that's a 64 core AMD model vs 28 core Intel model. Both of which have mostly comparable performance per core):



Total performance is up to 2x better on AMD due to more cores (see review link). Power draw is similar. Here is the bottom line from the review:

AMD hit an unqualified home run with its 2nd Generation EPYC platform. This is now the platform to get in the market. Intel has talented engineering teams and in 2020 they will be competitive again. In the meantime, this is AMD’s market. There are many reasons for still buying Intel Xeon Scalable, but for the vast majority of users, unless 2nd Generation Intel Xeon Scalable CPUs are being discounted very heavily, AMD gives you more performance per dollar and per watt. Plus, AMD has a more modern architecture that Intel looks to follow in 2020 with Cooper Lake that it announced this week having almost as many cores.

2. Desktop: Anandtech Zen 2 review

Actual power-draw @ load:

bottom line:

Perhaps the best arguments for the 3700X and 3900X is their value as well as their power efficiency. At $329 the 3700X particularly seems exciting, and gamers will want to take note that it posts the same gaming performance as the $499 3900X. Considering that AMD is also shipping the CPU with the perfectly reasonable Wraith coolers, this also adds on to the value that you get if you’re budget conscious.

The 3900X essentially has no real competition when it comes to the multi-threaded performance that it’s able to deliver. Here the chip not only bests Intel’s mainstream desktop designs, but it's able to go toe-to-toe with the lowest rung of Intel's more specialized HEDT platforms. Even AMD’s own Threadripper line-up is made irrelevant below 16 cores.

All in all, while AMD still has some way to go, they’ve never been this close to Intel in over a decade. This is no longer the story of the AMD that is trying to catch up to Intel; this is now the story of the AMD that is once more a formidable rival to Intel. And, if the company is able to continue to execute as well, we should be seeing even more exciting things in the future.

And, for these reasons, we are awarding AMD's 3rd generation Ryzen processors an AnandTech Editor's Choice Silver award for their value and energy efficiency. AMD has raised the bar indeed.

3. Laptop:

1. Thinkpad 14s AMD's versio better than Intel (in both perfomance and battery life).
2. As is HP Envy 360x vs Specter 360x

Or take a look at this video if you prefer:

TL;DR:
Starting from the 7nm generation, AMD actually holds a significant perf/power advantage

 

[A///]

Zen2 has been incredible in terms of power use. I feel people are still drawing on AMD's history in the dark years when it comes to performance. If Zen3 sees a power reduction again, then it should be a sweet comparison to RKL which has some juicy rumors that should push its power use up.

Apple OS is better than Windows most likely, (though not for games) but then you can also put unix based OS's on just about any laptop, so I think the OS thing is irrelevant unless you really want the mac OS in particular. And then there are hackintoshes kinda sorta.

Macbooks have pretty good build quality, but then the current ones are so hard to upgrade and repair...The older ones were much better than the new ones.

Yep, that's what I was going with. You need to want macOS to buy into it or like the "lifestyle" of the company. I wouldn't repair any high end slim Windows notebooks either to be honest. Not with my fumbling hands. I'd probably snap the screen.

While W10 has made it easier for people to use a PC, I'd still feel comfortable teaching someone how to use MacOS then Windows, ignoring the semantics that a Mac is a PC too.

OT, but everytime I see you post I think you're that rich Youtuber. Hah

 

[LightningZ71]

Getting back to the core of the topic of this thread:

AMD's biggest problem for much of the 2010s was not having the process advantage that Intel had. Intel was often using a significantly better node than AMD, both in ability to clock, and in ability to achieve good performance for lower power levels. It was often far more dense than what AMD was using as well. This left AMD in a significant dilemma: if they tried to compete with Intel on CPU performance in the laptop space, they were going to blow power and thermal budgets in every case. If they tried to compete in having more cores, they had to either sacrifice core size (bobcate/jaguar) or have almost no iGPU while still not achieving much in the way of performance. They had to find some niche that would allow them to show some sort of compelling reason to purchase their mobile products over an equivalently priced Intel product, and that reason was their iGPU. Essentially, if they couldn't beat intel in core performance, they were going to beat them on system value at their target price points by providing more SOMEWHERE without being too bad elsewhere.

AMD's problem, however, was that the price points where they could actually compete at all in were unprofitably low. OEMs that wanted to use their products had to be absolutely brutal in cutting costs, and they did so by using very poor cooling solutions, limited power delivery, and often outfitted APUs that, while capable of using two DRAM channels, were only configured with one. If AMD wanted to keep ANY volume, they had to go along with it. They couldn't demand more R&D at the prices those laptops would have been able to sell for.

You also have to look at the market environment that they were selling into. The mobile market was full of 2C/4T core based laptops and 2c/2t and 4c/4t Atom derived products. Four threads was the max for 90% of the market, and going for a higher thread count by potentially using 8C Bobcat/Jaguar cores was not going to get them anywhere. The places where having higher CPU performance really mattered, which are games and professional level software packages, were still largely single threaded with, at most, a helper thread or two to handle I/O. AMD was in no place to compete on single threaded performance in mobile. Bobcat and Jaguar just didn't have the ability to scale in clock speed, and the construction cores we know to be power hungry and still limited in absolute thread performance even when power limits were completely ignored.

The one thing that AMD had going for them was that you could build a laptop that could play casual games at modest quality settings well enough that you wouldn't need to have a low end dGPU as well. When properly configured, it was... enough. My son had one of those low end HP laptops with a cat core APU and it was OK in the games he tried to play. I had an i5 laptop from work of a similar era that didn't have a dGPU and his laptop always did better with games when I swapped in a personal hard drive in my work laptop to test its performance.

It is my contention that AMD's choice to use a large iGPU in their Mobile APUs that they marketed in mobile was not only a good decision, it was the ONLY choice that they had to stay even remotely relevant in the market. While the end result of limping through that era has been long term harm to their reputation and generally depressed ASPs on their SKUs as a result, it was one of the only reasons that they managed to survive that era at all.

 

[Shivansps]

Getting back to the core of the topic of this thread:

AMD's biggest problem for much of the 2010s was not having the process advantage that Intel had. Intel was often using a significantly better node than AMD, both in ability to clock, and in ability to achieve good performance for lower power levels. It was often far more dense than what AMD was using as well. This left AMD in a significant dilemma: if they tried to compete with Intel on CPU performance in the laptop space, they were going to blow power and thermal budgets in every case. If they tried to compete in having more cores, they had to either sacrifice core size (bobcate/jaguar) or have almost no iGPU while still not achieving much in the way of performance. They had to find some niche that would allow them to show some sort of compelling reason to purchase their mobile products over an equivalently priced Intel product, and that reason was their iGPU. Essentially, if they couldn't beat intel in core performance, they were going to beat them on system value at their target price points by providing more SOMEWHERE without being too bad elsewhere.

AMD's problem, however, was that the price points where they could actually compete at all in were unprofitably low. OEMs that wanted to use their products had to be absolutely brutal in cutting costs, and they did so by using very poor cooling solutions, limited power delivery, and often outfitted APUs that, while capable of using two DRAM channels, were only configured with one. If AMD wanted to keep ANY volume, they had to go along with it. They couldn't demand more R&D at the prices those laptops would have been able to sell for.

You also have to look at the market environment that they were selling into. The mobile market was full of 2C/4T core based laptops and 2c/2t and 4c/4t Atom derived products. Four threads was the max for 90% of the market, and going for a higher thread count by potentially using 8C Bobcat/Jaguar cores was not going to get them anywhere. The places where having higher CPU performance really mattered, which are games and professional level software packages, were still largely single threaded with, at most, a helper thread or two to handle I/O. AMD was in no place to compete on single threaded performance in mobile. Bobcat and Jaguar just didn't have the ability to scale in clock speed, and the construction cores we know to be power hungry and still limited in absolute thread performance even when power limits were completely ignored.

The one thing that AMD had going for them was that you could build a laptop that could play casual games at modest quality settings well enough that you wouldn't need to have a low end dGPU as well. When properly configured, it was... enough. My son had one of those low end HP laptops with a cat core APU and it was OK in the games he tried to play. I had an i5 laptop from work of a similar era that didn't have a dGPU and his laptop always did better with games when I swapped in a personal hard drive in my work laptop to test its performance.

It is my contention that AMD's choice to use a large iGPU in their Mobile APUs that they marketed in mobile was not only a good decision, it was the ONLY choice that they had to stay even remotely relevant in the market. While the end result of limping through that era has been long term harm to their reputation and generally depressed ASPs on their SKUs as a result, it was one of the only reasons that they managed to survive that era at all.

Cat cores IGP was not good at all... it was better than the HD2000, but the HD3000 in the Intel 2nd gen were already better than Zacate IGP, with a way more powerfull CPU core driving eveything.
Remember that Zacate APUs had a integrated HD5450 GPU limited to single channel DDR3.
Also, AMD Kabini IGP in desktop was worse than Haswell Pentium IGP, and Kabini was the desktop version of Jaguar, same IGP was also use in Beema and Mullins. They upgraded the IGP in Stoney Ridge again, but AMD never removed the single channel limit on cat cores, killing IGP perf.

Whats WORSE, AMD had a even SLOWER version of Bobcat, the "Z" APUs... a few people that i know brought some lenovo 14" notebooks with Z bobcats... you dont know what slow is until you have one in your hands. And Z versions, and every low clock and single core versions they did with cat cores was a huge mistake.

 

[Shivansps]

Also not sure why everyone is dismissing the big core mobile APUs, AMD had mobile versions for every one of the big core APU they did, this is a mobile Llano, the first big core APU:

The AMD Llano Notebook Review: Competing in the Mobile Market

www.anandtech.com

This was a few months after Bobcat launch, as everyone can see AMD Llano was a good product for notebooks at its time. So what exactly happen after this? Because everyone started using Bobcat instead.

 

[A///]

Ever heard of random sampling? All internet surveys are opt-in, so what's your point?

A lot of it has to do with the random sampling, ability to opt-out by power users, mixed data coming from cafes (Chinese cafes were removed from data), the use of old hardware and the inability to distinguish desktop from laptop. It doesn't measure new equipment, either. There's a lot of issues with the survey that don't make it very reliable.

 

[alexruiz]

Ok i wanted to make this thread to discuss about this matter ever since people started chearing when AMD cut down the IGP in Renoir with some people thinking that the idea was giving "CPU higher priority because IGP priority gave nothing to AMD".

Im going to go back several years to the first notebook i ever had, the MSI U230, this was the time that in small factors your choices were a increible slow Atom CPU+useless IGP, the ION platform (Intel Atom + Nvidia chipset with a OK igp), later replaced with the ION2 platform when Intel did not allowed Nvidia to make chipsets anymore, ION2 was Intel Atom+Intel chipset+Nvidia ION2 gpu in a x1 PCI-E connection used via optimus, and finally AMD Yukon platform, with AMD Neo CPUs (it were the same CPUs but with lower clocks and voltage).
The MSI U230 as one of the best you could get in the Yukon platform, with a dual core L335 and a RS780 chipset with a HD3200 IGP, AMD was having far more CPU perf than the ATOMs, and the HD3200 was hand to hand with the IONs, because the HD3200 was a little slower than Nvidia ION GPUs(specially ION2 that had VRAM), but the CPU perf difference was way too high, so in gaming the HD3200 was better, specially with dual channel, only the ION2 was at top a few times.

Then AMD came up with i think it is the worst they ever did, the small cores. The first APU was the E-350 that was a Dual Core with Bobcat cores and a integrated HD6310 (it was a HD5450 in a IGP), but with single channel ram.
I went ahead and brought one, a HP DM1z, i was planning to give my U230 away and replace it the DM1z... You guys have no idea how dissapointed i was with the DM1z, the Bobcat cores were slower than the ones in the L335, and the IGP was held back by the single channel ram, resulting only in a very small jump gaming in performace. Only batery life were about 1 hour longer.
And thats was the last notebook i owned, by that time i was already working in a computer distrubutor with access to a lot of notebooks and general hardware.

Everything went downhill there, Intel did a far better job in keeping OEMs from using the worthless Atom off the mainstream notebooks(until Baytrail), but AMD small cores were not restricted to small factors and started to go in every type of notebook, this damaged AMD image A LOT, because performance was BAD, really, really bad, even in IGP has Intel HD3000 was outperforming the 80CU IGP in the small core APUs. The small cores are a problem even today as unsold Carrizo-L and Stoney Ridge notebooks models are still around, even with a SSD these things feel slow. And SSD arent common on those models.

And that was not the only problem i saw, the number of small core notebooks was hidding the big core AMD APU models, the big core APU models became rare, but they did a lot better in performance vs the Intel notebooks, but even big cores APUs had both CPU and GPU performance problems vs Intel, as well as using more power. At the end AMD did not had a clear GPU lead over Intel in big cores in notebooks either.

This changed only with the first Ryzen Mobile APUs, but they still had turbo and power efficiency issues that AMD fixed in Renoir.
So what i think is was never about giving the IGP priority, it was about the mistakes they did with the small cores, and they never had a clear lead on the mobile big core APUs either.

You only post about "how bad AMD is" no matter the product
DON'T BUY them!

Better for us who want a laptop with in high-demand Renoir APU

 

[Kenmitch]

You only post about "how bad AMD is" no matter the product
DON'T BUY them!

Better for us who want a laptop with in high-demand Renoir APU

lol , but it does look that way.

 

[tamz_msc]

A lot of it has to do with the random sampling, ability to opt-out by power users, mixed data coming from cafes (Chinese cafes were removed from data), the use of old hardware and the inability to distinguish desktop from laptop. It doesn't measure new equipment, either. There's a lot of issues with the survey that don't make it very reliable.

All surveys are done through random sampling. There is no reason to believe that one set of users choose to opt out more often than the rest of the users. Chinese cafe PCs being counted multiple times is no longer an issue. The fact that it doesn't distinguish between laptops and desktops is not relevant to the premise of this thread. There is no reason for newer systems to be counted more often than older systems, after all it measures the components that Steam users have in their systems and the relative popularity of those components with respect to the overall user base, it's not supposed to measure the distribution of sales. For that you have things like Mindfactory data.

 

[A///]

All surveys are done through random sampling.

Most surveys that require the most accurate answers have preemptive questions that filter out bad results. This is the norm when taking a survey, at least in higher education. This is called a "screener."

There is no reason to believe that one set of users choose to opt out more often than the rest of the users.

Power users typically care about privacy. While data is likely anonymized, people who are more technically inclined prefer to remain as anonymous as they can, and will choose to limit this data. You can see for yourself you wander over to Firefox chat rooms or boards. Firefox doesn't sell user data, they collect it for their own use, but there's always been a backlash at not allowing users to opt in rather than having to opt out.

Chinese cafe PCs being counted multiple times is no longer an issue.

Please learn to read carefully:

mixed data coming from cafes (Chinese cafes were removed from data)

The fact that it doesn't distinguish between laptops and desktops is not relevant to the premise of this thread.

It makes a world of a difference. The survey doesn't see a mobile variant of a processor. It sees cores and threads. If the distinguation was not relevent to the premise of the thread neither was your bringing up the use of laptops in casual gaming.

There is no reason for newer systems to be counted more often than older systems, after all it measures the components that Steam users have in their systems and the relative popularity of those components with respect to the overall user base, it's not supposed to measure the distribution of sales.

Ah, so the age does not matter. Do you not think it's not an accurate representation with dual cores are from 2006 may be counted in along with new dual cores, or quad cores from 2007 along with modern quad cores? Or AMD's old 6 core systems that are more than twice as slower performing than modern 6 core processors by AMD? All you get is a very generic idea of what's being used. The minute you see more data than what's provided you can accurately paint a picture.

For that you have things like Mindfactory data.

Mindfactory is a small German etailer. That's sales data, not active use data. We're looking for lone systems with the intended hardware. Not 4-5 components purchased and used in various computers. You'd still want an accurate idea of what people are using while accessing your gaming network.

The only viable data Steam offers up are gaming resolution and video cards in use. CPU speeds and CPUs (cores) are unhelpful.

 

[Shivansps]

You only post about "how bad AMD is" no matter the product
DON'T BUY them!

Better for us who want a laptop with in high-demand Renoir APU

it is a hardware discussion forum you cant tell me what to post or what opinion i should have about a product.

 

[Shivansps]

Most surveys that require the most accurate answers have preemptive questions that filter out bad results. This is the norm when taking a survey, at least in higher education. This is called a "screener."


Power users typically care about privacy. While data is likely anonymized, people who are more technically inclined prefer to remain as anonymous as they can, and will choose to limit this data. You can see for yourself you wander over to Firefox chat rooms or boards. Firefox doesn't sell user data, they collect it for their own use, but there's always been a backlash at not allowing users to opt in rather than having to opt out.

Please learn to read carefully:




It makes a world of a difference. The survey doesn't see a mobile variant of a processor. It sees cores and threads. If the distinguation was not relevent to the premise of the thread neither was your bringing up the use of laptops in casual gaming.


Ah, so the age does not matter. Do you not think it's not an accurate representation with dual cores are from 2006 may be counted in along with new dual cores, or quad cores from 2007 along with modern quad cores? Or AMD's old 6 core systems that are more than twice as slower performing than modern 6 core processors by AMD? All you get is a very generic idea of what's being used. The minute you see more data than what's provided you can accurately paint a picture.

Mindfactory is a small German etailer. That's sales data, not active use data. We're looking for lone systems with the intended hardware. Not 4-5 components purchased and used in various computers. You'd still want an accurate idea of what people are using while accessing your gaming network.

The only viable data Steam offers up are gaming resolution and video cards in use. CPU speeds and CPUs (cores) are unhelpful.

Guys this is very simple, it is the only source of information we got, we cant dismiss that on someone opinion without any data to support it.

The main problem i see with steam survey is that every user is asked once a year to participate, so it always has old data that may be months old and not longer accurate.
There is nothing to say about the opt-in.

 

[Shivansps]

lol , but it does look that way.

Well, this is not really a critic, as we are talking about old stuff, this is a subject that have been mentioned several times in several threads.

 

[Hitman928]

Guys this is very simple, it is the only source of information we got, we cant dismiss that on someone opinion without any data to support it.

The main problem i see with steam survey is that every user is asked once a year to participate, so it always has old data that may be months old and not longer accurate.
There is nothing to say about the opt-in.

Does it ask every user once per year? I've had years where it has asked me to participate several months in a row and just came off of ~2 years without being asked once. I have friends who have never been asked (that they can recall). Do you know how Steam actually conducts the survey?

 

[Shivansps]

Does it ask every user once per year? I've had years where it has asked me to participate several months in a row and just came off of ~2 years without being asked once. I have friends who have never been asked (that they can recall). Do you know how Steam actually conducts the survey?

When the asia cyber cafe issue came out, Valve said it was once per year, and to me it has been once per year so far.

Valve does not give the exact information on how the system works but they said once per year to each user, and each month a random numbers of users are asked to participate.

What i dont know is what happens if you refuse or if you miss the window, if you asked again in what time or never again.

 

[Markfw]

Guys this is very simple, it is the only source of information we got, we cant dismiss that on someone opinion without any data to support it.

The main problem i see with steam survey is that every user is asked once a year to participate, so it always has old data that may be months old and not longer accurate.
There is nothing to say about the opt-in.

This survey, just the fact that you can opt in or out alone nullifies ANY real useful data. By looking at flawed data, it is worse than no data at all. Sales figures, even to distributors is better. If you have been on these forums long, you will realize that this question has come up, and been argued over more times than I can count. And it has NEVER come close to having a winning decision that it is of any real use, just a flamebait topic.

 

[A///]

This survey, just the fact that you can opt in or out alone nullifies ANY real useful data. By looking at flawed data, it is worse than no data at all. Sales figures, even to distributors is better. If you have been on these forums long, you will realize that this question has come up, and been argued over more times than I can count. And it has NEVER come close to having a winning decision that it is of any real use, just a flamebait topic.

JP is a good source if your company can afford it, but I think their quarterly analysis is free?

 

[jpiniero]

This survey, just the fact that you can opt in or out alone nullifies ANY real useful data. By looking at flawed data, it is worse than no data at all. Sales figures, even to distributors is better. If you have been on these forums long, you will realize that this question has come up, and been argued over more times than I can count. And it has NEVER come close to having a winning decision that it is of any real use, just a flamebait topic.

You don't understand how surveys work. Random sampling is effective if you have enough of a sample size. There's no reason to think people using AMD GPUs are more or less willing to take the survey.

 

[Markfw]

You don't understand how surveys work. Random sampling is effective if you have enough of a sample size. There's no reason to think people using AMD GPUs are more or less willing to take the survey.

Yes I ubderstand how surveys work, but as I said the fact that you can out out, nullifies any useful data. Someone here said this first. A particular group of people can most often say NO to a survey. I for one don't want to share any personal information, and always refuse them. I have 16 video cards active, and 10 not in use currently ! I am sure there are others like me. Miner as we all know use a LOT of cards. I doubt most of them are even on steam (I for one am NOT on steam). I am sure we can thing of hundreds of groups that are not on steam or don't want to share their information. And as somebody else said, is the survey reset every month ? or six months or ???? What about gamers that don't use steam at all ?

Need I go on. This is stupid to argue about this, but its a useless source of ANY valid information about what is in use today.

As to your comment on sample size, here is a scenario that would prove my point....(not saying this is happening) So lets just say that 80% of the steam users are between ages 12 and 20 and play 5 different games for the most part. Well, even a 100% sample size would surely be WAY off of the number of GPUs in ANY category.

100% of crap is still crap.

Was just thinking, recalling from my high school statistics class..(for those that want a more technical reason).. A random sample with a 95% confidence is fine, so long as your population is representative. What I described above guarantees that its NOT a representative population, so there is no confidence about the results.