[Misleading] AMD market share up 10.4% in Q2 2017 [Updated]

[DocSwag]

22 minutes ago, zMeul said:

1. not a flame war
2. PassMark report is based on crumbs while STEAM has a hugely bigger user base; and even if only 1/2 of that user base submitted the survey it's still vastly larger than PassMark
3. to my knowledge, neither AMD nor Intel disclose actual CPU sales/shipments numbers - so, analysts have nothing to analyze 

1. I was exaggerating  

2. I agree, steam is more trustworthy than Passmark, but it's still not good enough. You're still looking at a very targeted group of people.

3. That's quite untrue. You can make estimates of market share based off of revenue and whatnot. Read this:

http://m.wikihow.com/Calculate-Market-Share

 

Naturally, these aren't super precise exact numbers, but they're good estimates and are much better than estimates based off of steam, an application that is primarily used by gamers and that most people don't use.

[mr moose]

9 hours ago, Jito463 said:

Where I work, we sell refurbished systems in addition to doing custom builds.  There are a lot of 2nd & 3rd gen Core i5 / i7 systems that we still sell, because they're sufficient for your average user.  So, I suppose Intel's stagnation has been good for us, as it helps enable the refurb market.

Not too sure how big the used market is, suffice to say that can only add to how impressive 1-2% is, given that's another entire demographic of users who aren't going to buy a new CPU. 

[leadeater]

4 hours ago, DocSwag said:

3. That's quite untrue. You can make estimates of market share based off of revenue and whatnot. Read this:

http://m.wikihow.com/Calculate-Market-Share

 

Naturally, these aren't super precise exact numbers, but they're good estimates and are much better than estimates based off of steam, an application that is primarily used by gamers and that most people don't use.

AMD, Intel, Nvidia are all very good at breaking down their revenue reports to market segments etc so yea it's not that hard to gauge market share based on those numbers. Also you can gather sales data reports from major distributors too which you can combine those two data sources to get a more accurate picture of what specific SKUs are selling the best etc.

[LAwLz]

On 7/4/2017 at 9:54 AM, tom_w141 said:

 

@Taf the Ghost. Lastly as I said that would indicate that AMD share has gone down since Ryzen, now I know its zMeul and you love to hate on AMD but I'm sure even you can see that Ryzen won't have had a negative impact on AMD's share.

I have no problem believing that. Ryzen is still rare or non-existing in pre-builts and laptops. That means that those (which are the vast majority of PCs sold) will be Intel. 

 

Remember, in order to gain marketshare you need to sell more than your competitor. Do you really think the Ryzen sales from people who build their own PCs will be enough to outweigh all the laptop and pre-built sales, on top of all the custom builds with Intel? I wouldn't be surprised if more FX chips than Ryzen chips are still being sold. 

 

Also, I'd take the Steam survey over PassMark any day of the week. Sure it's not perfect but it should be a fairly decent representation of what gamers have. 

I think it's pretty funny that when PassMark showed huge marketshare gains everyone is happy and takes jabs at Intel. Then more reliable data shows a bleaker picture for AMD and everyone gets mad and tries to say it is not accurate. For crying out loud someone actually said WCCFTech was more reliable than the Steam Survey. It's like people only believe things they want to be true and rejects everything else. 

[Grinners]

9 minutes ago, LAwLz said:

Remember, in order to gain marketshare you need to sell more than your competitor.

To clarify; a company that sells 1/10th of a product will increase it's market share if it sells 2/10ths of a product. Its competitor can still sell 8/10ths of the product (and therefore 'sell more').

 

It's the % change that matters, not who sells more, that leads to market share changes. 

[LAwLz]

11 hours ago, Grinners said:

To clarify; a company that sells 1/10th of a product will increase it's market share if it sells 2/10ths of a product. Its competitor can still sell 8/10ths of the product (and therefore 'sell more').

 

It's the % change that matters, not who sells more, that leads to market share changes. 

Actually you're right. I'm an idiot.

AMD "only" needs to sell a higher percentage than they currently hold. It would be interesting to know what percentage that is though. I would not be surprised if it's lower than the 19% the Steam survey shows considering that Intel is so dominant in laptops and offices.

 

Edit: I got a feeling people click "agree" for the "I'm an idiot" part... ;_;

[tom_w141]

1 hour ago, Grinners said:

To clarify; a company that sells 1/10th of a product will increase it's market share if it sells 2/10ths of a product. Its competitor can still sell 8/10ths of the product (and therefore 'sell more').

 

It's the % change that matters, not who sells more, that leads to market share changes. 

Thanks for saying this before I did. Saved me the trouble of explaining something so simple  

[Taf the Ghost]

51 minutes ago, LAwLz said:

Actually you're right. I'm an idiot.

AMD "only" needs to sell a higher percentage than they currently hold. It would be interesting to know what percentage that is though. I would not be surprised if it's lower than the 19% the Steam survey shows considering that Intel is so dominant in laptops and offices.

I'm pretty sure I've seen 10% come up a lot in the pre-Ryzen launch window.  In the server space, it rounds to 0%. 

 

There's sort of a deep hilariousness to AMD surviving so long because Intel was still paying them for illegally forcing them out of most of the Laptop market.  Though AMD does seem to have a pretty sizeable presence in China, something not to be discounted going forward.

 

And AMD really needs to get Raven Ridge out. That's the chip that'll really put them in a good space in Mobile.

[mr moose]

3 hours ago, LAwLz said:

Actually you're right. I'm an idiot.

AMD "only" needs to sell a higher percentage than they currently hold. It would be interesting to know what percentage that is though. I would not be surprised if it's lower than the 19% the Steam survey shows considering that Intel is so dominant in laptops and offices.

 

I have also seen quite a resurgence in laptops with AMD chips in recent months.  I wouldn't be surprised if that market is helping the current figures in steam as well as elsewhere.  

[Demonking]

Steam hardware survey is inaccurate because corporation that i like who don't give a damn about me is losing lol. Let AMD be in Intel position and they would be doing the same shit you give Intel shit for why because they are a corporation not your friend.

[cj09beira]

22 hours ago, Taf the Ghost said:

I'm pretty sure I've seen 10% come up a lot in the pre-Ryzen launch window.  In the server space, it rounds to 0%. 

 

There's sort of a deep hilariousness to AMD surviving so long because Intel was still paying them for illegally forcing them out of most of the Laptop market.  Though AMD does seem to have a pretty sizeable presence in China, something not to be discounted going forward.

 

And AMD really needs to get Raven Ridge out. That's the chip that'll really put them in a good space in Mobile.

hopefully they have it on stores in time for the back to school sales

[Jito463]

2 hours ago, Demonking said:

Steam hardware survey is inaccurate because

It's inaccurate because of unknown sampling size.  There is literally no way to know how many users submitted the surveys, or even how many were polled in the first place (not everyone gets the surveys).  Surveys - like polls - are inherently inaccurate.  They can give some indicators, but there's literally no way to know if the numbers are true.

 

As Samuel Clemens is often quoted as saying, there's "Lies, d***** lies and statistics".  Surveys/polls are nothing more than statistics, and it's very easy to manipulate statistics to suit whatever message you wish to convey.

 

Maybe it's true, maybe it's not.  The problem is, we can never know for sure, and especially when we have no idea how random or how large the sampling size is.

[LAwLz]

29 minutes ago, Jito463 said:

It's inaccurate because of unknown sampling size.  There is literally no way to know how many users submitted the surveys, or even how many were polled in the first place (not everyone gets the surveys).  Surveys - like polls - are inherently inaccurate.  They can give some indicators, but there's literally no way to know if the numbers are true.

 

As Samuel Clemens is often quoted as saying, there's "Lies, d***** lies and statistics".  Surveys/polls are nothing more than statistics, and it's very easy to manipulate statistics to suit whatever message you wish to convey.

 

Maybe it's true, maybe it's not.  The problem is, we can never know for sure, and especially when we have no idea how random or how large the sampling size is.

Come on... You can't be serious. Inaccurate and "100% perfect" are not synonymous.

Something can be accurate even though we do not know the exact sample size. If I say "blue is the most common favorite color" then would you claim that's inaccurate just because you don't know how many people I asked? What if it turns out I asked every single person on the planet? The conclusion of a survey does not change just because you have more or less insight to the survey. The conclusion always stays the same. You can not claim something is inaccurate just because you lack knowledge about it.

 

Even if we don't know the exact numbers it is safe to say that it is a very big one. Most likely several millions.

Steam has 125 million active users. If we assume that only 1% has gotten the survey (which seems unlikely based on just common sense as we as anecdotal evidence) that's still 1.25 million users. That's a pretty big sample size.

It is also pretty absurd to assume that Valve would manipulate these statistics to favor Intel or AMD. They do not stand to gain anything from it, and it would be quite a bit of work for no apparent reason.

 

If you're going down the "don't trust it because we can never be sure", then you will quickly end up in a situation where you can never be sure of anything ever (except of course, that you exist). That's why you can't look at things in a black and white manner, and have to make reasonable assumptions.

 

I can't find anything wrong with the Steam survey. It is most likely an accurate representation of people who has Steam installed on their computers. The problems only start appearing when you try to apply the Steam statistics, which are sampled from a a specific subgroup of all computer users, and apply it to the general public. Steam's hardware survey is in no way an accurate representation of the average computer user, but it is an accurate representation of the average Steam user (which is basically the same as the average PC gamer).

[79wjd]

4 hours ago, Demonking said:

Steam hardware survey is inaccurate because corporation that i like who don't give a damn about me is losing lol. Let AMD be in Intel position and they would be doing the same shit you give Intel shit for why because they are a corporation not your friend.

Punctuation. Please.

 

But you're completely right, at the end of the day, AMD is a business and as such their main purpose in life is too maximize their profits. However, (some) people like to pretend that AMD is literally Jesus and exists solely to save all mankind.

[Jito463]

1 hour ago, LAwLz said:

Come on... You can't be serious. Inaccurate and "100% perfect" are not synonymous.

Something can be accurate even though we do not know the exact sample size. If I say "blue is the most common favorite color" then would you claim that's inaccurate just because you don't know how many people I asked? What if it turns out I asked every single person on the planet? The conclusion of a survey does not change just because you have more or less insight to the survey. The conclusion always stays the same. You can not claim something is inaccurate just because you lack knowledge about it.

Okay then, unknown accuracy (though I would argue that if you can't know how accurate something is, that inherently makes it inaccurate).  It still doesn't alter that we don't know anything regarding sampling size or number of responses they received during the survey, and thus can't make even a rough estimation on how closely the survey reflects real world usage.

 

1 hour ago, LAwLz said:

If you're going down the "don't trust it because we can never be sure", then you will quickly end up in a situation where you can never be sure of anything ever (except of course, that you exist). That's why you can't look at things in a black and white manner, and have to make reasonable assumptions.

If it helps, I don't trust any survey/poll.  Ever.  They're almost never right and usually lead to false assumptions.  The poll could have shown AMD with 50% market share, and I would have dismissed it just as quickly.

[Okjoek]

I like to see AMD doing well. I know that Kaby Lake and Ryzen are doing great, but I've been considering the option of not upgrading from Ivy Bridge for yet another generation. For gaming and entertainment tasks I do it still "just works"

 

When I do upgrade next I'm going to go super small form factor using only mITX, M.2 storage and low profile or half length GPU.

[LAwLz]

53 minutes ago, Jito463 said:

I would argue that if you can't know how accurate something is, that inherently makes it inaccurate

Then you do not know what "inaccurate" means.

Inaccurate means incorrect or untrue.

Just because I don't know that 34664 * 2343 is ~81 millions doesn't mean it isn't ~81 millions.

You can't say something is false just because you don't know if it is true. Maybe you meant to say that the data is unreliable?

 

57 minutes ago, Jito463 said:

It still doesn't alter that we don't know anything regarding sampling size or number of responses they received during the survey

True, but it would be completely illogical to assume that the sample size is so small it does not hold any value. Like I said before, even if they only get 1% of their users to click agree to the survey that's still over 1 million. That's a lot of users.

 

1 hour ago, Jito463 said:

and thus can't make even a rough estimation on how closely the survey reflects real world usage.

I think it is silly to say that the Steam survey isn't even good enough for a rough estimate of what Steam users run. I would be very surprised if there was a large deviation between the hardware survey and what people were running.

To say that you can't even use the Steam hardware survey to gauge what Steam users are running, not even a rough estimation, is nonsense.

 

1 hour ago, Jito463 said:

If it helps, I don't trust any survey/poll.  Ever.  They're almost never right and usually lead to false assumptions.  The poll could have shown AMD with 50% market share, and I would have dismissed it just as quickly.

So you always dismiss all surveys and polls? It doesn't matter the question, the sample size or any other factors?

That's... not a very bright thing to do.

[Taf the Ghost]

Surveys measure "something". Whether that "something" tells you anything beyond who took the survey is a very hard question to deal with.

[Jito463]

2 hours ago, LAwLz said:

Then you do not know what "inaccurate" means.

Inaccurate means incorrect or untrue.

Just because I don't know that 34664 * 2343 is ~81 millions doesn't mean it isn't ~81 millions.

You can't say something is false just because you don't know if it is true. Maybe you meant to say that the data is unreliable?

I said can't know, not don't know.  I can do the math and confirm if the above is true.  If you said: 34,664 x ????? = ??????, then I can't possibly know the answer.  No different with these surveys.

 

2 hours ago, LAwLz said:

So you always dismiss all surveys and polls? It doesn't matter the question, the sample size or any other factors?

That's... not a very bright thing to do.

That's your opinion.  Like I said, they're fine for an indicator, but I would never assume they're accurate.

[LAwLz]

45 minutes ago, Jito463 said:

Like I said, they're fine for an indicator

Ehm, you said Steam's survey was so inaccurate we couldn't even use it as a rough estimate.

That's why I reacted to your posts. If you just said "Steam's hardware survey is fine as an indicator but I don't think it is 100% accurate" then I would have been totally fine with your posts.

[Jito463]

10 hours ago, Jito463 said:

It's inaccurate because of unknown sampling size.  There is literally no way to know how many users submitted the surveys, or even how many were polled in the first place (not everyone gets the surveys).  Surveys - like polls - are inherently inaccurate.  They can give some indicators, but there's literally no way to know if the numbers are true.

*ahem*

 

@LAwLz

Edited July 6, 2017 by Jito463
Added notify tag for Lawlz