[ H ]: AMD Gains 4.8% Market Share in Q1 2016

[sontin]

Mercury's numbers are totally different from AMD's published one:
http://www.mercuryresearch.com/graphics-pr-2016-q3.pdf

So, AMD sent a press release with fake numbers to keep the real numbers from the public?! Or did they just want to pump their share price?

 

[Erenhardt]

Mercury Research isreporting the shares as a four-quarter volume-weighted average

What that means:
for 2016Q2 there are included numbers from:
2016Q2, 2016Q1, 2015Q4, 2015Q3
For 2016 Q1:
2016Q1, 2015Q4, 2015Q3, 2015Q2
For 2015Q4:
2015Q4, 2015Q3, 2015Q2, 2015Q1
For 1015Q3:
2015Q3, 2015Q2, 2015Q1, 2014Q4

The increase in 2016Q2 is taking into account the decrease which happened in 2015 and 'smooths out' the spike . I color-coded quarters to help understand what is going on.

 

[sontin]

Doesnt make any sense. Mercury is reporting numbers for the quarter. Their market share for Q2 2016 shows the shipment for Q2.

Mercury is not reporting that AMD has won 7,2 points in the desktop segment. Mercury calls AMD wins "modest growth".

 

[Erenhardt]

It is in the paragraph above the table:
http://www.mercuryresearch.com/graphics-pr-2016-q3.pdf
It does exactly what it says on the tin. Smooths spikes.

If for the next 2 quarters market share stays the same, they will still report increase because in those 4 quarter avg. will not include low numbers from 2015 sales, but a stable state of 2016 and a share spike in Q2

 

[USER8000]

So, AMD sent a press release with fake numbers to keep the real numbers from the public?! Or did they just want to pump their share price?

Nope:

http://www.amd.com/en-us/who-we-are/newsroom
http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-irhome

You seen any press releases about market share there??

It is in the paragraph above the table:
http://www.mercuryresearch.com/graphics-pr-2016-q3.pdf
It does exactly what it says on the tin. Smooths spikes.

If for the next 2 quarters market share stays the same, they will still report increase because in those 4 quarter avg. will not include low numbers from 2015 sales, but a stable state of 2016 and a share spike in Q2

So its an average of four quarters reported in that document.

Mercury Research is
reporting the shares as a four-quarter volume-weighted average to smooth
the noise of seasonal inventory cycles and reveal ongoing share trends.
Additional details are available in the full report supplied
to clients last
week.

What is more surprising is that these numbers won't include Polaris but would include Pascal,as they will be upto the end of June this year,meaning two months of Pascal sales.

 

[sontin]

It is in the paragraph above the table:
http://www.mercuryresearch.com/graphics-pr-2016-q3.pdf
It does exactly what it says on the tin. Smooths spikes.

If for the next 2 quarters market share stays the same, they will still report increase because in those 4 quarter avg. will not include low numbers from 2015 sales, but a stable state of 2016 and a share spike in Q2

This is not what Mercury reports. They are showing the last 5 quarters from a volume perspective (units). They dont add the numbers from the last four quarters and divide them...

Nope:

http://www.amd.com/en-us/who-we-are/newsroom
http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-irhome

You seen any press releases about market share there??

AMD sent it to the press via e-mail. It is an official statement from them.

 

[USER8000]

This is not what Mercury reports. They are showing the last 5 quarters from a volume perspective (units). They dont add the numbers from the last four quarters and divide them...



AMD sent it to the press via e-mail. It is an official statement from them.

The mercury report says the following:

Mercury Research is
reporting the shares as a four-quarter volume-weighted average to smooth
the noise of seasonal inventory cycles and reveal ongoing share trends.
Additional details are available in the full report supplied
to clients last
week.

You are getting confused - its a 4 quarter average in that document,ie,they took the last four quarters and made a weighted average. So what you do is you add the last 4 quarters,and make a weight average.

The report is looking at the average increase over a year,not quarter to quarter shipping changes.

That means it includes last year,when AMD had some pretty bad quarters with the Maxwell releases.

There is more information in the rest of the report and the other figure is the change from last quarter to this quarter in shipping figures.

Plus,there is no indication this is a press release as HardOCP does not say it is one - it means either HardOCP has a subscription or were sent it from Mercury Research.

Its not even stated in the HardOCP article and AMD publishes all its PR releases on the website.

 

[Erenhardt]

This is not what Mercury reports. They are showing the last 5 quarters from a volume perspective (units). They dont add the numbers from the last four quarters and divide them...

That is exactly what they do.
Weighted (by units shipped) average of 4 months.

 

[raghu78]

Key points are AMD gained unit volume but Nvidia made more significant revenue gains due to record setting asp and complete dominance of the USD 400+ desktop GPU market.

http://www.mercuryresearch.com/graphics-pr-2016-q3.pdf

"AMD’s average total share increased for the first time in more than four years. The last time the company experienced an increase was the first quarter of 2012. Both desktop and mobile shares were up this quarter.

Nvidia lost share in the both segments, but average desktop share remains up more than five points on year. Record-setting improvements in average selling prices due to high-end GPU demand allowed Nvidia to increase revenues substantially even with fewer units shipped."

 

[frozentundra123456]

Wow, that is some very "creative" number crunching there. What actually happened based on the total market share:
Year over year Q2 15 to Q2 16: Total share down 1.7%, desktop down 0.8%, laptop up 1.8%.

Thanks Sontin for giving us the real numbers.

 

[Erenhardt]

Wow, that is some very "creative" number crunching there. What actually happened based on the total market share:
Year over year Q2 15 to Q2 16: Total share down 1.7%, desktop down 0.8%, laptop up 1.8%.

Thanks Sontin for giving us the real numbers.

No, again. Those are a year average values. We are not discussing that here. Concentrate on the quarter results, or skip derailing/trolling.

 

[Abwx]

Wow, that is some very "creative" number crunching there.


Thanks Sontin for giving us the real numbers.

Lol...

Mercury Research is reporting the shares as a four-quarter volume-weighted average to smooththe noise of seasonal inventory cycles and reveal ongoing share trends.

So it s an average based on 4 quarters, hence to gain 2% on this basis require 8% more markeshare on the 4th quarter assuming that the number was flat the 3 previous quarters, it could be of course 4% for each of the last 2 quarters...

There s 2.2% MS difference between the average of the 3 first quarters and the 4th one, so they actually took at least 8.8% marketshare in the last quarter to get to the final number, so much about stats creativity without knowing basic arithmetic..

 

[frozentundra123456]

No, again. Those are a year average values. We are not discussing that here. Concentrate on the quarter results, or skip derailing/trolling.

I know what the numbers are. That is why I said it was a very "creative" way to calculate market share. Year over year change is a very common way to look at sales results, and tells a more accurate story than artificially devising a moving average so that it includes a very low quarter in order to make the most recent numbers appear better. Even then, AMD did not increase their share 4.8% of the total market, only 4.8% from their dubiously calculated previous share.

 

[Erenhardt]

I'm fairly certain you have that one wrong again.
Pure quarter numbers are even higher than the creative data from sontin's link, where the spike in the last quarter is flattened with very weak last 3 quarters and a year back numbers are inflated thanks to better share 5, 6 and 7 quarters ago.

 

[sontin]

These numbers dont make sense when you calculate them. You can see it with the Q4 2015 and Q1 2016 numbers which are mathematically impossible to "achieve".

So either Mercury's calculation is way off or they are just showing the actual shipments.

 

[nurturedhate]

oR TY

These numbers dont make sense when you calculate them. You can see it with the Q4 2015 and Q1 2016 numbers which are mathematically impossible to "achieve".

So either Mercury's calculation is way off or they are just showing the actual shipments.

Or you don't like the news from a site you can't hope to refute so it makes no sense to you.

 

[ShintaiDK]

Seems its based on this:
http://www.mercuryresearch.com/graphics-pr-2016-q3.pdf

It only raises more questions.

 

[Headfoot]

I know what the numbers are. That is why I said it was a very "creative" way to calculate market share. Year over year change is a very common way to look at sales results, and tells a more accurate story than artificially devising a moving average so that it includes a very low quarter in order to make the most recent numbers appear better. Even then, AMD did not increase their share 4.8% of the total market, only 4.8% from their dubiously calculated previous share.

Do you disagree with their reasoning for using the moving average (to smooth out seasonal sales spikes)? This is a fundamental problem with year-over-year. If you measure back-to-school to back-to-school (e.g. August to August) laptop sales will overwhelm other sales and give you incorrect analysis. If AMD sold lots of desktop cards and few laptop cards you would incorrectly miss the desktop sales on a Aug. to Aug YoY. I would venture to say laptops weigh heavily during back to school but discrete graphics comes back for Holiday season and Spring post-Computex

 

[Abwx]

These numbers dont make sense when you calculate them. You can see it with the Q4 2015 and Q1 2016 numbers which are mathematically impossible to "achieve".

So either Mercury's calculation is way off or they are just showing the actual shipments.

You miss the fact that these are aggregated numbers, wich means that the actual numbers for each quarter are concealed by JPR, you can of course extract the real numbers but this require a little more than averaging a succession of quarters..

Anyway you are aknowledging that you understand nothing about the underlying maths yet you pretend that AMD numbers from OC are irrelevant, rofl...

 

[exar333]

At the end of the day, I would argue market share of sales is more important than units, but it is very difficult to grow ASP without strong unit growth. This is promising for AMD and will help grow their user base and hopefully lead to continued growth, sales and profit. There is plenty of market for both companies to do well, and healthy completion is great for us (the community).

 

[Headfoot]

The thing is its not either or. We're not limited to considering only market share by $, or market share by units. We can look at both. Thats exactly what the report did. They are both valuable metrics for different purposes

 

[sontin]

You miss the fact that these are aggregated numbers, wich means that the actual numbers for each quarter are concealed by JPR, you can of course extract the real numbers but this require a little more than averaging a succession of quarters..

Anyway you are aknowledging that you understand nothing about the underlying maths yet you pretend that AMD numbers from OC are irrelevant, rofl...

Everybody can calculate the numbers from the Mercury report. The problem is that the Mercury calculation is not designed to fabricate the "current" quarter in an isolation.

I went back and calculated AMD's market share over the last 8 quarters based on the informations (4 quarter average) from Mercury:

Q3: 27,9%
Q4: 18,5%
Q1: 10,7%
Q2: 0,8%
Q3: 1,8%
Q4: 8,3%
Q1: 16,1%
Q2: 29,4%

Guess i missed nVidia's press statement that they had nearly 100% of the "discrete desktop" market...

 

[kawi6rr]

Steam survey = invited to participate.

How can one call it the end all for gpu market share?

There is no option to participate initiated by end users, no option to edit ones results when changing gpus, etc.

Seems silly too me at least to call it a accurate assessment of gpu market share.

Exactly! I have steam on my machine but I've never participated in their survey so my 290 hasn't been counted. I see all the anti AMD fanboys are out in full troll mode again on the forums.

 

[Abwx]

Everybody can calculate the numbers from the Mercury report. The problem is that the Mercury calculation is not designed to fabricate the "current" quarter in an isolation.

And what did i say here ? :

wich means that the actual numbers for each quarter are concealed by JPR,.

I went back and calculated AMD's market share over the last 8 quarters based on the informations (4 quarter average) from Mercury:

Q3: 27,9%
Q4: 18,5%
Q1: 10,7%
Q2: 0,8%
Q3: 1,8%
Q4: 8,3%
Q1: 16,1%
Q2: 29,4%

Guess i missed nVidia's press statement that they had nearly 100% of the "discrete desktop" market...

What you seemed to miss is basic maths courses..

This is total BS, how could you get 0.8% since the 3 previous quarters 4 months average are over 10%..?..

This say that the numbers in isolation cant be lower than 10 but i suppose that thread crapping the thread thanks to such gross mistakes, that personaly i wouldnt dare posting a tenth of it, is not of any concern for you, neither is the fact that it show you as not being credible for the slightest...

 

[tviceman]

If AMD is moving more discrete units, that bodes well to their future and to consumers who like competition. I don't really see what there is to argue or be upset about.