Was this market share data wrong? CpuBenchmark.Net

[Nocte]

Hello,

 

I am writing a University paper on AMD and Intel, and I was wondering if the data shown here is wrong: http://wccftech.com/amd-takes-10-4-cpu-share-intel-q2-2017-largest-single-quarter-share-gain-history/

I am unsure as if you now go on https://www.cpubenchmark.net/market_share.html the 10.4% difference is not shown, so I am thinking that at the time of the article the data shown by cpubenchmark.net was wrong.

 

What do you think? I cannot afford to provide false data.

[iRileyx]

The one on userbenchmark is only based on the number of cpu's that are tested by people who've ran their benchmarks, its by no means an accurate representation of the actual marketshare.

[Nocte]

Just now, RKRiley said:

The one on userbenchmark is only based on the number of cpu's that are tested in their benchmarks, its by no means an accurate representation.

I know it is not accurate, but it is the closest you can get to an accurate representation.

All surveys are carried out by considering a sample of the whole thing, it would be unrealistic to go and check what all systems in the world are running.

[GDRRiley]

this is just 1 section of the numbers. were not really sure at this point. http://store.steampowered.com/hwsurvey/processormfg/

[Wailmer]

Userbenchmark is an okay site to compare performance, but they are only based on what users provide (like the name says). There are more than a couple of times where the information might be incorrect or the sample size is small enough to skew results.

 

If you're writing a college level paper I'd look for more credible results.

[Spuriae]

cpubenchmark is Passmark, which I'd imagine to be even worse than userbenchmark for accurate market share data. Market share is based on the number of benchmarks run with each processor in a given period of time.

[Nocte]

3 minutes ago, Wailmer said:

Userbenchmark is an okay site to compare performance, but they are only based on what users provide (like the name says). There are more than a couple of times where the information might be incorrect or the sample size is small enough to skew results.

 

If you're writing a college level paper I'd look for more credible results.

Do you have any recommendation?

I will keep looking.

[Wailmer]

7 minutes ago, Eibe said:

Do you have any recommendation?

I will keep looking.

Does your school provide anything like EBSCO host or other scholarly sources? Google Scholar is a decent place to start too.

[Wailmer]

4 minutes ago, Wailmer said:

Does your school provide anything like EBSCO host or other scholarly sources? Google Scholar is a decent place to start too.

Also look for the sources provided in articles, rather than the editorials themselves.

Some quick googling and following some links provided some official numbers:

http://jonpeddie.com/press-releases/details/amd-increased-market-share-in-q2-2016-despite-overall-gpu-shipments-declini

[Rolling Potatoe]

The numbers are based on scores submitted quarterly.

The graph in the article is for 1st of july, so the second quarter of 2017 is finished but the 3rd quarter data only consist of a few days of benchmarks

and is constantly changing (+ this is around where many people took their new ryzen cpus for a spin, so that data point is inflated).

If you look at the website again now the third quarter is finished and data for Q4 is being collected.