First results from the GZDoom 3.3.0 survey

Here, developers communicate stuff that does not go onto the main News section or the front page of the site.
[Dev Blog] [Development Builds] [Git Change Log]

Moderator: GZDoom Developers

Re: First results from the GZDoom 3.3.0 survey

Postby m8f » Wed Mar 28, 2018 1:39 am

Graf Zahl wrote:One more interesting thing is that we got 3x as many Linux users as we have downloads of the binary, suggesting a large number of people who self-compile.

Do people who install GZDoom via repository (http://debian.drdteam.org) count as who downloaded the binary?
User avatar
m8f
 
 
 
Joined: 29 Dec 2017

Re: First results from the GZDoom 3.3.0 survey

Postby Graf Zahl » Wed Mar 28, 2018 2:36 am

I don't think so. The counter is attached to the link on the downloads page.
User avatar
Graf Zahl
Lead GZDoom Developer
 
Joined: 19 Jul 2003
Location: Germany

Re: First results from the GZDoom 3.3.0 survey

Postby Caligari87 » Wed Mar 28, 2018 8:10 am

Yeah, Linux users who don't compile (like myself) seem to like repositories. Directly downloading a binary is generally a no unless unavoidable. I'd say that probably 90% of your untracked Linux downloads are using debian.drdteam.org.

8-)
User avatar
Caligari87
I'm just here for the community
User Accounts Assistant
 
Joined: 26 Feb 2004
Location: Salt Lake City, Utah, USA
Discord: Caligari87#3089

Re: First results from the GZDoom 3.3.0 survey

Postby Matt » Wed Mar 28, 2018 4:21 pm

When it comes specifically to GZDoom on my home PC I always compile, but when installing GZDoom anywhere else, just as a matter of habit I'll use the repositories and download the binary only as a last resort.
User avatar
Matt
Putting the XD into *xdeath since 2007
 
 
 
Joined: 04 Jan 2004
Location: Gotham City SAR, Wyld-Lands of the Lotus People, Dominionist PetroConfederacy of Saudi Canadia

Re: First results from the GZDoom 3.3.0 survey

Postby Graf Zahl » Fri Mar 30, 2018 1:49 am

I already noticed yesterday that the numbers were starting to change, with a lot more 32 bit binaries being downloaded.
And checking the database today confirmed that by seeing that the number of OpenGL 2 users was increasing far stronger than the rest. It looks like now the Brutal Doom users start to catch up and they seem to run on low end hardware to far bigger proportions than the early downloaders.

As for the rest, not much has changed. Software rendering is still at 10%, Windows XP is insignficant (21 reports from 17 users out of 2347 reporting , one of which is clearly bogus) and still no signs of ancient pre OpenGL2 / Shader model 2 hardware being used at all.


At the low end of supported hardware things are more interesting:

We are still at 0 users using Shader Model 1.4 hardware and 15 users reporting software rendered 2D. Of these 15 users, 9 also report use of the hardware renderer so they can be assumed to have at least semi-modern hardware. Of the remaining 6 users, 5 are on Linux which gives a high probability of being stuck on a system that cannot do hardware accelerated 2D due to the minimum requirement of OpenGL 3, but could if that backend was fixed to run on GL 2 - and only one user on Windows.

So it takes no rocket science that with such low numbers and the feature being one major roadblock, its days are numbered, this will easily be one thing that can be removed without worrying too much, so that the entire 2D code can be cleaned up and consolidated. There's currently 4 different backends here, each with different features and limitations, which makes it very hard to work on it.
User avatar
Graf Zahl
Lead GZDoom Developer
 
Joined: 19 Jul 2003
Location: Germany

Re: First results from the GZDoom 3.3.0 survey

Postby Zenon » Fri Mar 30, 2018 5:16 am

I think the people who are still running Windows XP are the ones that believe the meme that windows 10 is a botnet, thus I'm pretty sure they didn't trust the survey.
They probably don't realize that by not participating, they've shot themselves in the foot and will be victims of GZDoom's potato PC purge :twisted:
User avatar
Zenon
Rabid
 
Joined: 20 Apr 2006
Location: New Zealand

Re: First results from the GZDoom 3.3.0 survey

Postby Graf Zahl » Fri Mar 30, 2018 5:33 am

If that was the case there would be a stronger deviation between 32 bit user engagement and 64 bit user engagement in the survey. But that isn't the case. On the contrary:

User engagement from 64 bit users is 1726 users for 3604 downloads reporting, that's 48%, and user engagement from 32 bit users is 396 users for 725 downloads reporting, that is 54%!

So no, I think we can safely discount that myth. It rather looks like users of old hardware are actually more active in taking part in the survey because they realize it's in their best interest to let their specs known.
The hopelessly paranoid are most likely such a small minority that it doesn't make any difference. And these people cannot be helped anyway, so why bother?
User avatar
Graf Zahl
Lead GZDoom Developer
 
Joined: 19 Jul 2003
Location: Germany

Re: First results from the GZDoom 3.3.0 survey

Postby GFD » Fri Mar 30, 2018 12:49 pm

Those user engagement figures really are shockingly low to me. I consider myself passably knowledgeable on the subject of information security, and I had no issues at all with this whole operation. I really just want to know what the thought process was for all the users who declined, to see if anything can be done to improve this. Are they simply wary of what they don't fully understand? I thought the dialogue box explained everything in fairly plain language.
User avatar
GFD
My brain's probably worth a lot of money!
 
Joined: 31 May 2010
Location: Canada

Re: First results from the GZDoom 3.3.0 survey

Postby Graf Zahl » Fri Mar 30, 2018 12:59 pm

You can ask the same thing to people who do not participate in elections. It seems to be Human nature that some do not care about anything that happens outside their small comfort zone, but once these things actually do affect them, they are the ones who complain the loudest. Yeah, they should have acted earlier when they had the chance.
User avatar
Graf Zahl
Lead GZDoom Developer
 
Joined: 19 Jul 2003
Location: Germany

Re: First results from the GZDoom 3.3.0 survey

Postby Cacodemon345 » Sat Mar 31, 2018 12:00 am

There should be a way to check if there are people using software-rendered OpenGL to run GZDoom. That way, the survey will expose more real numbers. Because there are people using Mesa Software OpenGL implementation to run GZDoom, I believe.
Cacodemon345
 
Joined: 22 Dec 2017
Discord: Cacodemon345#9151

Re: First results from the GZDoom 3.3.0 survey

Postby Kinsie » Sat Mar 31, 2018 12:02 am

Cacodemon345 wrote:There should be a way to check if there are people using software-rendered OpenGL to run GZDoom. That way, the survey will expose more real numbers. Because there are people using Mesa Software OpenGL implementation to run GZDoom, I believe.
I doubt that this is trivial to detect as separate from "real" OpenGL, and also that the numbers of users would break the single digits.
User avatar
Kinsie
A Concept Utterly Obsolete
 
Joined: 22 Oct 2004
Location: MAP33
Discord: Find Me...
Twitch ID: thekinsie

Re: First results from the GZDoom 3.3.0 survey

Postby Cacodemon345 » Sat Mar 31, 2018 12:06 am

Kinsie wrote:
Cacodemon345 wrote:There should be a way to check if there are people using software-rendered OpenGL to run GZDoom. That way, the survey will expose more real numbers. Because there are people using Mesa Software OpenGL implementation to run GZDoom, I believe.
I doubt that this is trivial to detect as separate from "real" OpenGL, and also that the numbers of users would break the single digits.

It is required to make sure they also appear in the survey as separate numbers and to make sure the survey exposes more real numbers. Because like I said, there are people who run GZDoom using Software-rendered OpenGL.
Cacodemon345
 
Joined: 22 Dec 2017
Discord: Cacodemon345#9151

Re: First results from the GZDoom 3.3.0 survey

Postby wildweasel » Sat Mar 31, 2018 12:13 am

Cacodemon345 wrote:It is required to make sure they also appear in the survey as separate numbers and to make sure the survey exposes more real numbers.

It's been stated previously that GZDoom only collects the supported OpenGL version - not whether it's accelerated OpenGL, and not the hardware being used.
User avatar
wildweasel
「お前はもうトースト」[you are already toast.]
Moderator Team Lead
 
Joined: 15 Jul 2003

Re: First results from the GZDoom 3.3.0 survey

Postby Rachael » Sat Mar 31, 2018 12:40 am

Cacodemon345 wrote:It is required to make sure they also appear in the survey as separate numbers and to make sure the survey exposes more real numbers. Because like I said, there are people who run GZDoom using Software-rendered OpenGL.

For our purposes, this is 100% irrelevant and muddies the waters for what we're trying to accomplish.

We really don't give a flying fuck how a person gets OpenGL running on their computer - if it's GPU powered, or rendered via Mesa or Swiftshader - as long as it gets the result they want, it's still using the same code on GZDoom's side regardless. It may be in that person's best interest to run without said ABI's for the purposes of decision making in the survey results, but if the ABI makes them happy then that is the result we really care about in the end.

That shit is only relevant when they have a problem and it becomes a technical issue or if they try to report a bug using an unsupported ABI - but it has no effect on the purpose of the survey data.
User avatar
Rachael
Webmaster
 
Joined: 13 Jan 2004
Discord: Rachael#3767
Twitch ID: madamerachelle
Github ID: madame-rachelle

Re: First results from the GZDoom 3.3.0 survey

Postby Graf Zahl » Sat Mar 31, 2018 1:00 am

Why should we care? The numbers clearly show that old computers are a minority here. Why should we split that up even further? Did you know that OpenGL on old Intel chipsets is actually just a software emulation for many features? Those chips have no capabilities aside from displaying an image. Should we try to break those out as well?

The purpose of this survey was to have an idea of how many users are running a system that is below our recommended specs. And that information is precisely what we got. Because now we know what stuff can be discontinued without adversely affecting the reporting user base.
But regardless of that, it should be clear to anyone that a system running on an Intel integrated chipset with OpenGL 2 support only is not going to cut it anymore if you want more than just run the engine at its most basic settings. This is the bottom of the barrel of currently supported systems and they are known to have performance issues. Right now their number is still too large to discontinue support, but seeing that they are in decline, sticking to such a computer is not doing their users any good. Just like Windows XP has almost disappeared, even among GZDoom users, so will these computers sooner or later.
User avatar
Graf Zahl
Lead GZDoom Developer
 
Joined: 19 Jul 2003
Location: Germany

PreviousNext

Return to Developer Blog

Who is online

Users browsing this forum: No registered users and 1 guest