Unity 5.3 Performance issues

That's indeed the big question. I've been a Pro user since 2010. Then, being a Pro user meant something. Now, it's basically just a fee you pay to Unity to do their QA for them: your bug reports get priority if you're on the latest beta.

Oh, and you also get a few minor Asset Store items for free every month, and no splash screen at startup. The latter may well be the most important aspect of them all.

 

And this is exactly what I feel and been seeing. Get customers to do the QA work for them!!.. I already challanged and put forward some good replies in this thread.. right now I have some hope for 5.4 but until I visually see improvements in performance and stability.. The gates are open..

 

I think that we can dream .... nothing will change ...
Since unity 5 , the engine is good on the paper only ....

 

As long time unity developer (since 2.5), I have to agree with the notion that it is getting very hard to handle unity upgrading. And it's not just the bugs on Unity's part, but also with regards to the plugins so many people use. It's dreadful to think what a huge amount of time I have spent massing around with plugins issues and unity issues.

If it is not clear to Unity - I'd be very worried, but Unity has to be changing strategy regarding the engine development.

Such a strategy might be coming out with a service for all unity developers, through which you could send your game to a secure sandbox environment, which will automatically run the game and check for issues on upcoming unity version - the point is for unity to detect before hand possible issues. It might even be sent automatically, or done with all unity projects which use the cloud build service.
In regards to plugins, I think you could even automatically send the plugin developers notifications warning them their plugin will have issues with upcoming versions (as well as their users until an upgrade comes, maybe even have an automatic "ready for Unity 5.4" flag in the asset store).
Yes - it's a monstrous massive project, but some kind of massive change is needed.

If you take in account the costs for all unity developers in the world to handle these upgrades - unity and its developers can save tremendous time with such changes.

Unity crossplatform abilities are probably the best you can find. But continuing to enable that and letting the developers concentrate on the actual game development (content wise) - is a must for Unity to stay ahead.

 

Folks from Unity: any thoughts on this:
http://forum.unity3d.com/threads/un...rd-and-viking-village-sample-projects.375643/
?
Here's a link to the youtube clip showing the glitching (mentioned in the thread) at it worst:

 

@JeffersonTD, I didn't see any performance issues in that video.

 

I certainly did. There was all kinds of popping below the copper colored ring.

 

Do you see something written about performance issue on the post of @JeffersonTD?
No, so why you answer this post with "I didn't see any performance issues in that video." ?
This thread, is not really related to performance issue only now. With the complains of everybody, it's most a global post of this new boring engine version.

 

Most of the posts in this thread are still related to performance issues and there is no reason to just add everything to this thread.

 

One of the 5.3.1 bugs that I thought was fixed in P1 has reared it's (ugly) head again. Somehow Render Texture mipmaps are broken again. My characters have random garbage all over them unless you're right up close on them.

 

That's truly ominous. Are the gates going to let the goats escape?

I guess it's a framerate issue. I would suggest buying an inkjet printer.



While there are problems, it's best to report bugs and not indulge in gates or paper.

 

True. That's why I originally didn't come to this thread, but after MrEsquire pointed to this thread I got the impression that there's exceptionally lot of problems with version 5.3, so as I haven't gotten any ideas about what's wrong I've grown pretty frustrated, and this thread seems like a place where Unity staff might react.

What's more, I actually downgraded to Unity 5.1, and you know what? The shadow rendering issues with the Viking Village demo are now gone! And I see that the effects in the Unity Labs demo weren't working correctly either in Unity 5.3. but do work now in Unity 5.1.

I was planning on checking the Courtyard demo with 5.1., but apparently it needs version 5.2.2. So I guess I'll upgrade to that. Let's see if it works on that version. If it does, it starts to seem that the issues really are related to 5.3. and even though not related to performance per se, thematically really close to what's handled in this thread.

 

If you dont like this thread, stop following it and replying, like we told you earlier.

I refuse for this thread to die out as this thread is the WAKE UP CALL to Unity.

It sure discusses performance issue and moves onto issues why we keep seeing these performance drops.
From the amount of people who have replied to the thread and liked postings I can only say you trying to divert the attention. Make another thread if your not happy and leave us to discuss the problems.

Please give @JeffersonTD a break. He has a valid issue that just shows that Unity not even testing/updating there own assets/demos to work fully with latest releases 5.3, hence it is performance related.

 

So you see us "Customers" like a goats ?
So professional .....

I will explain you one time my point of view, because like all the unity staff, you just browse quickly the forum and then don't really check what we write.

I do not work for unity, so i don't care about your bugs report. I'm a customer and i don't have to submit a report for something that can be done with a blank project or with the default assets.
If you can't find your errors yourself, then employ the right guy who will find it and remunerate these guy.
A lot of new engine/option exist and a lot of peoples continue to use unity because they have a big project started on it. So some of us don't have the choice to stay on unity because of contract,investment ect ...

But the world turn and your dear goats will maybe eat a better grass soon ...

 

Just wanted to pop in and say, Hippo doesn't work for Unity. He's a volunteer.

Right now I'm imploring everyone to submit to feedback.unity3d.com

 

Exactly this, otherwise there would have been a constructive reply to each point raised by each forum poster..A reply by some Unity staff, I think one reply from Unity member Brett is not enough, more Unity members need to reply..

 

But this is for feature requests?

 

Sort of. Its basically a way Unity can tally feedback for specific requests/issues/features. Unity grew really fast and they need numbers to sort things out. Post a feedback, spread your link around, and get votes. It will make an impact.

 

Huh... Did some uninstalling, downloading and reinstalling. Went to Unity 5.1., then Unity 5.2.2 and then even back to Unity 5.3.1. to make sure that there just wasn't something messed up the previous time I upgraded to Unity 5.3.1 (well, the difference between going to Unity 5.1 and Unity 5.2 compared to going to Unity 5.3.1 is that for Unity 5.3.1 I took the update package instead of the full install, but that shouldn't make a difference right?). In any case here are the results:

• Viking Village
  • Unity 5.1 and Unity 5.2 working as it should
  • Unity 5.3. shadows messed up in a really unacceptably ugly way.
• Courtyard
  • Unity 5.1. not supported. Fair enough.
  • Unity 5.2. It works!!
  • Unity 5.3. The ridiculous mess shown in the video above.
• Unity Labs
  • Unity 5.1. works nicely.
  • Unity 5.2. and Unity 5.3. works in a way that seems okayish, but more ugly compared to how it is in Unity 5.1. Not sure if there is more to it, but at least it seems the corridors and areas are fully lit, whereas looking at them with Unity 5.1. they shouldn't be.
• My own proto project (which I'm of course the most interested in):
  • Unity 4.6.8. a shader I have is working consistently the way I want it to work.
  • Unity 5.1, 5.2. and 5.3. The shader works inconsistently in a similar fashion to what is shown in the Unity 5.3. Courtyard video I've posted earlier (in other words in a really cryptic, angle and distance dependent way)!

So in summary:

• Unity 4.6.8. Things that are supported, work.
• Unity 5.1. My simple shader works inconsistently
• Unity 5.2. My simple shader works inconsistently and Unity Labs demo doesn't look as good as it does in Unity 5.1.
• Unity 5.3. My simple shader works inconsistently, Unity labs demo doesn't look as good, Viking Village demo shadows are broken and Courtyard demo is totally messed up.

So the newer the version the more messed up things are! What's going on, seriously?! Is it that no-one cares about any backward compatibility? And in addition no Unity essential demos are updated to work properly with newer versions? If either of these is true, that really blows, and even if both were true, I still don't understand what's going on with the shader in my project or the Courtyard mess.

 

Thank you for trying to help, but I think this is not needed in this case.
If you read the whole thread from start to this post (yes maybe boring)
You understand abit more where some of the concerns are coming from.

This thread not seeking any new features, or addressing a very specific Unity bug.
The thread addressing a wider issue with Unity performance (engine performance)
Then the conversation topic is why after each release we see bad performance and bugs which we feel could have easily been seen by proper QA testing by Unity.

I know your trying to help but I dont think the forum moderators and community team know the full workings of the different Unity teams and cannot provide feedback on Unity 5.3 performance, the initial starting point of the thread.
So please understand where some of the regular community member points are coming from and of course thoose customers who do not visit forums as much.

Thanks

 

I'm only recently a Community Manager. I was full indie for two years (as a programmer) and my game paid my salary. I have kept close tabs on this thread and it both resonates and breaks my heart.

What I am telling you is as a forum thread you don't have a ton of pull with the dev team. We, the community team, have been told we are "cherry picking" tweets and forum posts. The dev teams want numbers. If you want your voice heard should truly heed my advice.

A user recently wrote a gamasutra blog post, but it didn't get much more than a glance even though it was tweeted from the Gamasutra account because 1) it was a personal blog and 2) it didn't see much traction.

I want your voice heard in a way that will make heads turn.

 

Thank you for sharing your story and being honest.

But I dont understand what numbers you wish, as this not really a numbers tread.

Im not seeking to turn heads or change how Unity work, I already see that fellow community members agree with how I feel.Im seeking Answers, so why is it so hard to get some?? This is not a attack/rant thread at all, but some of us feel failed, but still love Unity and the tool and will continue to use it, but really hard after each release.

Maybe we not all lucky to have games that pay our salaries..

EDIT: if you want numbers then read this thread, made well before 5.3 even entered beta.

http://forum.unity3d.com/threads/where-are-the-unity-performance-testers.356395/

 

I have many successful indie friends in the same boat. We have (Unity) teams asking if its just a few vocal community members or if the problems are actually wide spread. They are looking for data to back that up. Getting votes on a Feedback post is one way to show numbers that's agnostic of me "cherry picking"

 

Okay I got it, well I sure have to word the feedback post in a proper manner and make sure its not personal to anyone or a specific team.

But this really does shock me, I understand the need for some more proof about performance, but its not clear to me why the QA team cannot catch simple bugs or fail to test mobile project properly or why there own assets/demo projects do not perform well.

Are you really serious that they can claim everything is bright and sunny especially in the mobile development world.

 

Hollie, what you just wrote is very worrying. If we as users don't have the attention of the developers in any other way than through the feature voting channel, and if the community team moreover doesn't have their attention than through sheer numbers and statistics, then I get the impression that nobody owns the product and that the direction is set exclusively by popular vote. This can't be so; there has to be some sort of strategy devised and maintained by someone, otherwise Unity as a product is completely adrift. It's the people at that level that should listen, not necessarily the devs.

Unity, as a company, has done everything right for a good number of years now. That's what has brought the company to the position it occupies today. The company was once quite closed; not much information seeped out. However, Unity did turn around to embrace a much more open attitude, which I think is one of the factors behind Unity's success. The communication lines became much more open, and everyone has benefitted from it.

However, it seems as if internal communication has broken down, or at least is very strained. This is not uncommon when a company grows rapidly. But it's also a strong factor in the demise of many a successful company. On the other hand, we've seen positive signs too, such as the VP:s postings. It's no secret that things aren't ideal at the moment. It's also becoming obvious that many at Unity share this opinion. If the developers have had to shield themselves from the users to the point where they no longer listen, then that's very bad news for everyone.

We can't express our concerns in numbers only; we need to vent things openly with other Unity users. It's part of the process and allows everyone to see the bigger picture. Just voting for features through the normal channels doesn't cut it as it's just details.

(And of course you should cherry pick. Isn't the entire point of being on the community team that you have a wider perspective on the product and its usage than a single developer focussed on one particular feature in one particular subsystem?)

 

I am following this thread, because I am interested to understand what kind of performance issues exists. Titles have the purpose to give an idea about the content. Telling everyone with unrelated issues to join the thread can only lead to a general rambling, which unfortunately seems to be what you would like to achieve. If someone tries to understand what kind of performance issues exist in 5.3 and maybe get an idea of certain workarounds, there is no more place to find that. Thank you!

 

No I didn't imply any such thing, sorry for language barrier. I am a user like you. My job (voluntarily) is to assist by drawing attention to problems like yours (which helps get them solved) and also clean spam, moderate queues and a lot of unpaid work. From time to time I also make light-hearted jokes about gates and goats (which had nothing to do with your post).

You're probably frustrated at the bugs. Again, the best way to deal with a bug is not to take to the forums, but to formulate a bug report. Everyone at Unity and all the users that surround you and me, appreciate bug reports.

A report from you could stop a bug in my own game, so thank you if you choose to report

In general, people who report bugs are seen very highly and are respected because it takes time to make a proper bug report and that time is valuable. Unfortunately, posting on a forum hasn't managed to fix much so far.

 

@MrEsquire @PeterB
I still strongly encourage open venting It can be a very positive thing. I don't think anyone at unity thinks things are bright and cheery (personal opinion) (note 2: except for the made with unity initiative, that is a heavenly gold mine of positivity).

With everyone having an opinion its not at all bad to get stats/analytics. Its hard to be unbiased when fingers are being pointed. As game devs we have the "dont read the comments" quib because it hurts to read things that say your product is buggy. It's the same in any development (web, software, etc).

Don't see this as bad or forlorn, but instead if you can work proactively to help get that data, then things can move forward. I'm going to put a smiley face here because I know everyone at Unity wants to get back on track. If you can help, please do so

 

If someone creates a Feedback Post, please share the link so we vote it.

 

I think it's important to remember that despite the fact that there are some major bugs and performance issues with recent Unity releases, it's far from all doom and gloom. I'm sitting here working on my project right now. Nothing is holding me up. My game is running smoothly, and I'm making some great progress. I think that the quality level of releases has slipped significantly lately and that's something that Unity need to urgently address, BUT, in the meantime I'm able to get on with my work and that's what's most important at the end of the day.

It frustrates me that recent Unity releases have ALWAYS broken something in my projects, but in hindsight, I've always managed to work around the problems pretty quickly. Unity is still the product that has enable me to become a "real" Game Developer - and for that alone, they've earnt quite a bit of leeway with me. They'll need to f$%k things up a hell of a lot more to lose me as a customer at this point.

I'm not angry, I'm just disappointed

 

Good for you, however some of us are royally f*cked as are our clients.

 

unity 5.4 better have performance improvements, or else unreal engine wins
we all dont want that to happen.... I mean who is really going to be using unity after a buggy 5.4?

 

Really? That's your contribution to the thread? Might want to be a bit realistic.

 

You have to admit that Unity 5 hasn't been the most stable, and the new development process seems to be pushing out releases way too fast -- so much so, that the only stable releases are the patches themselves, and even then after a few iterations of patches. Which is a big problem, as you can tell by the other posts, Many (most?) people don't even know about the patch releases.
Unity really needs to re-evaluate this. All fixes from the previous release need to be in the first version of the point release. And all new features should actually work out of the box. And there should *never* be a feature lost (like baked shadows from trees), or a performance degradation (like the 98% perf drop that I saw in 5.3.1, which turned into only a 40% perf drop in 5.3.1p1).

I admit that a bad 5.4 wouldn't be enough to make me switch. But I'm getting tired of upgrading to get things working, only to have different things broken.

Unity really needs to do a full-on bug fix version - get everything stable again, so we can all move to it and catch our breath.

 

Well, given that Unity 5 is changing so much, it is unsurprising how things are developing. WE users demanded the Earth, and are now complaining it isn't perfect in less than 7 days. What is realistic is seeing where Unity is going with this. Instancing is now in 5.4 and this is quite a game changer among other really great things.

Sure, at Christmas time when staff finally need time off with relatives, some rushed mistakes. I don't think it qualifies for a hanging. Honestly, I don't. I prefer to pull my punches for a time when actual slacking is happening, like 4 did slack a bit

We (collectively as a majority) demanded more frequent releases. We demanded a roadmap with accurate dates and times. This meant that releases were going to be a bit more hit and miss. The key take away is: is the product improving over time? I think it is. A performance regression is one of the more serious things Unity worries about, so I'm confident this one will get resolved, and it doesn't require beating them up over.

 

A performance regression is one of the more serious things Unity worries about, so I'm confident this one will get resolved, and it doesn't require beating them up over.[/QUOTE]

well yes thats true

 

"geometry instancing is something that typically is applied for very specific purposes. It isn't a 'oh we just slap this on everything and yay its faster"

is that quote true?

 

Yes quite true. It will however fix one of the biggest slowdowns in Unity: draw calls - where possible Everyone at some point gets upset when their 5,000 blades of grass don't seem to be that nippy.

But if you're doing a lot of bullets, debris, similar enemies, foliage etc... then yes, expect some sweet wins. You still pay the cost of drawing and so on gpu side, but that cost is pretty constant. What instancing does is free cpu up to get on with the good stuff while the GPU just goes and draw them all.

It's like proper dynamic batching, without silly limits and without dynamic batching costing cpu time (which is why it has silly limits). Although I bet instancing has silly limits too

Personally I have big plans for the feature, I do need to draw a lot of the same mesh.

 

Question: Am I building my project wrong?

Previous Unity 5 builds would build in a blink of an eye. 5.3.1 takes a dog's age to build the same 5-scene test build. My project has over 40,000 files in it along with a Resources folder, but for now only uses and renders Sprite assets while none of my 3D models are even used yet. Memory usage on my laptop skyrockets. Is this a bug? Because testing gameplay, when the time comes, will take a very long time.

 

@holliebuckets
Right now, this topic has 289 replies and 11,5 thousand views, aren't those numbers that may indicate interest?
Also, while I like the idea of the feedback system, it has one major drawback: The votes don't replenish. A few months ago, I voted "all in" for a feature I thought would be useful but now I am left with no votes at all. I don't want to start a whole new discussion about this, though (there probably already is one somewhere anyway). I just want to point out that currently, the feedback system may not be as useful as it could be. Not to mention that users who are actually willing to change the priorities for this software by voting will lose their votes very quickly and then they cannot vote on any new ideas.
Also, I agree with @SteveJ that we can at least keep working to some extent (besides the 5.2 Editor performance issues, which made it hard to keep going). But the thing is that this won't help when you are planning to release soon. I wouldn't want to release with crucial bugs scattered all over the place. Currently, I am stacking about 8 apps that are mostly finished. But what is going to happen when I finish those? I can't release like that. I just hope that one of the next patch releases will be somewhat ready.

 

I've posted the auto-testing idea in the feedback.

You are welcome to vote if you see fit.

https://feedback.unity3d.com/suggestions/use-existing-projects-to-test-unity-upgrades-and-plugins

 

Are you making a MOBILE game? if so, how has your performance been especially with Android over each release?

 

Very well said, I think the system is just another Unity tickbox ( I voted all my 10 points for WatchOS support) totally forgot about this, and seems no updates on it for months or Unity feedback.

The topic created by @GuyTidhar although this is not really how I would have worded it as it not going to happen what Guy has requested. Sorry dude but things don't work like this.. We don't want customers to do the work, we want Unity QA department or whatever department to the job they suppose to, this includes performance testing.

Unit testing on one part of the engine although works well and you find specific bugs and issues with one part of the code, the break down comes with integration testing and manual testing. It seems when things are integrated together things start to fail, its not easy to test these things. Still many issues would have been resolved with simply doing some manual testing.

See this thread before 5.3 even came out...
http://forum.unity3d.com/threads/where-are-the-unity-performance-testers.356395/

90 percent of companies want to remove Manual QA testers and replace with automation testing, maybe there are no manual testers on the Unity team..I can only speculate as still not had any proper reply from Unity team (holidays are no excuse) because this topic has come up well before 5.3, same things where asked when 5.2 came out and 5.1, so dunno why some people shocked or surprised.

 

Absolutely we have everything we asked for! Unity has done so many efforts in so many ways! Unity has everything : great features, great doc, great community, great services, great price. It's really the ultimate game-making tool, there aren't many softwares like this out there. The only problem is that since v5, the app is on a chaotic road. Fortunately, the Unity team is aware of that and working on it. Solutions are on the way.

 

@holliebuckets: Regarding the importance of "hard numbers", have you considered something as simple as a Customer Survey? It should probably be directed towards Pro users only. It doesn't need to be complex. A few multiple choice questions and a section reflecting the type of user would provide you with all the leverage you need.

Also, Pro users would jump at the chance to get their voices heard, and it would show Pro users that Unity as a company cares about their opinion. Win-win.

 

Why just target at pro users? Have a survey question asking that as another data point, then they can spin the data however they want. Otherwise, this seems like it would be a pretty valuable method for collecting the type of data they're looking for.

 

Call me cynical if you want but I learned a long time ago that the feedback site rarely has any impact. Even the feedback items that do happen to get implemented often don't get updated on that site.

I already have all my 10 votes used up. I wish I had more because there are just too many fundamental things that Unity needs to fix. How can Unity let a popular feedback item about a fundamental problem like this go on for so long? And Unity is now building other features on top of this faulty serializer (JSON Utility), making these new features just as faulty. This makes me worried.

The point is, the feedback site might be an okay place to vote for the next big feature, but I really don't think it's a good place for providing the "get your S*** together" feedback that this thread is trying to get at.

What if it's the few vocal community members who are letting Unity know about wide spread problems? I rarely submit bug reports anymore because it's too much work to try recreate the bug in a new repro project. I don't even submit crash reports because the bug then just sits open forever. I've noticed a continual degradation in performance and stability since 4.6 relating to mobile development, so we just don't update Unity anymore. And if we're not updating Unity, that means we won't be buying Unity 6 when it comes out.

 

We send out surveys to a random set of users every month. If anyone gets one please fill it out.

Moving this to discussion, cause you know, it's a discusion now.

 

As well as Desktop...