Friday Facts #261 - Performance + New player interaction

[SilverWarior]

Also, I found out that game is way more stable in performance in tight resource situations (for me it's playing game, recording in OBS for stream via CPU and in parallel recording via nvenc video card a local file) when I set core affinity (I usually give Factorio 2 cores out of 4 - works out brilliantly). Could Factorio do that on start maybe?

That would be due the fact that your recording software is heavily loading one of the CPU cores which is forcing the CPU to stay at max speed instead of throttling down for power saving. So you don't experience lag spikes caused by the delay which is required for CPU to speed up in order to be able to handle bigger load when needed.

Now when I'm plying Factorio I allow it to use all 8 cores of my AMD FX 8350 CPU. Spreading the load to all 8 CPU cores allows my CPU to remain in low power state almost all the time. Granted I still haven't reached building of megabases yet,
I do experience slight stutter once in a while but from my observation this is most often caused by some other background application suddenly needing lots of processing power.

[SilverWarior]

I'd suggest altering the sleep timer to sleep for a shorter period of time, then run a spinlock on awake to time the next update more accurately. Using a spinlock waker also allows to pre-warm execution threads , as it takes time for the OS to wake them all up again.

Nice idea. But if not implemented properly could simply prevent CPU from going into power saving mode (lowering its frequency) and this could be really bad for laptop owners since their CPU's would be heating more due to them constantly being at high speeds. Not to mention that you could then forget about playing of Factorio on battery power al together as it would be drained super fast.

[aweecz]

Hello,

Can't stop thinking if i start new game will it work with all those new things in 0.17? Especially new terrain generation.
Don't know, is it worth to start right now? Or just simply wait until new release since it pretty neat and big update.


Thanks for response.

[darkfrei]

Can you add autoscaling of all pictures? 10-100 MB pictures are very bad too.

[Fahrradkette]

I hope 0.17 won't come too soon so we can read another performance themed post about the introduction of grade separation for rails so networks can be separated and its pathfinder potentially moved to another CPU core.

[Useless_Pony]

...
For the solar panels the correct message is 'doesn't connect to wire' and for connectables, 'wire can't reach' or similar.
...
re: The Tutorial: The messages are so useful, but so clunky. If not for the gameplay hooking me, the message delivery would have lost me.

I'd recommend the message "incompatible with circuit networks" or simply "incompatible" for the circuit network wire to non-circuit network stuff.

I agree, the game play is so good that I can tolerate the bad message delivery.

[Jap2.0]

Hello,

Can't stop thinking if i start new game will it work with all those new things in 0.17? Especially new terrain generation.
Don't know, is it worth to start right now? Or just simply wait until new release since it pretty neat and big update.


Thanks for response.

Saves are forwards-compatible, if you start a game in 0.16 and then load it in 0.17 most things will be the same (occasionally a few things change and need to be fixed, but I don't know of any for 0.17). After you explore a few chunks outside of the edge of your map there will be a line where it changes from 0.16 to 0.17 map generation, so there will be a mismatched area of map generation but no real issues.

[Dirkadin]

Regarding your performance section. This is something that I have never experienced in Factorio; however, I do have the "speed step" setting turned off in my bios (not sure if AMD has a different name for this). What this does is keep the CPU running at the currently set max speed. For example I have a i7-5820k and over clocked to 4.0Ghz and with speed step off it stays at 4.0Ghz no matter what. Perhaps this could also be a solution to your CPU speed fluctuation problem.

[uscolumbia]

Just my 2c on the tutorial. I didn't find the pausing of the game distracting. If you wanted to you could just skip each notice and plug on through by hitting the tab (I think) button. I find this much better than some voice over or cut scene that you can't skip.

The hardest part for me to learn was stringing builds together to make a complex component. Science and green circuits took me a while to understand how to build on my own. Science needs belts and inserters; belts: iron and gears; while inserters: gears, plates, and green ciruits; green circuits: plates and copper wire, etc. I kept getting confused trying to figure out the whole string before starting instead of taking each part individually.

I guess that is really the learning curve for me. Getting the Factorio mindset that then leads to an addiction

[Paul17041993]

I'd suggest altering the sleep timer to sleep for a shorter period of time, then run a spinlock on awake to time the next update more accurately. Using a spinlock waker also allows to pre-warm execution threads , as it takes time for the OS to wake them all up again.

Nice idea. But if not implemented properly could simply prevent CPU from going into power saving mode (lowering its frequency) and this could be really bad for laptop owners since their CPU's would be heating more due to them constantly being at high speeds. Not to mention that you could then forget about playing of Factorio on battery power al together as it would be drained super fast.

Yes that's very true, thus it's ideal to keep the spinlock time as short as possible, within the last millisecond at the most. Spinlocks can also have a severe negative impact on performance when used incorrectly, thus should only be used for specific tasks such as pre-emptive waking by a parent or timing thread.

They don't always have a readable impact on battery life though, as like I mentioned the CPU will jump in and out of low power of it's own accord, depending on both the battery's and CPU's power limits, along with how much the GPU (both dedicated and integrated) is being utilised. Although that does bring up an idea for a low power mode in the game, or a half-framerate mode if you will...

[Tev]

... if the first 15 minutes of a game feels shitty, there is big chance that the player will not play any further.

A million people tried the demo, learned from the tutorial, and went on to buy the game.

There's nothing fundamentally wrong with the first 15 minutes and in my opinion the first 15 minutes are excellent.

+1

I mean, you can improve it. Somehow. But reading some comments here, your ideas and all the tradeoffs involved ... What actually needs improving? Current version works. Overhaul will consume a lot of time for who knows what benefit.

Maybe a few tweaks is all that is needed. E.g. some in-game wiki, that seems like straightforward & common suggestion.

[Blacky007]

regarding the error messages:
please add the information or error message - "to many wire connections" if you are trying to add more then 5 wires to one powerpole

[Sander_Bouwhuis]

Regarding the tutorial: DON'T use modal dialogs.
The reason for this is that you can't act upon the information while the dialog is open. And, after you close it, you can't look back to see what it said if you can't properly do the steps.
I would STRONGLY suggest showing the information of every step constantly on the left 1/3 or so of the screen so that a new player can refer to it.

I can't count the amount of times I clicked a dialog away only to think, "Wait, what EXACTLY did it say again?"

[FasterJump]

I suggest/approve few things that others may have already said:
-a "give me a hint!" button
-a button available at the end of a "mission"/"map" that open a list of what you learnt in this part of the campaign/tutorial. This button could be named something like "Review gameplay tips" or "Open checkup"
-if you close a tutorial pop-up, it may be great to be able to open it back at any time.

[sporefreak]

I like the floating text, but as a old player I usually know why something isn't working, Please leave the option to keep the warnings in the bottom corner out of the way, I know to check there if I have questions but I know for newer players the floating text will help a lot! I personally would just find it annoying after a while.

[Paul17041993]

I like the floating text, but as a old player I usually know why something isn't working, Please leave the option to keep the warnings in the bottom corner out of the way, I know to check there if I have questions but I know for newer players the floating text will help a lot! I personally would just find it annoying after a while.

With wires and such it's supposed to give you a red crossout when you cant connect them, so is it really an issue there? (although yes, it's a bit broken in .16)
Or are you referring to building lines and it occasionally crosses out of range or over existing machines/blocks?

[abregado]

realistically the lack of balance in the tutorial left me with playing a heavily unbalanced game until i learned this way would be better. luckily your game can be played that way as it will just work, even if it doesn't work best. so that is a major plus. but i think learning the balance of some things would be helpful in the tutorial.

One of my main missions with the new content it to increase the quality of the lesson that the early game gives. Just forcing the player to make each interaction once in a row its a great way to say as a developer "Hey! We put a tutorial in our game so dont complain", but I think Factorio deserves better.

i will be testing the new tutorial when 17 experimental is released and will give you feedback for sure.

I look forward to your feedback, but I would also enjoy you watching someone who never played Factorio play the new early game. It can be very frustrating for the observer, but if you can manage to say nothing during the experience (and even watch them suffer through parts they dont understand) then your feedback will be even more valuable.

I found the existing tutorial effective, even though I did find the constant modal pop ups very disruptive. I think having them be non-modal, so the user can continue interacting with the world while the tutorial text stays visible, would be a solid improvement.

With the current tutorial, the modal popups are 100% necessary, as they need to show you things that will complete very quickly (eg, craftng queue, smelter progress). The new system does not have modal popups, but now suffers from being unable to show these things very well.

Have you considered collecting metrics during the campaigns (time for a player to reach the progression milestones, and a full replay of their experience), and having an OPT-IN button for the player to send it to Wube for review?

The new version has anonymous metrics built in that are stored in the savegame (but only generated during the tutorial scenario). Currently this only includes result and timing data, but I would like to extend it to process data. Let us see if we put it into the 0.17 expermental release. NB4 data privacy shitstorm: this data would not be sent to us automatically unless you send us your savegame (which I would love to have happen), and is not going to be added to freeplay.

My experience as an instructor over the years and as a trainer is that the best way for people to learn something and have it stick the best is not to show them or give them a very regimented step by step, but coach them through it. Let them handle the controls and work with it. People will pick it up at varying speeds, but for most, it will stick better with them running things with a little prompting and help when needed.

This is more the direction we are heading. Two sections: The first one where the player is introduced to new concepts one after the other, and a second section where we "assess the learning" by having the player complete a "mini freeplay". It is also important for the data that we have metrics not only for how the player interacts with the lesson, but how they apply the knowledge that we expected them to learn.

Give the player a choice before they start as to what kind of tutorial they want

I saw this suggestion a few times and wanted to mention something. Players already have this option in Factorio (and basically every other game). They can have the tutorial where they play the tutorial or they can have the tutorial where they skip the tutorial and figure it out for themselves (or use some other learning resource such as wikis and youtube).

Giving further choices can be dangerous as a new player has no context in which to make this decision. As game designers, it is our responsibility to find a happy medium where new players have the right amount of choice. Equally important is to create a scenario that works for both Players of a Similar Genre and Humans without Game Literacy (plus everything in between).

The hardest part for me to learn was stringing builds together to make a complex component. Science and green circuits took me a while to understand how to build on my own. Science needs belts and inserters; belts: iron and gears; while inserters: gears, plates, and green ciruits; green circuits: plates and copper wire, etc. I kept getting confused trying to figure out the whole string before starting instead of taking each part individually.

I guess that is really the learning curve for me. Getting the Factorio mindset that then leads to an addiction

This was the other reason that the tutorial was originally slated for a rework. The current tutorial does not have that "Factorio Feel" of encouraging creativity. It is actually a good tutorial, but it gives you an impression that this game will tell you what to do. We all know that once you load Freeplay, the game will give you no direction apart from launching a rocket.

... if the first 15 minutes of a game feels shitty, there is big chance that the player will not play any further.

A million people tried the demo, learned from the tutorial, and went on to buy the game.

I mean, you can improve it. Somehow. But reading some comments here, your ideas and all the tradeoffs involved ... What actually needs improving? Current version works. Overhaul will consume a lot of time for who knows what benefit.

Maybe a few tweaks is all that is needed. E.g. some in-game wiki, that seems like straightforward & common suggestion.

I agree it has been a grand undertaking. The benefit is unfortunately not one that anyone reading this forum right now will benefit from (ok, maybe a bigger playerbase is better for everyone). Come release, there will be a spike of new players who have been either holding off on buying or will see the game for the first time. These players need a bigger hook than what is currently offered in the demo/tutorial.

The other component is that the current tutorial was designed before research was added to the game. Looking back, it also seems that the current version was added at the same time as Assembling Machines. These two things have become a much much larger part of the game and deserve to be shown.

Asking your community about the tutorial is like asking a yeti ...

Instead of telling people their opinions are not valuable I prefer to let people know how to make their opinions more useful. If anyone REALLY wants to give me some useful intel about the tutorial (new or old), then please invite someone over to play the tutorial while you watch and follow this easy 10 point program (its not easy).

• Do not under any circumstances give them help yourself
• Encourage them to ask questions out loud while they play (but do not answer)
• Have them play the tutorial and then drop them into Freeplay
• Let them know they can quit at any time (being frustrated and quitting is part of the process)
• In their freeplay session, see how long it takes for them to Automate Red Science (if they do)
• Stop after 3 hours total (tutorial + freeplay)
• Observe their behaviors and learn something for yourself how new players actually play this game.
• Write down your observations and send them to me (ben.buckton@factorio.com)

This is a pretty excruciating process for the observer. You will want to help them. Seeing as the best introduction to the game is having a friend teach you, this process is effectively the opposite and may just stop them playing the game again ever. BUT! You will now have a really good viewpoint from which to build tutorial ideas.

Actually, this is the reason why it was Kovarex writing the FFF this time. He came to watch some of our in house testing sessions and immediately saw a few problems he could solve. The new additions are great and we already see them working.

[Deadly-Bagel]

Please stop using red over green and brown, those arrows are very difficult for me to see. Additionally in this situation, red is generally an indicator of "something is wrong" and could potentially be misinterpreted.

Very happy for the option of the wide theme, though the "User" section on the right of each post is far too wide (over double the width it needs to be).

Definitely looking forward to the finished game ^_^ there have been so many awesome improvements that it's going to be a totally different experience - again!

[0111narwhalz]

In regards to the "quick reminder for the player who didn't pay attention in the tutorial," why not a few simple overlays to point out the functions of various UI elements? Fuel bars, the crafting queue, the equipment array, and so on. Just short snippets of text and arrows/boxen/highlights.

[Oktokolo]

Please stop using red over green and brown, those arrows are very difficult for me to see. Additionally in this situation, red is generally an indicator of "something is wrong" and could potentially be misinterpreted.

They look extremely visible to me. If they would be more visible, i would have to wear sunglasses to avoid getting blinded by their in-your-face-ness.
But they might indeed be less obvious for some color-blind people. Wonder if they would look better and would be more accessible if they where bright yellow, purple, or cyan...