Performance comparison for CPUs

[Alekthefirst]

So i have an idea for how we can set up a list of the ingame performance for different people`s computers. This list can then be used by new users to see how well they can expect their game to run when the entity count goes up, or for competetive purposes...

I do not know wether the entity count in a map can skew the results, but i propose using the easiest to acess map for everyone to acces: Campaign\First Steps.
For the most similar testing conditions we let the game window run at it`s default size.

How to:
1. Make sure you have NO MODS
2. Launch the game with its default window size
3. Set the autosave interval to "never"
4. Enter the first map in the campaign; "First Steps"
5. Tab through all the messages until you are left with objective: Use W, A, S and D jeys to move (or something like that)
6. Enable the FPS counter from f4 menu
7. Zoom in as much as you can
8. Enter this command:

Code: Select all

/c game.speed = 1000

and wait until the UPS stabilizes
This will set the theoretical maximum UPS to 60 000, i dont expect you to break this barrier, but if you do, make the number in the command higher. Now look at your UPS, and approximate an average from what you see. Also note down your FPS if you want to, but it is rather irrelevant as it is UPS we want to look at. The UPS is your "score"
9. Take screenshot of game window (Alt + Print screen on windows)
10. RESET YOUR AUTOSAVE INTERVAL!!!!!!!!!!!!!!!
11. Post here (Go to bottom of this post, or just see the next posts, for an example on how it CAN be done, feel free to submit in another way if you think yours is better)

Results: Keep them coming!
Performance chart(top score to lowest) (i round to the nearest 100)


Some people are interested in the Score/RAM speed chart, here is that

I am collecting the data in a google sheet, which compares the scores on a scale, go here for more details, like what the most popular cpu is!
https://docs.google.com/spreadsheets/d/ ... sp=sharing

Presenting Your System
This is how i will present my system:
In your post also include your system specs and (if you think it is necessary) a comment about special things that could hurt/improve your UPS

Factorio Version: 0.12.19
Specs:
OS: Win 10 Home
CPU: C2Q Q9300 oc to 3GHz
GPU: GTX 570
RAM: 4G 1333 MHZ
Storage: 120G SSD (Kingston V300)

Average UPS: 15100 (i am listing the score without overclock, which matemathically should be around 12600)
Average FPS: 15
Screenshot:

[Alekthefirst]

Just posting my laptop here

Factorio Version: 0.12.19
Specs:
OS: Win 10 Pro
CPU: Core i7 2670QM
GPU: GT 630m
RAM: 12G 1333 MHZ
Storage: 240G SSD (Some external Transcend SSD)

Average UPS: 18750
Average FPS: 18
Screenshot: (Forgot to zoom in)

[Alekthefirst]

Me here again, posting my 3rd computer

Factorio Version: 0.12.19
Specs:
OS: Win 10 Pro
CPU: Intel Core i3 3120m
GPU: Intel HD 4000
RAM: 6G 1333 MHZ
Storage: 120G SSD (not sure what model, but its running on SATA 2)

Average UPS: 17000
Average FPS: 17
Screenshot:

[Hellrespawn]

For the record, that map is First Steps, not New Hope. If you do this on New Hope, biters are gonna ruin your day real fast.

Factorio Version: 0.12.19
Specs:
OS: Win 10 Pro
CPU: Core i5 4690
GPU: GTX970
RAM: 8GB 1600 MHz
Storage: 4TB Seagate Desktop

[Alekthefirst]

For the record, that map is First Steps, not New Hope. If you do this on New Hope, biters are gonna ruin your day real fast.

I derped, thanks.

[sillyfly]

It would be interesting to see how these results differ on a map with actual things happening (maybe make a simple saved-game with a belt and inserter loop, and a few logistic bots flying around...)

[Smarty]

here are my results

Factorio Version 0.12.19
OS: Win 10 64x
CPU: AMD FX-8350 @ 4.0 GHZ I had it at 4.6 but due to stability problems I had to set it at default again
GPU: GTX 760
RAM: 8GB @ 1333MHZ
Storage: 1TB crappy toshiba drive

Average FPS: 12
Average UPS: 12800

[Alekthefirst]

CPU: AMD FX-8350 @ 4.0 GHZ

You got 13000 UPS?, i`d expect a CPU that beats my Q9300 in every (single core) benchmark by some 30-40% and has double the amount of cores to wreck my main system in factorio

[Smarty]

to be honest i dont know whats the problem but i dont care either. I have that processor over a year and never had problems with it.
going to try to change settings like max threads see if that makes a difference

[pedter]

Lots of other things running but it shouldn't have affected the core handling UPS. Thought it might be capped at 30k for awhile but the UPS wanders and the core sits at 100%.

Factorio Version: 0.12.19
Specs:
OS: Win 7 Pro
CPU: Intel Core i7 4770k (standard 3.50GHz)
GPU: GTX 770
RAM: 16GB @ 1866MHz
Storage: 250GB SSD (Samsung 840 EVO)

Average FPS: 30.0
Average UPS: 30000

[Smarty]

no difference at all with difference amount of threads

My cpu sits at a cozy 18% usage
14% is from factorio

[keyboardhack]

Noticed a trend with RAM speed
Obviously this could just be a coincidence.

[pedter]

no difference at all with difference amount of threads

CPU monitor notes that Factorio is single-threaded; the only thing more cores will do is offload other work to other threads. Helps a little, but not much.

Noticed a trend with RAM speed
Obviously this could just be a coincidence.

Try plotting CPU speed vs. UPS instead? Seems like a much more likely trend; I really hope standing there doing nothing isn't asking much of the RAM!

[keyboardhack]

Try plotting CPU speed vs. UPS instead? Seems like a much more likely trend; I really hope standing there doing nothing isn't asking much of the RAM!

Seems to be a lot more random when plotting against CPU speed



Could make sense if the map doesn't fit in cache. Factorio would have to fetch from RAM every UPS which could explain why RAM speed is more important.

[Nemoder]

Factorio Version: 0.12.19
Specs:
OS: Debian Testing
CPU: Core i7 4770k
GPU: GTX 960
RAM: 16G 1333 MHZ

Average UPS: 31500
Average FPS: 31.5

[Alekthefirst]

If we get multiple submissions for the same CPU (which i use in the charting right now), should i list them separately or do the average?
EDIT: im doing the average

[Alekthefirst]

Factorio would have to fetch from RAM every UPS which could explain why RAM speed is more important.

Im upgrading from my 4G of 1333 MHz to a 8G 1600Mhz kit soon, i can test then to see if there is noticable difference

[bublabab]

Factorio Version: 0.12.19
Specs:
OS: Windows 7 Pro
CPU: Ivy I7-3770K no oc
GPU: AMD R9 R290x
RAM: 16G 1600 MHZ
Storage: 120G SSD

Average UPS: 24714
Average FPS: 24
Screenshot:

[DaveMcW]

There is not much CPU used in the first map of the campaign, so it makes sense that the bottleneck would be RAM. You need a bigger factory to test CPU.

[Alekthefirst]

There is not much CPU used in the first map of the campaign, so it makes sense that the bottleneck would be RAM. You need a bigger factory to test CPU.

The true test of that will happen when i get my new ram kit in the mail, i have also noted that one user on reddit with 8G RAM (don't know his speeds though) with a i5-3570K beat your 16G 1600MHz and a i7-3770K, where the 3770K is supposed to be able to hande memory intensive applications better and have slightly higher single core performance.

Also other results match the expected performance when looking at single threaded benchmark results.
Example: remove 20% of my q9300s score (because i overclocked), then you get the expected UPS at stock clock speeds: 12600, now compare that to that 4690 test, which got 28500. Which is 2.26 times more than my q9300. Now look at the comparison here: http://www.cpu-world.com/Compare/625/In ... -4690.html, It says that the single core performance of the 4690 is 2.16 times higher that. So the scores check out.
Now compare my q9300 (still at stock speeds) with the FX-8350. The score is almost the same (12600 vs 13000) and according to cpu-world, the single core performance of the AMD chip is juuuuuuust slightly higher. Scores still check out.