This week, with 1.1 experimental just released, we take a look at two of the things it brings with it. First, Conor_ investigates what the new train stop limit allows him to do in his TSM-based factory. Then, Therenas explores what the multithreading update to belts means, both in theory and in practice. Careful, you might just learn something!
Continue reading: https://alt-f4.blog/ALTF4-15/
Alt-F4 #15 - Investigating 1.1
-
AlternativeFFFF
- Long Handed Inserter
- Posts: 58
- Joined: Fri Aug 28, 2020 12:07 pm
- Contact:
-
ferblabadub
- Manual Inserter
- Posts: 1
- Joined: Wed Dec 02, 2020 4:00 pm
- Contact:
Re: Alt-F4 #15 - Investigating 1.1
Thanks for the work on the multi-threaded belt logic investigation.
What CPU was this performed with? Is there any indication of performance on low- and high-thread count CPUs, and how well this scales?
What CPU was this performed with? Is there any indication of performance on low- and high-thread count CPUs, and how well this scales?
Last edited by ferblabadub on Wed Dec 02, 2020 5:01 pm, edited 1 time in total.
Re: Alt-F4 #15 - Investigating 1.1
Hi, author here, glad you enjoyed it!
Regarding your question, the measurements were run on a i5-8600k@4.8 GHz, with memory clocked at 2133MHz. I want to emphasize again though that the results are not very precise, and are only intended to show the general trend of the improvements.
Regarding scaling, I don't have any particular results to answer your question. In general though, I think the same heuristic for running Factorio apply. High single core speed is likely king, with memory latency and speed being the bottleneck in most situations, according to the devs. I can imagine that anything more than four cores has a rapidly declining performance advantage. For anything more detailed, you'd need to run further tests with a much more precise testing setup than me just using my PC :p
Regarding your question, the measurements were run on a i5-8600k@4.8 GHz, with memory clocked at 2133MHz. I want to emphasize again though that the results are not very precise, and are only intended to show the general trend of the improvements.
Regarding scaling, I don't have any particular results to answer your question. In general though, I think the same heuristic for running Factorio apply. High single core speed is likely king, with memory latency and speed being the bottleneck in most situations, according to the devs. I can imagine that anything more than four cores has a rapidly declining performance advantage. For anything more detailed, you'd need to run further tests with a much more precise testing setup than me just using my PC :p