- --
Viewing Issue Advanced Details
ID | Category [?] | Severity [?] | Reproducibility | Date Submitted | Last Update |
---|---|---|---|---|---|
09057 | Speed | Major | Always | 21 days ago | 13 days ago |
Tester | Augusto | View Status | Public | Platform | MAME (Unofficial Binary) |
Assigned To | Resolution | No change required | OS | Linux (64-bit) | |
Status [?] | Closed | Driver | |||
Version | 0.271 | Fixed in Version | Build | 64-bit | |
Fixed in Git Commit | Github Pull Request # | ||||
Summary | 09057: all games using cedar magnet: high cpu usage and slowdown | ||||
Description |
Hello. Thanks for reading my topic. OS Linux kernel 6.8 Ubuntu 24.04 and CPU AMD FX 6 cores 3500 Mhz The games using cedar_magnet driver are having much slowdown for example mag_xain ( Xain'd Sleena SC 3.0 magnet system) That game use 5x Z80 5 Mhz and not use netlist. The games are simple not having 68k or other better cpus. If using cpu power profile power saver is 24 % speed and profile high performance is 70 % speed. Overclocking all CPUs to 200 % is 45 % speed. Perhaps not is related with Z80 changes. Have an good week and new happy year. |
||||
Steps To Reproduce | |||||
Additional Information | |||||
Github Commit | |||||
Flags | |||||
Regression Version | |||||
Affected Sets / Systems | all games using cedar magnet | ||||
Attached Files
|
|||||
Relationships
There are no relationship linked to this issue. |
Notes
9
No.22725
hap Developer
21 days ago
edited on: 21 days ago |
Like tokio: expectation, not a bug. Yes, if you say it has 5 Z80, each 5MHz, and I see config.set_perfect_quantum(m_maincpu); in the driver, it's going to be slow. |
---|---|
No.22730
Augusto Tester
21 days ago
|
Only few games use config.set_perfect_quantum ? tokio are in the list ? Z80 changes not does slowdowns in several others games. |
No.22731
hap Developer
21 days ago
edited on: 21 days ago |
It depends on how many Z80s the hardware has. tokio has 3, and two of them at pretty high clock speed (6MHz), therefore it got pretty taxing to emulate. - when using perfect quantum, that is. |
No.22732
Augusto Tester
21 days ago
|
That is the problem. Bubble Bobble use the same hardware with exact 2 Z80 at 6 MHZ plus YM3526 and not does much slowdown as tokio. The other games using multiple parallel Z80 not having slowdown even using the new Z80 core. |
No.22738
Robbbert Moderator
21 days ago
|
My machine is quite old (i7-8700) and doing a bench run gave 143.63% - running it normally gives a steady 100.00% >mame mag_xain -bench 90 Average speed: 143.63% (89 seconds) So I cannot confirm any problem. |
No.22740
hap Developer
20 days ago
|
He's just saying the driver got slower after the Z80 rewrite. Which is expected. |
No.22741
Robbbert Moderator
20 days ago
|
Oh, yeah, that is expected. So it's not a bug, maybe better discussed on a forum. Augusto unless you can prove a bug, I'm going to close this. |
No.22745
Augusto Tester
20 days ago
|
Sincerely I not understand about as internally is done timing for emulated cpus and thus being Z80 is as being much "simple" to emulate at point being strange for me only having 5x Z80 being that high slowdown was as an performance regression. I have loaded an high number of games using multiple Z80 and confirm the recent new Z80 core not does slowdown for several games. If an game in source line use "config.set_perfect_quantum(m_maincpu)" and so require high cpu usage that slowdown is really expected. However I say again the Z80 changes not are doing slowdown for several games. |
No.22785
JimCarlTay Tester
14 days ago
|
Consistent 100% game speed on my computer. Unable to reproduce. |