thanks Martin. I appreciate you are willing to test, even if i’s just to fill in holes. as the saying goes “I’ve started, so I’ll finish”
feel free to post whenever you have time. I don’t have many rockchip devices. Thankyou
I’m sure they will be but part of the reason why I want to add those boxes is because you might have a sale on for one box, and it allows others to know those boxes exist and have same chipset and same performance 
@ emveepee
that’s another reason why I want a complete list
THX
I have an H96 s905x2 but the test wont run or it doesn’t show results.
thanks Bazzle, it seems for some boxes it won’t show results but i’s still run the script. It’s a UI issue and the developer has spent a couple of days trying to fix , he first sends an update in his test repo
we are on version 1.2.7 (compared to version 1.1.8) so he’s been trying
you can find his test repo and the thread here
https://forum.kodi.tv/showthread.php?tid=314081&page=7
once a user has said it works I’ll let you know, so you can try again please.
I appreciate the effort thankyou.
1 Like
hey thanks for trying, sorry about the issues, another user has mentioned this issue here
https://forum.kodi.tv/showthread.php?tid=314081&page=7
@Lunatixz is trying to fix this issue, what happened before was in the logs it will give you the past for the score, so it’s running but showing the result is causing some issue.
today we’ll get more news on the issue. I’ll keep you updated on any developments or you can check yourself on that thread. as soon as this issue is fixed we’ll be able to run the test again . Thanks for your patience
No, it’s not showing any result in logs, only errors I mentioned before:
- 2025-05-10 19:27:08.716 T:4142 error <general>: GetDirectory - Error getting plugin://script.pystone.benchmark/
- 2025-05-10 19:27:08.733 T:4027 error <general>: CGUIMediaWindow::GetDirectory(plugin://script.pystone.benchmark/) failed
@ Kju you’re welcome BUT remember this is what the developer wrote
"> CPU Benchmark uses pystone a single core python module to benchmark native python implementation.
Brut python performance is calculated using a standard metric “pystone” . Users can formulate real word expectations comparing Kodi plugin performance.
The pystone score does not reflect hardware capabilities, nor does it benchmark them. pystone does not take into account multi-core processing, which can greatly impact the performance of plugins that supports it … Enjoy!!."
the single core python scores will have a big play in addons but some addons might utilise multicore more etc.
I still think the scores do really well to show relative performance for addons.
the developer has an I/O Benchmark addon
https://kodi.wiki/view/Add-on:I/O_Benchmark
I’ve ran it and got results that are all over the place, so I’m asking the developer to have a look.
once it’s ready and we combine these REA/WRITE scrores with the single core python scores, we’ll get a slightly better picture,
so boxes with the same SOC, might give you different READ speeds. these read speeds have an impact on how fast artwork will be displayed, I also want random read speeds if the author can manage.
so user can run this and report back so I can add to the collumn. will be fun 
if you have any results you want to add to the community table, I’d be happy to take them
So the “somewhat faster” from the first post is a bit misleading. When I replaced my N2+ with a Homatics R 4K Plus I felt a big performance degradation. But I thought, it was due to the EMMC vs USB. But now I think it’s more likely that it’s the CPU itself.
I mentioned this countless times to ht2tweak that saying something like the N100 is same speed as AM6B+ is not fair representation. big reason why I looked around for an addon to give us better objective measurement.
the SK1 is a huge surprise though, it’s 90% faster single core python speed. THX
Yes, I know, but that makes the 922 even more powerful. The 922 has 2 little and 4 big cores, the 905 has 4 little cores (successor of the little cores from the 922 with ~15% better performance).
I think, the Linux read/write buffer might play a role here. That’s a pitfall many people run into when testing I/O performance on Linux.
nice catch , so you are saying the big cores take over in the python test.
I was running windows laptop, and the results where very inconsistent. Good to have another person who understands this better than I and glad it’s given community some value
They should. That’s what the big cores are there for. If there is cpu heavy work, the Linux scheduler should put the work on the big cores. I may double check it when my AM6B arrives in the next few days.
1 Like
this is my scores from vontar x4
https://paste.kodi.tv/noyocahoga
1 Like
The following Boxes were tested on Latest Nightlys Piers NO ** The exception is the HK1 RBOX using latest Nightly NE 5.4.210
With default Estuary Skin
Tanix TX5 Plus Deluxe S905X3
https://paste.kodi.tv/ujimaxufif
Kinhank G1 S905X4-J (using ugoos x4q DTB)
https://paste.kodi.tv/dikoluqele
HK1 RBOX S905X4
https://paste.kodi.tv/ujudedowiz
X96 X10 PRO 8/64 S928X-J
https://paste.kodi.tv/geqomikimi
I added Tanix Box because is Daily Driver in Bedroom, it has a horrible pystone score,yet Media Playback is near Flawless
Happy Testing
1 Like
legend! thankyou, finally I have the Vontar. I’ve added that you are using CoreELEC because your log is identical OS to coreELEC, let me know if it’s another OS. it alligns exactly with Homatics and Tox3. if you have any other boxes le me know thanks
absolute legend! thankyou for going through all those boxes! must have taken some time.
I wanted all of them, haven’t heard f the tanix TX5.
Did anything surprise you. I’m glad they are aligning with other boxes with similar chipset.
subjectively do the scores align with your experience with the boxes? the x96 x10 has 191k while the G1 has 58,477
we also have kinhank with android 11 and android 14, and having it on coreelec is faster.
if you’d like to add any more boxes, happy testing!
The G1 is S905X4-J
The X96X10 PRO is S928X-J
That is why huge difference!!!
X96X10 Pro is now Daily Driver instead of Ugoos Am6b Plus.
Probly because my Dolby Vision Profile 7 FEL library is small / still Building.
Both are hooked up to Same Dolby Vision TV
Iam just really impressed with X96X10 PRO, and price was under $100 US when i purchased
that’s insane price, thanks for sharing. I do believe the x96x10 is going to be the king of S928X, team_CoreELEC has that specific box and worked with X96 (I think)
the scores for single core python is almost double that of S922X, it’s weird because it’s not supposed to be like that, but they are consistent. Good stuff
Does it run CE? If not then this result doesn’t belong here.
3 Likes
thanks for the score. vpeter is right that since this is CoreELEC forum, we should discuss CoreELEC hardware. If anyone wanted to share none CoreELEC devices, you have a reddit link in the sheets that takes you to kodi subreddit to post your scores.
Sorry vpeter I should have made it clear that only accepting CoreELEC results here, hope it’s okay to keep that borygo score here so users are aware what is and isn’t allowed to be discussed here for obvious reasons, thanks
