You should also use fps per Watt value to make performance more comparable. This could tell use if the increased power draw is porportional to the increased fps
So did you just ran your games on default power settings? On a steam deck with SteamOS or in another handheld using one of the deckyloader plugins on bazzite you could just put hard limits on the system TDP. Granting you full control over the battery life. In theory it's a relatively easy problem to fix.
@@cncdanyeah that’s a good plan theres also tons of other performance tweaking you can do by changing config files etc if you really want to get the best out of linux
Thanks! I actually did expect this may be the case, as I've had a similar experience when running old Laptops on Linux. In most cases the windows drivers are much more mature than the linux drivers are. Since most of these AMD APU's ship with windows on a battery opperated device, its in the manufacturers interests to ensure they are as efficient as they possibly can be! What I didn't expect is that Linux could perform better than windows out of the box. I might have to re-visit this one after some tweaks to both OS's to see if there are any improvements to be made to either!
I've run tiny10 on one of my past projects. It definitely helped a bit since that system was very low powered but it would be interesting to see if it helps much here!
Thankyou! I'm not sure yet, it's an idea I've toyed with (probably as a kit) but I would need enough interest and a substantial number of pre-orders to be able to justify the cost of producing molds for the housing as I don't really think printed housings are a viable option for a finished product
5:45 Yeah, I have a different display module (one that I bought cheap from AliExpress with "ears" mounting holes on the control board that are the same size as the screen) and it reacts the same - it restarts every time I tap into game mode. do you find how fix it or its just driver issue and cant be fixed?
I've had a couple of people comment in the last few days that there was a recent fix for this issue. I haven't had a chance to try it again have a go of updating to the latest version as it may fix your issue.
Apparently, the latest major version of Nobara OS is much better for handhelds, compared to Bazzite, and in a few cases even better than Steam OS. Please let us know how that one works, compared to Bazzite.
interesting vid.. loving the handheld build. wondering tho as i have a gpd win max 2 with a 6800 in it and it matters a whole lot on the tdp of the system for speed. as you getting so much better battery life in windows i wonder if the tdp is setup lower and "hampering" your gpu/cpu unit in windows, where linux in bazzite is setting the tdp much higher (as i saw on a intel laptop i have tried bazzite on ) worth a look
Possible. I'm not sure how windows power management handles TDP but I haven't got any extra software running to cap it. It is suspicious that the power consumption isn't the same between the two systems at full load so I'll have to check and see if there's any settings I can adjust. I'm sure there's also some software and services I can disable in windows to help improve performance. This test just gives an "out of the box" experience comparison for both systems.
@@cncdan Alrighty :) I dont know what it takes on windows to use fsr or anything, i know on my deck i just hit the button with three dots and turn on fsr. improves fps alot when upscaling from a lower resolution.
I have been looking around to see if a diy kit for a steam deck was even possible, and absolutely no luck. Really i just want to make a retro handheld, i just have no idea where or how to start the process for it. Any tips for someone wanting to start making handhelds?
Sure, don't start with something this complex! Grab a raspberry pi, install retropie on it and hook it up to a tv or monitor first and get a feel for how it all works. Next, do some research on the different communication standards that screens use and pick one you think you can handle. Then you just need a way to add controls and a way to power it up. I'm considering doing another simple handheld build like this in the future, just to show the process!
Maybe. I'm considering making this a kit but it will depend on how much interest I get. I would need a substantial number of pre-orders to be able to fund making the molds for injection molding the housing and it may all just end up being too expensive to be worthwhile. If that happens, I'll just open source the design and move on to the next project!
I believe when you tested this the patch for touch gestures for gamescope was broken hence the restart on touch. i think this might be fixed soon as touch gestures are being decoupled from gamescope and handled by HHD instead
I guess this shouldn't be surprising to me since I usually run 2 scripts (one created by a Linux focused PC maker called Slimbook) to help manage power consumption on my Linux laptops. I guess the conventional thought around lightweight Linux vs. Windows isn't based on how power efficient the OS is rather than how little it throttles raw performance of hardware. Usually when I see a video on how a Linux distro "breathed new life" into an old/underpowered laptop, I typically observe increased performance, not efficiency. Having granular control on CPU and GPU TDP must be the key to achieving power efficiency and is probably too tall a task for a small distro development team to pull off on their own, given the wide variety of hardware they might have to target. Valve only have 2 devices to target in their case. In comparison, most hardware manufacturers will target compatibility with one OS, Windows.
Possible, but I haven't made any adjustments to any of the power settings on either side so that's how they perform "out of the box". I would expect Linux to clock down the same way windows does when it's not under full load though but obviously it was unable to achieve the same level of efficiency at lighter loads which a TDP limit shouldn't effect.
Good question. ChimeraOS is different from BazziteOs, and is configured differently etc. Apart from ChimeraOS, there is also WinesapOS, and Nobara Project worth considering. I had no problem with these distributions on the AMD platform in desktop, and laptop versions.
I imagine most of the SteamOS Linux distro's are using similar drivers. I wouldn't expect there to be a big performance difference between them but I may test some others anyway to see if my touchscreen issue is still present.
I did a whole video on it! th-cam.com/video/trL_QeV8_ao/w-d-xo.html It's actually bigger than it looks, because some of it is hidden within the grip. Azoteq (the company that makes this trackpad) make a few different sizes if you need a bigger one. The trackpad is just connected to an Arduino which communicates with it and becomes a "USB Mouse" to the system its connected to.
Yeah I probably should've mentioned that, both systems are just in the default power mode/settings etc. I will have a fiddle with some of the settings in windows and rerun the Forza test and see if it improves over my previous result.
I switched windows into "Performance mode" in the power management settings last night and ran the same Forza 4 benchmark. Results were within a few percent of the last test so I would say it has made no real difference.
My take away is Linux is consuming more power to provide more performance .. so essentially performance wise the bottle neck for windows is windows itself
@@cncdan As others have mentioned in the comments VSYNC tends to not work properly or the same as it does on windows and I think you even mentioned in the video a game ignored it or w/e and wasn't capped with VSYNC on and I also think even when VSYNC does work properly from our point of view maybe it still trying to provide as much performance it can and just not seeing those frames due to the cap
Forza didn't ignore it, I just don't understand how to read it's results properly as they are quite complex. Perhaps another way to approach this test would be to configure a TDP cap on both Linux windows and see how they perform with the same TDP limit
Linux and vsync. When you want it on its usually off and wheb you dont want it you cant turn it off. Be it in firefox for youtube, minecraft, other games, vsync just doesnt work relliably as in windows.
It worked in two out of the three games I tested, at least! I'm still not sure about Forza 4 as it has an entire page of benchmark results, and about 4 different columns that are measured in FPS alone!
very cool looking handheld pc would look cool with a rgb ring around the track pad and instead of blue triggers and bumpers make them clear with rgb inside you could even make them turn red when pulled and id make the face buttons and the d-pad alilbit bigger like a xbox controller but other wise i want one lol oh did you get the shell made out of metal or plastic? anyways thanks for the great content love it.
The housing has been SLS 3D printed in Nylon. As for the RGB, I tend to find it a little distracting if there's too much of it. The triggers can't really be seen from the front of the device so there probably isn't much point to those and sadly, there just isn't enough room around the trackpad for something like that. As for the buttons, I covered some of my options earlier in the series but I ended up settling on these switch lite buttons as I feel they are large enough for my hands and I would've had to increase the size of the device even further to accomodate larger buttons and that just isn't something I was keen on doing.
On my device at least, it performed a bit better than windows for most of the titles I ran but it also used a bit more power doing so, even at reduced "sensible" settings. There's probably a lot more optimisation that could be done on both OS's still but I didn't want the video to end up being too long!
As far as I've seen, linux does not offer better performance or battery life with modern PCs. Most of the performance benefits come from low end PCs that are already starved for performance. Battery life also isn't any better and in some ways could be worse depending on the distro. Now there are some games that run better in Linux and some games run better in Windows so it's not really a clear winner since most games perform identically
This statement is untrue, and repeats an old stereotype about Linux. In this case, it is also necessary to mention, and understand what BazziteOs is and how it differs from other distributions. Because BazziteOs is based on Fedora Silverblue or Fedora Kinoite, and uses the Distrobox bazzite-arch container. Support for specific software including specific drivers can only be added by maintainers. This is very important, and overlooked by most people. For example, a special version of the Linux kernel or additional applications is required for the proper operation of devices such as Microsoft Surface, Asus, Framework laptops, etc. So it's not fair to judge that something doesn't work out of the box. The author of the video downloaded an Iso image that is dedicated to another handheld, not to this specific mini pc board. So it does not contain any specific fixes or optimizations. So you can't expect everything to run correctly. There are also a lot of videos on TH-cam with handhelds or other hadware showing better performance in games and emulation on Linux. For example, game emulation on GPD Win Mini on Linux works better than on Windows 11. The battery also holds relatively better, and the device does not heat up as much as on Windows 11. You can also see on Linux players' channels that the performance does not differ from Windows or even exceeds.
Thanks @timnowak8573, you've pretty much covered everything I was going to say! I know for a lot of the commerically available handhelds, Linux can improve performance AND battery life. I'm not suprised about the slightly worse power consumption here as this is just a generic mini pc with an APU that AFAIK isn't even used in any other handheld, making the fact that it performed better than windows in every test I tried even more impressive! I'm sure with some driver optimisations the power consumption could be sorted!
@@cncdan Yeh, exotic or specific hadware, configuration can cause problems on Linux, but there are also problems on Windows.. By the way, I think it is worth contacting the maintainers, in this case BazziteOs. They are very helpful in case of problems, and on their website you can find information about various devices, configuration, etc. I also think that it is worth reading a bit about the system, in this case Bazzite, which is based on Fedora, which is Atomic and Immutable. It has its pros and cons. For people who do not have unusual configurations, and have little experience, it seems to be a suitable system. To sum up, I'm a fan of the project, and I'd love to see how other Linux distributions I mentioned in the comments work 👍.
The system restarting when touching the touchscreen *in gamemode* happens to me too, it's not just you. I'm using an AYANEO Air. It does not happen on my GPD WIN 3 though.
My current setup for my AYANEO Air is HoloISO because almost everything works fine. I usually flash micro SD cards when I want to test different operating systems.
Good to know! I haven't had a chance to try HoloISO yet but I can definitely give it a go. Sadly the Beelink SER5 I used in this device doesn't have an SD card slot. I could flash a USB stick and boot from that but I didn't have any spare USB 3.0 drives and I REALLY didn't want to do it over USB 2.0 🤣
I am surprised at your results. I would have thought windows would use more power for sure. Thank you for testing this!
As am I!
Windows is pretty on par with Linux on power use on the Deck as well
same. knowing how bloated Windows is nowadays
My guess is the windows power management is more aggressive than whats on this version of linux.
Why does this look soo cool... im gonna cry from its awesomeness
You should also use fps per Watt value to make performance more comparable. This could tell use if the increased power draw is porportional to the increased fps
Have you ever run batocera on it? Your work is always a bliss to see, thanks for the video.
Thankyou! I have not tried batocera on this system, but I'm sure it would run wonderfully though!
I feel like once this releases and more people can try bazziteos the performance could be optimized more
So did you just ran your games on default power settings?
On a steam deck with SteamOS or in another handheld using one of the deckyloader plugins on bazzite you could just put hard limits on the system TDP. Granting you full control over the battery life.
In theory it's a relatively easy problem to fix.
Yeah, it's definitely something I can take a look at. Maybe I should compare the FPS both systems are capable of when Capped to the same TDP?
@@cncdanyeah that’s a good plan theres also tons of other performance tweaking you can do by changing config files etc if you really want to get the best out of linux
Its just wierd that windows is consuming less, something none of us expected. Great job!
Thanks! I actually did expect this may be the case, as I've had a similar experience when running old Laptops on Linux. In most cases the windows drivers are much more mature than the linux drivers are. Since most of these AMD APU's ship with windows on a battery opperated device, its in the manufacturers interests to ensure they are as efficient as they possibly can be! What I didn't expect is that Linux could perform better than windows out of the box. I might have to re-visit this one after some tweaks to both OS's to see if there are any improvements to be made to either!
Be interested to see the stats with AtlasOS applied to windows.
I've run tiny10 on one of my past projects. It definitely helped a bit since that system was very low powered but it would be interesting to see if it helps much here!
The quality of this handheld is insane! Are you planning to produce them or something?
Thankyou! I'm not sure yet, it's an idea I've toyed with (probably as a kit) but I would need enough interest and a substantial number of pre-orders to be able to justify the cost of producing molds for the housing as I don't really think printed housings are a viable option for a finished product
@@cncdan Kit's sound great, I'd be interested.
@@cncdanjust at the top of your head, how much do you think it would cost? Depending on the price, I'd definitely be interested!
@@cncdaninteressted too :-)
5:45 Yeah, I have a different display module (one that I bought cheap from AliExpress with "ears" mounting holes on the control board that are the same size as the screen) and it reacts the same - it restarts every time I tap into game mode.
do you find how fix it or its just driver issue and cant be fixed?
I've had a couple of people comment in the last few days that there was a recent fix for this issue. I haven't had a chance to try it again have a go of updating to the latest version as it may fix your issue.
@@cncdan i download and instal latest version - at same day as i comment above) so maybe for my screen its not fixed(
Apparently, the latest major version of Nobara OS is much better for handhelds, compared to Bazzite, and in a few cases even better than Steam OS. Please let us know how that one works, compared to Bazzite.
Any thoughts on making a deck with a framework motherboard?
Take a look at TommyB's channel, he's done just that!
interesting vid.. loving the handheld build. wondering tho as i have a gpd win max 2 with a 6800 in it and it matters a whole lot on the tdp of the system for speed. as you getting so much better battery life in windows i wonder if the tdp is setup lower and "hampering" your gpu/cpu unit in windows, where linux in bazzite is setting the tdp much higher (as i saw on a intel laptop i have tried bazzite on ) worth a look
Possible. I'm not sure how windows power management handles TDP but I haven't got any extra software running to cap it. It is suspicious that the power consumption isn't the same between the two systems at full load so I'll have to check and see if there's any settings I can adjust. I'm sure there's also some software and services I can disable in windows to help improve performance.
This test just gives an "out of the box" experience comparison for both systems.
Wonder how clean arch would compare to the distro you chose
You've got amazing skills. 👍
Thankyou!
Did you use any system wide fsr settings or anything when running these tests?
@@LearningToThrive just default, out of the box settings on both systems.
@@cncdan Alrighty :) I dont know what it takes on windows to use fsr or anything, i know on my deck i just hit the button with three dots and turn on fsr. improves fps alot when upscaling from a lower resolution.
I have been looking around to see if a diy kit for a steam deck was even possible, and absolutely no luck. Really i just want to make a retro handheld, i just have no idea where or how to start the process for it. Any tips for someone wanting to start making handhelds?
Sure, don't start with something this complex! Grab a raspberry pi, install retropie on it and hook it up to a tv or monitor first and get a feel for how it all works. Next, do some research on the different communication standards that screens use and pick one you think you can handle. Then you just need a way to add controls and a way to power it up. I'm considering doing another simple handheld build like this in the future, just to show the process!
@@cncdan I would absolutely love to see that!
Will there be a production run eventually??
Maybe. I'm considering making this a kit but it will depend on how much interest I get. I would need a substantial number of pre-orders to be able to fund making the molds for injection molding the housing and it may all just end up being too expensive to be worthwhile. If that happens, I'll just open source the design and move on to the next project!
I believe when you tested this the patch for touch gestures for gamescope was broken hence the restart on touch.
i think this might be fixed soon as touch gestures are being decoupled from gamescope and handled by HHD instead
Good to know! I will eventually get the linux ssd back into it and test again!
I guess this shouldn't be surprising to me since I usually run 2 scripts (one created by a Linux focused PC maker called Slimbook) to help manage power consumption on my Linux laptops.
I guess the conventional thought around lightweight Linux vs. Windows isn't based on how power efficient the OS is rather than how little it throttles raw performance of hardware.
Usually when I see a video on how a Linux distro "breathed new life" into an old/underpowered laptop, I typically observe increased performance, not efficiency.
Having granular control on CPU and GPU TDP must be the key to achieving power efficiency and is probably too tall a task for a small distro development team to pull off on their own, given the wide variety of hardware they might have to target. Valve only have 2 devices to target in their case. In comparison, most hardware manufacturers will target compatibility with one OS, Windows.
very interesting results.Was Linux running in a higher power mode ? It could be why the results were better and the power used was more.
Possible, but I haven't made any adjustments to any of the power settings on either side so that's how they perform "out of the box". I would expect Linux to clock down the same way windows does when it's not under full load though but obviously it was unable to achieve the same level of efficiency at lighter loads which a TDP limit shouldn't effect.
What about ChimeraOS?
It is still linux
Good question. ChimeraOS is different from BazziteOs, and is configured differently etc. Apart from ChimeraOS, there is also WinesapOS, and Nobara Project worth considering. I had no problem with these distributions on the AMD platform in desktop, and laptop versions.
I imagine most of the SteamOS Linux distro's are using similar drivers. I wouldn't expect there to be a big performance difference between them but I may test some others anyway to see if my touchscreen issue is still present.
Out of curiosity, where did you get the small track pad? And are there different sizes? And how does it work?
I did a whole video on it! th-cam.com/video/trL_QeV8_ao/w-d-xo.html
It's actually bigger than it looks, because some of it is hidden within the grip. Azoteq (the company that makes this trackpad) make a few different sizes if you need a bigger one. The trackpad is just connected to an Arduino which communicates with it and becomes a "USB Mouse" to the system its connected to.
just for interest sake, have you put windows in performance mode? that will impact power usage and performance
Yeah I probably should've mentioned that, both systems are just in the default power mode/settings etc. I will have a fiddle with some of the settings in windows and rerun the Forza test and see if it improves over my previous result.
I switched windows into "Performance mode" in the power management settings last night and ran the same Forza 4 benchmark. Results were within a few percent of the last test so I would say it has made no real difference.
I wonder how distribution specific this is, if at all (the power consumption I mean)
My take away is Linux is consuming more power to provide more performance .. so essentially performance wise the bottle neck for windows is windows itself
But if that were the case, wouldn't it also then consume the same or less power as windows when limited by VSYNC and lower settings?
@@cncdan As others have mentioned in the comments VSYNC tends to not work properly or the same as it does on windows and I think you even mentioned in the video a game ignored it or w/e and wasn't capped with VSYNC on and I also think even when VSYNC does work properly from our point of view maybe it still trying to provide as much performance it can and just not seeing those frames due to the cap
Forza didn't ignore it, I just don't understand how to read it's results properly as they are quite complex. Perhaps another way to approach this test would be to configure a TDP cap on both Linux windows and see how they perform with the same TDP limit
I kinda expected this all the CNC machines computer at my work ran on these lil nuggets & they all used a stripped down windows
Why had name deck
Because of the steam deck. I asked for naming suggestions in one of the earlier videos and this was one of the better suggestions so I went with it!
Linux and vsync. When you want it on its usually off and wheb you dont want it you cant turn it off. Be it in firefox for youtube, minecraft, other games, vsync just doesnt work relliably as in windows.
It worked in two out of the three games I tested, at least! I'm still not sure about Forza 4 as it has an entire page of benchmark results, and about 4 different columns that are measured in FPS alone!
very cool looking handheld pc would look cool with a rgb ring around the track pad and instead of blue triggers and bumpers make them clear with rgb inside you could even make them turn red when pulled and id make the face buttons and the d-pad alilbit bigger like a xbox controller but other wise i want one lol oh did you get the shell made out of metal or plastic? anyways thanks for the great content love it.
The housing has been SLS 3D printed in Nylon. As for the RGB, I tend to find it a little distracting if there's too much of it. The triggers can't really be seen from the front of the device so there probably isn't much point to those and sadly, there just isn't enough room around the trackpad for something like that. As for the buttons, I covered some of my options earlier in the series but I ended up settling on these switch lite buttons as I feel they are large enough for my hands and I would've had to increase the size of the device even further to accomodate larger buttons and that just isn't something I was keen on doing.
I'm sop stupid, where was it better than windows?
On the Steam Deck, ROG Ally, and Legion Go
On my device at least, it performed a bit better than windows for most of the titles I ran but it also used a bit more power doing so, even at reduced "sensible" settings. There's probably a lot more optimisation that could be done on both OS's still but I didn't want the video to end up being too long!
please i'm begging you, try putting a 4060 in it :D i really want to see what you can get with a 75W total power envelope
Second!
bro invent smach Z and did it better.
As far as I've seen, linux does not offer better performance or battery life with modern PCs. Most of the performance benefits come from low end PCs that are already starved for performance. Battery life also isn't any better and in some ways could be worse depending on the distro. Now there are some games that run better in Linux and some games run better in Windows so it's not really a clear winner since most games perform identically
This statement is untrue, and repeats an old stereotype about Linux.
In this case, it is also necessary to mention, and understand what BazziteOs is and how it differs from other distributions. Because BazziteOs is based on Fedora Silverblue or Fedora Kinoite, and uses the Distrobox bazzite-arch container. Support for specific software including specific drivers can only be added by maintainers. This is very important, and overlooked by most people.
For example, a special version of the Linux kernel or additional applications is required for the proper operation of devices such as Microsoft Surface, Asus, Framework laptops, etc. So it's not fair to judge that something doesn't work out of the box.
The author of the video downloaded an Iso image that is dedicated to another handheld, not to this specific mini pc board. So it does not contain any specific fixes or optimizations. So you can't expect everything to run correctly.
There are also a lot of videos on TH-cam with handhelds or other hadware showing better performance in games and emulation on Linux. For example, game emulation on GPD Win Mini on Linux works better than on Windows 11. The battery also holds relatively better, and the device does not heat up as much as on Windows 11.
You can also see on Linux players' channels that the performance does not differ from Windows or even exceeds.
Thanks @timnowak8573, you've pretty much covered everything I was going to say! I know for a lot of the commerically available handhelds, Linux can improve performance AND battery life. I'm not suprised about the slightly worse power consumption here as this is just a generic mini pc with an APU that AFAIK isn't even used in any other handheld, making the fact that it performed better than windows in every test I tried even more impressive! I'm sure with some driver optimisations the power consumption could be sorted!
@@cncdan Yeh, exotic or specific hadware, configuration can cause problems on Linux, but there are also problems on Windows..
By the way, I think it is worth contacting the maintainers, in this case BazziteOs. They are very helpful in case of problems, and on their website you can find information about various devices, configuration, etc.
I also think that it is worth reading a bit about the system, in this case Bazzite, which is based on Fedora, which is Atomic and Immutable. It has its pros and cons. For people who do not have unusual configurations, and have little experience, it seems to be a suitable system.
To sum up, I'm a fan of the project, and I'd love to see how other Linux distributions I mentioned in the comments work 👍.
The system restarting when touching the touchscreen *in gamemode* happens to me too, it's not just you. I'm using an AYANEO Air. It does not happen on my GPD WIN 3 though.
My current setup for my AYANEO Air is HoloISO because almost everything works fine. I usually flash micro SD cards when I want to test different operating systems.
Good to know! I haven't had a chance to try HoloISO yet but I can definitely give it a go. Sadly the Beelink SER5 I used in this device doesn't have an SD card slot. I could flash a USB stick and boot from that but I didn't have any spare USB 3.0 drives and I REALLY didn't want to do it over USB 2.0 🤣
@@cncdan I completely understand. I am also short on USB thumb drives haha. Maybe try an external SSD or HDD?
@@cncdan I just updated bazzite on my AYANEO Air and the touchscreen doesn't cause crashing anymore. Maybe it's worth another try?