I was expecting there to be a tiny advantage to linux. I think a lot of people might associate android with bluetooth controllers which can introduce more delay.
@@dandiaz19934 Bluetooth has made some incredible strides. I think the Dualshock 4 uses BT 4.0 and not even the faster 5.0, and it still manages to beat wired speeds for the same controller by a few ms.
With some modern Android devices Linux has a huge advantage. E.g. XP Plus and especially Odin Pro has quite a lot more latency than Taki's devices here.
Actually in my experience run-ahead doesn't reduce latency. But it does skip ahead so that _when_ the reaction comes - e.g. a shot - the bullet is already in the air. It's like the game fast forwards a little but the fast forward still happens delayed (e.g. 123 msecs on Odin Pro) - but sort of catches up in that instant.
@@DavidFilskov Excatly, that's the feeling I get when I play with run-ahead settings. We can see how run-ahead works by using higher values. It's as if we fast forward to a snap shot in time.
Run ahead is good for removing the game's inherent (internal, original, best case) lag. Then that plus a minorly laggy controller or display, will bring you near to if you had a crt and good controller
I notice there's substantial input latency on my RP2+ when running PS1 games through Duckstation, but not nearly as much so when running through PCSX Rearmed. Not sure you'd want to do a whole video on the subject, but I think it's interesting how much latency one emulator has over the other.
Have you tried the espxe standalone app on the retroid? For me it was pretty good, nearly instantaneous and lower than some other emulators I use on it, not perfect though and it seemed like the polling rate was sometimes inconsistent
Android has gotten a lot better. I used to have a micro to mini USB cable so I could wire the dual shock 3 controller directly to my phone back in the day and that totally destroyed any Bluetooth controller I tried at the time. I really think Bluetooth was the limiting factor but that was 3 or 4 versions of Bluetooth ago.
@@steel5897 yeah, I haven't tried but if you had a usbc to usbc cable can you wire a 8bitdo controller to a phone? 8bitdo also makes some really good phone clams that snap onto their controllers. If you got a really short usbc cable you would be set if it is possible. I'm not sure if it is though, it might still force you to use Bluetooth.
The majority of comments I've seen about Android-based input latency on Android devices with built-in controls hovered around the original Retroid Pocket 2 with its stock OS. Were you able to test that device?
Well that's due to that being a extremely old OS and very anemic hardware. The hardware of the original RP2 waa comparable to a phone back in 2013 or even a Xperia Play.
@@13JacksonMr I agree, which is what interested me in seeing how the input lag compares on a device that gets a lot of comments about input lag. I've found the RP2+ to be nice and responsive, so I wasn't surprised to find similar results to Linux devices.
@@RetroGameCorps On the Odin in docked mode, I’m curious if the 8bitdo pro controller has more, less, or the same input lag when connected to the device’s Bluetooth vs. using 8bitdo’s receiver connected using USB on the dock.
That's about what I expected. I've used both Linux and Android and never could detect a difference, and certainly not to the level that those who complain about lag would have people believe. I've noticed a bit of lag on certain TVs or with Bluetooth controllers on a phone, so I know it what it feels like. Retroarch works so well in how it handles the display that I was expecting maybe a slight difference if there was one at all. I appreciate seeing the test to demonstrate this and like how you used the LEDs to track the button presses.
I would assume that Android performance has gotten to the point where a physically connected controller wouldn't have unsatisfactory latency. Things should only get better from here one way or another
Not surprised, all emulation devices I've tried so far have felt like they have a ton of input latency compared to original hardware, regardless if they're linux or android. Please keep testing input latency on new devices you get! I'd love to have a playable retro emulation device sometime in the future(all mine are collecting dust after about 20 minutes of play). Something with close to similar ms to originals would be fabulous
I've never felt there was inherently more "lag," per se, on Android. It's always come down to components. Slower/unoptimized processors, less or slower RAM, etc. seems directly proportional to input response. For example, IMO Dead Cells is unplayable on the RP2+ while running smooth as silk on a Snapdragon 865. I don't mean graphics (which are serviceable on the Retroid), I mean the twitch inputs required to play the game properly, and compared against performance on multiple platforms. Beyond that, a basic disclaimer that Bluetooth on Android particularly sucks bears repeating. Some manufacturers have finally gotten around to trying harder with their drivers, but they've historically shipped as 'good enough' and based on Google's lazy base code.
@@SpinneretteAlex In comparison with PC or consoles, BT connections on AOSs are almost always laggy to me (besides being prone to general connectivity issues). Gamepad input isn't even a priority in Android, let alone solid BT drivers for gamepads. Personally, the only Android-based systems I've used with decent bt-controller performance are the Shield TV and phones with Snapdragon 750-ish or better procs.. To me the experience was on par with wired.
I've only ever heard of excessive lag on bluetooth controllers, perhaps Android has unfairly gotten the reputation of high input lag because more often than not, emulators on Android are being controlled with bluetooth controllers.
I'm not surprised with the results. Android has improved a ton over the years, I had both Android/Linux as equal. But the title of the video made me laugh, semi clickbait.
You are correct. If you aren't using the auto option, you would have to find the setting that works the best for your hardware. The auto setting is better, but you need the processing power to be able to use it.
Excellent video. It would be worth measuring iOS as there is a software element to it as well as Apple touchscreen being miles ahead and also when screen is off - some kind of music app.. Running the audio at 48 kz or 98 khz instead of 44k should help also. It was amazing that you found two tactile button controllers and with a LED as well.
My issue is input latency isn’t in comparing emulation devices against each other, it’s when compared to original hardware. Obviously, an additional 30ms of latency isn’t that bad for most games, but it makes certain types of games a rougher, and in rare cases, unplayable experience. For example. My favorite GBA game is a game called Rhythm Tengoku, which is a rhythm-based WarioWare style game. It is literally unplayable on any portable emulation device I have ever used, which is very unfortunate. But that additional 30ms of latency is enough to throw off the timings in the game. I’m hoping when I receive my Analogue Pocket it’ll solve those issues for me, but who knows.
Can we get an update on your experience? Coming from a novice who just got into emulating I noticed the input lag almost immediately since, as you mentioned, games that require precise timing are almost unplayable. Which is unfortunate because those are the games I really enjoyed playing, Megaman Zero, Warioware Twisted, Battle Network, and Rhythm Tengoku which I never had the opportunity to play but will definitely be checking it out. Thinking about getting my first handheld emulator and wanted to get it right the first time, do you have any recommendations? I've been looking at the new Miyoo Mini+ which I believe is Linux based, but maybe there are better choices out there for input lag.
RetroArch's run-ahead latency reduction is amazing when the cores you're playing with support it and your hardware is powerful enough to run it. I haven't tried GBA cores with it but for stuff like Mega Drive... I prefer emulation over original hardware now.
For future videos on new devices, please include input latency analysis when possible. And, specifically, go over your best settings to reduce latency as much as possible. For example, I still don’t know how to properly configure run-ahead for most devices.
Historically Android had really horrific touch latency, and I think that legacy has stuck with them. I remember reading about it on Anandtech, if I remember correctly the touch latency was 114 ms.
I even tried cable... Made sure No settings Like picture Optimasation is disrupting... But Nothing helps... Display has 1 MS Response time... I tried different Games... Its Just Bad
love to use my android phone with BT gamepad for retrogaming on the go, personaly I'm using retroarch for emulation 2d era consoles it has amazing "run-ahead" option to reduce input lag to a minimum, absolutely love this feature, a gameplay become very similar to original hardware
To make a real fair comparison between Linux and Android you would have to compare it on the exact same handheld and not two different hardwares. It would be really helpful if you did this latency test for every new device you review.
The problem is that there are not any handheld that is optimized for both Android and Linux. At best, those that can have dual booting has more effort into running 1 more than the other. It wouldn't be reliable.
This is great information and much appreciated! The ultimate test for me is beating Soda Popinski in Super Punch Out. Impossible for me on Pi3B+ or in windows on intel core i5 with emulation station, too much lag! But in this perticular game, almost every frame counts regarding some timed punches. I can beat him fairly easy on my NT Mini Noir, so FPGA is still number 1 for me. But software emulation i use daily as well, both ate great!
It's good to see your conclusion. I have seen too many people exaggerate the problems with software emulation saying that the lag is huge and that even simple feats(like levels in platform games) are nearly impossible. It's funny, really. Yes, FPGAs will be even better and it's good that it exists for people who want that. Certainly, Punch Out is well known as a game that would require that approach. I was interested in FPGAs for gaming, but by the time those devices were around, I had enough time to realize that software emulation served my needs extremely well.
I don't really notice the input latency on my RP2+, but I do notice some sound latency. It's not distracting, but compared to the real hardware, the sound latency is very apparent. The GB / GBA emulators on my PSP even have less sound latency. I have tried setting the sound latency option as low as possible, but it won't help.
Great job really 👍👍👍 as always ! Looking at the ultimate devices it would be very interesting to find out if it's true that those systems running Android lose performance in time while a Linux system is going to perform exactly the same way over the years !! In my opinion it is crucial choosing a device you want to keep and use over the years. Thank you man 😀
I like watching these videos but the main issue of introducing around 2 extra frames of lag over original hardware is still my biggest concern that stops me from buying any of these devices. Runahead-mode can help but some times this works better on some games and console cores better than others. Sometimes the audio gets broken using runahead even if the game runs with less lag.
I was always under the impression that it was Bluetooth in particular that Android struggled with. I use a bluetooth controller on my PC and it's fine but on Android it can be rough.
I think it's good to remember that there's latency inherent to the game engine / game system itself, like the game will receive your A press to make Mario jump but Mario won't start moving upward that same frame. You can lookup retroarchs run ahead feature for more info on that.
WTF HUHu saying there isn imput lag lmfao there 0 lag in my 351p only alittle lag on retrop2 hahahaahhaaa iduno wtf u talking about lag in the system in hairited his parents waelth bull hsit
It would be nice to see if there are ways for manufacturers to build devices to help cut down on latency just like PC gaming companies build controllers and screens that optimize for latency. Perhaps this will be the next big focus of Retroid, Ambernic and Powkiddy in the future. Perhaps we can narrow the gap between original hardware and emulation even further. It would be nice to see 40 and even 30ms or less even in a cheap handheld by raising the bar for screen and PCB quality. The good news? Most of these emulation handhelds are probably the same when it comes to the classic games. Nothing is more off-putting then a platformer or fighting game with a TON of lag.
Pre-reveal expectation: since it's RetroArch running on both and RetroArch goes all out on having settings for every little thing that might affect input delay I would expect there to be no difference beyond what the hardware in each device might have. I base this on my experience with OS X and linux and Windows being the exact same in my experience even though for desktop use OS X feels laggier for input.
Hypothesis before watching: Android has more lag but it is less than someone could notice outside of extremely high performance scenarios like tournament level fighting games. Thoughts after watching: Makes sense. I've never been able to notice differences in input lag between these devices but some ROMs have sluggish feeling controls compared to native. Always sluggish in a way that was ambiguous and left me wondering if I was just getting old.
When you've gotten used to 70 ms and below (for FPGA) the usual 80 to 110 ms for Android feels a bit sluggish - and Odin Pro's 140 ms feels very wrong. 1-2 frames of run-ahead or pre-emptive frames on Odin makes it more acceptable but still feels jerky / sluggish compared to 80 ms with with run-ahead.
I've had high input latency with many platforms regardless of OS; system overhead usually isn't an issue. The input latency usually comes from either the program itself, or from the bluetooth controller. I wouldn't be surprised if the 351p was actually slower with at least some programs. After: Wow, I really thought the input lag would vary more than that. I thought it would vary with no rhyme or reason, but it looks like these emulators have been optimized to the point that there isn't much more they can do to lower the input latency outside of specific optimizations and runahead use.
I really noticed a difference on SMB3, Megaman 2 and DKC on the Retroid Pocket 1. This made me switch to RG351m and it was a lot better. This test however makes me think it might have been a problem with Retroid Pocket 1 rather than Android. Maybe I'll give the RP3pus a chance
I thought RG351P would be slower since it used additional uC to handle inputs, opposed to other 3326 handhelds using GPIOs directly, I'd love to see a comparison with other 3326 devices though, like the original OGA.
What's strange is I've seen people say that they got rid of their Retroid Pockets specifically because the input lag seemed much worse, and I've seen other videos that tested it and had much worse results.
I recently tested the Retroid Pocket 2, an LG V40 with a Razer Kishi controller and a few Anbernic devices (RG351P, RG351MP, RG552) and noticed the Retroid Pocket 2 and the LG had a higher latency in SNES emulation than the other devices. All other tested emulators and games seemed the same across the board. I didn't go in-depth as you did, but I agree with your findings here. Great job!
@@TakiUdon yes sir, the non plus version. I did like the device but that latency was a deal breaker for me since I love me some SNES! I ended up keeping the RG351MP and have been enjoying my time with it. 😁
This doesn't surprise me whatsoever. I've actually used android emulation and nothing feels different. Input delay in general is overblown; i barely feel delay when i remote play my series s on my phone
My prediction was that I didn't know. The Retroid runs android, but has a more powerful chip to go with it. I figured that if the Android OS was slower, the more powerful CPU would still make any direct comparison impossible. Between those two devices, I'd probably choose the Retroid. I'd love to see a comparison between a gba, a fpga based system like the analogue pocket, a linux system, and then an android system with the same soc as the linux one. My hunch is that a properly stripped linux system would be much better than android, given they run on the same device. But then if you strip android down to it's core you might just end up with the same result. The problem is that I don't know much about the codebase of either kernel, though I know the android kernel was forked from linux.
We are talking about milliseconds here lol I think people get too caught up in the tech analysis. If it feels bad to play that's one thing but if it feels good doesn't make a difference.
Do you have a tutorial up on how to reduce the input latency of GBA emulation on the Retroid Pocket 2? Because I bought that device specifically to play GBA and I almost never use it because every GBA emulator I’ve tried on it just has too much input latency. I’ve spent hours swapping emulators and configuring them to death with no good result.
Trying to play the android port of sotn and I can literally see the time it takes between pressing a button on my xbox controller and alucard performing and action
As someone who hasn't obtained one of these devices yet, I must say that any kind of input lag compared to the original hardware is rather disappointing to see.
Thanks! My main concern right now relates to the input lag while using wireless controllers as it relates to the docked mode of the Odin and Steam deck, both of which I have on pre order. I have a cat that destroys cables, so a wireless controller is a must, but it seems that Bluetooth controllers have crazy input lag, and alternative wireless controller options seem to be limited. One question I have, on the Odin for instance, what is the lag when using the 8bitdo pro controller when connected to the Odin’s Bluetooth versus connecting to one of 8bitdo’s wireless receivers plugged into the dock via USB. Would it be the same, or which would be faster? The theory being that using the BT+USB interface is faster than android’s handling of the built in Bluetooth. Different topic, why isn’t anyone creating super fast wireless controllers with minimal latency using proprietary wireless tech like all of these Logitech mice and keyboards devices use? If Bluetooth is forever an input lag problem, why not use something different? People are obviously willing to pay a premium for nice controllers.
Great video! - What's that small black device at 1:47 ? - I'm looking for a device that can run N64 properly with a D-pad above the joystick - not below :) - Its CPU is SD800, right? - I can't seem to find that handheld it online, though.
I'd especially want to retest my reflexes against Kirby Super Star in all cases to see where my reflexes stand these days. I hit 11's and occasionally 10's on original hardware of both Quick draw on Kirby's adventure and Samurai Kirby on Kirby Super Star in SNES. As a kid of course. But with emulators, the best I was getting was 18. Ew
hello, can you please do a follow up video. ive been looking into buying a device for my brother but i want the input delay to be at least 16ms to 45ms. so please make a follow up video i am highly interested in the topic and in these products. the community needs to get together and make this a industry standard for low input delay devices that can cost around $70 or $120 thanks you and have a great day
2 роки тому
Is there some kind of filter used in RG351P (linux) it seems like there might be some kind of CRT type filter. Since it adds another element to the pipeline it might be possible that it introduces some latency. Of course i am nor sure if it is significant but I think it is important to run these kinds of experiments with every controllable parameter is close as possible. I am not sure if I can perceive it every game but input latency is a very interesting topic for me. Especially for systems that don't have inherent frame buffers. For 3D games it usually less important for me. Thank you for the technical dive in into this topic.
Whenever is use a bluetooth controller in aethersx2 on android, my game lags when used for a long time. Touch controls seems to be fine but the bluetooth controllers and i want to know why?
I expect there to be a difference (which comes out on top, I have no clue), but that functionally, the difference will be inconsequential towards overall user experience.
I emulate burnout 3 on my OnePlus 8t playing with a gamesir T1s controller and I feel no input lag at all doing those burning lap events w the f1 car, which is hell to handle if you're not pinpoint precise. So nope no input lag at all edit: emulator is aethersx2
I just checked the Retroid Pocket 2+… 35 days delay on shipment according to their website, according to the Discord even 2 or 3 months now. Maybe because it’s a small company or supply shortages, who knows….
I had imput latency with gba emulation with retroid pocket 2. It really stops me using it. Strangely enough, n64 emulation (I tried only mario 64) didn't have any noticeable latency.
late is the default option, and that is what I had it set to for this. For the end of the video, I tried to get the 2+ to have a lower value. Going from late to normal had no impact.
Awesome video 👍 can you make a video about HDMI/display lag when connected to the TV from an Android device? I can seriously notice it when playing midnight club 2
I've been using a PS3 controller with an Nvidia Shield and there is definitely some input lag sometimes - I think it might be down to battery levels in the controller... EG the lower the battery gets in the controller the worse the input lag gets.
To be fair, the Shield is highly overrated. And it's sometimes hard to get accurate info on it. Look around UA-cam and you'll find people still claiming that it can run Gamecube despite the emulator being taken off the Play store years ago.
I have an arcade and the system is emuelec running on android tv with usb controllers of course, when playing marvel vs capcom (i play it every day on fightcade) there is a noticeable lag. Then switch to raspberry pi 3 b+ with retropie and theres no noticeable lag at all... maybe android tv are design to enhance tv quality and add lag for some reason
That was very interesting for sure. I didn't expect a big difference but thought it would be a bit more than it was. It would be interesting to widen the list of devices to see if the results remain the same on a larger scale. For example even between different Linux handhelds I've noticed very slight differences in input latency. Could be a placebo effect but the OpenDingux based systems feel like they have lower latency than my RK3326 handhelds. Plus even my RK3326 handhelds seem to vary in input latency very very slightly.
Xbox series and xone controllers have really bad latency when connected to Android via Bluetooth, it's night and day difference wired Vs Bluetooth. No idea why
The further we strayed away from dead-simple Commodore 64 and NES 8-bit design, the more laggy af video game controller controls became in general, lol. Gone are the days of GOOD controls.
My expectation is that Android will have just a smidge higher latency but not enough to be perceptible or negatively impact the experience. I'm curious about comparing older Android versions like 4.4 on the OG GPD XD, and with Bluetooth controllers, see if we can figure out if there was historical basis for the assumption Android input latency is bad. Assuming I'm right that it is not an issue on the RP2+...
I would love to see this using a PS1 emulator. Maybe it's just in my head but I feel like I notice lag on my RP2 vs my 351P so much so that it made me stop using my RP2 for this.
Hi taki, I wonder if the Android device uses a Bluetooth controller, such as Feizhi Wee, how much will the delay increase? and if it use a otg controller such as Gamesir X2? will if much more input lag compared to the native controller on rp2+
I found Retroid Pocket 2+ has significantly less latency (Retroarch) than Retroid Pocket 2's default simulator. Could be default simulation software issue though. I was really disappointed with RP2's latency before. RP2's latency issue is echoing with people's bad memory regarding Android latency, although maybe totally different causes.
L A G - *08:22** It should say NES, not SNES*
L A G
L
AG
L A G
LAG
I was expecting there to be a tiny advantage to linux. I think a lot of people might associate android with bluetooth controllers which can introduce more delay.
There was another dude on youtube testing the latency of bluetooth vs wired and it's negligible
@@dandiaz19934 Bluetooth has made some incredible strides. I think the Dualshock 4 uses BT 4.0 and not even the faster 5.0, and it still manages to beat wired speeds for the same controller by a few ms.
This! Also, a lot of the clamshell android controllers have horrible deadzones which make it feel like there's lag
@@dandiaz19934 not for Xbox controllers
With some modern Android devices Linux has a huge advantage. E.g. XP Plus and especially Odin Pro has quite a lot more latency than Taki's devices here.
The next video I would like to see is how to reduce or optimize input latency on such devices (run ahead, etc.).
This!
Actually in my experience run-ahead doesn't reduce latency. But it does skip ahead so that _when_ the reaction comes - e.g. a shot - the bullet is already in the air. It's like the game fast forwards a little but the fast forward still happens delayed (e.g. 123 msecs on Odin Pro) - but sort of catches up in that instant.
@@DavidFilskov Excatly, that's the feeling I get when I play with run-ahead settings. We can see how run-ahead works by using higher values. It's as if we fast forward to a snap shot in time.
I think y'all are running ahead more frames than the game code natively supports. Ie: most NES games can only run ahead one frame
Run ahead is good for removing the game's inherent (internal, original, best case) lag. Then that plus a minorly laggy controller or display, will bring you near to if you had a crt and good controller
I notice there's substantial input latency on my RP2+ when running PS1 games through Duckstation, but not nearly as much so when running through PCSX Rearmed. Not sure you'd want to do a whole video on the subject, but I think it's interesting how much latency one emulator has over the other.
Have you tried the espxe standalone app on the retroid? For me it was pretty good, nearly instantaneous and lower than some other emulators I use on it, not perfect though and it seemed like the polling rate was sometimes inconsistent
@Coregee - Good point.
Android has gotten a lot better. I used to have a micro to mini USB cable so I could wire the dual shock 3 controller directly to my phone back in the day and that totally destroyed any Bluetooth controller I tried at the time. I really think Bluetooth was the limiting factor but that was 3 or 4 versions of Bluetooth ago.
@@steel5897 yeah, it isn't ideal. Can't all controllers that charge through USB C be wired to a phone that charges with USB C?
@@steel5897 yeah, I haven't tried but if you had a usbc to usbc cable can you wire a 8bitdo controller to a phone? 8bitdo also makes some really good phone clams that snap onto their controllers. If you got a really short usbc cable you would be set if it is possible. I'm not sure if it is though, it might still force you to use Bluetooth.
The majority of comments I've seen about Android-based input latency on Android devices with built-in controls hovered around the original Retroid Pocket 2 with its stock OS. Were you able to test that device?
Bing bang boom ty. for how hot that device was, it deserves a test.
Well that's due to that being a extremely old OS and very anemic hardware. The hardware of the original RP2 waa comparable to a phone back in 2013 or even a Xperia Play.
@@13JacksonMr I agree, which is what interested me in seeing how the input lag compares on a device that gets a lot of comments about input lag. I've found the RP2+ to be nice and responsive, so I wasn't surprised to find similar results to Linux devices.
@@RetroGameCorps On the Odin in docked mode, I’m curious if the 8bitdo pro controller has more, less, or the same input lag when connected to the device’s Bluetooth vs. using 8bitdo’s receiver connected using USB on the dock.
~83ms
That's about what I expected. I've used both Linux and Android and never could detect a difference, and certainly not to the level that those who complain about lag would have people believe. I've noticed a bit of lag on certain TVs or with Bluetooth controllers on a phone, so I know it what it feels like. Retroarch works so well in how it handles the display that I was expecting maybe a slight difference if there was one at all. I appreciate seeing the test to demonstrate this and like how you used the LEDs to track the button presses.
I would assume that Android performance has gotten to the point where a physically connected controller wouldn't have unsatisfactory latency. Things should only get better from here one way or another
Not surprised, all emulation devices I've tried so far have felt like they have a ton of input latency compared to original hardware, regardless if they're linux or android.
Please keep testing input latency on new devices you get! I'd love to have a playable retro emulation device sometime in the future(all mine are collecting dust after about 20 minutes of play).
Something with close to similar ms to originals would be fabulous
I've never felt there was inherently more "lag," per se, on Android. It's always come down to components. Slower/unoptimized processors, less or slower RAM, etc. seems directly proportional to input response. For example, IMO Dead Cells is unplayable on the RP2+ while running smooth as silk on a Snapdragon 865. I don't mean graphics (which are serviceable on the Retroid), I mean the twitch inputs required to play the game properly, and compared against performance on multiple platforms.
Beyond that, a basic disclaimer that Bluetooth on Android particularly sucks bears repeating. Some manufacturers have finally gotten around to trying harder with their drivers, but they've historically shipped as 'good enough' and based on Google's lazy base code.
Have you played on anything with good bt? How did compare to wired/usb?
@@SpinneretteAlex In comparison with PC or consoles, BT connections on AOSs are almost always laggy to me (besides being prone to general connectivity issues). Gamepad input isn't even a priority in Android, let alone solid BT drivers for gamepads. Personally, the only Android-based systems I've used with decent bt-controller performance are the Shield TV and phones with Snapdragon 750-ish or better procs.. To me the experience was on par with wired.
@@steel5897 I haven't had any problems on PC unless trying to play with two controllers.. Then it turns into a stutter and hitch fest.
I've only ever heard of excessive lag on bluetooth controllers, perhaps Android has unfairly gotten the reputation of high input lag because more often than not, emulators on Android are being controlled with bluetooth controllers.
I'm not surprised with the results. Android has improved a ton over the years, I had both Android/Linux as equal. But the title of the video made me laugh, semi clickbait.
8:29 isn't this why RA now has a frame delay option? In which it polls the input as late as possible?
You are correct. If you aren't using the auto option, you would have to find the setting that works the best for your hardware. The auto setting is better, but you need the processing power to be able to use it.
Excellent video. It would be worth measuring iOS as there is a software element to it as well as Apple touchscreen being miles ahead and also when screen is off - some kind of music app.. Running the audio at 48 kz or 98 khz instead of 44k should help also. It was amazing that you found two tactile button controllers and with a LED as well.
This is something I've wondered about for a while now. Thank you for diving into this topic, Taki!
Another thing: I wonder what kind of SCREEN the Gameboy had compared to the Android/Linux handhelds. That lag can sometimes have an impact as well
My issue is input latency isn’t in comparing emulation devices against each other, it’s when compared to original hardware. Obviously, an additional 30ms of latency isn’t that bad for most games, but it makes certain types of games a rougher, and in rare cases, unplayable experience.
For example. My favorite GBA game is a game called Rhythm Tengoku, which is a rhythm-based WarioWare style game. It is literally unplayable on any portable emulation device I have ever used, which is very unfortunate. But that additional 30ms of latency is enough to throw off the timings in the game.
I’m hoping when I receive my Analogue Pocket it’ll solve those issues for me, but who knows.
Can we get an update on your experience? Coming from a novice who just got into emulating I noticed the input lag almost immediately since, as you mentioned, games that require precise timing are almost unplayable. Which is unfortunate because those are the games I really enjoyed playing, Megaman Zero, Warioware Twisted, Battle Network, and Rhythm Tengoku which I never had the opportunity to play but will definitely be checking it out. Thinking about getting my first handheld emulator and wanted to get it right the first time, do you have any recommendations? I've been looking at the new Miyoo Mini+ which I believe is Linux based, but maybe there are better choices out there for input lag.
RetroArch's run-ahead latency reduction is amazing when the cores you're playing with support it and your hardware is powerful enough to run it. I haven't tried GBA cores with it but for stuff like Mega Drive... I prefer emulation over original hardware now.
Runahead is a game changer, also we need more 2.4ghz controllers!
For future videos on new devices, please include input latency analysis when possible. And, specifically, go over your best settings to reduce latency as much as possible. For example, I still don’t know how to properly configure run-ahead for most devices.
Thanks for the video!
It would be very interesting to see a comparison against native hardware for newer devices (so many new handhelds on 2022).
Historically Android had really horrific touch latency, and I think that legacy has stuck with them. I remember reading about it on Anandtech, if I remember correctly the touch latency was 114 ms.
I even tried cable... Made sure No settings Like picture Optimasation is disrupting... But Nothing helps... Display has 1 MS Response time... I tried different Games... Its Just Bad
To me 75ms and 66ms are both unacceptably high input lag for a gaming system for action games (fine for strategy/rpg though)
Lol
Clicked on a Taki video and got an ad for Tali's. Coincidence? I think not.
love to use my android phone with BT gamepad for retrogaming on the go, personaly I'm using retroarch for emulation 2d era consoles it has amazing "run-ahead" option to reduce input lag to a minimum, absolutely love this feature, a gameplay become very similar to original hardware
To make a real fair comparison between Linux and Android you would have to compare it on the exact same handheld and not two different hardwares. It would be really helpful if you did this latency test for every new device you review.
The problem is that there are not any handheld that is optimized for both Android and Linux.
At best, those that can have dual booting has more effort into running 1 more than the other. It wouldn't be reliable.
Agreed. For lag sensitive gamers it's great to know which OS is the best for each device.
I've finished a lot of games with this input lag and bluetooth controllers.Never bothered me.
This is great information and much appreciated! The ultimate test for me is beating Soda Popinski in Super Punch Out. Impossible for me on Pi3B+ or in windows on intel core i5 with emulation station, too much lag! But in this perticular game, almost every frame counts regarding some timed punches. I can beat him fairly easy on my NT Mini Noir, so FPGA is still number 1 for me. But software emulation i use daily as well, both ate great!
It's good to see your conclusion. I have seen too many people exaggerate the problems with software emulation saying that the lag is huge and that even simple feats(like levels in platform games) are nearly impossible. It's funny, really. Yes, FPGAs will be even better and it's good that it exists for people who want that. Certainly, Punch Out is well known as a game that would require that approach. I was interested in FPGAs for gaming, but by the time those devices were around, I had enough time to realize that software emulation served my needs extremely well.
I don't really notice the input latency on my RP2+, but I do notice some sound latency. It's not distracting, but compared to the real hardware, the sound latency is very apparent. The GB / GBA emulators on my PSP even have less sound latency.
I have tried setting the sound latency option as low as possible, but it won't help.
Very useful and from what I could see on screen, there seems to be little at all to worry about in general.
Great job really 👍👍👍 as always !
Looking at the ultimate devices it would be very interesting to find out if it's true that those systems running Android lose performance in time while a Linux system is going to perform exactly the same way over the years !! In my opinion it is crucial choosing a device you want to keep and use over the years.
Thank you man 😀
I like watching these videos but the main issue of introducing around 2 extra frames of lag over original hardware is still my biggest concern that stops me from buying any of these devices. Runahead-mode can help but some times this works better on some games and console cores better than others. Sometimes the audio gets broken using runahead even if the game runs with less lag.
I was always under the impression that it was Bluetooth in particular that Android struggled with. I use a bluetooth controller on my PC and it's fine but on Android it can be rough.
I think it's good to remember that there's latency inherent to the game engine / game system itself, like the game will receive your A press to make Mario jump but Mario won't start moving upward that same frame. You can lookup retroarchs run ahead feature for more info on that.
WTF HUHu saying there isn imput lag lmfao there 0 lag in my 351p only alittle lag on retrop2 hahahaahhaaa iduno wtf u talking about lag in the system in hairited his parents waelth bull hsit
Agreed. I think most retro games have 1 frame of latency, though, which is only ~17 ms.
It would be nice to see if there are ways for manufacturers to build devices to help cut down on latency just like PC gaming companies build controllers and screens that optimize for latency. Perhaps this will be the next big focus of Retroid, Ambernic and Powkiddy in the future. Perhaps we can narrow the gap between original hardware and emulation even further. It would be nice to see 40 and even 30ms or less even in a cheap handheld by raising the bar for screen and PCB quality. The good news? Most of these emulation handhelds are probably the same when it comes to the classic games. Nothing is more off-putting then a platformer or fighting game with a TON of lag.
I haven't seen anyone say it so - Android is based on the Linux Kernel...
But Linux doesn't have that heavy input processing layer that Android has "thanks" to Google.
Pre-reveal expectation: since it's RetroArch running on both and RetroArch goes all out on having settings for every little thing that might affect input delay I would expect there to be no difference beyond what the hardware in each device might have. I base this on my experience with OS X and linux and Windows being the exact same in my experience even though for desktop use OS X feels laggier for input.
Hypothesis before watching: Android has more lag but it is less than someone could notice outside of extremely high performance scenarios like tournament level fighting games.
Thoughts after watching: Makes sense. I've never been able to notice differences in input lag between these devices but some ROMs have sluggish feeling controls compared to native. Always sluggish in a way that was ambiguous and left me wondering if I was just getting old.
When you've gotten used to 70 ms and below (for FPGA) the usual 80 to 110 ms for Android feels a bit sluggish - and Odin Pro's 140 ms feels very wrong.
1-2 frames of run-ahead or pre-emptive frames on Odin makes it more acceptable but still feels jerky / sluggish compared to 80 ms with with run-ahead.
I've had high input latency with many platforms regardless of OS; system overhead usually isn't an issue. The input latency usually comes from either the program itself, or from the bluetooth controller. I wouldn't be surprised if the 351p was actually slower with at least some programs.
After: Wow, I really thought the input lag would vary more than that. I thought it would vary with no rhyme or reason, but it looks like these emulators have been optimized to the point that there isn't much more they can do to lower the input latency outside of specific optimizations and runahead use.
How well does the "run ahead" feature of retroarch work on these device? I'd love to see that tested ♥. Thank you!
I really noticed a difference on SMB3, Megaman 2 and DKC on the Retroid Pocket 1. This made me switch to RG351m and it was a lot better.
This test however makes me think it might have been a problem with Retroid Pocket 1 rather than Android. Maybe I'll give the RP3pus a chance
I thought RG351P would be slower since it used additional uC to handle inputs, opposed to other 3326 handhelds using GPIOs directly, I'd love to see a comparison with other 3326 devices though, like the original OGA.
Need to find a second-hand OGA...
I guess the time taken for the external uC to poll and then transmit is still less than 2 frames
@@ChrisRX13 They adds up, not the shortest board theory.
@@TakiUdon An RGB10 would do then.
The RG351MP uses GPIOs directly too.
What's strange is I've seen people say that they got rid of their Retroid Pockets specifically because the input lag seemed much worse, and I've seen other videos that tested it and had much worse results.
I recently tested the Retroid Pocket 2, an LG V40 with a Razer Kishi controller and a few Anbernic devices (RG351P, RG351MP, RG552) and noticed the Retroid Pocket 2 and the LG had a higher latency in SNES emulation than the other devices. All other tested emulators and games seemed the same across the board. I didn't go in-depth as you did, but I agree with your findings here.
Great job!
RP2 (the original)? It is possible. That device has something in the signal chain, so it could be one frame slower than the RG351p.
@@TakiUdon yes sir, the non plus version. I did like the device but that latency was a deal breaker for me since I love me some SNES! I ended up keeping the RG351MP and have been enjoying my time with it. 😁
Great video. I am more interrested in the sound latency between emulation (android) and the real hardware.
Love the lemmino music in the background
Cheers Taki, interesting tests.
This doesn't surprise me whatsoever. I've actually used android emulation and nothing feels different. Input delay in general is overblown; i barely feel delay when i remote play my series s on my phone
My prediction was that I didn't know. The Retroid runs android, but has a more powerful chip to go with it. I figured that if the Android OS was slower, the more powerful CPU would still make any direct comparison impossible. Between those two devices, I'd probably choose the Retroid. I'd love to see a comparison between a gba, a fpga based system like the analogue pocket, a linux system, and then an android system with the same soc as the linux one.
My hunch is that a properly stripped linux system would be much better than android, given they run on the same device. But then if you strip android down to it's core you might just end up with the same result. The problem is that I don't know much about the codebase of either kernel, though I know the android kernel was forked from linux.
We are talking about milliseconds here lol I think people get too caught up in the tech analysis. If it feels bad to play that's one thing but if it feels good doesn't make a difference.
Do you have a tutorial up on how to reduce the input latency of GBA emulation on the Retroid Pocket 2? Because I bought that device specifically to play GBA and I almost never use it because every GBA emulator I’ve tried on it just has too much input latency. I’ve spent hours swapping emulators and configuring them to death with no good result.
The original RP2?
@@TakiUdon yes
@@Arcad3n try myboygba it had alot better latency compared to retroarch
So underrated topic, I'm surprised in some ways about the results
(
for a very basic test, it was a interesting video :)
Trying to play the android port of sotn and I can literally see the time it takes between pressing a button on my xbox controller and alucard performing and action
As someone who hasn't obtained one of these devices yet, I must say that any kind of input lag compared to the original hardware is rather disappointing to see.
Thanks! My main concern right now relates to the input lag while using wireless controllers as it relates to the docked mode of the Odin and Steam deck, both of which I have on pre order. I have a cat that destroys cables, so a wireless controller is a must, but it seems that Bluetooth controllers have crazy input lag, and alternative wireless controller options seem to be limited. One question I have, on the Odin for instance, what is the lag when using the 8bitdo pro controller when connected to the Odin’s Bluetooth versus connecting to one of 8bitdo’s wireless receivers plugged into the dock via USB. Would it be the same, or which would be faster? The theory being that using the BT+USB interface is faster than android’s handling of the built in Bluetooth. Different topic, why isn’t anyone creating super fast wireless controllers with minimal latency using proprietary wireless tech like all of these Logitech mice and keyboards devices use? If Bluetooth is forever an input lag problem, why not use something different? People are obviously willing to pay a premium for nice controllers.
Why LED on 351P does not flash up?
The LED isn't as powerful as the one wired to my 2+. It's faint, but I can see it.
Great video! - What's that small black device at 1:47 ? - I'm looking for a device that can run N64 properly with a D-pad above the joystick - not below :) - Its CPU is SD800, right? - I can't seem to find that handheld it online, though.
OH it's a Z-Pocket Game! - ua-cam.com/video/Fij379z0nkw/v-deo.html
I am not too surprised, it comes mostly down to how well emulators are programmed.
I'm not that suprised by the results as android has gotten significant better over the last few years. Great video btw, I love this tech stuff!
I'd especially want to retest my reflexes against Kirby Super Star in all cases to see where my reflexes stand these days. I hit 11's and occasionally 10's on original hardware of both Quick draw on Kirby's adventure and Samurai Kirby on Kirby Super Star in SNES. As a kid of course. But with emulators, the best I was getting was 18. Ew
hello,
can you please do a follow up video. ive been looking into buying a device for my brother but i want the input delay to be at least 16ms to 45ms.
so please make a follow up video i am highly interested in the topic and in these products.
the community needs to get together and make this a industry standard for low input delay devices that can cost around $70 or $120
thanks you and have a great day
Is there some kind of filter used in RG351P (linux) it seems like there might be some kind of CRT type filter. Since it adds another element to the pipeline it might be possible that it introduces some latency. Of course i am nor sure if it is significant but I think it is important to run these kinds of experiments with every controllable parameter is close as possible.
I am not sure if I can perceive it every game but input latency is a very interesting topic for me. Especially for systems that don't have inherent frame buffers. For 3D games it usually less important for me. Thank you for the technical dive in into this topic.
No filter
Whenever is use a bluetooth controller in aethersx2 on android, my game lags when used for a long time. Touch controls seems to be fine but the bluetooth controllers and i want to know why?
Ooh I love the intro. Is it new?
I expect there to be a difference (which comes out on top, I have no clue), but that functionally, the difference will be inconsequential towards overall user experience.
I emulate burnout 3 on my OnePlus 8t playing with a gamesir T1s controller and I feel no input lag at all doing those burning lap events w the f1 car, which is hell to handle if you're not pinpoint precise. So nope no input lag at all edit: emulator is aethersx2
havent noticed any issues with delay or frame times when using xbox controller to my pixel 3xl
I would love to see this same comparison done with battery consumption. It feels like Android drains the battery faster than Linux but it could be me.
What's the first gamepad showing on the video ? model name ?
Finally someone set this straight
I use Dualshock 4 and honestly haven't noticed any maybe a little bit with Citra or dolphin but everything else plays great.
I just checked the Retroid Pocket 2+… 35 days delay on shipment according to their website, according to the Discord even 2 or 3 months now. Maybe because it’s a small company or supply shortages, who knows….
I’ve been waiting two months now, it said the 35 day thing when I ordered it. Wermp
I had imput latency with gba emulation with retroid pocket 2. It really stops me using it. Strangely enough, n64 emulation (I tried only mario 64) didn't have any noticeable latency.
Have you tried to put input polling from "late" to "normal" in Retroarch?
late is the default option, and that is what I had it set to for this. For the end of the video, I tried to get the 2+ to have a lower value. Going from late to normal had no impact.
@@TakiUdon thanks!
Awesome video 👍 can you make a video about HDMI/display lag when connected to the TV from an Android device? I can seriously notice it when playing midnight club 2
I must be really sensitive to it because I couldn't keep using my RP2 (not + mind you) because SNES felt like it had so much more lag than my RG350M.
how come the linux led doesnt light up?
What is the model name in the middle of thumnail? I've never seen that. looks like psp.
Try an TWS and Bluetooth controller playing a game using wifi/data and boy you got a field day 😩
LOL
I've been using a PS3 controller with an Nvidia Shield and there is definitely some input lag sometimes - I think it might be down to battery levels in the controller... EG the lower the battery gets in the controller the worse the input lag gets.
To be fair, the Shield is highly overrated. And it's sometimes hard to get accurate info on it. Look around UA-cam and you'll find people still claiming that it can run Gamecube despite the emulator being taken off the Play store years ago.
@@Magus12000BC When did anything being taken off the play store stop anyone?
Thanks taki that's answered my curiosity
Can you tell me, where the led is to be soldered on and which one I need?
Thank You!
I have an arcade and the system is emuelec running on android tv with usb controllers of course, when playing marvel vs capcom (i play it every day on fightcade) there is a noticeable lag. Then switch to raspberry pi 3 b+ with retropie and theres no noticeable lag at all... maybe android tv are design to enhance tv quality and add lag for some reason
That was very interesting for sure. I didn't expect a big difference but thought it would be a bit more than it was. It would be interesting to widen the list of devices to see if the results remain the same on a larger scale. For example even between different Linux handhelds I've noticed very slight differences in input latency.
Could be a placebo effect but the OpenDingux based systems feel like they have lower latency than my RK3326 handhelds. Plus even my RK3326 handhelds seem to vary in input latency very very slightly.
Xbox series and xone controllers have really bad latency when connected to Android via Bluetooth, it's night and day difference wired Vs Bluetooth. No idea why
The further we strayed away from dead-simple Commodore 64 and NES 8-bit design, the more laggy af video game controller controls became in general, lol. Gone are the days of GOOD controls.
how to lower the input lag in RA? whats the best settings
did these tests run with or without the run ahead feature on the android side?
Would disabling the built in vsync in an Android OS help with input delay?
Next video you do, hit emulation beginners with a best emulation seatings for the Odin pro
My expectation is that Android will have just a smidge higher latency but not enough to be perceptible or negatively impact the experience.
I'm curious about comparing older Android versions like 4.4 on the OG GPD XD, and with Bluetooth controllers, see if we can figure out if there was historical basis for the assumption Android input latency is bad. Assuming I'm right that it is not an issue on the RP2+...
That would be a great idea!
The scientific method of telling that "the speed of a machine does not affect end result" three times in a row is what made me a believer :)
The problem is, they haven't send them to me yet 😭😭been months now
I would love to see this using a PS1 emulator. Maybe it's just in my head but I feel like I notice lag on my RP2 vs my 351P so much so that it made me stop using my RP2 for this.
Note he’s using an RP2+ and not the original so it’s possible that’s the difference.
Hi taki, I wonder if the Android device uses a Bluetooth controller, such as Feizhi Wee, how much will the delay increase? and if it use a otg controller such as Gamesir X2? will if much more input lag compared to the native controller on rp2+
0:35 what's that handheld??
Android runs in a VM. There is always a very slight delay regarding touch input, even on high end phones
I found Retroid Pocket 2+ has significantly less latency (Retroarch) than Retroid Pocket 2's default simulator. Could be default simulation software issue though. I was really disappointed with RP2's latency before. RP2's latency issue is echoing with people's bad memory regarding Android latency, although maybe totally different causes.
That's due to the fact that the RP2+ has better hardware and software than the OG RP2