I don't personally mind the slight inconsistencies as long as there is no obvious incompatibility or observable issues. I might get one in the future for another setup.
I'm not sure how long it would take but, wouldn't be better to compare both boards individually each with the actual original hardware console, let's say the PS1 and one game for example? Would that help figure out the frame "behaviour" between the two boards and, at the same time, see which of the two boards is closer to the original hardware? Do you understand where I'm going with this?
I know it would probably be time consuming for you and likely wouldn't result in as many views, but I would *LOVE* to hear your thorough breakdown of fields, frames, and frame timing. I feel like I'm already familiar with the math and concepts, but it never hurts to learn something new. Great content as always.
One thing to keep in mind is that the crystal oscillators are not going to be exactly the same board to board, and there could be as much as ±50 Hz deviation. Seeing a frame difference of ±5 frames makes sense over a long period. Also, it's important to remember that FPGA bit streams are not like software - if the FPGA wasn't of high quality, the system wouldn't even run. The FPGA's silicon floor plan has a specific timing constraint, which the bit streams are carefully generated to match.
Well, as someone who plays only on original consoles and for who Mister was just an affordable way to play arcade games, which I could not buy other ways, it means nothing why clone differs from DE10 and why is still somehow good. For me, the problem is that there are differences. Buying clone boards for people like me is not a good idea. After all, all mister vores are developed and optimized for DE10. So, if they aren't perfect even on DE10, I still expected slightly more accuracy on the original, and for me, it is still a deal worth extra money.
water waves in Waverace64 are not pre-rendered to appear at the same exact time. i'm not sure if it's real time, but i'll at least it's pretty damn close! So the pre-recorded paths the racers are following can differ a bit everytime you let the intro play. Basically, the intro sequence isn't a pre-recorded replay. recorded replays, to be very precise, must record the waves patterns, because there IS a timer in actual races, so everything have to be recorded. While in the game intro, it's not necessary to record the waves patterns. So basically, in the intro, water behaviors will always be different, even if it's just slightly.
Some PS1 games with issues on MiSTer FPGA you could test Constructor (Europe) 2D visual glitches Warhammer: Shadow of the Horned Rat (USA) black screen after completing the second mission Nuclear Strike (USA) black pixels in parts of level polygons. Thrasher Presents Skate and Destroy & Dave Mirra Freestyle BMX (USA) Random hangs from music streaming. I think some of these issues might have been related to RAM speed and seeing as the Taki Clone has different speed RAM it may show different results.
because of the side-by-side nature of the footage you can cross your eyes, merge the images and the differences will shimmer. (like the sunglasses on the logo) also works for those spot-the-differences games at the bar.
Thanks for taking the time to make this video! Have you had a chance to compare the analogue video out on the older 16-bit IO board vs the Taki 24-bit one? Curious to know if there's a noticeable difference when using a PVM.
Thank you for making the video. I'm having a hard time hearing the audio and music for all of these games because there's too much narration. The most important thing for me with any of these devices is to hear how the audio syncs up with the action on screen. I scanned through a ton of the video searching for that kind of analysis and didn't find anything. Any insight into that specifically?
Not to diminish the effort you put into the frame by frame analysis on the timeline, but for me the most compelling evidence in favor of Taki's clone comes when you play ESP RaDe near the end. If the output of the clone doesn't conflict with your muscle memory in that sort of game where frame timing is crucial, what else can you ask for? I look forward to running some similar tests with V-ISM combo timing in Alpha 3 if I manage to snag a board from batch 2. Based on the results you've presented, I doubt I'll be disappointed. On that note, as a Capcom guy (going all the way back to Champion Edition), I've always had a soft spot for Garou. Great game, very much looking forward to the release of its sequel next year. I'm also very much looking forward to any updates you're able to provide in future videos regarding guidance from Porkchop or based on your own testing on safely pairing the Taki clone with the Mister Addons analog boards.
Here's a potential future topic... for those of us with multiple MiSTer FPGA systems, how can one setup one card and clone it to other cards... so one doesn't have to do every card from scratch? Is there way? If so, I haven't figured it out. I tried using dd on Linux once and that didn't seem to work. I was able to capture an image and then write it out to another card of the same make/model/size... but then it didn't work.
You can migrate to a smaller SD card with the MiSTer utility, just as long as the total data used on the source card doesn’t exceed the size of the destination card. It’s a pretty handy and clever script. You’ll need a usb SD card adapter for the destination card and be prepared to for it to take a *long* time if you have a lot of data.
I got one of the second release!! I’m so excited. I’ve had an eye on the mister project for years and now I’m finally jumping in. Can’t wait for jt to arrive.
I noticed that there's a firmware file for the misterpi on the store website. Is this frame rate difference something that can be adjusted with a firmware update so that it more closely matches the DE10-Nano behavior? Also, when is he going to release the Hyper Neo Geo 64 clone board / core combo?
Thanks for the comparison video! I wonder how you setup the recording: Recording via HDMI? Recording in OBS? I myself have tried get good recordings/stream via HDMI both in NTSC and PAL modes. I really don't get it, how to set it up so the video will be super smooth. I mean the SNES core runs at 60.1 Hz in NTSC mode. How do I get perfect recording of that, is it even possible?
Can someone explain the allure of an FPGA? I get that it more accurately creates the electrical configuration of a retro device among having other uses, but once you output to hdmi or use usb controllers doesn't that change things electrically anyway and add lag? Don't get me wrong, I love retro stuff and electronics but I can't seem to find the niche this device fills.
All modern TVs will introduce lag. As will monitors. You can do SNAC on MiSTer and not introduce lag to the system via those adapters. As far as response time is concerned MiSTer is as good as it gets outside real hardware
The draw for me is that I can output the exact refresh rate from the original hardware. No screen tearing, dropped or duplicated frames. Emulation always bothers me with the stuttering or tearing.
@@CultOfTheGlendaAll modern PCs are capable of outputting arbitrary refresh rates and most modern emulators take advantage of that. Stuttering and tearing would be a configuration issue at your end.
I can't believe how you have new clips all the time 24/7(!). -Do you ever sleep? As for Wave Race 64: Sometimes when i leave game running for a while for nonstop title screen(s), there are differences how scene behaves on actual real hardware. Since it's realtime intro, it actually seems to be that under variable wave conditions, A.I racers sometimes make mistakes. -I've seen them even fail some jumps on intro(!). Just like when you play the game, it is never certain that all the waves behave exactly same as when you previously played it.
When one or the other goes ahead, are their respective audio in sync with their video? That's all that would matter to me, that the audio is in sync with its own video. Doesn't matter that one might load seven frames faster.
One thing that may be overlooked in this video is how the MiSTer Pi handles inputs and input latency. A good test would be to get a controller where you can program a long macro sequence and compare the footage after the button press for the whole macro sequence. Excellent job though, you did a crazy amount of work in a short amount of time!
I imagine it would perform similar to the original mister fpga since it is a clone board. I would be interested in seeing the durability of this Taki board but no idea how you would test it.
This is a great test! Keep in mind that the clock frequency of real consoles will always be slightly varying to some degree, especially so if the console is subjected to increasing heat from internal heat buildup and high ambient temperatures, which could also cause some variations to how the frames get produced. In this case I would overall expect the Mister hardware, both the originals and the Taki clones, to act much more consistent than even the original consoles. Ultimately there is only one way to find out, time to bring out the hairdryer. 😅
@@VideoGameEsoterica You could also try placing them into an oven at lower settings, as you can more easily control and monitor the exact temperature that way. 🤣
In theory, the Taki clone may actully perform better under higher temperatures as they are using the automotive variant of the SoC which tolerates higher temperatures than the industrial variant on the DE10-nano.
Great vid, and agree with the saturn, n64 and ps1 core stressing out mister devices the most. Maybe fun to also look at and compare the various turbo cores? ps1, n64, neo geo......?
I wonder if this differences with the psx core is because the Taki clone board has a better sd card reader than the original board so the iso disc reads are quicker on the loading parts.
I'm curious how the 2 boards compare to a second Terassic board and against real hardware. I bet all perform within the same margin of error. Even the real hardware always has the slightest of variations because every component that goes into them has a specific manufacturing tolerance. Those differences will all add up to a positive or negative performance difference to the theoretically perfect device that you would never see without comparing them side by side.
Thank you. It has 24bit video output. But i read that original mister can't output hdmi and vga simultaneously. Old version with 18bit can handle 2 video outputs. Is it true?
@@VideoGameEsotericathank you and thanks for your extensive knowledge and coverage on mister and FPGA. You’re my go to for anything and everything mister.
Some of the raspberry pi ones are good (ironically). I got a third party pi5 one in the UK. It's 5V4A and 20W plus has an in-line power switch. Taki said anything over 5V2A is fine and mentions it on his website
When you're talking about the 59.94/60Hz thing, does that mean you're using vsync_adjust=0 in your .ini file? Do you know how they compare in vsync_adjust=2 or does your capture hardware not support it?
Thank you for your hard work and effort. The differences you found between the Terasic DE-10 and Taki MiSTer Pi are negligible and are absolutely not a concern for me at all. I'm so happy and fortunate I was able to get my Taki Mister Pi order in yesterday -- my first MiSTer FPGA ever. Now, I'm torn on what display to pair it with -- a CRT tv/monitor, or a modern HDMI display (then I will need something like retro tink?). What is your favorite choice of display and connection setup with your MiSTer FPGAs?
You could use mister with a Tink 4k to great results. Otherwise, mister scales to 1080p out of the box, so you'll still get good digital output with no additional hardware required.
@@derekhaines1199 At $750 for the Tink 4k it's way overpriced for what it does. I think I will connect the MiSTer Pi to my CRT. Honestly, for $750 I think I'd even rather spend it on a large AV receiver that can act as a hub for multiple game consoles and then output to a modern tv or even CRT.
@@lordbacon4972 totally valid! The great thing about mister is it has great analog and digital output. So you should be able to plug it into a CRT or a modern display and get a quality picture either way. Obviously there are benefits to using a CRT, so I think if you go that route you’ll have an amazing experience.
Any wall wart that delivers 5V/3A will be fine. I’m using the Steam Deck charger and it negotiates the power down as the board is compatible with USB PD
I'm having a horrible experience trying to get their mega bundle. Taki's website is a glitch fest. Batch #1 I had the mega bundle in my checkout cart and was stuck there unable to proceed because the website stated that my zip code is invalid. Batch #2 I had it in my checkout and the exact same error occurred this morning! I had it in my check out for both batches and I'm not able to purchase it! Sent messages to "customer support" never received a response! My zip code was typed perfect (it is only 5 numbers lol) and I quadruple checked it because I was pissed I couldn't checkout......now they are out of stock for the second time.
@@VideoGameEsoterica no worries I was not judging or anything. Been trying to get my first for a while. Now I’m waiting for the clone, once I can get my hands on one.
I'm waiting for the handheld to dock. Steamdeck with RetroArch is fun, but the third-party docks are a pain for analog. I didn't keep these CRTs for nothing!
@@VideoGameEsoterica Been thinking about GUN4IR, but seemed too complicated to purchase, setup, etc. Siden has been no picnic for sure, only time it EVER worked right was a batocera build with steam deck
if you created an account beforehand, that was the issue. when prompted for shipping address just hit *add new and enter it manually. that worked for me!
taki has a great alternative to the de10nano. But the big problem he has is that he has a store that is green and fails in everything. When you select the pack, it goes to the cart, yes, but if there is another one behind it that buys it faster, it disappears from the cart. Or errors with postal codes, etc.
I remember a SNES game that used discrepancies in oscillators for RNG purposes, I think it was one of the Ultima games. That was problem on bsnes, since the emulator wouldn't take those discrepancies into account, making the RNG predictable. So, not even real hardware is 100% accurate.
Well I'm surprised you measured any difference between the two! What I thought would be the difference is if your capture device is less exact in making sure it gets every frame, or if an SD card could be inconsistent based on when memory access in it's own system is available, and perhaps there is another SD card more accurate across both MiSTer hardware. I've learned it's just never possible to digitize anything in our analogue world with 100% accuracy, and that would include the old analogue video when being converted to digital frames in a different system like hdmi and then you're converting again when sending that to a video capture device.
Does anyone know how to get a case for this if you don’t have a 3D printer? Would a normal 3 stack case work or does it have to be from Taki’s 3D print file?
There are some 'cases' which are just a top and bottom so offer some protection but also give full access to the sides and ports. So they'd be compatible with Taki's clone and something in the interim until full cases are more readily available
Sorry to add to your pain... but how about adding in original hardware to the comparison? Possibly also different revisions of the same console hardware - I wouldn't be surprised if you see some minor variations like these even on original hardware
Another variable is the core itself -if you're mainly seeing differences in PS1 games and not so much on other platforms, it could always be something non-deterministic in the way that core has been written. To test that you'd want to compare several Terasic DE10s to each other.
I have a theory. How does the PS1 core load assets? It must start from SD but because it's a disk based system, in order to copy original ps1 hardware it loads stuff as it goes right? Depending on the cut-scene or game it can load assets in different ways. So maybe it's just the SD card slot is better quality or the route it takes to RAM is shorter... I imagine this will popup with any disk based console depending on how it was programmed. Something like NEOGEO/GB I believe it loads everything into ram before it boots (on mister). I could be wrong on that. Then there's bank switching which is how old games handled accessing large rom files which needs to be instant for emulation or else it's noticeable. So my theory is your theory that the SD card reading is faster, and it will only be noticeable on large CD games. There's also something on the PSX mister documentation that talks about the using Dual Channel RAM on the PS1 being ever so slightly MORE accurate. So I think it's something to do with that. There's also an option in the PSX core to "Pause when CD slow". If it doesn't crash disabling this might cause the cutscene to be better synced and prove that the taki sd is faster.
Love Mark of the Wolves. I think Capcom has the edge, but only slightly. :) At least up through CPS2. The Last Blade 2 is my favorite "serious" Neo Geo fighter. Otherwise Waku Waku 7 or Samurai Shodown IV would be my other favorites. Looking like these new MiSTer boards are great! Now to wait for ready availability. :) I'll probably grab a few full sets.
Capcom has the edge? Thats crazy talk. SNK for life! Haha but seriously…I do love Waku Waku 7. I wish I’d picked a cart up back in the day when it was “cheap”
@@VideoGameEsoterica It plays amazing too, imo. I was playing it earlier today. It's very strategic, a thinking man's game. People think the cast is a bit strange but I see why they were created. The short dude with the backpack may look unimpressive but he can't be hit with some standing kicks and can do lots of close attacks.
The difference in frame rate for the PSX core on the two boards is certainly interesting. I'm positive it has something to do with whatever subtle differences the Taki's chip has with the DE-10, although if someone can find another reason then I'll certainly be surprised. I'd say its still worth getting the Taki clone despite this small difference, though. A couple frames difference isn't gonna be that noticeable to most people, IMO.
No if I didn’t show the difference nobody would see it and it’s still as possible it’s some indeterministic behavior of PSX in and of itself. No real definitive reason as to the “why” yet. Only core that had any genuine offset. Everything else was 1:1 perfect
davinci resolve FTW. Differences could be caused by loading on the SD card, Memory speed, non 60fps fractional fps on capture hardware, and maybe the crystals on the boards governing the clocks. Cuz the FPGA chip is the same
Yeah I don’t think 1 frame matters even after accounting for refresh rate. The speed running community probably won’t allow FPGA to compete in the same category as real console and that’s the only situation I can think of where 1-2 frames over several minutes actually matters. No doubt the input lag on Taki’s board is the same and that’s one of the biggest draws to FPGA imo aside from accuracy. There’s nothing better than getting instantaneous response on 8 and 16 bit generation games.
I have to wonder if on the N64 itself there may be slight differences between different runs, as I recall Robert mentioning that the N64 core on the MiSTer had less variability in RAM access times than the real hardware in some of the tests he did. PSX and N64 are probably right on the edge between previous, more deterministic consoles and later consoles that had a lot more variability due to hardware and BIOS revisions.
@@VideoGameEsoterica It also randomizes what racer is there, how many racers are there, etc. It’s really weird when both racers seem to synchronize when they get a similar movement seed.
@@VideoGameEsoterica the correct one seems to be the one on the right, the title screen wiggles a little bit when it gets "printed" on the screen. Just tested on real hardware
I know the video is focusing on frame rates, but the thing that stuck out the most to me is the difference in color reproduction between the two boards. I think 15:31 is the most obvious example - just compare the blues in the ocean on the planet between the two shots. May be something to consider for a future comparison video
FPGA designs are designed to run at a certain spec and they will either run that spec or malfunction, but they will not delay themselves or run ahead. It's possible for delay to be due to external factors, such as RAM being in use or slower/faster, but not for anything internal.
Thank you for the video, always love your content and updates. I did notice visual bugs on Die Hard Arcade on the clone. The helicopter blades in the intro were flickering and the right wall in front of the elevators had black lines running through it.
Great comparison. Brings up so many questions though: 1. For PSX, what’s closer to original hardware? Terrasic or Taki? 2. How does the QMTECH board compare? 3. Is there variance between Terrasic boards? Is there variance between Taki boards? Hell, is there variance between PSX models? These are all interesting questions, but moreso due to the science, not the gameplay experience.
it may even be inferior because Taki made sure to have better vendor for the RAM. i know there was something with QMtech board that had higher latency than the OG de10 nano but that wasn't the RAM.. just can't remember exactly what it was
With first batch sold out in 9 min and the second 3 it is not much sense to know all this details. Regular user will not have a possibility to buy it for a loooooong time anyway.🤷♂
The Wave Race intro very likely uses recorded controller inputs to drive the racers. This is common through all gaming history. Famously, the introduction to Super Mario Bros does the same thing and can desync in certain situations, causing mario to softlock or die. If not everything in Wave Race is deterministic, same as SMB1, these slight differences in cutscene playback will occur.
I don't personally mind the slight inconsistencies as long as there is no obvious incompatibility or observable issues. I might get one in the future for another setup.
Zero incompatibilities and again the differences could be down to something within PSX hardware / Crash 3 too
I'm happy with what I saw on the MiSTer Pi, the subtle nuances and fluctuations do not bug me at all. Thank very much VGE
You are welcome :)
@@VideoGameEsoterica ♥
I keep hearing Mister Pi... is this based on the raspberry pi or something? Does it mean it's not a 'real' mister?
@pmgf911 no it’s just the name they picked and it’s causing some confusions. It’s just a MiSTer
No it’s real
20:14 I never tire of hearing this timeless vampiric philosophy! 😄
😂
I'm not sure how long it would take but, wouldn't be better to compare both boards individually each with the actual original hardware console, let's say the PS1 and one game for example?
Would that help figure out the frame "behaviour" between the two boards and, at the same time, see which of the two boards is closer to the original hardware?
Do you understand where I'm going with this?
Yes I know what you mean. Maybe in the future
I’ll look in the future
Super useful comparison. Thanks. Did you do a frame by frame comparison with the QMTech clone out of interest?
I have not as of yet no
I know it would probably be time consuming for you and likely wouldn't result in as many views, but I would *LOVE* to hear your thorough breakdown of fields, frames, and frame timing. I feel like I'm already familiar with the math and concepts, but it never hurts to learn something new. Great content as always.
Maybe I’ll tuck something into a future video
thanks for all the time you spent making this video! can't wait for mine to arrive!
Happy to do it. Enjoy it when it arrives!
One thing to keep in mind is that the crystal oscillators are not going to be exactly the same board to board, and there could be as much as ±50 Hz deviation. Seeing a frame difference of ±5 frames makes sense over a long period. Also, it's important to remember that FPGA bit streams are not like software - if the FPGA wasn't of high quality, the system wouldn't even run. The FPGA's silicon floor plan has a specific timing constraint, which the bit streams are carefully generated to match.
It could be down to fluctuations in the oscillation. Which you can see in real hardware too
50 Hz? - that's one whole second of more/less gameplay here in the PAL region ;)
@@DavidFilskov 50Hz on a multiple megahertz crystal, of course.
@@Cicirifu Sure! - just joking :)
Well, as someone who plays only on original consoles and for who Mister was just an affordable way to play arcade games, which I could not buy other ways, it means nothing why clone differs from DE10 and why is still somehow good. For me, the problem is that there are differences. Buying clone boards for people like me is not a good idea. After all, all mister vores are developed and optimized for DE10. So, if they aren't perfect even on DE10, I still expected slightly more accuracy on the original, and for me, it is still a deal worth extra money.
water waves in Waverace64 are not pre-rendered to appear at the same exact time. i'm not sure if it's real time, but i'll at least it's pretty damn close! So the pre-recorded paths the racers are following can differ a bit everytime you let the intro play.
Basically, the intro sequence isn't a pre-recorded replay. recorded replays, to be very precise, must record the waves patterns, because there IS a timer in actual races, so everything have to be recorded. While in the game intro, it's not necessary to record the waves patterns. So basically, in the intro, water behaviors will always be different, even if it's just slightly.
That’s why I mentioned peculiar behaviors in the intro sequence. Kind of a random seed situation
Some PS1 games with issues on MiSTer FPGA you could test
Constructor (Europe) 2D visual glitches
Warhammer: Shadow of the Horned Rat (USA) black screen after completing the second mission
Nuclear Strike (USA) black pixels in parts of level polygons.
Thrasher Presents Skate and Destroy & Dave Mirra Freestyle BMX (USA) Random hangs from music streaming.
I think some of these issues might have been related to RAM speed and seeing as the Taki Clone has different speed RAM it may show different results.
I’ll look into it
@@VideoGameEsoterica Did you find out?
because of the side-by-side nature of the footage you can cross your eyes, merge the images and the differences will shimmer. (like the sunglasses on the logo)
also works for those spot-the-differences games at the bar.
Haha not a technique I’ve ever heard
Thank you for the video.
Question: You have a couple of MiSTer builds on the DE10-nano, are the psx cores the same on those performance-wise?
I can test on the 2.5V ram version but that would be an outside the norm test as it’s the oldest “modern” ram version
This is what I thought... if you ran 2 de10 nanos with same stock parts would they match up
I may look in the future
Thanks for taking the time to make this video! Have you had a chance to compare the analogue video out on the older 16-bit IO board vs the Taki 24-bit one? Curious to know if there's a noticeable difference when using a PVM.
You’ll notice a diff in cores that support it. The color banding will be near to eliminated
Thank you for making the video. I'm having a hard time hearing the audio and music for all of these games because there's too much narration. The most important thing for me with any of these devices is to hear how the audio syncs up with the action on screen. I scanned through a ton of the video searching for that kind of analysis and didn't find anything. Any insight into that specifically?
I left a few sound samples. Audio is 1:1
@@VideoGameEsoterica thank you for your response and your efforts.
Not to diminish the effort you put into the frame by frame analysis on the timeline, but for me the most compelling evidence in favor of Taki's clone comes when you play ESP RaDe near the end. If the output of the clone doesn't conflict with your muscle memory in that sort of game where frame timing is crucial, what else can you ask for?
I look forward to running some similar tests with V-ISM combo timing in Alpha 3 if I manage to snag a board from batch 2. Based on the results you've presented, I doubt I'll be disappointed.
On that note, as a Capcom guy (going all the way back to Champion Edition), I've always had a soft spot for Garou. Great game, very much looking forward to the release of its sequel next year. I'm also very much looking forward to any updates you're able to provide in future videos regarding guidance from Porkchop or based on your own testing on safely pairing the Taki clone with the Mister Addons analog boards.
That’s what I put the test in there :) show it’s possible
Best looking and playing fighter on Neo Geo
At 12:22, the "Die Hard Arcade" on the title screen animates only on Taki's board. Why do you think that is?
We’re looking into it. No clue. Were thinking maybe there’s a random variant in the attract sequence
Solved: YT compression stupidity. Both animate on the export.
Here's a potential future topic... for those of us with multiple MiSTer FPGA systems, how can one setup one card and clone it to other cards... so one doesn't have to do every card from scratch? Is there way? If so, I haven't figured it out. I tried using dd on Linux once and that didn't seem to work. I was able to capture an image and then write it out to another card of the same make/model/size... but then it didn't work.
There is a utility you can add to MiSTer that will duplicate your card to an equal or larger size card
You can migrate to a smaller SD card with the MiSTer utility, just as long as the total data used on the source card doesn’t exceed the size of the destination card. It’s a pretty handy and clever script.
You’ll need a usb SD card adapter for the destination card and be prepared to for it to take a *long* time if you have a lot of data.
True you can go smaller too
@@VideoGameEsoterica What is the name of that utility and if you have the URL for it handy, please share.
Links get eaten but google “natrox mister FPGA utilities github”
I wonder, is such minor variability also seen between different actual hardware skus? Like, between older and newer PSX models?
That would be the next step if there is a part 3
@@VideoGameEsoterica i wonder if this would happen between 2 de-10 nano boards as well
There are revisions so it’s possible
Taki's version has slighly more vivid color output. You can notice this in the pink on Bomberman
I’ll have to take a closer look
I got one of the second release!! I’m so excited. I’ve had an eye on the mister project for years and now I’m finally jumping in. Can’t wait for jt to arrive.
Congrats. Welcome to the club!
Fantastic comparison, very useful. Good shmup skills there too 👌
Haha thanks. I do love my shmups
I noticed that there's a firmware file for the misterpi on the store website. Is this frame rate difference something that can be adjusted with a firmware update so that it more closely matches the DE10-Nano behavior?
Also, when is he going to release the Hyper Neo Geo 64 clone board / core combo?
No that really isn’t part of this. And yes we need HNG 64!
@@VideoGameEsoterica Any chance that it could be related to his IO Board or the ram chip?
Thanks for the comparison video! I wonder how you setup the recording:
Recording via HDMI?
Recording in OBS?
I myself have tried get good recordings/stream via HDMI both in NTSC and PAL modes. I really don't get it, how to set it up so the video will be super smooth. I mean the SNES core runs at 60.1 Hz in NTSC mode. How do I get perfect recording of that, is it even possible?
HDMI into Elgato 4K Pro MK2 into the Elgato software
Is it a CRT shader you're using with Street of Rage 2?
Yes it’s just part of that core
Have you done similar tests between 2 de10 nano boards?
Not yet
@@VideoGameEsoterica Cool, it would be interesting to see if there any variances between de10 nanos as well :)
I’ll def check into it
Why didn’t you also compare to QMTech’s board?
That would just be a diff vid
Can someone explain the allure of an FPGA? I get that it more accurately creates the electrical configuration of a retro device among having other uses, but once you output to hdmi or use usb controllers doesn't that change things electrically anyway and add lag? Don't get me wrong, I love retro stuff and electronics but I can't seem to find the niche this device fills.
All modern TVs will introduce lag. As will monitors. You can do SNAC on MiSTer and not introduce lag to the system via those adapters. As far as response time is concerned MiSTer is as good as it gets outside real hardware
The draw for me is that I can output the exact refresh rate from the original hardware. No screen tearing, dropped or duplicated frames. Emulation always bothers me with the stuttering or tearing.
A very nice feature
@@CultOfTheGlendaAll modern PCs are capable of outputting arbitrary refresh rates and most modern emulators take advantage of that. Stuttering and tearing would be a configuration issue at your end.
I can't believe how you have new clips all the time 24/7(!).
-Do you ever sleep?
As for Wave Race 64:
Sometimes when i leave game running for a while for nonstop title screen(s), there are differences how scene behaves on actual real hardware.
Since it's realtime intro, it actually seems to be that under variable wave conditions, A.I racers sometimes make mistakes. -I've seen them even fail some jumps on intro(!).
Just like when you play the game, it is never certain that all the waves behave exactly same as when you previously played it.
Haha I sleep. Sometimes 😂
just ordered one!!! cant wait to set it up :D its my first one...already have tons of roms waiting
Welcome to the club :)
When one or the other goes ahead, are their respective audio in sync with their video? That's all that would matter to me, that the audio is in sync with its own video. Doesn't matter that one might load seven frames faster.
Audio is in sync
is mister powerful enough to upscale ps1, Saturn and n64 games?
No you won’t really be getting any internal rendering resolution bumps like you think of with software emulation. PSX has SOME improvements
One thing that may be overlooked in this video is how the MiSTer Pi handles inputs and input latency. A good test would be to get a controller where you can program a long macro sequence and compare the footage after the button press for the whole macro sequence.
Excellent job though, you did a crazy amount of work in a short amount of time!
Yes I managed the build and initial review and this comparison in about 28 clock hours haha
@@VideoGameEsoterica always impressed with your dedication!
I love doing it :)
Great analysis. I’m 100% in for one now. 👍
@craigreaper7295 👍
I imagine it would perform similar to the original mister fpga since it is a clone board. I would be interested in seeing the durability of this Taki board but no idea how you would test it.
Durability is only something we will know with time but I have no reason to think it would be an issue
This is a great test! Keep in mind that the clock frequency of real consoles will always be slightly varying to some degree, especially so if the console is subjected to increasing heat from internal heat buildup and high ambient temperatures, which could also cause some variations to how the frames get produced. In this case I would overall expect the Mister hardware, both the originals and the Taki clones, to act much more consistent than even the original consoles. Ultimately there is only one way to find out, time to bring out the hairdryer. 😅
Lolol hairdryer. Very scientific 😂
@@VideoGameEsoterica You could also try placing them into an oven at lower settings, as you can more easily control and monitor the exact temperature that way. 🤣
That would really make it a MiSTer Pi wouldn’t it? Baking it 🤣
In theory, the Taki clone may actully perform better under higher temperatures as they are using the automotive variant of the SoC which tolerates higher temperatures than the industrial variant on the DE10-nano.
Really wouldn’t be running long enough at all to show a variable
Great work! It would be great if you could test accessories such as Mistercade with the clone board in a future video.
I may very well do that :)
Great vid, and agree with the saturn, n64 and ps1 core stressing out mister devices the most. Maybe fun to also look at and compare the various turbo cores? ps1, n64, neo geo......?
Maybe turbo in a future vid :)
I wonder if this differences with the psx core is because the Taki clone board has a better sd card reader than the original board so the iso disc reads are quicker on the loading parts.
Due to the fade ins being the same loading is doubtful
How similar comparison looks for qmtech?
Will compare shortly
If one is faster or slower it must be in any loop since it is the same hardware, if any core varies it is for other reasons
Yes for PSX it could be something the hardware naturally does in FPGA or on real hardware
I'm curious how the 2 boards compare to a second Terassic board and against real hardware. I bet all perform within the same margin of error. Even the real hardware always has the slightest of variations because every component that goes into them has a specific manufacturing tolerance. Those differences will all add up to a positive or negative performance difference to the theoretically perfect device that you would never see without comparing them side by side.
I may do a three or four way shootout
Someone made such comparison. 2 DE10s vs MisterPi:
ua-cam.com/video/9E-8Uz9qI3M/v-deo.html
This is some great testing work! Thanks so much for this.
Happy to do it
Really want one for competitive street fighter on cps1 and 2 can u test those?
Next test will be arcade focused
Lol you low key flexed your shmup skills. Thats a creative way to benchmark things.
Haha it’s the best way. Pure feel. All that matters. In a bullet hell it either has the feel or it doesn’t and you can tell immediately
No one's talking about the two clumps of white powder in the background ?? 😂
lol it’s candle wax or table salt
Do you recommend Mister Pi for CPS2 games? it's the same as MiSTER?
In my casual “just playing” yep it’s rock solid good. Everything is
Thank you. It has 24bit video output. But i read that original mister can't output hdmi and vga simultaneously. Old version with 18bit can handle 2 video outputs. Is it true?
Not true. All normal triple stacks can output analog and HDMI at the same time. Only thing that can’t is dual ram builds
Garou's only Capcom rival is Darkstalkers don't @ me
Is there a USERIO port on the clone and does it work with snac and snax?
Yes and yes
@@VideoGameEsotericathank you and thanks for your extensive knowledge and coverage on mister and FPGA. You’re my go to for anything and everything mister.
Happy to do it and plenty more coming :)
Got hit with 50$cdn import fees through DHL. Anyone else or just unlucky? Still excited though.
I’m sure it’s just down to where you live
What's a good power supply i can buy? Any recommendations.
Some of the raspberry pi ones are good (ironically). I got a third party pi5 one in the UK. It's 5V4A and 20W plus has an in-line power switch. Taki said anything over 5V2A is fine and mentions it on his website
Anker is a good brand.
I’ve been using the Steam Deck one as the PD negotiates down to 5V with a tiny bit of heat. I’ve also used the Pi adapter
Very good
Thank you for the info.
When you're talking about the 59.94/60Hz thing, does that mean you're using vsync_adjust=0 in your .ini file? Do you know how they compare in vsync_adjust=2 or does your capture hardware not support it?
It just means the standard for video refresh rates in the modern tv / capture era
@@VideoGameEsoterica Perhaps I didn't explain very well. I'm asking what setting you're using in your MiSTer.ini setting file for vsync_adjust.
Adjust=0 across both tests
I think the aliens from 16:42 are from Omicron Persei 8.
That does sound like Lrr now that you mention it. Good catch
My main interest would be if the arm side is more performant on mister pi
It wouldn’t be. Same exact chip
Thank you for your hard work and effort. The differences you found between the Terasic DE-10 and Taki MiSTer Pi are negligible and are absolutely not a concern for me at all. I'm so happy and fortunate I was able to get my Taki Mister Pi order in yesterday -- my first MiSTer FPGA ever. Now, I'm torn on what display to pair it with -- a CRT tv/monitor, or a modern HDMI display (then I will need something like retro tink?). What is your favorite choice of display and connection setup with your MiSTer FPGAs?
And the differences may be down to something in PSX hardware still. Both boards are 10/10s
You could use mister with a Tink 4k to great results. Otherwise, mister scales to 1080p out of the box, so you'll still get good digital output with no additional hardware required.
@@derekhaines1199 At $750 for the Tink 4k it's way overpriced for what it does. I think I will connect the MiSTer Pi to my CRT. Honestly, for $750 I think I'd even rather spend it on a large AV receiver that can act as a hub for multiple game consoles and then output to a modern tv or even CRT.
@@lordbacon4972 totally valid! The great thing about mister is it has great analog and digital output. So you should be able to plug it into a CRT or a modern display and get a quality picture either way. Obviously there are benefits to using a CRT, so I think if you go that route you’ll have an amazing experience.
Can you recommend how to get power to this thing? I’m confused on that part and mine is coming in the mail
Any wall wart that delivers 5V/3A will be fine. I’m using the Steam Deck charger and it negotiates the power down as the board is compatible with USB PD
@@VideoGameEsoterica okay so the Rog ally one would work then.
I haven’t tested that but check the stats
Can the fast frame issue be resolved through a software update?
Fact is the root cause is still not certain. We’d need hundreds of independent tests across so many variables to even know a why
I'm having a horrible experience trying to get their mega bundle. Taki's website is a glitch fest. Batch #1 I had the mega bundle in my checkout cart and was stuck there unable to proceed because the website stated that my zip code is invalid. Batch #2 I had it in my checkout and the exact same error occurred this morning! I had it in my check out for both batches and I'm not able to purchase it! Sent messages to "customer support" never received a response! My zip code was typed perfect (it is only 5 numbers lol) and I quadruple checked it because I was pissed I couldn't checkout......now they are out of stock for the second time.
Yes seems like some real growing pains going on
Sir! How many Misters do you have?
Too many. Six (but obv for testing and channel purposes)
GOOD MORNING SIRS
Haha
@@VideoGameEsoterica no worries I was not judging or anything. Been trying to get my first for a while. Now I’m waiting for the clone, once I can get my hands on one.
Haha didn’t seem judgmental at all
I wonder how the QMTECH board stack up
I may throw that in the mix in the future
Davinci Resolve
Resolve is all I’ll ever use
I was on the store page and it sold out in zero seconds while I was waiting in line for checkout.
Yeah I heard it was fast this morning
I'm waiting for the handheld to dock. Steamdeck with RetroArch is fun, but the third-party docks are a pain for analog. I didn't keep these CRTs for nothing!
Gotta keep those CRTs!
i was one of the lucky ones, able to snag one last night. Totally want to do a sinden build
I’d recommend GUN4IR over Sinden. Better software support without custom cores and much better accuracy without any borders
CRT + Guncon, my dude
Well obv I have that too lol but he’s talking LCD
@@VideoGameEsoterica yeah, my last crt was a massive 42" Sony flat screen, replaced with a 50" plasma (THOSE were great tv's).
@@VideoGameEsoterica Been thinking about GUN4IR, but seemed too complicated to purchase, setup, etc. Siden has been no picnic for sure, only time it EVER worked right was a batocera build with steam deck
I think you would get the same result comparing 2 DE10nanos
I believe I would too. Even two actual PS1s
Had one in my cart but it wouldn't accept my address for some reason. Hopefully that isn't the case tomorrow morning.
Day zero bugs that hopefully get ironed out
if you created an account beforehand, that was the issue. when prompted for shipping address just hit *add new and enter it manually. that worked for me!
Good tip!
taki has a great alternative to the de10nano. But the big problem he has is that he has a store that is green and fails in everything. When you select the pack, it goes to the cart, yes, but if there is another one behind it that buys it faster, it disappears from the cart. Or errors with postal codes, etc.
Growing pains with all things new
@@VideoGameEsoterica Have you also tried the fps delay on the qmtech one?
Not as of yet
@@VideoGameEsoterica Will you do it to compare de10nano with the clones?
Probably
I remember a SNES game that used discrepancies in oscillators for RNG purposes, I think it was one of the Ultima games. That was problem on bsnes, since the emulator wouldn't take those discrepancies into account, making the RNG predictable. So, not even real hardware is 100% accurate.
Never is. So many revisions and minor variations. Nothing is consistent
Well I'm surprised you measured any difference between the two! What I thought would be the difference is if your capture device is less exact in making sure it gets every frame, or if an SD card could be inconsistent based on when memory access in it's own system is available, and perhaps there is another SD card more accurate across both MiSTer hardware.
I've learned it's just never possible to digitize anything in our analogue world with 100% accuracy, and that would include the old analogue video when being converted to digital frames in a different system like hdmi and then you're converting again when sending that to a video capture device.
Yes there are such a huge amount of variables it’s impossible to predict how everything works in conjunction
Does anyone know how to get a case for this if you don’t have a 3D printer? Would a normal 3 stack case work or does it have to be from Taki’s 3D print file?
Just search for a 3d printing service. It's cheap 😅
Libraries, online print services. Google your area and 3D printing :)
Yes tons exist
There are some 'cases' which are just a top and bottom so offer some protection but also give full access to the sides and ports. So they'd be compatible with Taki's clone and something in the interim until full cases are more readily available
Sorry to add to your pain... but how about adding in original hardware to the comparison? Possibly also different revisions of the same console hardware - I wouldn't be surprised if you see some minor variations like these even on original hardware
Another variable is the core itself -if you're mainly seeing differences in PS1 games and not so much on other platforms, it could always be something non-deterministic in the way that core has been written. To test that you'd want to compare several Terasic DE10s to each other.
Haha I dunno how far that rabbit hole I want to go
Correct and I left that as an option on the board
I have a theory. How does the PS1 core load assets? It must start from SD but because it's a disk based system, in order to copy original ps1 hardware it loads stuff as it goes right? Depending on the cut-scene or game it can load assets in different ways. So maybe it's just the SD card slot is better quality or the route it takes to RAM is shorter... I imagine this will popup with any disk based console depending on how it was programmed. Something like NEOGEO/GB I believe it loads everything into ram before it boots (on mister). I could be wrong on that. Then there's bank switching which is how old games handled accessing large rom files which needs to be instant for emulation or else it's noticeable. So my theory is your theory that the SD card reading is faster, and it will only be noticeable on large CD games. There's also something on the PSX mister documentation that talks about the using Dual Channel RAM on the PS1 being ever so slightly MORE accurate. So I think it's something to do with that. There's also an option in the PSX core to "Pause when CD slow". If it doesn't crash disabling this might cause the cutscene to be better synced and prove that the taki sd is faster.
The dual ram just affects audio related items so that wouldn’t have much of a diff here
I want one just to have a dedicated dual ram setup.
Honestly currently dual ram is basically irrelevant. Maybe not in the future. But currently
I have dual ram and let me tell you that you are probably better off using those FPGA pins for analog video out.
Love Mark of the Wolves. I think Capcom has the edge, but only slightly. :) At least up through CPS2. The Last Blade 2 is my favorite "serious" Neo Geo fighter. Otherwise Waku Waku 7 or Samurai Shodown IV would be my other favorites. Looking like these new MiSTer boards are great! Now to wait for ready availability. :) I'll probably grab a few full sets.
Capcom has the edge? Thats crazy talk. SNK for life! Haha but seriously…I do love Waku Waku 7. I wish I’d picked a cart up back in the day when it was “cheap”
Art Of Fighting 3 is awesome but it doesn't get the praise it deserves. I think it's up there with the best.
@simon41978 amazing animations. Just not the biggest roster and by that point people wanted dozens of characters to choose from
@@VideoGameEsoterica
It plays amazing too, imo. I was playing it earlier today. It's very strategic, a thinking man's game. People think the cast is a bit strange but I see why they were created. The short dude with the backpack may look unimpressive but he can't be hit with some standing kicks and can do lots of close attacks.
The short characters that you whiff a high kick on drive me insane
Bought one for my buddy. Can't wait for him to get it.
What a friend!
@@ninjatemper8797 It sold out in like 5 seconds. I couldn't get one. another round is going up tomorrow at 8am EST. Good luck.
More orders tomorrow 8 am EST
There will be batches after these initial 2 as well
Dayyyuuummmmm dawg!
🌮
Just a couple comparisons. Like 12 hours worth of work style 🤣
He brings the taco, you bring the cheese (the second tier cheese)
Why would I bring VT cheese though? I live right here in WI, so top-tier cheese all the way baby!
@@VideoGameEsoterica if you look at the numbers Wisconsin has won many more cheese competitions than VT!
Of course i'd love a test against real hardware, and that's an other good reason to use your expensive scalers again
Haha maybe in the future
The difference in frame rate for the PSX core on the two boards is certainly interesting. I'm positive it has something to do with whatever subtle differences the Taki's chip has with the DE-10, although if someone can find another reason then I'll certainly be surprised.
I'd say its still worth getting the Taki clone despite this small difference, though. A couple frames difference isn't gonna be that noticeable to most people, IMO.
No if I didn’t show the difference nobody would see it and it’s still as possible it’s some indeterministic behavior of PSX in and of itself. No real definitive reason as to the “why” yet. Only core that had any genuine offset. Everything else was 1:1 perfect
@@VideoGameEsoterica Guess we won't know for sure until more tests are done, like you said.
You’d need a data set across 100s of unique setups to really narrow down a cause
@@VideoGameEsoterica does the game play faster or just cutscenes?
Game is 1:1
the question is, which one is more accurate…. ?
The age old question
davinci resolve FTW. Differences could be caused by loading on the SD card, Memory speed, non 60fps fractional fps on capture hardware, and maybe the crystals on the boards governing the clocks. Cuz the FPGA chip is the same
Yes. Many diff variables could be at play
Yeah I don’t think 1 frame matters even after accounting for refresh rate. The speed running community probably won’t allow FPGA to compete in the same category as real console and that’s the only situation I can think of where 1-2 frames over several minutes actually matters. No doubt the input lag on Taki’s board is the same and that’s one of the biggest draws to FPGA imo aside from accuracy. There’s nothing better than getting instantaneous response on 8 and 16 bit generation games.
Oh yes latency is dead on the money. Gameplay across both is amazing
I have to wonder if on the N64 itself there may be slight differences between different runs, as I recall Robert mentioning that the N64 core on the MiSTer had less variability in RAM access times than the real hardware in some of the tests he did. PSX and N64 are probably right on the edge between previous, more deterministic consoles and later consoles that had a lot more variability due to hardware and BIOS revisions.
There’s also random seed differences in the intro
Then its settled; no one should ever buy a DE-10 in the future. The clone is a much better value for the money.
If stock can keep up then yes it’s the economical choice
Waverace 64 has a lot of randomness to the intro though the RNG seed was similar between those.
Yes that was my point with that comparison. Not all differences are meaningful
@@VideoGameEsoterica It also randomizes what racer is there, how many racers are there, etc. It’s really weird when both racers seem to synchronize when they get a similar movement seed.
Lucky to get such a close one. Vs the fighting games where it took forever to get the same stage twice
Nobody noticed the different rendering of the title screen at 12:22?
Wow no…what is that?? Same file, same core, same game. Thats wild
@@VideoGameEsoterica now I need to power on my Saturn to see which is the accurate one :D
I’m so confused by that haha
@@VideoGameEsoterica the correct one seems to be the one on the right, the title screen wiggles a little bit when it gets "printed" on the screen. Just tested on real hardware
I posted it on the Saturn channel. I’m wondering why it’s diff. Maybe there are multiple attract sequences?
I know the video is focusing on frame rates, but the thing that stuck out the most to me is the difference in color reproduction between the two boards. I think 15:31 is the most obvious example - just compare the blues in the ocean on the planet between the two shots. May be something to consider for a future comparison video
I noticed this too and was surprised it wasn't mentioned, to my eyes the DE-10 nano had slightly more vibrant colours.
YT can slightly shift color reproduction when compressing side by sides
@@VideoGameEsoterica ohhh interesting! Thanks for clarifying 🙏
Yes the hard part is the final compression which I am not able to control
Could this difference in psx be down to 'silicon lottery' the way the same graphics cards can produce different results?
I wouldn’t expect it to be that
FPGA designs are designed to run at a certain spec and they will either run that spec or malfunction, but they will not delay themselves or run ahead. It's possible for delay to be due to external factors, such as RAM being in use or slower/faster, but not for anything internal.
It’s also possible PSX exhibits loading speed variations per run too on real hardware
Excellent video! Thanks for doing these. Love it and keep up the great work. Your videos always make my day ❤
Aww thanks. Plenty more coming
@@VideoGameEsoterica Hell yeah 😎
So, the frame differences cannot be framed as hardware failures. 😁
Haha dad joke
Why the heck does it have "Pi" in the name though? How confusing.
Just what they went with 🤷🏻♂️
If the name is too confusing for some people then maybe this product isn't for them I guess...
@@delmaralvarado7776 Or maybe, just maybe, the name is just plain bad despite the product being good.
At the end of the day the name really doesn’t affect anything
- $3.14 off
Good sign for the eventual handheld? I have zero interest in the clone other than how it will relate to a portable mister.
Fascinating stuff, tell us about more things you have zero interest in.
@@tog2842fr people always adding dumb shit to their comments lol
All signs point to an excellent handheld
lol be nice
same here
Can't wait for mine to get here.
Hope it goes fast!
Thank you for the video, always love your content and updates. I did notice visual bugs on Die Hard Arcade on the clone. The helicopter blades in the intro were flickering and the right wall in front of the elevators had black lines running through it.
I’ve found that YT stepped on the compression on this clip. I posted the export clip on discord. YT compression does odd things sometimes.
Hey, your voice is much improved in this video.. keep up the great content! 😄👍
Seems like the same voice to me
Great comparison. Brings up so many questions though:
1. For PSX, what’s closer to original hardware? Terrasic or Taki?
2. How does the QMTECH board compare?
3. Is there variance between Terrasic boards? Is there variance between Taki boards? Hell, is there variance between PSX models?
These are all interesting questions, but moreso due to the science, not the gameplay experience.
1: I’d have to check
2: next batch of tests
3: yes there are multiple revisions of the Terasic board
4: absolutely differences in PSX revisions
@@VideoGameEsoterica Sigh. Capturing....
(This gives me an excuse to use my finally working PSX capture setup!)
Haha like we need excuses to try something new with capture
After watching this, all I've got to say is I've never been more grateful to be an owner of the superior original OG Mister board.
lol I never said it was superior
it may even be inferior because Taki made sure to have better vendor for the RAM. i know there was something with QMtech board that had higher latency than the OG de10 nano but that wasn't the RAM.. just can't remember exactly what it was
The CAS on the QMTech was technically slower than on the DE10
With first batch sold out in 9 min and the second 3 it is not much sense to know all this details. Regular user will not have a possibility to buy it for a loooooong time anyway.🤷♂
There will be more batches coming
i drubnk a shot when he saied that oneframe difiernce is notmal... I DOnt feeel fo good...
lol
They sold out in like 2 min. I couldn’t get one.
I heard it went even quicker today
Didn’t know you where such a cinema/film guy would love to know some of your favourite movies!
Haha maybe one day but yes I’ve worked in film for two decades now
His favourite film is Die Hard.
It’s up there
The Wave Race intro very likely uses recorded controller inputs to drive the racers. This is common through all gaming history. Famously, the introduction to Super Mario Bros does the same thing and can desync in certain situations, causing mario to softlock or die. If not everything in Wave Race is deterministic, same as SMB1, these slight differences in cutscene playback will occur.
That’s what I said “intro does something weird even on real hardware” as you can have minor variances in the same in engine cutscene