MiSTer FPGA Forum

just watched this, its really interesting in relation to lag or the lack of it on Mister.

yes, that is a great real world example of how well FPGA gaming matches original hardware, regarding input lag.
kind of drives home the point that FPGA gaming is as close as you can get to the real thing (and IMO, more versatile and subjectively better)

heres a reddit post that compares shinobi running on original hardware and compares it to some common emulation youd find today.
real hardware, Pi3 with runahead, Pi3 without runahead, and the MiSTer

https://www.reddit.com/r/fpgagaming/com ... _runahead/


from their findings, the MiSTer matches original hardware in input lag, and the Pi3 with 1 frame of runahead is almost 2x laggier, and even worse without any runahead.

mind you, i have no idea how optimized this person tried to make their Pi3, because i have had decent results with some settings (video_threaded=false, video_driver=dispmanx, video_max_swapchain_images=2)
but, it is still amazing how the MiSTer matches original hardware in input lag

Audio lag is also important. There is no emulator without audio lag.

annette wrote: ↑Sun Sep 12, 2021 3:34 pm just watched this, its really interesting in relation to lag or the lack of it on Mister.

Input lag is the major killer feature where no matter what trick is used software emulation cannot and probably never will beat FPGA. 90% of the time four frames of latency might not really matter, but even if we can't easily perceive four frames of latency visually it is still 66.4ms of time (at 60fps) which is an eternity in a twitch platformer and it certainly will not feel right to play, shows itself clearly in edge cases like this where "the lag is cumulative".

I've been dabbling in emulators since the Kgen, Zsnes, UltraHLE days but something 'off' meant I'd play for 10 minutes and get bored, still today even with a decent Retroarch setup. MiSTer genuinely changed that and I've played a few titles to completion, I realise now that it was the lag that was killing my enjoyment..

Input lag it's what makes the game "feel" right and, in many of those NES-era games, very precise timing was required to successfully play some levels.

Of course, even the real hardware had some lag, it's not possible to have zero lag from input to screen reaction, because the input must be proceed and the game code will take some time to react to it but, the main issue those test are proving, is that FPGA has the SAME input lag as the original hardware, which makes playing the game easier, because it was designed that way!

Also, not having audio lag because the audio circuits runs in parallel with everything else on FPGA, also helps a lot, because the audio cues are better in sync with what you see on screen.

Putting both together, it's clear why you can really game *seriously* on FPGA ( playing a game to completion, or competitively ), while with other emulation methods it's all more casual.

I also have a RPI4 in a Gameboy DMG case, which is very nice and it's portable but, sometimes I feel overwhelmed by the extreme flexibility of Retropie/Retroarch, there are just too many options to configure, with dozen of different emulators for the same system, let alone all the extra stuff like Attract mode, Scraping for game descriptions, setting up borders, flyers, etc, which will result in spending way more time configuring than playing the games.

annette wrote: ↑Sun Sep 12, 2021 3:34 pm just watched this, its really interesting in relation to lag or the lack of it on Mister.

fascinating watch, thanks for posting, brilliant example

The uploader of that video mentioned that he didn't actually use an original controller with his MiSTer setup to beat that level. It was a Hori Fighting Commander, which is nowhere near the best in terms of input lag from what I've seen in the spreadsheet. That's a bit of a relief for me, because I was dreading the thought of needing to buy original controllers and SNAC adapters for EVERYTHING.

El_Syd wrote: ↑Tue Oct 19, 2021 11:40 am The uploader of that video mentioned that he didn't actually use an original controller with his MiSTer setup to beat that level. It was a Hori Fighting Commander, which is nowhere near the best in terms of input lag from what I've seen in the spreadsheet. That's a bit of a relief for me, because I was dreading the thought of needing to buy original controllers and SNAC adapters for EVERYTHING.

Yeah! And it's worth noting that as you get into the "B-tier" and higher in the spreadsheet, the differences get smaller and smaller.

For example, I have an 8bitdo Gbros adapter I use in wired mode with my SNES Classic controllers. I also made a Daemonbite NES controller adapter.
The Gbros has ~6 ms measured latency on average while the Daemonbite has ~1 ms. So mathematically, the Daemonbite has 1/6 the latency of the Gbros, which sounds like a lot less!

...but when you get to real-world applications, a difference of 5 ms is less than 1/3 of an NTSC frame. Even for a speedrunner, it would be pretty hard to feel the difference between the two (although both might still feel a little different from real hardware.)

SNAC is amazing but if you want to keep things on a lower budget, you're not missing that much by using a decent-rated USB controller. It's a whole lot of fun to strive for perfection but definitely not necessary.

EmK_IronFist wrote: ↑Wed Oct 20, 2021 9:11 am SNAC is amazing but if you want to keep things on a lower budget, you're not missing that much by using a decent-rated USB controller. It's a whole lot of fun to strive for perfection but definitely not necessary.

I am a perfectionist, especially when I'm creating a setup like a retro gaming room with a MiSTer and CRT. I'll see how things go with the iBuffalo SNES controller which is supposedly the best USB one. Dunno why I don't see many people mention it.

El_Syd wrote: ↑Thu Oct 21, 2021 2:06 pm Dunno why I don't see many people mention it.

It has the best latency on paper based on limited testing but some people don't like the feel/build quality and it's been discontinued so the price has gone up a bit. Originally it was $10 USD shipped. Now going prices for used ones are often $25-$40 (usually at least $30 when you include shipping) and there's competitive alternatives at those prices. When the "best" is only microseconds ahead of its closest competitors, it really comes down to preference and what's available, and I think people tend to go for newer options that are a little easier to find. Honestly though, can't go wrong either way as long as you like the feel.

Please note that the Retroarch core being used is Nestopia. MESEN core (Retroarch) on a good HW is as much lagfree as MiSTer. MiSTer has one core for each system which matures overtime that makes it great. Software emulation if done correctly on a good HW would be equally good.

Also PI3/4 are not Retrogaming devices nor are powerful. Comparing MiSTer with PI will not give you the real answer of HW Vs SW emulation. You need a good multi core CPU with high boost clocks.

Ofcourse if cores are there on MiSTer, its hands down the best alternative since its 15 WATT power consumption. But for games / cores not on MiSTer, a mature emulator on Retroarch (non pi) can be good enough.

suverman wrote: ↑Tue Oct 26, 2021 3:39 amOfcourse if cores are there on MiSTer, its hands down the best alternative since its 15 WATT power consumption. But for games / cores not on MiSTer, a mature emulator on Retroarch (non pi) can be good enough.

Thank you for saying this. I thought I was the only person who cared about the power consumption.

I once asked on Facebook if MiSTer was a good "green" (well as green as gaming gets) platform compared to PC emulation overkill, and I was laughed at and told "who f*cking cares" which kinda shut me up speaking about it.

caffeinekid wrote: ↑Thu Oct 28, 2021 12:07 am Thank you for saying this. I thought I was the only person who cared about the power consumption.

It was absolutely a big deal to me. The fact it sips power where my PC chomps through it, especially in the summer, is a big difference.

A lot of that is personal comfort but it does feel better to use some cool and efficient.

Larkhainan wrote: ↑Thu Oct 28, 2021 2:26 am A lot of that is personal comfort but it does feel better to use some cool and efficient.

Yes! It's often ignored but power consumption on FPGA is usually lower than the original machines. It's even lower jf therr us no ARM present (I forgot what it was on MiST but its lower than the DE10 Nano). By that standard an accurate FPGA core is a much better choice for long term use, or for portable setups.

That said, RPis are also low power consumption so power alone is not the only criteria for comparison.