Same here. I'd be suprised if I even knew what day it was.  

 

I don't think you'd be able to tell.

I did a test earlier, and posted about it in another thread.

The short of it is, had two Super Mario Bros. games running side by side.
One NES, one AVS.  Started them at the same time and the games were in sync at first.
Mario has a timer. It has 400 faster than normal seconds that count down.  Slowly the audio become unsynced between the two systems, and near the end of the countdown, there was a difference of 1 Mario Second between them, the AVS playing ever so slightly slower.

I highly doubt this could be noticeable to a human being, if the two weren't running side by side with clocks.

During play, there's no way you'd notice it. No way.
 

I do plan to notify him, I was just hoping to have it confirmed first (I guess it could be some issue with my NES). Ichinisan got nothing when he mentioned a Ninja Gaiden glitch so I wanted to bring up all three issues together (third issue is that 720p doesn't work on all TVs).

What HDTVs don't support 720p, besides the earliest of pre-2005 HD CRTs and rear projection sets (which also presumably don't take 1080p or any progressive format above 480p, creating compatability issues with a lot of modern devices)?

The TVs actually do support 720p signals, just not the Hi-Def NES' 720p. I should have worded it differently:
"third issue is that [Hi-Def NES'] 720p doesn't work on all [720p] TVs"

A PlayStation 3 or other 720p source works fine while the TV never initializes with the Hi-Def NES set to 720p. One is my Sony XBR Trinitron KV-32HS510. I'll also be testing on my KV-30XBR910 in a few hours. Kevtris used an HDMI analyzer that was supposed to ensure that the kit outputs standard signals for maximum compatibility with any TV. I suspect he paid a bit more attention to 1080p than 720p and compatibility may have suffered as a result.

Though some are pre-2005, these TVs are probably the absolute best to pair with a Hi-Def NES if 720p works. Unfortunately, the 510 seems to lock 16:9 aspect using vertical compression for 720p signals anyway (no zoom/stretch functions), so it's not the ideal TV. That wouldn't be an issue on the 16:9 sets and some of the other 4:3 sets in the series.

I've had my 910 since I paid $2,300 in 2003 and it still blows people away. There are actually people who periodically remind me to call if I ever decide to get rid of it.   The 510 was an $8 Goodwill find from several months ago and it still looks (and SOUNDS) AMAZING. If you ever see one of these TVs for a reasonable price, BUY IT! For classic gaming I made a list of the most desireable 4:3 models:
Sony KV-40XBR800
Sony KV-36XBR800
Sony KD-36XS955
Sony KV-36HS510
Sony KV-32XBR800
Sony KV-32HS510

Speaking of glitchy stuff, how about a Classic Controller Pro glitch performed on the XBR910 (though an old iPhone 4 recording doesn't do it justice):



 

I did some testing with Mario as well but using intro loops.

I tried comparing my NESRGB NES to the AVS and the NESRGB was ahead by quite a margin by the 10 minute mark.

I then tested 2 more of my unmodified NES' against it with the same result.

Currently looping an unmodified NES Vs. the NESRGB and it seems to be 1 to 1 or so close together that it will take hours to desync.

Hopefully this is something that can be fixed via firmware update.

I also found a couple glitched games on the 1.20b5 AVS using an Everdrive at the moment:
Gradius 2 - Can only see the bottom bar + top 1/4 of the screen, Intro also has a messed up ship image on the opening scene.
Gremlins 2 Japanese Version - A couple of the end video scenes are pretty glitched up.

Looks like after nearly 8 hours my original NES was .5-1 second ahead of my NESRGB modified NES.

That sounds like any difference is caused by the timing Xtal. These things are manufactured to within 5 digits of accuracy. A one second drift in 8 hours is 8 * 3600 or 28800 +/-1, within five digit territory.

The AVS locks the ouput to 59.94Hz over HDMI compared to 60.1 or whatever the NES does. Since it is slowing down the console ever so slightly, likely halting the PPU very briefly during Vsync to give the NES that extra half scanline worth of scan it lacks being 240p (262 scans) native versus standard 480i (262+263 scans). I'm not sure if the AVS halts the CPU, scales the timing slightly, or lets it run for a few extra cycles at the end of each frame.

Regardless, human players are not going to notice a .1%-.2% slowdown, but an HDTV might. We don't have quartz timing crystals embedded in our brains after all.

I fail to understand if the matter being debated is:
1. fidelity to the original gaming experience; or
2. legitimacy of score records.

Point 1. in my opinion is not even debatable: you have a closed bag with 600 pennies in one hand, and another closed bag with 599 pennies is the other hand: can you tell which one weights more? If you can, it is because you guessed, not because you knew by evaluating their weight in your hands. Same with 59.94Hz compared to 60.0?Hz. The frequency difference is on a range of about 0.1Hz on a nominal frequency of 60Hz, so that's about 0.16% (or like Star says above: 0.1% to 0.2%). I think that even 10x these values, I mean a difference of 1% or 2% in speed, would be hard to detect just playing the game (I mean, without comparing real time with another screen where the game runs 1%-2% faster or slower).

Point 2. is * maybe * arguable, if considering that the reaction time (from when an hint is on screen - or a sound is performed - to the deadline for a player to react) is defined by some specific arbitrary number of frames: if such frames are 0.1% 0.2% slower, a player has a slightly longer time to react, * maybe * ? Can this make a difference in real life gaming? I really can't tell.

Any further explanation on what exactly is being debated here is of course welcome.  

It is not a debate at all.

Merely pointing out factual differences between the original system and the AVS in the hopes that it can become an even better system that is as close to the true NES experience as possible.

As for your examples though a highly skilled speed runner could easily tell you the difference between those 2 bags and tell that the reaction times required are different.

60.1Hz, 1 frame = 0.0166389s
59.4Hz, 1 frame = 0.0168350s

Let's assume, for instance, that in a game you have 10 frames to react to something.

60.1Hz, 10 frame = 0.166389s
59.4Hz, 10 frame = 0.168350s

Difference in reaction time: if my math is correct, about 0.002s. No one can be aware of that.

Then, if your argument is: "but you do have these 2/1000 of a second bonus", sure you do. But performing a blind test, not knowing if the game is running 60.1Hz or 59.4Hz, I'm ready to bet that no one can tell just playing it, at which of these two possible speeds the game is running. You can guess, and be lucky for once, but statistically, on big numbers of tests, you will have about 50% of correct calls. No matter who you are.

Two extra milliseconds in reaction time is an advantage? Perhaps yes.
It is something that you will detect and be aware of playing? Definitely not.

I don't wish to antagonize, I'm just sharing some opinions based on common sense and experience.

So, just my random two cents on the matter!  

Edit: and of course, even extending the planned reaction time to, for instance, 20 frames, and consequentially 4 milliseconds extra running at 59.4Hz, the concept won't change: you can claim to have an advantage at 59.4Hz maybe, but, blindly testing, you won't be able to detect, just playing the game, if you are having such advantage or not.

Some of the NES speed run tricks are frame perfect.

I recognize to be ignorant on the matter, I know little about it: I speak from a logic perspective.  

Unless I completely misunderstand the matter here, I think that the number of frames displayed is identical.
What changes is the rendering time interval (+0.1%/+0.2%) between frames (from a frame to the next):
60.1Hz, 1 frame = 0.0166389s
59.4Hz, 1 frame = 0.0168350s
So, I think that two identical speed run still identical frame number wise: at 59.94Hz is a tiny, slightly, humanly undetectable longer "real time wise", this since the refresh speed from a frame to the next was 0.0002s (0.2ms!) slower. It is not like some frames are skipped or dubled. Or it is? And however, the gaming experience itself, is reasonably identical, in any case.

I believe that is true but why not make them exactly the same if it is possible?

Also the 10 minute loop of mario's intro on the NESRGB looked like it was around 2 seconds ahead already so the difference may be more like .33% on the current beta firmware (1.20b5).

NES runs at 21.477272MHz (oscillator) / 4 (clock divider) / 89341.5 (pixels per frame) = 60.0988Hz = 16.639ms
AVS runs at 74.25MHz (oscillator) / 1650x750 (pixels per frame) = 60.0000Hz = 16.666ms

So a ~0.02ms difference per frame. Top Tetris players start noticing gamepad lag around 6-8ms, so that is FAR beyond anyone's reaction times. Absolutely nobody will notice even on frame perfect moves. To get "accurate" speed run numbers it just needs that fixed scaling factor to convert time into frames. If you are counting frames on a video then you already need to scale depending the video capture device rate. You could be playing at 60.1Hz, recording at 59.94Hz, then playback at 60.0Hz. If you are hand timing with a stop watch then you need to add in the 0.1-0.3s for the human like on the track. If you are playing on an emulator then who knows what the actual rate is.

Take a longer world record, like Zelda at 28:50. 28:50 = 1730s = 1730000ms = 103972.5 NES frames. That number of frames would take 28:52.8 on the AVS. Around 3 seconds difference, but the record is already throwing away 1 second of precision by only specifying whole seconds. Looks like the record is coming down 30-60s at a time, so 2s is insignificant for now.

Maybe it makes a difference with a shorter game, like SMB, where fractional seconds actually matter. 4:57.260 = 297260ms = 17865.25 NES frames. AVS time would be 297742ms = 4:57.742, so it does need the scaling factor if you are in the top 20 of all time. Beyond that the scores are already not precise.

With around 1s extra per 10 mins, if your record is under 10 mins and you aren't keeping fractional seconds then your accuracy is already off by as much as the AVS is. For records in the 5 min range a fixed multiplier will get the wall clock time with no benefit to the player. For anything in the 20+ minute range the records aren't likely a few seconds difference anyways.

Hi there BunnyBoy,

Thanks for the info.

Is it possible to adjust that rate and are there other factors that may be coming into play such as how the AVS handles specific games etc...?

Easy to change, but whether it is useful depends on your TV. My Hannspree will accept anything including horribly off spec signals. My first Samsung will barf if the video is 1 pixel per frame off. No idea if capture devices would accept it.

I would be happy to test it for you on a few capture devices I have here if you have time to give it a shot.

There is quite a large speed running community on Twitch so it may be worth the time if it doesn't take too long to implement.

Also not sure you saw the message about the glitches I came across there but here they are again:
Using an Everdrive -
Gradius 2: The top 1/4 of the ship on the intro screen is offset, Only the top 1/4 of the screen is visible when the game is started.
Gremlins 2: The ending scenes have some strange glitches.

These issues are not inherent to the Everdrive since I tested it on my regular NES and they were fine.

It's actually 59.94, so the margin of difference is even smaller than your calculation suggests.

60.1 / 59.94 = 1.00267. That's barely a .2% difference.

599 pennies in a bag is a good analogy. In fact even if you weighed the pennies, you have to consider wear. Heavy wear = less weight. Also if there are older pure copper pennies in the bag as opposed to copper plated zinc, this difference would reflect as a minute weight discrepancy, so even weighing the pennies on a scale may be within the margin of error and not sufficient to detect the missing penny.

EDIT: Bunny confirmed the AVS is locked to 60Hz, not 59.94 (or did he round it off? seriously who cares about a .1% difference...   ) so it's even closer to a real NES than NTSC broadcast spec.

Yes, my math was off by a 10x factor (* against my own point *), and the starting data was uncorrect too. My bad, my apologies about that. Everyone, please ignore my post that Star quoted, bunnyboy pointed out that the correct math is 0.02 milliseconds per frame.

Also, if I knew that bunnyboy was going to post few hours later, I would just shut up. I was mostly trying to understand (out of personal curiosity) what people was really concerned about, and trying to explain my logic, I ended up sounding like a teacher. Bad position, when you have little specific knowledge on the matter. Still, my conceptual point (which was: speaking on reasonable terms, the gaming experience is identical), given the exact values, is even more correct and accurate.

Again, all, my apologies for the uncorrect data which I provided.  
 

Actually, given that the AVS (or Hi Def NES Mod) runs .16% slower than the NES, it may be at a disadvantage to individuals looking to break records with speed runs. Say the time it takes to beat game X on the real NES is 1 hour, 3,600 seconds. On the AVS, all other things being equal, the same game would take 3,605 seconds to beat. The guy with the AVS must play a faster game by 6 seconds to beat the guy with the NES. When competition is really fierce, those extra seconds may mean the difference between breaking a record and being an also-ran. If a game can be completed under ten and a half minutes, then it will not matter, but every 10.6 minutes adds an extra second to the AVS clock.

No need to appologise user. I agree though the difference is miniscule enough to not matter. People don't realise how small .2% really is.

@Great Hierophant: Funny you bring up world records. Gameboy is off by a few cycles every frame to sync with Game Boy Player and Super Game Boy, yet SpeedDemosArchive requires you to use the Game Boy Player for capture instead of original hardware. Not sure what the rules are for Twin Galaxies.

The timing difference playing Game Boy on a Game Boy Player or Super Game Boy is comparable to the AVS versus NES. So if AVS were real Nintendo hardware, the scorekeeping and speed running authorities would have accepted it. But it's not (except through scoreboard) so nobody throw out your toasters. Tube TV and VCR or DVD recorder is still the best way to record game footage for verification purposes.  

Another thing that seems when talking with lag / slowdown - noticed yesterday that PAL video setting seem to lag a lot more, for example Mr. Gimmick original cart has huge lag and slowdown. Music and game runs on correct speed, but controller input lag is almost half a second (response seem sluggish when moving and jumping etc). Don't know why this happens, and it only happens with PAL settings. Gimmick! on NTSC settings works like a charm. (And PAL Mr. Gimmick works fine on PAL NES I have). I have reported it to bunnyboy though, hoping to get that sorted out as well in the upcoming firmware updates.

I noticed this too! I'm not familiar with PAL consoles so I didn't know if it was just a botched port of Gimmick, or what. But it does make sense that this is not normal.

I also noticed the controller outright dropping inputs when I was using the Everdrive in PAL mode

SpeedDemosArchive allows you to use a Super Game Boy, a Super Game Boy 2 or a Game Boy Player to record Game Boy games.

The frame (refresh) rate of the Game Boy, Super Game Boy 2, Game Boy Color and Game Boy Advance is 59.7275Hz and the Super Game Boy is 61.1679Hz.  The Super Game Boy thus is 2.4% faster than the real Game Boy and SpeedDemosArchive multiplies your time by that percentage to come up with a true time.  For regular Game Boy and Game Boy-compatible Color games, the Super Game Boy 2 is the best bet for competitive record setting.

While SpeedDemosArchive says "you're going to have to use" the Super Game Boy, Super Game Boy 2 or Game Boy Player for a Game Boy capture, the Game Boy Player is the "suggested" method to use for a Game Boy Advance capture.  There are other devices that can play Game Boy Advance games on a TV.  

The Game Boy Player is rather interesting.  It uses the same crystal as the Game Boy Advance and thus outputs at the same refresh rate, 59.7275.  However, the Gamecube outputs video from the Game Boy Player in 480i or 480p at 59.94Hz, the NTSC standard.  With the dropped frames, the added latency required to upscale the graphics and the 480i flicker or 480p blur, you will experience a less-than-ideal image for record setting.  

However, there is unofficial software called the Game Boy Interface for the Game Boy Player.  This has an Ultra Low Latency version that will output in a proper 240p at a 59.7276Hz, almost identical to the handhelds' rate. SpeedDemosArchive does not specifically prohibit or allow this.