EXAーARCADIA [New Arcade Platform/Publisher]
Re: EXAーARCADIA [New Arcade Platform/Publisher]
At least today I learned what an arcade game is.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
I learned that MrJBRPG is an EXA fanboy.
<Aquas> EDMOND DROPPED OUT OF HIGH SCHOOL TO SMOKE COPIOUS AMOUNTS OF OPIUM
<Zeether> shoe failed college again <croikle> credit feed
<Zeether> shoe failed college again <croikle> credit feed
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Sumez wrote:At least today I learned what an arcade game is.
光あふれる 未来もとめて, whoa~oh ♫
[THE MIRAGE OF MIND] Metal Black ST [THE JUSTICE MASSACRE] Gun.Smoke ST [STAB & STOMP]
Re: EXAーARCADIA [New Arcade Platform/Publisher]
It is currently running at 2f lag all-together on exA, the same as the original PCB and 2f less than the 360 port.Bananamatic wrote:cv1k port to a platform with 3 frames of default lag, good luck I guess
@trap0xf | daifukkat.su/blog | scores | FIRE LANCER
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
wait! are they porting the cv1k version of the game on this thing? why the fuck
Re: EXAーARCADIA [New Arcade Platform/Publisher]
That's an odd conclusion to come to.
@trap0xf | daifukkat.su/blog | scores | FIRE LANCER
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
-
Bananamatic
- Posts: 3530
- Joined: Fri Jun 25, 2010 12:21 pm
Re: EXAーARCADIA [New Arcade Platform/Publisher]
I'll believe you once you explain the lag reported in A&B on a CRT (or worse, in the input test mode)
Re: EXAーARCADIA [New Arcade Platform/Publisher]
A&B's lag comes from Unity primarily I believe, and I think is something like 4f in total... a bit unfortunate, but seems like an engine issue. That said, I haven't done any actual investigation into A&B, so I don't know any specifics really. I'm not sure why the input test is so fucked, maybe a lack of care since it's just for testing that controls work, but it is much worse than in-game for sure.
Infinos EXA is also 2f, and is out now, so you could check me by checking that
Infinos EXA is also 2f, and is out now, so you could check me by checking that
@trap0xf | daifukkat.su/blog | scores | FIRE LANCER
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
-
CStarFlare
- Posts: 3003
- Joined: Tue Feb 19, 2008 4:41 am
Re: EXAーARCADIA [New Arcade Platform/Publisher]
That's very cool. Thanks for confirming.trap15 wrote:It is currently running at 2f lag all-together on exA, the same as the original PCB and 2f less than the 360 port.Bananamatic wrote:cv1k port to a platform with 3 frames of default lag, good luck I guess
-
schleichfahrt
- Posts: 141
- Joined: Sun Jun 12, 2016 12:50 pm
Re: EXAーARCADIA [New Arcade Platform/Publisher]
I have no knowledge of A&B's code, but it is very easy to produce extra lag when using unity.trap15 wrote:A&B's lag comes from Unity primarily I believe
I fell into that particular trap a little while ago when making something non-game related using unity, after it became a problem a rather simple fix cut the latency in half.
When you ruin some enemy, add to score points.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Okay...trap15 wrote:A&B's lag comes from Unity primarily I believe, and I think is something like 4f in total... a bit unfortunate, but seems like an engine issue. That said, I haven't done any actual investigation into A&B, so I don't know any specifics really. I'm not sure why the input test is so fucked, maybe a lack of care since it's just for testing that controls work, but it is much worse than in-game for sure.
Infinos EXA is also 2f, and is out now, so you could check me by checking that
I am intrigued by the conversation on the input lag discussion, and I believe that input lag is defined on processing the input and game logic before displaying changes on the screen.
We know that Exa Arcadia, through the CEO / Founder's own words, that maximum input lag limit is 4 frames and minimum recorded is 2 frames.
To verify the input lag measurement, is it the following:
1. JVS IO board and game software logic to the motherboard unit, with display monitor excluded
2. combined total of JVS IO board, game logic, and rendering output to the monitor from the motherboard
3. game software logic only to the motherboard, which the IO board and the monitor are excluded, which does explain additional input lag from external devices
It's great to know so developers can be more hardcore on optimizing towards lower input lag.
Last edited by MrJBRPG on Fri Mar 20, 2020 12:51 pm, edited 1 time in total.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
A game can have as much input lag as it wants, technically. I would hope developers care about their input lag situation, but it's up to them in the end. Minimum could potentially be 1 frame with some trickery, I think 2 or 3 is a respectable amount.
I said all-together, so I mean comparing switch activation with on-screen action.
I said all-together, so I mean comparing switch activation with on-screen action.
@trap0xf | daifukkat.su/blog | scores | FIRE LANCER
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
<S.Yagawa> I like the challenge of "doing the impossible" with older hardware, and pushing it as far as it can go.
-
mikejmoffitt
- Posts: 629
- Joined: Fri Jan 08, 2016 7:26 am
- Location: Tokyo, Japan
Re: EXAーARCADIA [New Arcade Platform/Publisher]
I spent basically all of my time after work today doing some massive changes to the video backend for Gimmick in order to bring lag down to 2f, as I was dissatisfied with the average 3-4f from the loctest.
By adding a frame-by-frame advance feature, I was able to validate that I wasn't erroneously introducing buffer stages between what the game logic perceives as input, and the act of running the logic. That is to say, if I put it in frame advance mode, hold jump, then advance one frame, Yumetaro jumps on the very next rendered frame. This gives me confidence that the game logic is not a problem. Runahead emulation does not apply in this case, anyway, so that wouldn't fix anything if there was game logic lag here.
What remains is the double-ended challenge of the input systems that feed into the game logic, and the video system that follows it.
For the input, JVS takes time to poll. It takes a lot of time for some IOs. Even better, some IO boards will take their sweet time to respond to your request, and then give you old data! The Sega "Type B" IO with the built-in JAMMA is pretty awful in this regard. The Capcom one is a bit better, but it double-buffers internally still. So far, the best has been the Sega "Type 3" Lindberg IO, and some of the Taito Vewlix ones. Two variables at play, out of our control: 1) how long the IO takes to give us data, and 2) how often, and how deeply buffered, the IO's internal updating process runs.
For the output, there are some enemies presented by many modern graphics APIs. It may be difficult to mitigate these the further you are away from the actual underlying graphics calls and drivers; working with OpenGL calls directly is a very close way that gives you a lot of control, while working with a prebaked engine (Game Maker Studio, Unity, etc) may take you a lot further away from it - or at least hide a lot of the nice knobs.
In 2020, sometimes it is necessary to jump through some absurd hoops to get the input lag under control. Everything modern wants to buffer everything for you in queues deeper than you'd like, and wants you to do everything asynchronously! If you want to be in precise control of latency, it is sometimes necessary to defeat these mechanisms of convenience.
I don't want to go into detail with it, but I rewrote Gimmick's video today to defeat some problems with a library I was using, and switched to another wholesale.
--
When measuring JAMMA boards, we can usually remove concerns about the input complexity. Most often, it's just a raw uint8 or something similar mapped to a memory address. Your delay is just gate propagation, not worth mentioning (some more complex boards like STV have a microcontroller doing the input reading, and sometimes it can fuck it up).
With JVS... it's difficult to get a clear measurement without closer introspection. You can analyze the bus to see how long it's taking doing TX and RX (and time between), but it still doesn't tell you if the data is out of date at a glance.
--
As of today, this is how I've measured my results in improving Gimmick's latency:
There's a useful free application called Is it Snappy, which lets you scrub through a 240fps recording from a phone camera and mark Input and Output frames. It will then tell you the difference.
I have an LED hooked up to Button 2 (Jump, in this game). When the button is pressed, it illuminates (it is buffered by a 74LS04 inverter to achieve this).
I have the phone in portrait, so that the rolling shutter of the camera does not cause a substantial discrepancy between Yumetaro in the middle and the LED on the bottom. Otherwise, we might get an additional 1/240th of a second error.
This first screenshot is the frame before[i/] the button is pressed. Here, a frame is 1/240th of a second; four camera frames (4 x 4.16ms) is one game frame (16.67ms).
This second screenshot shows the LED illuminated. The button has been pressed. The raster is about 1/2 completed; the game requests inputs from JVS at the start of vblank. This is our Input frame.
This third screenshot is the frame right before we have positive confirmation that Yumetaro has jumped.
This last screenshot shows that Yumetaro has jumped. This is our Output frame.
The difference between the Output and Input frames show just about two frames (33.3ms).
By adding a frame-by-frame advance feature, I was able to validate that I wasn't erroneously introducing buffer stages between what the game logic perceives as input, and the act of running the logic. That is to say, if I put it in frame advance mode, hold jump, then advance one frame, Yumetaro jumps on the very next rendered frame. This gives me confidence that the game logic is not a problem. Runahead emulation does not apply in this case, anyway, so that wouldn't fix anything if there was game logic lag here.
What remains is the double-ended challenge of the input systems that feed into the game logic, and the video system that follows it.
For the input, JVS takes time to poll. It takes a lot of time for some IOs. Even better, some IO boards will take their sweet time to respond to your request, and then give you old data! The Sega "Type B" IO with the built-in JAMMA is pretty awful in this regard. The Capcom one is a bit better, but it double-buffers internally still. So far, the best has been the Sega "Type 3" Lindberg IO, and some of the Taito Vewlix ones. Two variables at play, out of our control: 1) how long the IO takes to give us data, and 2) how often, and how deeply buffered, the IO's internal updating process runs.
For the output, there are some enemies presented by many modern graphics APIs. It may be difficult to mitigate these the further you are away from the actual underlying graphics calls and drivers; working with OpenGL calls directly is a very close way that gives you a lot of control, while working with a prebaked engine (Game Maker Studio, Unity, etc) may take you a lot further away from it - or at least hide a lot of the nice knobs.
In 2020, sometimes it is necessary to jump through some absurd hoops to get the input lag under control. Everything modern wants to buffer everything for you in queues deeper than you'd like, and wants you to do everything asynchronously! If you want to be in precise control of latency, it is sometimes necessary to defeat these mechanisms of convenience.
I don't want to go into detail with it, but I rewrote Gimmick's video today to defeat some problems with a library I was using, and switched to another wholesale.
--
When measuring JAMMA boards, we can usually remove concerns about the input complexity. Most often, it's just a raw uint8 or something similar mapped to a memory address. Your delay is just gate propagation, not worth mentioning (some more complex boards like STV have a microcontroller doing the input reading, and sometimes it can fuck it up).
With JVS... it's difficult to get a clear measurement without closer introspection. You can analyze the bus to see how long it's taking doing TX and RX (and time between), but it still doesn't tell you if the data is out of date at a glance.
--
As of today, this is how I've measured my results in improving Gimmick's latency:
There's a useful free application called Is it Snappy, which lets you scrub through a 240fps recording from a phone camera and mark Input and Output frames. It will then tell you the difference.
I have an LED hooked up to Button 2 (Jump, in this game). When the button is pressed, it illuminates (it is buffered by a 74LS04 inverter to achieve this).
I have the phone in portrait, so that the rolling shutter of the camera does not cause a substantial discrepancy between Yumetaro in the middle and the LED on the bottom. Otherwise, we might get an additional 1/240th of a second error.
This first screenshot is the frame before[i/] the button is pressed. Here, a frame is 1/240th of a second; four camera frames (4 x 4.16ms) is one game frame (16.67ms).
This second screenshot shows the LED illuminated. The button has been pressed. The raster is about 1/2 completed; the game requests inputs from JVS at the start of vblank. This is our Input frame.
This third screenshot is the frame right before we have positive confirmation that Yumetaro has jumped.
This last screenshot shows that Yumetaro has jumped. This is our Output frame.
The difference between the Output and Input frames show just about two frames (33.3ms).
-
StarCreator
- Posts: 1941
- Joined: Mon Jan 12, 2009 2:44 am
- Location: Maryland, USA
- Contact:
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Hm. I'm using the Sega rev B boards myself. If they're as bad as you say here, I'm going to have to explore alternatives.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Will the ghost of Takakura Ken break into your house and chop off your hog if your game goes over 4f? :OMrJBRPG wrote:We know that Exa Arcadia, through the CEO / Founder's own words, that maximum input lag limit if 4 frames
光あふれる 未来もとめて, whoa~oh ♫
[THE MIRAGE OF MIND] Metal Black ST [THE JUSTICE MASSACRE] Gun.Smoke ST [STAB & STOMP]
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Discovered latest update from developer behind Vritra Hexa, with new stage and boss in progress (used translate machine) thanks to quick sketch of new boss.
https://twitter.com/saddy575/status/1240705200500830208
https://twitter.com/saddy575/status/1240705200500830208
Re: EXAーARCADIA [New Arcade Platform/Publisher]
An auspicious development - that turtle looks ready to bite an incautious handler's hog clean off :O
光あふれる 未来もとめて, whoa~oh ♫
[THE MIRAGE OF MIND] Metal Black ST [THE JUSTICE MASSACRE] Gun.Smoke ST [STAB & STOMP]
Re: EXAーARCADIA [New Arcade Platform/Publisher]
mikejmoffitt: thank you, great work! When I finally get the chance to play this, I'll be thinking about these after-work hours of yours. I mean it.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
https://exa.ac/en/2020/03/31/exa-arcadi ... 177369a3b7
I did suspect that Exa Arcadia had to cancel physical showings out of safety, and the announcement really helps. Glad that Exa Arcadia is more communicative.
I did suspect that Exa Arcadia had to cancel physical showings out of safety, and the announcement really helps. Glad that Exa Arcadia is more communicative.
-
cave hermit
- Posts: 1544
- Joined: Sat Sep 07, 2013 2:46 pm
- Location: Pennsylvania
Re: EXAーARCADIA [New Arcade Platform/Publisher]
So, are those cartridges literal usb flash drives or what?
Also I can't see this catching on anywhere outside of Asia.
Maybe Round 1 USA will have it? I doubt it since it seems like the bulk of their games are surplus units the Japanese branch didn't need for whatever reason, usually due to them being obsolete (which is why they're still on vanilla Tekken 7)
I can't see a lot of the arcades you see nowadays that have redemption games and Raw Thrills lightgun/driving games as their bread and butter wanting to buy this. Look at it through the eyes of a boardwalk arcade owner: Why would you pay thousands of dollars for these weird galaga clones with constant japanese girl chattering that random tourists will just look at confused and walk away from when you could buy another stacker machine and rack in the dough from suckers?
Also I can't see this catching on anywhere outside of Asia.
Maybe Round 1 USA will have it? I doubt it since it seems like the bulk of their games are surplus units the Japanese branch didn't need for whatever reason, usually due to them being obsolete (which is why they're still on vanilla Tekken 7)
I can't see a lot of the arcades you see nowadays that have redemption games and Raw Thrills lightgun/driving games as their bread and butter wanting to buy this. Look at it through the eyes of a boardwalk arcade owner: Why would you pay thousands of dollars for these weird galaga clones with constant japanese girl chattering that random tourists will just look at confused and walk away from when you could buy another stacker machine and rack in the dough from suckers?
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Oh god! I just realised, you won't be able to do this in an arcade.cave hermit wrote:constant japanese girl chattering
-
pablumatic
- Posts: 307
- Joined: Tue Jun 03, 2008 6:39 pm
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Looks like its either proprietary or an SSD drive in a plastic shell.cave hermit wrote:So, are those cartridges literal usb flash drives or what?
https://youtu.be/6XjaJAmWsa0?t=413
-
brokenhalo
- Posts: 1393
- Joined: Wed Mar 26, 2008 4:11 am
- Location: philly suburbs
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Looks similar to an msata drive connection, but with a few less pins than standard. So probably a small ssd with slightly modified connector.pablumatic wrote:Looks like its either proprietary or an SSD drive in a plastic shell.cave hermit wrote:So, are those cartridges literal usb flash drives or what?
https://youtu.be/6XjaJAmWsa0?t=413
Re: EXAーARCADIA [New Arcade Platform/Publisher]
I wonder if any of you played NOISZ before on Steam? If not, then it might be good to rev up those skills; because an enhanced version, with new stages, is in development.
https://exa.ac/en/2020/04/10/noisz-arc% ... nouncement
I can remember playing one other rhythm color match music game, Magical Beat, with standard arcade controls, but that was on the Taito NesicaxLive when played at Round 1 Southland Mall.
https://exa.ac/en/2020/04/10/noisz-arc% ... nouncement
I can remember playing one other rhythm color match music game, Magical Beat, with standard arcade controls, but that was on the Taito NesicaxLive when played at Round 1 Southland Mall.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Another Steam game? Let me know when Border Down 2 shows up.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Someone please play Infinos EXA if they get a chance, god damn. Buy it yourself or something, I dunno. Poor game deserves better.
Rage Pro, Rage Fury, Rage MAXX!
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Sure, just let me empty my entire bank accountDespatche wrote:Someone please play Infinos EXA if they get a chance, god damn. Buy it yourself or something, I dunno. Poor game deserves better.
<Aquas> EDMOND DROPPED OUT OF HIGH SCHOOL TO SMOKE COPIOUS AMOUNTS OF OPIUM
<Zeether> shoe failed college again <croikle> credit feed
<Zeether> shoe failed college again <croikle> credit feed
Re: EXAーARCADIA [New Arcade Platform/Publisher]
Or you can get Infinos Gaiden on steam for around 10 Euros. Sure the EXA version will have some exclusive content.Zeether wrote:Sure, just let me empty my entire bank accountDespatche wrote:Someone please play Infinos EXA if they get a chance, god damn. Buy it yourself or something, I dunno. Poor game deserves better.
But if you want to play the game and don't wanna empty your bank account just get the steam version.
Re: EXAーARCADIA [New Arcade Platform/Publisher]
I have it on location, but since the world's shut down at the moment, it sits and waits. Casual people were pretty big into it actually.Despatche wrote:Someone please play Infinos EXA if they get a chance, god damn. Buy it yourself or something, I dunno. Poor game deserves better.
-
mikejmoffitt
- Posts: 629
- Joined: Fri Jan 08, 2016 7:26 am
- Location: Tokyo, Japan
Re: EXAーARCADIA [New Arcade Platform/Publisher]
The exA version has 2f total input lag thanks to trap15's efforts, while the steam version will have a frame or two more than that.z0mbie90 wrote:Or you can get Infinos Gaiden on steam for around 10 Euros. Sure the EXA version will have some exclusive content.Zeether wrote:Sure, just let me empty my entire bank accountDespatche wrote:Someone please play Infinos EXA if they get a chance, god damn. Buy it yourself or something, I dunno. Poor game deserves better.
But if you want to play the game and don't wanna empty your bank account just get the steam version.