What's new
By:

CPS A-Board Differences

I thought the speed of the crystal oscillator / CPU dictated the processing power not the actual gameplay speed? SFII Champion Edition and Hyper Fighting require the faster processor to reduce slowdown not so that the gameplay is faster.

I’ve heard that Champion Edition can show slowdown when on the 10 MHz crystal & CPU leading some players to exploit and perform combos that wouldn’t be possible on the 12 MHz boards.
 
kikaso said:
I thought the speed of the crystal oscillator / CPU dictated the processing power not the actual gameplay speed? SFII Champion Edition and Hyper Fighting require the faster processor to reduce slowdown not so that the gameplay is faster.

I’ve heard that Champion Edition can show slowdown when on the 10 MHz crystal & CPU leading some players to exploit and perform combos that wouldn’t be possible on the 12 MHz boards.
Click to expand...
But overal gameplay just feels faster/smoother when there is no slowdown when you pull a combo or when multiple enemies appear on screen simultaneously.
 
YesAffinity said:
The compatibility and any differencs is one of the central discussions here. All games will run on either speed of crystal/processor. What the differences may be when a particular game is played at one speed versus the other has been extremely hard to concretely document, outside of some subjective assertions of how the gameplay "feels".

I'm not disputing people's assertions. I myself will assert that, having spent A LOT of time on sf2 hyper fighting, that it definitely plays differently on a 10mhz crystal than a 12mhz crystal. But again, concretely documenting and replicating the differences has been extremely challenging.
Click to expand...
I think for there to be credibility to it, there needs to be double blind testing done with a sample of as many people as possible.

I haven't played the games on the wrong speed processor enough to say I would be able to distinguish which is which. It seems I would just be guessing if it feels like what I'm used to or not.
 
KaPH33n said:
I think for there to be credibility to it, there needs to be double blind testing done with a sample of as many people as possible.

I haven't played the games on the wrong speed processor enough to say I would be able to distinguish which is which. It seems I would just be guessing if it feels like what I'm used to or not.
Click to expand...
Heres something that might provide some quantifiable testing. When 1cc'ing sf2hf, you get the credit roll sequence at the end. Pressing any button during this sequence causes it to run very fast (I assume unthrottled). The game will then return to normal speed when the sequence ends and it transitions back to attract mode.

Might be worth checking this credit roll sequence at max speed on a 10mhz vs 12 mhz. If both run equally fast, and I suspect they would, then I think it can be safely said that no cps1 sf runs any different on 10mhz vs 12mhz. That's strictly audio and graphics of course.

Somebody/anybody feel free to poke holes in my theory.
 
Nimmers said:
I just don't think SF2 does enough to work the CPU to the max very often. If there's a tell, I reckon it will be in 3 player Captain Commando or Cadillacs & Dinosaurs.
Click to expand...
I believe this is absolutely correct.

I tested the unthrottled credit roll sequence, on both 'A' boards. Great results! Also adding to the extremely interesting results is the winning podium sequence just before the credit roll sequence actually runs faster on a 12mhz board, as well.

When playing side-by-side from the start of the winner's podium sequence, the winner's podium sequence ends 49 frames sooner on a 12mhz 'A' board, and the credit roll sequence ends 314 frames sooner. Subtract the 49 frames from the 314, and the credit roll sequence effectively runs 265 frames faster on a 12 mhz 'A' board than it does on a 10 mhz 'A' board.

If anybody knows a good free piece of software that will provide the frame count while the video is playing, and has the ability to step through the video frame by frame, I can provide more data on the total frames of each sequence, and possibly that will give us some insight into just how much the speed difference equates to.

View: https://youtu.be/MkP1xmcEBxc
 
YesAffinity said:
If anybody knows a good free piece of software that will provide the frame count while the video is playing, and has the ability to step through the video frame by frame, I can provide more data on the total frames of each sequence, and possibly that will give us some insight into just how much the speed difference equates to.
Click to expand...
I have had good success in the past using free tools to do this type of thing... just off the top of my head I believe you can use vlc to output every frame of the video into a seperate jpg or png. then use imagemagick or something to draw a frame number on it... then ffmpeg or something to combine the frames back into a video again. makes sense to do it all with a script so that you can replicate it with any video.
 
YesAffinity said:
I tested the unthrottled credit roll sequence, on both 'A' boards.
Click to expand...
Credit sequence, attract mode, etc. are unaffected by the CPU speed. Difference is in game and has nothing to do with CPU load, difference behing exactly 20%, even in 2P mode with idle characters.
YesAffinity said:
When playing side-by-side from the start of the winner's podium sequence, the winner's podium sequence ends 49 frames sooner on a 12mhz 'A' board, and the credit roll sequence ends 314 frames sooner. Subtract the 49 frames from the 314, and the credit roll sequence effectively runs 265 frames faster on a 12 mhz 'A' board than it does on a 10 mhz 'A' board.
Click to expand...
It's because everytime you have a black transition screen, the 12MHz board loads RAMs faster.
It's also visible in attract, the 12MHz board will slowly desync every time there's a black screen, but animations in the attract play at the same speed.
 
The game loop is often* synchronised to vblank, so if all work gets done in the 16ms it takes for 60Hz frame then you don't see a difference between 10Mhz and 12Mhz. If the work takes just over 16ms you may drop a frame in 10Mhz, and not drop a frame at 12Mhz.

* more complicated games may spread out work like pathfinding over multiple frames - I don't think many arcade games fall into that class, though. ;)
 
Fluffy said:
The game loop is often* synchronised to vblank, so if all work gets done in the 16ms it takes for 60Hz frame then you don't see a difference between 10Mhz and 12Mhz. If the work takes just over 16ms you may drop a frame in 10Mhz, and not drop a frame at 12Mhz.

* more complicated games may spread out work like pathfinding over multiple frames - I don't think many arcade games fall into that class, though. ;)
Click to expand...
What you say is correct, and me too thought game logic cycle was always tied to vblank. I've been proved wrong with CPS1.
I really need to make a new timing video.
 
Maybe some things are tied to timers that are driven by the CPU clock? I never found much use for them, because there are usually more things that need to be timed, and only one or two hardware timers.
Do fighting games take sub-frame button inputs?
 
Fluffy said:
Maybe some things are tied to timers that are driven by the CPU clock? I never found much use for them, because there are usually more things that need to be timed, and only one or two hardware timers.
Do fighting games take sub-frame button inputs?
Click to expand...
can't speak for all of them, but I am reasonably sure I remember SSF2T being 60 input checks per second.
 
Hackcell said:
My SF2HF comes with a 10 mhz CPU and xtal. Does anybody have one (original) with a 12mhz cpu/xtal?
Click to expand...
My Champion Edition came with a 10 MHz too. I had to track a CPS Dash (12 MHz) board down. It seems as it is as simple as swapping the 10 MHz crystal for a 12 MHz—which essentially over locks the 10 MHz CPU. Folks on this forum have made the mod with good results.
 
@Nimmers - answering your question from the Australian roll call thread, which answer seems more appropriate here.

Just captured sf2hf running on a 10mhz A board and a 12mhz A board, just simply letting the timer run with 1P and 2P idle. From the first visible frame of the round (following the transition from the character portrait screen and a few blank screens) to the last frame that the two characters are facing each other (the next frame that follows, they both turn and enter their "draw game" stance):
10mhz board - the round completes in about 60s
12mhz board - the round completes in about 55s

It's about a 8.5% speed difference.
 
YesAffinity said:
@Nimmers - answering your question from the Australian roll call thread, which answer seems more appropriate here.

Just captured sf2hf running on a 10mhz A board and a 12mhz A board, just simply letting the timer run with 1P and 2P idle. From the first visible frame of the round (following the transition from the character portrait screen and a few blank screens) to the last frame that the two characters are facing each other (the next frame that follows, they both turn and enter their "draw game" stance):
10mhz board - the round completes in about 60s
12mhz board - the round completes in about 55s

It's about a 8.5% speed difference.
Click to expand...
That doesn't jive with the video I made earlier and posted here, my timing came out the same. I am 1000% sure the hardware I used was right and even took pics with the videos. I time them and get the same time.
 
Nimmers said:
That doesn't jive with the video I made earlier and posted here, my timing came out the same. I am 1000% sure the hardware I used was right and even took pics with the videos. I time them and get the same time.
Click to expand...
You were testing CE? This is HF. :)

But that makes it all the more interesting, in my opinion.
 
I just read through this entire thread and I am just looking for clarity.
From what i gather, it is safe to assume that 10Mhz games running on a 12Mhz board (like Street Fighter 2 World Warrior), will run faster than they would on a 10Mhz A-Board. Is that...widely accepted at this point?
 
Jorel81 said:
I just read through this entire thread and I am just looking for clarity.
From what i gather, it is safe to assume that 10Mhz games running on a 12Mhz board (like Street Fighter 2 World Warrior), will run faster than they would on a 10Mhz A-Board. Is that...widely accepted at this point?
Click to expand...
I don't think it works that way. It's more about the behavior when the CPU is working hard.
 
You must log in or register to reply here.
Back
Top