Yeah there isn’t a lot to account for the discrepancy other than the mismatch in framerate, and that causing abnormal behavior, which really means that unless someone wants to tear the code apart the IGT is still largely a question mark, which doesn’t seem like any way to mark a run.

Like at current we know that two runs that’re 35 minutes could be very far apart and unless that’s all menu’ing there’s clearly something weird going on, whether it’s what I think or what you’re getting at or something else, it definitely seems like it’s not an accurate metric.

I just want to be clear about one thing with the IGT, the frame rate slowdown doesn’t make things fair it actually makes things less fair.

When I tested it, and the frame rate slowed down, the IGT was actually still ticking at 30fps. So if your game is running at 10fps due to slowdown every second is counting 3 seconds against you. That means that any mode where the game runs smoother is always going to be privileged over ones that don’t, by a pretty large margin. Three seconds per second is pretty hard to overcome, and multiplicatively more punishing than just having your framerate dip by the inverse of the frame drop.

I’m glad things got shifted around but this game isn’t slowing down the timer when your framerate chugs. It is, in effect, speeding it up. That was the core of my argument against IGT to begin with, because it’s actually disturbingly inaccurate, rather than extra accurate and extra fair.

So here are three quick facts about the IGT that I think support using RTA or at least splitting things out:

1. everyone starts their timers 20-30 seconds before IGT starts

2. everyone stops their timer 90-120 seconds after IGT ends

3. IGT pauses during menus.

When you combine all of that RTA should actually be several minutes longer than IGT, but that’s clearly not the case.

I hope your testing yielded the same results mine did: that IGT seems heavily tied to frame rate, something no one has any real control over, and something you can’t reliably predict with regards to IGT.

Also, in the spirit of cooperation, I’d appreciate if you’d post what your findings were here, in the same way that you asked me to do that.

I mean you know, we can all look back and say what could've caused it not to escalate in the first place. I think probably not accusing someone of being a liar would've helped with that as well ;P

He didn't post the values because he didn't have them, he was just relaying my findings, and I was obviously more than happy to post them once I saw the thread. Hopefully it helps clear things up.

Also if you want the memory values, I have them for psxjinv2.0.2:

00000 000C3A90 d s 0 Timer 00001 00000000 S S 0 ---------------------------- 00002 00063454 d s 0 Timer

You can load those into Memwatch to see for yourself, but they won't work in CE, because the values are in the RAM of the emulator

And in Bizhawk2.4 it's..

7FFC34E60CD4, which works just fine in CE.

So, I'm the tech-inclined friend and I can confirm all the stuff that Punchy said about the IGT, that was all stuff I found. The timer starts when you fade up from black for the first time, a few seconds before you get control, and stops right when you fade to black after the final Hypnos kill.

It does pause during menus, it doesn't pause during door loads, and it doesn't pause during the unskippable cutscenes that I tested out.

The value I isolated is the timer in frames, I know it's the right one because if you divide it by 30 it gives you your exact seconds, and I was able to manipulate it to get it to spit out predictable times (1 seconds for 30, 4 seconds for 120, etc.)

Currently, I think the big discrepancy between RTA and IGT is because it seems to be running at full speed even when your framerate is jacked, because it's running concurrently with whatever's going on in the game, but honestly, that doesn't explain everything, because IGT is still pretty janky even with that.

At this point I can reliably predict what Punchy's discrepancy is going to be, but the discrepancies seem to be individual, since they're at least partially due to menu'ing differences as well. The upshot, though, is that they're very different from Console to Emulator, and if they're different at all, that seems pretty problematic.