Question About Wiz Screen...

[chickendung]

I am quite confused about something that has been floating around in the forums. Does the Wiz have a screwed up screen or something, because I've heard that it is now in portrait mode, and there isn't a way to fix it. Can someone fill me in on some of this stuff? What will the position of the screen be when it is released? By the way, I reordered the wiz if you heard me saying earlier that I canceled my preorder...

Oh, one more thing. Will it be possible to take apart the wiz to rearrange the buttons in a snes arrangement, and if so, would someone be willing to change the firmware so the button prompts are different on the wiz? (Because of the new arrangement I would like). The only thing that would probably be a problem is those annoying sticks coming out of the buttons which are on some controllers, and prevent rearranging...

Thanks Everyone, and have faith with GPH!

 

[WarmFluffyUK]

I am quite confused about something that has been floating around in the forums. Does the Wiz have a screwed up screen or something, because I've heard that it is now in portrait mode, and there isn't a way to fix it. Can someone fill me in on some of this stuff? What will the position of the screen be when it is released? By the way, I reordered the wiz if you heard me saying earlier that I canceled my preorder...

Oh, one more thing. Will it be possible to take apart the wiz to rearrange the buttons in a snes arrangement, and if so, would someone be willing to change the firmware so the button prompts are different on the wiz? (Because of the new arrangement I would like). The only thing that would probably be a problem is those annoying sticks coming out of the buttons which are on some controllers, and prevent rearranging...

Thanks Everyone, and have faith with GPH!

The screen has a slight diagonal tearing problem that you will probably hardly ever notice.

Regarding the buttons, sorry but no you will not be able to move them around as the way the buttons are moulded means they need to stay in the positions they are designed for. The buttons have different heights because the Wiz has sloped sides.

 

[chickendung]

Do you think the tearing will be fixed by the time it is released in April?

 

[Exophase]

Do you think the tearing will be fixed by the time it is released in April?

No, I don't think it's going to be fixed. Instead programs will have to adapt to portrait mode which doesn't have the problem, and a good number of them will probably take a speed hit doing so (and Wiz doesn't really need a speed hit right now).

Right now most people who have a Wiz can't even start doing this because the code to go to portrait mode doesn't work.

 

[chickendung]

EvilDragon said that the tearing would be fixed. When you say programs will have to adapt, does that mean emulators will be in portrait mode? I'm going to be very upset if my wiz comes and it's all effed up.

 

[vampire8bit]

Do you think the tearing will be fixed by the time it is released in April?

No, I don''t think it''s going to be fixed. Instead programs will have to adapt to portrait mode which doesn''t have the problem, and a good number of them will probably take a speed hit doing so (and Wiz doesn''t really need a speed hit right now).

Right now most people who have a Wiz can''t even start doing this because the code to go to portrait mode doesn''t work.


How bad will this speed hit be? Will some of the best GP2X emus (like picodrive, mame, gngeo, temper and gpfce) still be able to run games at full speed with 0 frameskip in portrait mode?

 

[WarmFluffyUK]

EvilDragon said that the tearing would be fixed. When you say programs will have to adapt, does that mean emulators will be in portrait mode? I'm going to be very upset if my wiz comes and it's all effed up.

I talked to someone from GPH the other day and it will not be fixed as the problem is with the OLED panel they are using.

 

[fettouhi]

EvilDragon said that the tearing would be fixed. When you say programs will have to adapt, does that mean emulators will be in portrait mode? I'm going to be very upset if my wiz comes and it's all effed up.

I talked to someone from GPH the other day and it will not be fixed as the problem is with the OLED panel they are using.


So this like with the psp 3000, it's a feature then? :blink:

Regards

André

 

[craigix]

They have probably already bought the screens so can't change to a new one until MK2.

 

[Exophase]

EvilDragon said that the tearing would be fixed. When you say programs will have to adapt, does that mean emulators will be in portrait mode? I'm going to be very upset if my wiz comes and it's all effed up.

I talked to someone from GPH the other day and it will not be fixed as the problem is with the OLED panel they are using.


And like I've said before, I think without a more technical explanation GPH is just blindly passing the blame.

 

[sam fisher]

It's a strategy that seems to work

 

[Chip]

Not having seen the tearing issue for myself, it's impossible to make a really accurate call on just how much of a problem it is. However, since the problem is only noticeable in some situations and there is a software workaround for the situations where it is needed, then this hardly seems like a device killer.

Yes, there will be a bit of a performance hit rotating the screen in software to eliminate tearing, but how much of one? Could someone do an actual benchmark to determine if the performance difference is even noticeable? I know there is probably a bit of work involved, but if someone can demonstrate that the speed difference in something like Quake is only 1-2fps, then most people will realize that this is Not A Big Deal. If I'm wrong and the difference is more like 10fps, then maybe all the concern is legitimate. Either way, everybody is just making wild ass guesses until there are some hard numbers to look at.

 

[Exophase]

Not having seen the tearing issue for myself, it's impossible to make a really accurate call on just how much of a problem it is. However, since the problem is only noticeable in some situations and there is a software workaround for the situations where it is needed, then this hardly seems like a device killer.

Yes, there will be a bit of a performance hit rotating the screen in software to eliminate tearing, but how much of one? Could someone do an actual benchmark to determine if the performance difference is even noticeable? I know there is probably a bit of work involved, but if someone can demonstrate that the speed difference in something like Quake is only 1-2fps, then most people will realize that this is Not A Big Deal. If I'm wrong and the difference is more like 10fps, then maybe all the concern is legitimate. Either way, everybody is just making wild ass guesses until there are some hard numbers to look at.

People keep asking for something like this and it can't be done because there's a big difference between using a generic piece of code to rotate the screen vs integrating the rotation into your software. What the former means is that you're going to have a 320x240 buffer that must be cached, which means cache misses across the board for loading it back in (slow on Wiz) and thrashing the entire cache every frame. For 16bpp I would expect it to be around the 2ms neighborhood, not counting the amount the thrashing degrades the performance of the application which would vary.

And please don't use fps as a unit of measurement, they're all very relative. If your app is running at 5 fps then losing 1 fps is much worse than if it's running at 100 fps. Let's say an app runs at 16.67ms per frame, or 60fps (like many emulators will want). If you add 2ms to that then it'll be 18.67ms per frame, or 53.56ms.

I think the difference, for people not properly modifying the code to cope with this better, is going to be worse than the difference between using and not using mmuhack.

 

[Chip]

And please don't use fps as a unit of measurement, they're all very relative. If your app is running at 5 fps then losing 1 fps is much worse than if it's running at 100 fps. Let's say an app runs at 16.67ms per frame, or 60fps (like many emulators will want). If you add 2ms to that then it'll be 18.67ms per frame, or 53.56ms.

I only suggested FPS on a game that everybody is familiar with because that is something everybody understands.

If 2ms per frame is an accurate estimate, then that can be translated into something everybody can understand. If a game runs at 30fps without rotation, and rotation adds 2ms per frame, then the game should run at ~28.3fps with rotation. Still playable by any standards. If you double it to 4ms, you're still looking at ~26.8fps after rotation slowdown.

So the moral of this story is that for many programs, the tearing issue won't even be noticeable. For instances where it is noticeable, it can be corrected in software. In those cases where the software fix is used, there will be some performance loss, but not a lot. It sucks that this issue exists at all, but unless you're really picky, it shouldn't be a deal breaker.

 

[MWeston]

Copying data from a fake frame buffer to the real one can be a huge penality. Even on the OMAP3 I have seen frame rates drop from 100fps down to 60fps by taking a 320x240 and blowing it up to 640x480. Of course there is more work done there quadrupling each pixel but Exophase said these cache misses cause huge penalties (across all ARM platforms I imagine) so it might be a big problem. Someone with the right hardware will have to test it though. Perhaps take your 320x240 game and draw it at 240x320 just to see how it goes even though it won't look right on the screen. That will satisfy everyone on what the real penalty will be. Of course, who wants to waste their time trying to show how bad an error will look so it will probably have to wait until the kernel changes to 240x320 and people are forced to try it (if that happens).

There is a c64 video in the Wiz archive that shows the tearing quite well, especially right before some Karate fighting type of game I think. It's a huge download though.

 

[Prophet]

I was considering a Wiz, but not if it's going to have tearing, or take a big speed hit to avoid tearing, since it's already kinda slow for 2009. I hate screen tearing, it's a major pet peeve for me personally.

Guess I'll just look toward getting a Pandora (once they get credit cards working for the second batch).

 

[Exophase]

So the moral of this story is that for many programs, the tearing issue won't even be noticeable. For instances where it is noticeable, it can be corrected in software. In those cases where the software fix is used, there will be some performance loss, but not a lot. It sucks that this issue exists at all, but unless you're really picky, it shouldn't be a deal breaker.

And if you do it RIGHT in software the speed hit should be very limited. I just don't have a lot of faith in people to bother. But you never know! I'm sure Picodrive will at least.

 

[chickendung]

Well, if it is noticeable and it screws up fast games, then I will get a refund or something. It is unacceptable for a device to have these problems in such a massive amount of devices.

I do have a feeling that the wiz will be delayed again, because I think GPH will catch on and want to take the time to fix it. I'll bet that if there are more than two more delays past april 30th, they will cancel the wiz. It would serve them right if they keep on screwing up the device's release.

 

[Exophase]

I don't think so dude, GPH has clearly thrown their hands up in the air and have convinced themselves that there's nothing they can do about it so they should just leave it alone.

 

[Chip]

I do have a feeling that the wiz will be delayed again, because I think GPH will catch on and want to take the time to fix it.
That is highly unlikely. They have to go with an off-the-shelf component, they decided to go with an OLED display in the Wiz, and there aren't very many 2.8" OLED displays to chose from. The manufacturer is not likely to make any changes to the hardware just for a small run of PMPs, especially since it isn't really a problem from their perspective. The display isn't "broken", it just wasn't designed to be used in landscape mode. Either GPH didn't do enough research when they sourced the part, or they didn't think it was a big enough issue to use something else.

Like I said earlier, it's not that big a problem as long as programs are coded correctly. Though as Exophase pointed out, it remains to be seen how many actually are done correctly.