Problems with winfellow 0.5.1

[theelf_]

Hello, sorry my english


I download the new version of winfellow, the 0.5.1, but not work. Actually I use 0.4.4

I have a Athlon XP 1ghz, and when double click to the icon, nothing happen

Version 0.4.4 works great

I tested the same 0.5.1 winfellow in a C2D machine, WinXP (same XP from Athlon), and works OK, is not the file..

I check the XP from Athlon, and the XP from C2D, and both are SP3, same Directx version, etc

Any advice? the new version have some special requirements? (CPU, ram etc)


Thanks

[carfesh]

Hi,

could you please upload your fellow.log file from the system it is not working on? I have a guess about what it might be, I'd like to check that. You should find it under

%APPDATA%\WinFellow

We did not intentionally change the requirements, apart from that you now need XP to run WinFellow (so Win2000 is no longer supported).

[theelf_]

Originally Posted by carfesh

Hi,

could you please upload your fellow.log file from the system it is not working on? I have a guess about what it might be, I'd like to check that. You should find it under

%APPDATA%\WinFellow

We did not intentionally change the requirements, apart from that you now need XP to run WinFellow (so Win2000 is no longer supported).

Hi carfesh!! thank you for your reply, you are right, I did not attach a log, sorry, im feel idiot!

"Unhandled exception detected" mm... very generic error??

WinFellow alpha v0.5.1 r880 (SVN) (release build)

information retrieved from registry:

CPU vendor: AuthenticAMD
CPU type: AMD Athlon(tm) Processor
CPU identifier: x86 Family 6 Model 8 Stepping 1
CPU clock: 1000 MHz

operating system: Windows XP
parameters: OS 5.1 build 2600, Service Pack 3

number of processors: 1
processor type: 586
architecture: 0

total physical memory: 223 MB
free physical memory: 101 MB
memory in use: 54%
total size of pagefile: 339 MB
free size of pagefile: 226 MB
total virtual address space: 2047 MB
free virtual address space: 2009 MB

display: SiS Mirage Graphics
media: Realtek AC'97 Audio
unknown: Ethernet Controller
media: Audio Codecs
media: Legacy Audio Drivers
media: Media Control Devices
media: Legacy Video Capture Devices
media: Video Codecs


information retrieved from hardware detection:

01/13/13 07:32:42: Unhandled exception detected, write minidump to C:\Home\Admin\Application Data\WinFellow\WinFellow_0.5.1.880_20130113_063242.dmp...

[carfesh]

Could you please mail me the full logfile, and the dump file (if possible, compressed as zip), to carfesh@gmx.net?

I can then take a closer look into why it crashes in a debugger. Does it show no text between information retrieved from hardware detection: and the next line, or did you shorten the file?

It sounds like the CPU detection code may cause issues with the CPU model you use; that would be odd, as that code wasn't changed between 0.4 and 0.5.

[carfesh]

By the way: it's most probably not related, but why does your date show the 13th of January 2013??

[theelf_]

Originally Posted by carfesh

Could you please mail me the full logfile, and the dump file (if possible, compressed as zip), to carfesh@gmx.net?

I can then take a closer look into why it crashes in a debugger. Does it show no text between information retrieved from hardware detection: and the next line, or did you shorten the file?

It sounds like the CPU detection code may cause issues with the CPU model you use; that would be odd, as that code wasn't changed between 0.4 and 0.5.

Hi, thank for taking your time in helping me

jeje, about date, yes, u r right, I did not realize, I use this computer only for games, never check the clock

I put correct date, but still same problem

I did not short any file, just copy paste


I upload the dump to my website, is a direct download, I think is easy than mail

http://www.akihabara-online.com/Mate...o2013/dump.zip

[carfesh]

You have a PN; I have built a test fix and am in discussions with the rest of the team about it.

[theelf_]

Hi carfesh

Thank you for the private build, you can erase now, I already downloaded

The new build crash too

Here is the new dump

http://www.akihabara-online.com/Mate...2013/dump2.zip




To be sure is not a OS problem, I reinstalled windows to test this build, same problem


This is my computer, the CPU is a Geode NX1500, is a Athlon XP with lower watts, but totally x86 compatible


http://uk.ts.fujitsu.com/rl/services.../futros400.htm



Thanks

[carfesh]

Hm,

the first crash is gone, it now seems to crash at a later place. Sadly, as I had compiled a new build in the meantime, I don't have the debug info anymore to debug that 2nd crash. I uploaded a new file and am afraid I have to ask you to again provide a dump, if possible?

Seems like we have more serious issues on AthlonXP processor based systems; I've never had one to test myself.

[theelf_]

hi carfesh

I download new build, same crash, here is new dump

http://www.akihabara-online.com/Mate...2013/dump3.zip


Sadly I don't have another Athlon XP to test, but I will try to assemble one tonight, since I have some CPU and motherboard socket A in some place of my house


Thanks very much

[carfesh]

Hi Theelf,

thanks, I could now debug your crash, but am somewhat at a loss to interpret what is happening; the result doesn't make any sense to me right now. I have committed the change into the code and compiled a release (beta) build, r944 - maybe that one behaves differently; and if it doesn't, at the very least I will be sure that I have properly archived debug symbols, as they are always part of the archive. So a new dump I should be able to interpret more reliably.

Will provide a link to the beta download to you by PN.

[theelf_]

Hi carfesh!

Thank you very much for your efforts

Here is the dump of the last build you sent to me

http://www.akihabara-online.com/Mate...2013/dump4.zip

Hope is useful !!



I found a motherboard socket A, now im looking for a athlon, PC133 ram, etc to build a test machine

[carfesh]

Hi,

that is a lot of effort I did some reading, it looks like Visual Studio 2012 (which we are using since version 0.5 to build WinFellow) will, by default, target SSE2 processors. The Athlon XP does not offer that, which is probably why it fails. I'm building an SSE(1) build right now, will provide it to you shortly for testing.

[theelf_]

Hi carfesh!

Ok, you found the problem! great, good job, was the SS2 instructions

(I tested in my Pentium 2 machine, and not work, because for sure the SS1, but not care, for the P2 i have DOS Fellow that is fast and nice !! jaja)


Im working now, and I can test the build much, I will do tonight, but at least, open and I can play a game


By the way, is this build, or version 0.5.x is deadly slow compared to 0.4.x?


I tried 3 games, and all lagged badly compared to the old version


Thanks!!

[carfesh]

The build I provided is a debug build; those are always significantly slower - it was just easier for me to build. A proper build would be a release build, with optimizations. I can provide a proper release build to you tomorrow morning, as I'm leaving for the evening.

Let me discuss with the rest of the team what they think about changing the project back to SSE, I already asked for feedback; 90% of the processors today are able to do SSE2, depending on the speed difference we should decide if it is ok to configure all builds to SSE, or maybe provide two builds so that those older processors can still be used.

[theelf_]

Hi carfesh, thanks for all your efforts



My opinion, from the "user" side...


....kega fusion, snes9x, zsnes, final burn, jNES, BlueMSX, Neko2, ePSXe, X68K, mame, MagicEngine, WinUAE, etc etc, I can list all emulators I use perfectly in my Athlon XP machine

In fact, all this emulators still work in my old pentium 2!

I don't know how much improvement SS2 give, but it seems that's is not a must in any emulators



Please don´t misunderstand me, I use Fellow since DOS version, but I think, if someone have a SS2 capable machine, will prefer to use WinUAE, because high compatibility,AGA, CD32, filters, etc etc

I found Winfellow great for old systems more than new ones

Thanks!

[carfesh]

I found Winfellow great for old systems more than new ones

Well, I wouldn't put it quite like that, as it is certainly also great on new systems, while WinUAE is certainly not so great on old ones

But I'm not getting into that discussion..

SSE1 processors are quite rare today, most "modern" processors support SSE2. And with modern, I mean the ones from the last 10 years or so, so there's still some pretty old systems among them. My own PC is like 7 years old, and it supports SSE2. I'm convinced that as more and more applications switch to the newer Visual Studio, issues like that will be increasing for you.

However you're not getting any argument here in principal, and a solution will be provided. Still, I'd like to examine more closely if that solution should be to just change the project to SSE(1) in general, or if we provide two separate downloads for SSE1/2. We have done that in the past with Direct3/5 for example.

I will provide a release build to you soon via PN.

[theelf_]

jeje, well, I will tell you the truth, I use fellow since DOS version, and I prefer to use fellow even on my new machine..

Maybe is nostalgic feelings, but I love fellow. I will never forget this looooong amiga-emulated-fellow nights in my (on the time) new 233mmx !!


back to topic, I downloaded the beta you send me by MP, works great, thanks. No worry, no share

By the way, I love the new "2x" option, with scanlines, looks goodly in my 1280x1024 screen! sadly is slow to me! ahhhh ..jaja


About SSE problem, from my point of view, if Winfellow is capable to run pretty well in a pre-SSE2 machine, there is no reason to stop supporting them, of course, besides new compiler, new options, etc you need to use

In fact, besides, Saturn, Playstation 2, GameCube/Wii emulators, that anyways because CPU power is not possible to use, I never found any that don't work on my Athlon XP




The problem for me with "modern" computers, Pentium4 and up, is that they break all the "retro" feeling. I don't feel comfortable anymore..

Back on time, everything was more "difficult" from bad floppy/tape (all time!), copy protections , no internet to get games jaja, etc etc

Now is seems like, build a new computer is so easy and cheap, and then just double click to a icon, download a game from internet... and play

Not fun, I need something else!! fight with the computer, and emulation retro on "retro" is perfect for have fun

[carfesh]

By the way, I love the new "2x" option, with scanlines, looks goodly in my 1280x1024 screen! sadly is slow to me! ahhhh ..jaja

great to hear that you like it - if you say slow, how slow? How many FPS are you able to get (you can enable displaying it in the various menu, they appear in the top right corner in windowed mode), and what does task manager say about WinFellow's CPU usage?

I certainly get what you mean about newer computers; it's probably not so much caused by the computing power, but also that the Internet is always available, so just about everything is available with the click of a button. Also, even the emulators are becoming more and more user friendly, taking challenges away to make it more easy to enjoy those old Amiga titles.

[theelf_]

Hi carfesh

jeje, yes, I like user friendly emulators, but not too much he? jaja


About your question, the 2x option, is slower, I cant say how much, but, I can see that solid is much slower than scanlines

I tested 3 games, that I think are less, normal, and high CPU demanding


Lotus 3 = 640x400 - 45% CPU
Lotus 3 = solid 2x - 85%-100%?¿ CPU
Lotus 3 = scanlines - 65-80% CPU

RUFF 'N' TUMBLE = 640x400 - 65% CPU
RUFF 'N' TUMBLE = solid 2x - 90%-100%+ CPU (small-normal slowdown)
RUFF 'N' TUMBLE = scanlines - 85-90% CPU


Jim Power = 640x400 - 80% CPU
Jim Power = solid 2x - 100%++ CPU (big slowdown)
Jim Power = scanlines - 95-100%+ CPU (mm..20% slowdown?¿)


Here a screenshot if RUFF 'N' TUMBLE, the first CPU usage is 1280x1024, 2x, scanlines, and the second is 640x480 with scanlines






Well, for less CPU demmanding games I can use the 2x option + scanlines, that's why I say I love this new option, good job