| [Files] See AmigaOS4 in action | ANN.lu |
|
| See AmigaOS4 in action : Comment 251 of 269 | ANN.lu |
| Posted by Phill on 23-Sep-2003 14:44 GMT | In reply to Comment 248 (Fabio Alemagna):
>I said you should "trust me"? Perhaps, I don't rememember, what I remember,
> though, is that I explained _why_.
Well it should be there if you scroll back. I don't think I'm wrong, but I'll check if you like.
Your reason for _why_ was flawed, which is why when I disputed you knew what you were talking about, I also gave the reason _why_.
Phill |
|
| See AmigaOS4 in action : Comment 252 of 269 | ANN.lu |
| Posted by Fabio Alemagna on 23-Sep-2003 14:48 GMT | In reply to Comment 250 (Phill):
> remember when I said improving emulation can make them slower and you said I was
> wrong too?
See, you don't even know how to read. To say I explained it to you more than twice already.Now, say whatever you wish, I'll let you have the last word, if that is what you are looking for. |
|
| See AmigaOS4 in action : Comment 253 of 269 | ANN.lu |
| Posted by Phill on 23-Sep-2003 14:52 GMT | In reply to Comment 249 (Fabio Alemagna):
>Ok, then please list which are these components which are enabled AND are
>different between UAE 0.8.22 and WinUAE 0.8.22.
I can't tell you that, as you haven't told me which release of WinUAE 0.8.22 you are talking about. According to the UAE page the latest 0.8.22 was released on the 27/07/2002 & the latest WinUAE 0.8.22 was released 24/08/2003. There are alot of changes to the base emulation code made during that time.
You will probably say I am stupid for not knowing which version of WinUAE 0.8.22 you are running, because it should be obvious. Therefore I'm prepared for one of your obvious questions = stupid answers replies. But cest la vie.
Phill |
|
| See AmigaOS4 in action : Comment 254 of 269 | ANN.lu |
| Posted by Phill on 23-Sep-2003 14:56 GMT | In reply to Comment 252 (Fabio Alemagna):
>> remember when I said improving emulation can make them slower and you said I was
>> wrong too?
>See, you don't even know how to read. To say I explained it to you more than twice already.
Explained what twice already? This is what I referred to:
"> & point out that most of the time improving it makes it slower.
Riiight, I must be so stupid! For whatever else reason would one _improve_ picasso emulation, if NOT for making it *worse*? "
Or is this one of those "I'm going to say random things again in reply because I think the answer should be obvious"? Or do you just not remember what you argue because you make it up as you go along?
Phill |
|
| See AmigaOS4 in action : Comment 255 of 269 | ANN.lu |
| Posted by dammy on 23-Sep-2003 15:18 GMT | In reply to Comment 242 (Phill):
> As I said I prefer anonimity & it's irrelevant to the discussion. Who you are
> doesn't change whether you are right or wrong. Sometimes I am wrong & then I
> don't mind being told.
Nice attempt in changing the topic of my question. I really don't give a damn who you are "Phill", I was asking you exactly what emulation programs have you written from scratch? Your damning Fabio and I have yet to hear a listing of your emulation credits. Just what accomplishments makes you an expert over Fabio? You sure do talk the talk, I have yet to see you walk the walk. Funny that, huh?
Dammy |
|
| See AmigaOS4 in action : Comment 256 of 269 | ANN.lu |
| Posted by Phill on 23-Sep-2003 15:24 GMT | In reply to Comment 255 (dammy):
>> As I said I prefer anonimity & it's irrelevant to the discussion. Who you are
>> doesn't change whether you are right or wrong. Sometimes I am wrong & then I
>> don't mind being told.
>Nice attempt in changing the topic of my question. I really don't give a damn
> who you are "Phill", I was asking you exactly what emulation programs have >you written from scratch?
Well thats who I am. I also gave you a rough idea about what I have been working on recently. I have written emulators from scratch but I'm sure as hell not going to discuss it with you. Especially as you seem to have ignored the fact that I told you that I had written emulation from scratch & yet you choose to ignore that part of the message when quoting and the bit where I said you'd respond in this particular way.
> Your damning Fabio and I have yet to hear a listing of your emulation
> credits. Just what accomplishments makes you an expert over Fabio? You sure > do talk the talk, I have yet to see you walk the walk. Funny that, huh?
As I said, It doesn't matter whether I am an expert over Fabio. What I said was right. He is wrong, it doesn't matter whether his dad is bigger than mine or anything. It's facts. He said things that were wrong, so I can only take it that he doesn't understand what he is talking about. I have recently learned that Fabio says random things in response to questions, if he doesn't think it's worth answering. Therefore I might be wrong about him, it depends on whether he was doing the whole random answer thing or he thinks what he said was true.
Dammy |
|
| See AmigaOS4 in action : Comment 257 of 269 | ANN.lu |
| Posted by dammy on 23-Sep-2003 15:37 GMT | In reply to Comment 256 (Phill):
>Nice attempt in changing the topic of my question. I really don't give a damn
> who you are "Phill", I was asking you exactly what emulation programs have >you written from scratch?
[quote]
Well thats who I am. I also gave you a rough idea about what I have been working on recently. I have written emulators from scratch but I'm sure as hell not going to discuss it with you. Especially as you seem to have ignored the fact that I told you that I had written emulation from scratch & yet you choose to ignore that part of the message when quoting and the bit where I said you'd respond in this particular way.
[quote]
I asked *exactly* what emulations have you written from scratch. Not a rough idea, but exact. For someone being so superduper intelligent, you sure do act dumb.
> Your damning Fabio and I have yet to hear a listing of your emulation
> credits. Just what accomplishments makes you an expert over Fabio? You sure
> do talk the talk, I have yet to see you walk the walk. Funny that, huh?
[quote]
As I said, It doesn't matter whether I am an expert over Fabio. What I said was right. He is wrong, it doesn't matter whether his dad is bigger than mine or anything. It's facts. He said things that were wrong, so I can only take it that he doesn't understand what he is talking about. I have recently learned that Fabio says random things in response to questions, if he doesn't think it's worth answering. Therefore I might be wrong about him, it depends on whether he was doing the whole random answer thing or he thinks what he said was true.
[quote]
Pity your being too gutless on giving any creditability to your statements. Once a true Red Troll, always a Red Troll, right Dave?
[quote]
Dammy
[/quote]
Take more pain killers, they may help you to remember who your suppost to be next time.
Dammy |
|
| See AmigaOS4 in action : Comment 258 of 269 | ANN.lu |
| Posted by Phill on 23-Sep-2003 17:25 GMT | In reply to Comment 257 (dammy):
>Pity your being too gutless on giving any creditability to your statements.
> Once a true Red Troll, always a Red Troll, right Dave?
Dave? I think you've got the wrong person. If I tell you the emulators I've written then you won't believe me anyway, you don't even believe my first name.
I'm not the troll, Fabio is. His "irony" posts are definite trolls. It's only now he's even bothered to think about comparing WinUAE & UAE code, although I guess he is waiting for me to spend the next month disecting every change so he can tell me just by looking at it that it's irrelevant.
It does cause problems being anonymous, but with the high profile emulation I'm involved with I can't be too careful. Someone has previously used information they gained about me to cause me problems. I don't like to trade on who I am or what I have done in the past, the people on the relevant emulation mailing lists / irc channels know who I am and thats all that I care about.
I'll leave Fabio to take a look at the diff's between the latest WinUAE code and the year old UAE code. I got some bugs to fix in my latest rendering code before I can release it. Some of the new games that will then work are pretty cool.
Phill |
|
| See AmigaOS4 in action : Comment 259 of 269 | ANN.lu |
| Posted by Alkis Tsapanidis on 23-Sep-2003 18:40 GMT | In reply to Comment 239 (Analnymous):
Well, both words are Greek, both are valid, next one. |
|
| See AmigaOS4 in action : Comment 260 of 269 | ANN.lu |
| Posted by Alkis Tsapanidis on 23-Sep-2003 18:42 GMT | In reply to Comment 242 (Phill):
Fabio's game of "knowledge by association", but it would involve listing alot of the top emulator programmers.
--
Well, "knowledge by association" is better than "knowledge by masturbation",
which is the game you play right here... |
|
| See AmigaOS4 in action : Comment 261 of 269 | ANN.lu |
| Posted by hammer on 23-Sep-2003 20:37 GMT | In reply to Comment 228 (Phill):
@Phill
The point was that MS was busy with other projects NOT knowledge of history contest.
>The took Presentation manager out of OS/2 3.0, added Program Manager & it >became Windows NT 3.51. AFAIK even Windows XP will still run OS/2 command >line apps.
Not quite. Note that an earlier NT code based was released (April 1993) i.e. Windows NT 3.1 (1)(2). In another words (from the residual "OS/2 NT" project), the first Windows NT release was "MS Windows NT 3.1".
Windows NT 3.5 was release at 1994. Windows NT 3.51 is just fix/patch from the Windows NT 3.5 code base (similar to NT 5.0(Win2K) -> NT 5.1(WinXP) -> NT 5.2 (Win2K3) situation). The kernel is roughly based on VMS while OS/2 compatibility is just API layer. Figure_01.gif shows the time line.
References
1. http://www.winntmag.com/Articles/Index.cfm?IssueID=97&ArticleID=4494
2. http://www.winntmag.com/Files/4494/Figure_01.gif |
|
| See AmigaOS4 in action : Comment 262 of 269 | ANN.lu |
| Posted by Bernie Meyer on 24-Sep-2003 01:53 GMT | In reply to Comment 250 (Phill):
Goddamn it! How many times do I have to explain that Amithlon is *not* faster because it doesn't emulate the Amiga chipset?
Amithlon is faster than UAE-JIT, especially for small loops, because it has several optimizations done to the JIT engine. Period.
The part that apparently always confuses people is that these optimizations were only *possible* because Amithlon doesn't even try to emulate the chipset, *and* because it uses its own, custom-hacked linux kernel. That's why the optimizations *could* be done. Understand?
It's really quite simple --- *if* you have to allow for the possibility of custom chip emulation (and user-space derived timing) in the JIT compiler, then the JIT compiler will be a little slow. That's the case with UAE-JIT. If Amithlon were to use the very same JIT compiler, it would be just as slow. But Amithlon doesn't, because it has support from the kernel to do its timing, and it doesn't need to allow for custom chip emulation --- so it uses a *different* JIT compiler. |
|
| See AmigaOS4 in action : Comment 263 of 269 | ANN.lu |
| Posted by Phill on 24-Sep-2003 05:40 GMT | In reply to Comment 260 (Alkis Tsapanidis):
oh good one, so that isn't a troll.
Phill |
|
| See AmigaOS4 in action : Comment 264 of 269 | ANN.lu |
| Posted by Fabio Alemagna on 24-Sep-2003 06:33 GMT | In reply to Comment 262 (Bernie Meyer):
> Goddamn it! How many times do I have to explain that Amithlon is *not* faster
> because it doesn't emulate the Amiga chipset?
Oh dear, looks like someone needs to apologize :-) |
|
| See AmigaOS4 in action : Comment 265 of 269 | ANN.lu |
| Posted by Phill on 24-Sep-2003 06:46 GMT | In reply to Comment 262 (Bernie Meyer):
I had always assumed it was because you didn't need to go through read/write memory handlers in amithlon, as you didn't need to trap access to the custom chips as they were never there.
I guess I was wrong
Phill |
|
| See AmigaOS4 in action : Comment 266 of 269 | ANN.lu |
| Posted by Bernie Meyer on 24-Sep-2003 10:55 GMT | In reply to Comment 265 (Phill):
See, that's why you shouldn't assume --- especially considering I have explained this over and over and over, for two years now.The memory handlers are there. You cannot run 68k AmigaOS programs without at least a tiny amount of custom chip emulation, simply because the system-official includes have macros for ENABLE and DISABLE which write directly to the interrupt control registers.Of course, *just* like in UAE-JIT, most JIT-compiled memory accesses don't go through the memory handlers, but rather hit memory directly. And just like in UAE-JIT, the MMU is used to catch the (extremely rare) cases where hitting the memory directly was the wrong thing to do, and the MMU exception handler will then call the appropriate memory handler, fix everything up, and invalidate that translated block of code. *No* difference between UAE-JIT and Amithlon in that regard.The difference is that in UAE-JIT, time is counted in "cycles", by the user space process; Furthermore, "events" (such as horizontal sync or CIA interrupts) get scheduled to happen at a certain number of "cycles". So the CPU emulation must do an "if ((cycles+=cycles_needed_for_last_block)>=next_scheduled_event_cycles) then handle event". In Amithlon, time is kept by the CPU's internal clock cycle counter (a 64 bit counter incremented every clock cycle). Events are scheduled with the kernel, which uses the hardware timer to actually cause an interrupt at the required time. That interrupt, just like real hardware interrupts, gets passed on as a signal to the emulation process. The process can then handle things in the signal handler.If, in a signal handler, or as the result of some chipset-write, a 68k interrupt condition is required, then the MMU is used to make a certain page in memory read-only. JIT-compiled code, on the other hand, has a single write instruction into that page at the end of every block of instructions. The MMU exception handler in the kernel is hacked to recognize writes to that particular page, and if they cause an exception, will *not* run the usual code, but instead short-circuit things by changing the x86 program counter to the value which was attempted to write --- which just happens to be the address of a "break out of compiled code and deal with interrupts" routine. A horrible hack, but oh so useful.So, at the *end* of each block of translated instructions, UAE-JIT needs to load a global variable, add a constant to it, store it back, compare it with the value of another global variable, and depending on the result, do a conditional jump --- at least conceptually (in reality, I optimized things to not need the second variable, but still...). In contrast, Amithlon does a single, dependency-free memory write, which in 99.999% of cases will simply succeed. *That's* why it's faster (or at least, that's half the answer. The other half would take even longer to explain :) |
|
| See AmigaOS4 in action : Comment 267 of 269 | ANN.lu |
| Posted by pixie on 24-Sep-2003 13:55 GMT | In reply to Comment 266 (Bernie Meyer):
Only ann makes it possible! ;) |
|
| See AmigaOS4 in action : Comment 268 of 269 | ANN.lu |
| Posted by strobe on 25-Sep-2003 05:52 GMT | All this talk about JIT is ridiculous.
A non-JIT 68k emulator running on a 120mhz 604 was MUCH faster than a 33mhz 68040. This was Apple's crappy 68k emulator, not Connectix.
Anyway the AOS4 demo would have been better if there were apps to test for the purposes of compatibility.
Ah screw it all. I hate desktop computers! |
|
| See AmigaOS4 in action : Comment 269 of 269 | ANN.lu |
| Posted by Daniel Miller on 25-Sep-2003 08:17 GMT | In reply to Comment 262 (Bernie Meyer):
Bernie Meyer says:
> Goddamn it! How many times do I have to explain that Amithlon is
> *not* faster because it doesn't emulate the Amiga chipset?
Have you considered threatening everyone with an AK47 until they get this straight, Bern? ;) I can see it now, "read the FAQ or else!"
> Amithlon is faster than UAE-JIT, especially for small loops,
> because it has several optimizations done to the JIT engine. |
|
|
- User Menu
-
- About ANN archives
- The ANN archives is powered by #AmigaZeux. It was updated daily (news last: 22-Oct-2004; comments last: 18-May-2005).
ANN.lu was created, previously owned and maintained by Christian Kemp, www.ckemp.com.
- Contribute
- Not possible at this time!
- Search ANN archives
- Advanced search
- Hosting
- ANN.lu was hosted by Dreamhost. Sign up through this link, mention "ckemp" as referrer and he will get a 10% commission on any account you purchase.
Please show your appreciation for any past, present and future work on ANN.lu by making a contribution via PayPal.
|
