Guest Big Time Software Posted November 15, 2000 Share Posted November 15, 2000 Yeah, nobody should be seriously trying to propose a fix. The only fix is the one we put in and have working right now. And that is to do a system more akin to PBEM for TCP/IP play. It is, however, superior and quicker than PBEM since we can make assumptions about things in RAM (i.e. the map has to be sent back and forth in PBEM, but not in TCP/IP). If you geeks want to swap math/code crap, please feel free to do so But we aren't going to lose any more sleep or time over this issue. Steve Link to comment Share on other sites More sharing options...
degreesK Posted November 15, 2000 Share Posted November 15, 2000 [ahh forget it.] [This message has been edited by degreesK (edited 11-15-2000).] Link to comment Share on other sites More sharing options...
Rossdaddy Posted November 15, 2000 Share Posted November 15, 2000 BTS: Thanks for the polite responses in this thread. I wonder how many different ways you could say, "We are fixing it and don't actually need any technical help." (Some folks take a bit longer to get the message, I guess.) BTW, I'm really looking forward to this patch and am sure it will work great when you're done. -Jim ------------------ When angry, count four; when very angry, swear. --Pudd'nhead Wilson Link to comment Share on other sites More sharing options...
Fuerte Posted November 15, 2000 Share Posted November 15, 2000 <BLOCKQUOTE>quote:</font><HR>Originally posted by degreesK: With the help of a friend with a Mac I've managed to get the following chaotic function to give the same results on both Mac and PC for doubles. Wasn't able to get the same results when using floats. PC gives the same results on AMD/Intel and Win98/2000. <HR></BLOCKQUOTE> Exactly, I believe that with a good compiler and double precision it could work right away, even without any extra floating point tricks. But BTS have said that they are using float, and a CodeWarrior compiler, which I don't know at all. It might be good but I just don't know it (or anything about Mac development). I have also said previously that the decision to make it more like PBEM is a good one, and I still think that way. But the use of floats must be a mistake, there can't be any speed difference. Link to comment Share on other sites More sharing options...
Guest dirkd1976 Posted November 15, 2000 Share Posted November 15, 2000 Sssssssssss......pop! (The sound of my brain frying after reading this thread) ------------------ Never mistake motion for action - Ernest Hemingway Link to comment Share on other sites More sharing options...
Guest Big Time Software Posted November 16, 2000 Share Posted November 16, 2000 <BLOCKQUOTE>quote:</font><HR>I have also said previously that the decision to make it more like PBEM is a good one, and I still think that way. But the use of floats must be a mistake, there can't be any speed difference.<HR></BLOCKQUOTE> A long time ago floating point math was slower than integer, but that has not been the case for a long time now. Especially if you have a PowerPC G3 or G4. The G4 has (forgive my layman's understanding ) two pipelines for processing floating point, each of which is probably better than the single one on a Pentium class chip. And to "prove" it, check this out... Combat Mission makes the "host" be the machine that scores the highest on a series of floating point tests Charles cooked up. This makes sense to everybody, I am sure. But what might come as a shock to people is that CM determined that my G4 400MHz system was FASTER than the P3 800MHz system playing against me. So although the P3 has twice the clock speed, my system was determined to be faster (as far as CM is concerned). If CM used integer instead of floating point, both would be slower. However, my G4 would probably be 3 times as slow as it is right now. That is a nice little argument for using floating point math to me Steve [This message has been edited by Big Time Software (edited 11-15-2000).] Link to comment Share on other sites More sharing options...
Guest aaronb Posted November 16, 2000 Share Posted November 16, 2000 <BLOCKQUOTE>quote:</font><HR>Originally posted by Big Time Software: A long time ago floating point math was slower than integer, but that has not been the case for a long time now[...]. If CM used integer instead of floating point[...]my G4 would probably be 3 times as slow as it is right now.<HR></BLOCKQUOTE> Wow. In the old days (say, '78-'93), it was canon that integer was 1.5-4 times faster than floating point - but back then, some CPU's simulated floating point with microcoded integer ops. Thanks for the update - I had silently assumed that your first "floating is faster than integer" was a non-programmer confusing what a programmer (Charles) had told them. Never assume. Link to comment Share on other sites More sharing options...
Fuerte Posted November 16, 2000 Share Posted November 16, 2000 <BLOCKQUOTE>quote:</font><HR>Originally posted by Big Time Software: If CM used integer instead of floating point, both would be slower. However, my G4 would probably be 3 times as slow as it is right now. That is a nice little argument for using floating point math to me <HR></BLOCKQUOTE> Again a misunderstanding... I meant that you should use double precision floating point instead of single precision. You said previously that you are using "float" in C, not "double". But even that may have been a misunderstanding! But I still think that integer math is (always?) faster or as fast as floating point math on most CPUs. In the current CPUs both are probably executed in single cycle, and in good conditions even many operations in one cycle. But of course there is no sqrt operation on integers built into CPU, so if complex arithmetic is needed, then floating point is faster. Link to comment Share on other sites More sharing options...
aka_tom_w Posted November 16, 2000 Share Posted November 16, 2000 <BLOCKQUOTE>quote:</font><HR>Originally posted by Big Time Software: A long time ago floating point math was slower than integer, but that has not been the case for a long time now. Especially if you have a PowerPC G3 or G4. The G4 has (forgive my layman's understanding ) two pipelines for processing floating point, each of which is probably better than the single one on a Pentium class chip. And to "prove" it, check this out... Combat Mission makes the "host" be the machine that scores the highest on a series of floating point tests Charles cooked up. This makes sense to everybody, I am sure. But what might come as a shock to people is that CM determined that my G4 400MHz system was FASTER than the P3 800MHz system playing against me. So although the P3 has twice the clock speed, my system was determined to be faster (as far as CM is concerned). If CM used integer instead of floating point, both would be slower. However, my G4 would probably be 3 times as slow as it is right now. That is a nice little argument for using floating point math to me Steve [This message has been edited by Big Time Software (edited 11-15-2000).]<HR></BLOCKQUOTE> Well I just have to say somthing here. Clock speed is not the be all and end all of CPU and computer preformance. The Risc Processor and the Velocity Chip and the Back side cache and the high perforance data bus in the Mac G4's ROCK! Seriously, a 500 mHz Mac G4 is a screaming fast machine even though the clock speed only says 500 mHz the damn thing will rip Through Photoshop tasks and filters and math intensive task like crunching CM tank battle intensive turns, faster than MOST PC machines. Really. Mac Tech Geek, -tom w [This message has been edited by aka_tom_w (edited 11-16-2000).] Link to comment Share on other sites More sharing options...
Tactical Wargamer Posted November 27, 2000 Share Posted November 27, 2000 Thought I would bring this up to the top......any chance of a update? TCP/IP before Santa? Link to comment Share on other sites More sharing options...
chrisl Posted November 27, 2000 Share Posted November 27, 2000 <BLOCKQUOTE>quote:</font><HR>Originally posted by RCMP: Thought I would bring this up to the top......any chance of a update? TCP/IP before Santa?<HR></BLOCKQUOTE> Let this one die. There's another one about the good TCP/IP news, which is that there is likely to be a beta version with TCP/IP within the next week (or less). The game code will be the same (with a few more tweaks, it sounds like) but the TCP/IP stuff will be less fully tested it sounds like. ------------------ Slayer of the Original Cesspool Thread. Link to comment Share on other sites More sharing options...
Capt. Vance Astrovich Posted November 27, 2000 Share Posted November 27, 2000 Well, RCMP, way to go. I have been on here 3 or 4 times already today, waiting for the good word on this patch. It is (as of last comment by BTS guys) scheduled for release today or tomorrow. So I have been watching herre with bated breath, dreading another delay. I am sure many have. But you bringing this thread to the top...well, my heart stopped let me tell you! Very unkind of you. Yes, this thread should die, as within 36 hours we should be playing over the interent, right? ...Right? RCMP, want to try a game over the internet. I am Canadian too. (Suspecting you are). Capt. Vance Astrovich Link to comment Share on other sites More sharing options...
Recommended Posts