Query failed: connection to localhost:9312 failed (errno=111, msg=Connection refused). GCC updated for 970 and Power4... - Future Apple Hardware Discussions on AppleInsider Forums Toggle navigation All Forums Recent Posts Sign In GCC updated for 970 and Power4... marcus Posted: May 16, 2003 12:03PM in Future Apple Hardware edited January 2014 Here... Don't know if this is significant or not... Marcus(Mods - Please delete/lock if this is useless info.) «12345»Comments Reply 1 of 98 programmer Posts: 3,409member May 16, 2003 12:08PM Quote:Originally posted by Marcus Here... Don't know if this is significant or not... Something wicked this way comes... Reply 2 of 98 hmurchison Posts: 12,341member May 16, 2003 12:10PM I hope it makes it easy to optimize code for Altivec and take advantage of the new features of the 970. I'm assuming we'll need additional tweaks to take advantage of the superior FPU as well. Reply 3 of 98 amorph Posts: 7,112member May 16, 2003 12:11PM Confirmed! Panther will introduce Objective-LISP! Reply 4 of 98 smalm Posts: 667member May 16, 2003 12:25PM Quote:Originally posted by hmurchison I hope it makes it easy to optimize code for Altivec and take advantage of the new features of the 970. I'm assuming we'll need additional tweaks to take advantage of the superior FPU as well. With its long fetch/decode/regroupe pipeline and its strengts with out-of-order executions the 970 is perhaps the PPC chip with the least need of compiler optimization to date. But every bit is wellcomed Reply 5 of 98 thegelding Posts: 3,230member May 16, 2003 12:35PM so what the hell did that all say in human, non-geek headed "i sure miss my pocket protector" "if it was legal, i'd marry ms pacman" "wouldn't it be cool if they named tv dinners after real tv show"***, terms??g*** all phrases i have heard geeks utter at one time or another in my life Reply 6 of 98 amorph Posts: 7,112member May 16, 2003 12:37PM This'll really be the first PPC designed to run crap code, as Hannibal's pointed out. Since Apple's using gcc now, this is an important point.However, it's also worth noting gcc's advantage: Since nobody owns it, neither IBM nor Apple has to wait for, say, Metrowerks to update. IBM engineers can roll any needed or desired optimizations right into the codebase. I have a feeling that if gcc were not open source, Apple would not have had anywhere near the success they've had getting a compiler tailored for their platform. Granted, gcc 2.x PowerPC code was about as bad as it got, but the current versions aren't half bad, and it's significant that Apple, Mot and IBM engineers can all contribute optimizations and improvements at once. There's no vendor to wait for. Reply 7 of 98 amorph Posts: 7,112member May 16, 2003 12:39PM Quote:Originally posted by thegelding so what the hell did that all say in human, non-geek headed "i sure miss my pocket protector" "if it was legal, i'd marry ms pacman" "wouldn't it be cool if they named tv dinners after real tv show"***, terms??It means that gcc now knows what a 970 is, and what it's capable of, in broad terms. Reply 8 of 98 t'hain esh kelch Posts: 1,274member May 16, 2003 12:59PM What is GCC...? Reply 9 of 98 kickaha Posts: 8,760member May 16, 2003 1:02PM GNU Compiler Collection.The compiler used on MacOS X for most development of the system, and the default compiler under Project Builder. Reply 10 of 98 opuscroakus Posts: 316member May 16, 2003 1:24PM Quote:Originally posted by Amorph This'll really be the first PPC designed to run crap code, as Hannibal's pointed out. Since Apple's using gcc now, this is an important point. Amorph, could you clarify why the ability to run the "crap code" is benificial? Does crap code = poorly written apps? Reply 11 of 98 panther Posts: 64member May 16, 2003 1:52PM Quote:Originally posted by Amorph Confirmed! Panther will introduce Objective-LISP! Hello everyone. I just would like to start off by saying how wonderful it is to be here. A very special thanks to Amorph for inviting me. And now, I'd like to introduce a very special friend, Objective-LISP.Have a nice day. Reply 12 of 98 carson o'genic Posts: 1,278member May 16, 2003 2:57PM Let me guess, this is the thread that all the off-worlders are using to communicate among themeselve to prepare for the final invasion of Earth. How else to explain the distinct lack of English word (or even type-that last link!!).http://forums.appleinsider.com/images/smilies/1wink.gif Reply 13 of 98 amorph Posts: 7,112member May 16, 2003 3:00PM Panther: Thanks! I knew there were OO versions of LISP around (I think Scheme is one such, and that's probably what GCC uses), but it's still funny. Quote:Originally posted by opuscroakus Amorph, could you clarify why the ability to run the "crap code" is benificial? Does crap code = poorly written apps? "Crap code" is a combination of two things:* Not poorly written apps so much as generically written apps; i.e. not optimized for the particular platform they're running on (there's not much any hardware can do for sloppy coding), and;* Less than stellar compiler output. gcc's PowerPC output was famously poor before Apple and IBM got their hands on it. As of right now, it's... better. It can't even begin to compare to hand-tooled assembly, though. (Basically, what a compiler does is translate a textual language written to be understandable to a human - well, most of the time - into a stream of bytes understandable to a CPU. The better the compiler, the more reliably and efficiently it does this, and the better the output runs on the machine.)Mot designs the way they do because embedded code is frequently written well, carefully tailored to the platform, and hand-tooled (or at least, very carefully profiled and tweaked to overcome a compiler's limitations). Given this, Mot can leave off a lot of the tricks that the Pentium4 and 970 use and ship a small, lean CPU.One of the issues facing any desktop or workstation chip is that applications programmers are generally far less concerned with such issues, and they tend to use compilers like gcc that are concerned with other issues than producing optimal executable code (such as a platform- and language- agnostic structure). The G3 and the G4 did not handle this kind of code all that well (they didn't do poorly, either, they just didn't try to compensate). The 970, like the Pentium 4, throws all kinds of logic at the problem of making sloppy code run well. It means a bigger, uglier CPU, but it's an eminently pragmatic tradeoff in the 970's intended market. Anything that's going to run Linux had better be good at running crap code. Reply 14 of 98 t'hain esh kelch Posts: 1,274member May 16, 2003 3:01PM Quote:Originally posted by Panther Have a nice day. [/B]Haha, nice zig! And nice song! Reply 15 of 98 fred_lj Posts: 607member May 16, 2003 11:01PM So, in extreme layman's terms, this means it's pretty much a 90% probability (or higher) that the next iteration of Power Macs will have the 970 inside? Or at least, if Panther/other apps are written to accomdate it, then, something HAS to be coming.I want reassurance, that's all. I want pictures that are stripped off the fiend's website by angry attorneys!But this is cause for woo-hooing!!!...Right? Reply 16 of 98 ghost_user_name Posts: 22,667member May 16, 2003 11:18PM Quote:Originally posted by fred_lj So, in extreme layman's terms, this means it's pretty much a 90% probability (or higher) that the next iteration of Power Macs will have the 970 inside? Actually, this says absolutely nothing about the next PowerMacs using 970s.All this says is that GCC is readying to compile code for the 970 processor. Nothing more. Remember that GCC is used by all kinds of people and that IBM is a Linux vendor. It's just as likely that the GCC updates are so IBM can compile Linux and other sofware for the 970s.People here just like making huge leaps to conclusions regarding Apple.*reality check* Reply 17 of 98 davegee Posts: 2,765member May 16, 2003 11:46PM Quote:Originally posted by Brad People here just like making huge leaps to conclusions regarding Apple.Brad,You should know better than anyone... If it wasn't for making huge leaps atleast half of us would get ANY exercise at all! D Reply 18 of 98 thai moof Posts: 76member May 17, 2003 12:33AM Quote:Originally posted by Marcus Here... Don't know if this is significant or not... Marcus(Mods - Please delete/lock if this is useless info.) There is something I don't understand, maybe one of the tech-heads can clarify it. On the posted document, it references Power4 this...Power4 that, but not 970 (though it mentions 970 at the top). Does this mean that it is sort of generically for the 970, but not specifically? Or does it mean that the 970 is treated (by the compiler) as exactly the same as the Power4?Also, I think Amorph is exactly right about the open source nature of gcc being a big advantage, particularly for Apple. Metrowerks saved Apple (period) when Apple switched to the PPC by having a compiler ready. It is great that Apple can rely on an open source solution for such a critical element, as it seems that gcc can be updated faster than if it was owned by someone. Heaven knows that Metrowerks is not updating their Mac compilers as aggressively as they used to...now that they are owned by Moto! Reply 19 of 98 shaktai Posts: 157member May 17, 2003 1:16AM Quote:Originally posted by Marcus Here... Don't know if this is significant or not... I think it is significant, as already pointed out in other terms. Very significant. Reply 20 of 98 jupiter Posts: 18member May 17, 2003 4:02AM These clues pretty much sum up what's going on...GCC Power4 and 970 specification GCC mailing list with Apple employee contributing Apple is contributing POWER4/970 code to GCC, they must be doing that for a reason.Just afew weeks till we know all. «12345» Sign In or Register to comment.