------------------------------------- Message #461 in area AMY_TECH From: David Jones Sent To: Adam Sternberg 27 Jun 92 19:49:55 Subj: Re: (Reply to #272) (Has reply #1288) In a message posted 1992-Jun-28 7:16:42, Adam Sternberg wrote: >want to know WHY the Amiga will just do it faster! If there is anybody >out there that can help me out with my problem, I would greatly >appreciate it. > Motorola chips are faster than their Intel counterparts because... 1. More registers. The 680x0 has 8 data registers (32 bit) and 7 address registers (32 bit) plus a stack pointer (32 bit). The 80486 has 4 data registers (32 bit) which can also be used for address, and 3 address registers (32 bit), plus a stack pointer. Overall, there are twice as many registers on the 68000, which means that you can go longer before your compiler has to store variables on the stack. In addition, the 680x0 registers are all general-purpose whereas the Intel registers are special purpose. For example, certain arithmetic instructions work only on EAX. The Intel equivalent of the DBRA instruction works only on ECX. And some addressing modes used by 16-bit MS-DOS software (which is what most of these machines run) work only with BX. Remember that when you're running DOS, the 32 bit nature of the 486 is lost. It's as if the 486 was only a 16 bit chip. 2. No segmentation. Segmentation is not a problem with the 486 if you are running a true 32-bit operating system such as UNIX. However, under DOS, a pointer is only 16 bits wide, and then there's a "segment" part that's also 16 bits wide. In effect, programs that access more than 64K of data must juggle "segment registers" which is really generating 32 bit addresses on a 16 bit machine. Remember the 6502? --- Juliet b1.76 * Origin: AMY_TECH moderator (1:163/109.8) ------------------------------------- Message #464 in area AMY_TECH From: Kevin A. D. Carter Sent To: Adam Sternberg 28 Jun 92 16:32:41 Subj: Amiga vs. Intel (Has reply #567) In a message of <26-Jun-92 02:12:10> Adam Sternberg (1:209/700) wrote: AS> ... I have heard by many Amiga AS> enthusiasts and magazine articles that the 25 mhz Amiga 3000 (68030) AS> is faster at grafic work than a 33 mhz, IBM 486 DX. I am curious how AS> this can be true when the 486 is 2-3 mips faster AND is 8 mhz faster AS> than the Amiga. A 68030 @25MHz in *not* "faster" than an 80486 @33MHz or even 25MHz (which many people think) however the an A3000 is faster than standard '486 clones in many areas, including graphic manipulation. As I understand it from Skipper Smith from Motorola (wish he used this conference instead of Amiga International) and a few other sources: A 68030 *IS* faster (or at least equivalent to) a 60386 @50MHz due to the way the chip opperates. A Motorola processor polls and updates (wrong terms I know) every cycle whereas Intel-based processors "poll" one clock cycle and "update" the next. This makes the Motorola equivalent processor nearly twice as fast as its Intel counterpart. As of the 60486 line of processors this is no longer true. A 68030 @33MHz and a 80486 @33MHz should process at approximately the same rate given similar operating systems and computer architecture. This is where the stock Amiga remains superior. Our custom chipset can handle and manipulate graphics faster than a "standard" '486. There are, however, new IBM compatibles with dedicated video slots and there are accelerated video cards now available. As software support for these cards increases and the hardware becomes cheaper there will be fewer and fewer advantages to owning an Amiga. Experience, support, and quality software and hardware is what will keep the Amiga alive in the future as the technological gap closes between Amiga and IBM compatible architecture. If they ever manage to write a decent Windows or OS/2, however, we may be in trouble. Kevin A. D. Carter --- * Origin: Kevin's Private Point via CONAN (London, Ontario) (1:221/125.37) ------------------------------------- Message #567 in area AMY_TECH From: Michael Mac Nessa Sent To: Brian Dhatt 01 Jul 92 17:36:00 Subj: Re: Amiga vs. Intel (Reply to #464) (Has reply #691) In a message dated 30 Jun 92 15:04:00, Brian Dhatt wrote: BD> That is what is happening now. OS/2 2.0 has true multitasking, and is BD> 32-bit. 800,000 copies sold since March 31, 1992. Microsoft soon BD> will be releasing thier 32-bit OS, Windows NT. Just telling you, so BD> you could, "watch out!" OS/2's GUI requires 4 megs to work properly, and that's cramping it. It really wants about 8 Megs to REALLY do all it's advertized to do. Windows NT is proported to need even MORE ram. Sheesh...glad I don't own an IBM compatable. --- DLG Pro v0.992/DLGMail * Origin: Emerald Keep - Only Amiga Makes It Possible. (1:2250/2) ------------------------------------- Message #691 in area AMY_TECH From: Ed Brown Sent To: Michael Mac Nessa 4 Jul 92 08:48:08 Subj: Re: Amiga vs. Intel (Reply to #567) (Has reply #712) > OS/2's GUI requires 4 megs to work properly... Ever notice just how many Amigas have >4 megs of ram? Is there a reason for that? You betcha and the PeeCee is no different. Sure, we *can* limp along with a meg of ram, but who the heck wants ta??? Most A3000s have 6 megs of ram but if you start throwing 24-bit pix around with AdPro, you just might wish you had 10 megs instead. That was the *only* time I ever thought that 6 megs wasn't enough. I'm sure that similar things occur on the PeeCees... especially when running OS/2, which is a substantial OS, in many ways. -=- Ed --- TransAmiga BBS v1.11g1 * Origin: Powers Of Two BBS Vancouver,WA,USA! (206)750-9703 (1:105/135) ------------------------------------- Message #731 in area AMY_TECH From: Mike Manzano Sent To: Adam Sternberg 04 Jul 92 13:25:26 Subj: Re: Amiga vs. Intel (Reply to #712) (Has reply #793) The funny thing is, OS/2 still has the Windows idea of (um, I forgot what they call it, I think they call it resources or somethin'), whereby each app can only use so much of it, or other apps won't work right. In other words, the address space for displaying windows is limited by the creater's of the OS, unlike the Amiga, which is only limited by the amount of CHIP MEM, and nothing hard-coded into the OS. --- Star-Net v1.02 * Origin: USS Enterprise -Warp Speed Ahead- 206-377-3170 (1:350/32.0) ------------------------------------- Message #803 in area AMY_TECH From: Brian McGroarty Sent To: Brent Burkholder 04 Jul 92 00:22:11 Subj: Re: Amiga vs. Intel (Reply to #793) (Has reply #806) > Actually, what you probably heard was that an Amiga with an 030 > would do graphics work faster, ala animations. This is true because > of the blitter. The '030 is left free to do the uncompression of the > anim file and the blitter displays it. I just finished an animation > that essentially moves the entire screen at once, 40 frames, 3 megs > and some change in delta compression. The fact of the matter is that > an IBM couldn't display the same thing if all it was doing was > pageflipping the stills, its hardware just isn't geared for it > (heaven forbid they should actually do delta-compression.) Deluxe Animator for the IBM does delta-compression quite well. I use it extensively. On another note, while an IBM system might exectue a greater number of instructions in a second, many Motorola instructions correspond to multiple Intel instructions, and the Motorola has considerably more registers to play in than the 8086/8 family does, resulting in fewer memory accesses and, therefore, greater speed... Gadz, I love the Amiga - now if people would pay me to program for that beast instead of the PC I'd be in heaven... --- Star-Net v1.02 * Origin: Outlaw Tech. Zone *HST* (ADS) 815/654-9729 (1:2210/654.0) ------------------------------------- Message #806 in area AMY_TECH From: Brian McGroarty Sent To: Steve Lee 04 Jul 92 00:37:28 Subj: Re: Amiga vs. Intel (Reply to #803) (Has reply #807) > First off, the MHz don't mean a thing unless you are comparing > the same chip, I'm not sure exactly if the 68030 is faster than a Even then, you frequently have to take RAM speeds into consideration; it gets very messy... > 486, but if the 486 has more MIPS (millions of instructions per > second) than that would in fact indicate that the 486 is faster in > that area, MFLOPS (mega-floating point operations) tell how fast > the chip can handle math operations (also, if either has a FPU > [floating point unit, a math coprocessor] that will in fact make it > unfair)... I do know for a fact that the 68040 at 25 mhz is faster > than the 486 at 33 mhz, mayby that is what you mean? (The 68040 is > basically a 68030 with a built in 68882 math coprocessor, not sure > bout the 486), on the subject of ray-tracing, that mostly depends The 486 is little more than a 386 with the math co-processor on board, a larger cache and some instruction piping; the fun part is that there are certain applications where the 386 is FASTER than the 486 (although not by much...) - I think (I may be wrong here, I'll try to double-check this (in the mean-time, someone scream at me if I'm totally blundering)) that the 68040 actually reduces the number of cycles required for many instructions, brings a faster math co-processor on board, reduces the number of page-sizes available to the MMU, pulls in a larger cache and super-intelligent instruction piping... big leap for Motorola, baby-steps for Intel... > on the software, but some chips take more > clock cycles to perform math operations than others (ex: the 65xx in > a Commodore 64 takes less cycles than a 808x to perform simple math > operations [addition & subtraction]) Bingo. Check out a multi-register stack-push on the Motorola and write the equivalent Intel code... you can go for over two-dozen instructions on the Intel compared to just 1 on the Motorola. Blah. --- Star-Net v1.02 * Origin: Outlaw Tech. Zone *HST* (ADS) 815/654-9729 (1:2210/654.0) ------------------------------------- Message #819 in area AMY_TECH From: Frank Perricone Sent To: Ed Brown Tue 7 Jul 92 18:32 Subj: Re: Amiga vs. Intel (Reply to #817) (Has reply #837) In a message dated 4 Jul 92 08:48:08, Ed Brown wrote: > OS/2's GUI requires 4 megs to work properly... EB> Ever notice just how many Amigas have >4 megs of ram? Is EB> there a reason for that? You betcha and the PeeCee is no EB> different. Sure, we *can* limp along with a meg of ram, but EB> who the heck wants ta??? Most A3000s have 6 megs of ram but Not quite true. I have 4.5M here on my Amiga, and 8M on my OS/2 PeeCee at work, and believe you me, the Amiga has a LOT more memory. The OS/2 machine is "limping along" at 4M. Seems you're assuming the "required" 4M of OS/2 is enough to get some work done in, as well, which it definitely isn't. EB> if you start throwing 24-bit pix around with AdPro, you just EB> might wish you had 10 megs instead. That was the *only* time Graphics always eat up lots of memory on any platform, true. EB> I ever thought that 6 megs wasn't enough. I'm sure that EB> similar things occur on the PeeCees... especially when running EB> OS/2, which is a substantial OS, in many ways. I never run out at 4M at home, but my 8M at work seems cramped. I see the system going into virtual memory sometimes when I'm only running two big apps and a few little ones. At home I run a Fido front door, two nodes of a BBS, and usually something else like a compiler plus literally dozens of little apps, with no problem. "We are young... learning that we're only immortal for a limited time." --- Falcon CBCS v1.00r * Origin: Crystal Ship, Douglas Alaska, 907-364-3362 (1:355/814) ------------------------------------- Message #837 in area AMY_TECH From: Frank Lazar Sent To: Brent Burkholder 06 Jul 92 11:37:20 Subj: Re: Amiga vs. Intel (Reply to #819) (Has reply #913) In a message posted <28 Jun 92 00:23:00>, Brent Burkholder writes: > In a message dated 26 Jun 92 02:12:10, Adam Sternberg wrote: > the A3000 > AS> will do ray-tracing faster, but how can this be when all > ray tracing > AS> is is number-crunching which, I would assume, the 486 would Actually, the question also is Motorola vs. Intel. A very basic difference is the base chips from which all trace their lineage from, the 68000 series vs. the 8088. The 68000 from the start had 32-bit hooks built into it, which is why the Amiga, Macintosh, and the ST were properly considered 16/32 bit machines. The first Intel PC's however were 8-bit design which among other things set up that addressable 640k address limit. Even the 386/486 chips are actually 32/16 whereas the Motorolas are the other way around. For that matter, while I'm not completely sure, but ray-tracers on the Macintosh also outpace the IBM/clone versions as well. And at least part of the reason is memory addressing. --- RMB v1.30 / FastPoint 1.39 * Origin: The Hermit's Hut (1:2605/611.2) ------------------------------------- Message #839 in area AMY_TECH From: Frank Lazar Sent To: Michael Mac Nessa 06 Jul 92 11:47:28 Subj: Re: hmm (Reply to #725) (Has reply #886) In a message posted <31 Jun 92 23:25:00>, Michael Mac Nessa writes: > In a message dated 27 Jun 92 00:46:00, Todd Carpenter wrote: > > TC> I heard for a guy in a local computer store that CBM is > putting out a > TC> new complete chip set to get the Amiga up to par with the > faster > TC> IBMer's can anyone tell me if this is true or not and if > true whats > TC> the specs, or release date on it. thanks. Todd Carpenter > > It's the "AA" chipset, and it surpasses the IBM's graphics. > No release date as of yet, but I hear rumors that it will need > a new Workbench/AmigaDOS to utilize them. > Let's not compare Apples and oranges here. The IBMers are only faster if you're comparing a 68000 500 or 2000 up against a 386 or above. Now slot in a decent '030 or one of the new '040 boards and the shoe is very much on the other foot, even while doing something you can't do yet with Windows or System 7, i.e. MULTITASKING. As far as the "new chipset" goes, there is nothing but rumors at this point. My suspicions that if present Amigas are going to use a new chipset, it would probably require replacement of motherboards as opposed to a moderate chip upgrade like ECS. --- RMB v1.30 / FastPoint 1.39 * Origin: The Hermit's Hut (1:2605/611.2) ------------------------------------- Message #913 in area AMY_TECH From: Rafal Wiosna Sent To: Jack Radigan 05 Jul 92 12:54:00 Subj: Amiga vs. Intel (Reply to #837) (Has reply #1046) >< Jack! >< rian Dhatt zostal rozjechany w liscie: JR> The 386SX has a 16bit data bus. The 486SX has a 32bit data bus, but no JR> working FPU inside. Hehe, even better: 487 is a 486DX with changes in pin - assigment. So if you buy 486SX and then 487 you get two 486SX + math coprocessor. 8^) Maybe some Taiwan producers will learn how to use 487 as a main processor (487 is cheaper than 486DX). INTEL'S PHENOMENA... Bye4Now... Rafal Wiosna --- GeeMail 1.0 * Origin: Never stand between a fire hydrant and a dog (2:480/19.1) ------------------------------------- Message #1046 in area AMY_TECH From: Jack Radigan Sent To: Rafal Wiosna 12 Jul 92 17:36:17 Subj: Amiga vs. Intel (Reply to #913) (Has reply #1088) In a message of , Rafal Wiosna (2:480/19.1) writes: JR> The 386SX has a 16bit data bus. The 486SX has a 32bit data bus, but no JR> working FPU inside. > Hehe, even better: 487 is a 486DX with changes in pin - assigment. >So if you buy 486SX and then 487 you get two 486SX + math coprocessor. 8^) >Maybe some Taiwan producers will learn how to use 487 as a main processor >(487 is cheaper than 486DX). INTEL'S PHENOMENA... No, there's no way you can get the cpu to work in a 487, that's why it's one in the first place, the cpu is dead. Same goes for a 486SX, the math coprocessor doesn't work. -jack- --- msged 2.07 * Origin: Dynalogic Support BBS * 609-625-2453 (1:266/61) ------------------------------------- Message #1187 in area AMY_TECH From: Jack Radigan Sent To: Ken Tylman 16 Jul 92 20:24:58 Subj: Re: Amiga vs. Intel (Reply to #1088) In a message of , Ken Tylman (1:373/8) writes: >I read that Intel went into one of the 486 chips, maybe it was the SX or >could have been the DX, to purposely disable the Coprocessor. They did >this because they said some users don't want one. So, they did more work >on the chip so they could charge less money. Bright, huh? Not really. When the 486 was first put into production it suffered from low yields, most chips do until the kinks in wafer manufacture for dense chips like the 486 are ironed out. That's primarily why their initial price is so high too. Anyways, you have to give Intel credit for thinking up the marketing scheme to sell partially damaged chips as 486SX (no coprocessor) or 487 (no cpu) chips. It allowed them to deal with a low yeilds and still make money on otherwise useless silicon. If you notice, Motorola followed suit with the coprocessor-less variant of the '030, it makes marketing sense to do so. The 486SX created a market that now probably requires Intel to disable working coprocessors in order to fill demand for the chip. Either way, they're not losing out. -jack- --- msged 2.07 * Origin: Dynalogic Support BBS * 609-625-2453 (1:266/61) ------------------------------------- Message #1189 in area AMY_TECH From: Jack Radigan Sent To: Jay Grizzard 16 Jul 92 20:35:53 Subj: Re: Amiga vs. intel (Reply to #1098) In a message of , Jay Grizzard (1:379/100) writes: >actually, I *believe* that's incorrect. I remember reading that the >487 *IS* a fully functional CPU that shuts down the original CPU, >because if the FPU was a completely sperate unit, "the propogation >delays would negate any performance increase of the FPU". Ah, but is the original cpu *always* shut down, or is it only halted while the coprocessor is executing a computation? I think you'll find that the 487 acts just like a traditional coprocessor, it takes control of the bus while performing a calculation, or series of calculations, but releases control back to the main cpu otherwise. It also doesn't need a *fully* functional cpu in order to do this either. Some subset of operations have to be working, but not the entire cpu. -jack- --- msged 2.07 * Origin: Dynalogic Support BBS * 609-625-2453 (1:266/61) ------------------------------------- Message #1214 in area AMY_TECH From: Dean Ridgway Sent To: Ken Tylman 18 Jul 92 23:08:14 Subj: Re: Amiga vs. Intel (Reply to #1088) (Has reply #1251) In a message posted <14 Jul 92 01:29:14>, Ken Tylman writes: -=>I read that Intel went into one of the 486 chips, maybe it was -=>the SX or could have been the DX, to purposely disable the -=>Coprocessor. They did this because they said some users don't -=>want one. So, they did more work on the chip so they could -=>charge less money. Bright, huh? Actually I think that these chips just failed their quality control checks on the MPU and Intel just wanted to cut their losses and market them anyway. Dean Ridgway | FidoNet 1:357/1.103 | InterNet ridgwad@jacobs.cs.orst.edu | CIS 73225,512 | ridgwad@atlantis.cs.orst.edu --- RMB v1.30 * Origin: Leopard's Lair [HST] - I heard it thru the Hotline! (1:357/1.103)