whitewolf@gnh-starport.cts.com (Tae Song) (05/09/90)
Is the Apple II old??? Well, maybe the 8-bit Apples... but certainly NOT the GS! The issue isn't weather the GS is old or not, but about speed. It's got to be one of the slowest machines around with a clock speed of 2.5-2.7Mhz. That's only little more than half the clock speed of an IBM PC (4.77Mhz). The 65816 is extremely poorly designed in many ways. It only has 3 registers, lacks lot of logic circuits to shave cycles on many of the instructions. Has trouble running at high speeds. The GS design isn't so hot either... somewhere the GS engineering must've gotten confused. The GS can be expanded to 8M, but only 4 of those are DMA compatible. They must've thought of limiting to 4M like on the Macs... we (Apple) won't want a computer that can hold more memory than a Mac would we (Marketing). Almost every I/O operations must be bottlenecked to 1Mhz, including/during DMA accesses and even to built-in peripherals and video. 2.5 is a small# in any case... and if ASIC pulls through with the 20+Mhz 65816, I can't see WHY Apple couldn't make a killer machine. Even at 2.5Mhz the GS puts in a good show... 20Mhz is 8x faster... if Apple let's the engineers wring all the speed out with caches, DMAs, etc... I think seriously it would give EVERY machine a run for their money. Ok, I'm not a graceful writer... heck, all my English teachers would attest to that but I just hope you ALL understand what I'm trying to say. The GS is running at all of 2.5Mhz!!! It only takes two intelligent people to push that # up to maybe 20-25Mhz... then you'd better watch out!
toddpw@tybalt.caltech.edu (Todd P. Whitesel) (05/10/90)
whitewolf@gnh-starport.cts.com (Tae Song) writes: > The 65816 is extremely poorly designed in many ways. depends on your point of view, really. >It only has 3 registers, lacks lot of logic circuits to shave cycles on many of >the instructions. Has trouble running at high speeds. 3 registers: that's what direct page is for. There aren't a lot of internal registers, but the 65xxx can get at direct page and stack very fast. Since these often fit entirely in the data cache the higher clock speeds largely offset the lack of registers. Lack of cycle shaving: this is something Mensch would love to fix but until someone rich wants him to build the 65x32 he is not about to risk the money. Trouble at high speeds: I maintain that this is due to Mensch's "kitchen table" mask. I am fairly convinced that economics is what prevents him from fixing it. >The GS design isn't so hot either... somewhere the GS engineering must've >gotten confused. The GS can be expanded to 8M, but only 4 of those are DMA >compatible. They must've thought of limiting to 4M like on the Macs... we >(Apple) won't want a computer that can hold more memory than a Mac would we >(Marketing). Almost every I/O operations must be bottlenecked to 1Mhz, >including/during DMA accesses and even to built-in peripherals and video. All true. And all so easy to fix. >2.5 is a small# in any case... and if ASIC pulls through with the 20+Mhz 65816, >I can't see WHY Apple couldn't make a killer machine. Even at 2.5Mhz the GS >puts in a good show... 20Mhz is 8x faster... if Apple let's the engineers wring >all the speed out with caches, DMAs, etc... I think seriously it would give >EVERY machine a run for their money. My point exactly. The "Apple //f" paper I wrote proposes one such machine. Apple already has most of the technology, they just need to be willing to put it into the new design. > The GS is running at all of 2.5Mhz!!! It only takes two intelligent people >to push that # up to maybe 20-25Mhz... then you'd better watch out! It took two people to design and lay out the 65816. It's taken two others with industry connections to reimplement it with state of the art technology. It's about time the Apple // kicked some butt for a change. Todd Whitesel toddpw @ tybalt.caltech.edu
crew@pro-harvest.cts.com (Chris Wicklein) (05/22/90)
In-Reply-To: message from THINGVOL@LAX.WISC.EDU > I have heard stories about faster 65816's. The designer, Mr. Mensch, had the > chip going 40MHz and over 100MHz with GaAs. Wow! Where _did_ you hear that? Either something weird is afoot, or you heard wrong. I haven't heard about Mensch converting to GaAs arrays, but he once said he could/would/might if someone was interested. Maybe someone is. Those speeds sound possible, but unlikely. They would be on par with a 150/375MHz 8086 respectively, at least by the benchmarks I've seen. (Boy, benchmarks are weird things. I can't see anyone building (even if it was possible) a 375MHz 8086 machine.) UUCP: crash!pro-harvest!crew ProLine: crew@pro-harvest ARPA: crash!pro-harvest!crew@nosc.mil INET: crew@pro-harvest.cts.com BITNET: crew%pro-harvest.cts.com