CompuServe Thread

#OS3.1 – 68020+

16 messages in this thread
#209333From: Ian J. EinmanAug 22, 1995 2:00 PM
Well, the 68020 has a lot of advantages internally, and they (unfortunately) are most useful with things like compression/decompression, which datatyoes datatypes use in the picture formats (JPG, GIF, TIFF). There are few things that require 030, 040, 060 etc. but the extra instructions available with the 020 are sufficient to nudge programmers in that direction. If you really want to extend the life of your A500, I'd recommend an 020 or 030 accelerator, and a 68881 or 882 math coprocessor. Even a fairly low speed 68020 would give you a jump over your current configuration, consider a 68020 at 14 MHz, like the a1200, is equivalent to a 28MHz 68000. If you can find a board, 68020 and 68030 just aren't expensive anymore, the problem would be getting a board from someone. But I would do it if I were you, since they are hard to find for the A500. Don't be surprised if a future version of the OS does not even support the 68000 at all. If you get a 020/881 combo, you will have a lot more speed, compatibility, and ability, and your A500 will last you quite a while longer. If you don't you will see more and more people release code requiring the 68020, simply because its more efficient and faster code (and smaller), and nowadays a huge number of users have an '020 or better, as will any potential new Amiga users, since all 1200s and 4000s will have 68020s – 68060 chips.
#209436From: Phillip WoollerAug 23, 1995 10:41 AM
I understand that the 020 is better than a 68k, in fact I have an a1200. But I need to upgrade the memory and disk and monitor before I can use it. I might also get an 060 upgrade, I might wait and buy an a4000 though. If you could buy a graphics card for the a1200 then I would definately (probably) stick with it. What is needed is a 16bit sound board and 64bit graphics adaptor to fit in the pcmcia slot 🙂 My a500 is about to die, I have to keep the lid off so I can reseat the chips (regularly). I don't want to upgrade it, but it works most of the time anyway. You said that new versions of the OS might not support 68k, don't hold your breath for new versions of the OS though 🙂 Phill
#209451From: Ian J. EinmanAug 23, 1995 1:52 PM
>>don't hold your breath for new versions of the OS though I think one revision is fairly guaranteed – mainly because Commodore had already had some stuff for it, and Escom started working on it right away. But development after that… we'll see. But I would be fairly certain Escom will make at least ONE version with their name on it. Note that all new 1200s and 4000s made will come with Workbench 3.1. They say it has been "optimized for speed", but nothing new added.
#209543From: Phillip WoollerAug 24, 1995 5:51 AM
I didn't know that ESCOM had started work on a OS revision, don't be suprised though if the OS in the 1200/4000 is the same 3.1 that you can buy today. OS3.1 is faster than 3.0 at some things (slower at others), mainly bug fixes. Commodore had alot of things in development, most will not be picked up by escom though. AAA would be good :-). The one thing I never heard about AAA is whether it was AGA/ECS compatible. If it was and CBM had prototypes working I can't see why we can't have a machine now with OS support later. Phill
#209596From: Ian J. EinmanAug 24, 1995 9:45 PM
Actually Escom themselves hasn't been working on a new OS. They have been subletting the work out to other people. Some people at Scala have been working on the look and feel of the new OS, including some ex-CBM employees. Latest news is that Escom is going to have another company do all software – not owned by Escom. This is actually probably more good news than bad. We'll see. AAA probably won't be continued. It was AGA/ECS compatible – but not plug compatible. A "buy a new computer" thing again. Existing OS would work well with it./exit I'm too dumb to figure out how to issue command to this thing. Who is reading what I am typing? post
#209607From: Norman AustinAug 24, 1995 10:41 PM
I am Ian, and thanks for the posts. Regards, Norman Austin 70743,1535
#209681From: Tom BarkerAug 25, 1995 4:51 PM
Why Ian…I always read your posts!<g> Tom
#209718From: Asha DeVelderAug 26, 1995 1:24 AM
Ian, <Who is reading what I am typing?> I am, and I suspect a lot of folks are! Thanks for doing it (the typeing that is…)!! Asha @}–`–,-
#209723From: Ian J. EinmanAug 26, 1995 2:05 AM
Well, I was trying to conference with someone. It got out in the wrong place.
#209729From: Rafael VicariaAug 26, 1995 4:37 AM
I am too, Ian. Your posts are greatly apreciated. You really rise s/n ratio here, really. >/exit I'm too dumb to figure out how to issue command to this thing. Who is reading what I am typing?< Are you replying online? Warmest regards Rafa <Motorola Inside> from Berlin, thanks to AP1.80 >…olive oil is the way to go.<
#209604From: Asha DeVelderAug 24, 1995 10:32 PM
Phillip, < The one thing I never heard about AAA is whether it was AGA/ECS compatible. If it was and CBM had prototypes working I can't see why we can't have a machine now with OS support later.> According to Lew Eggebrecht at the '92 WOA in Pasadena, AAA was not backwards compatible with AGA, but it was with OCS/ECS. I suppose that's all moot now as it doesn't look like AmigaTech is going to go with the AAA chipset, but rather go directly to RISC. Asha @}–`–,-
#209771From: John G. AtaAug 26, 1995 3:50 PM
Ian, You write: >consider >a 68020 at 14 MHz, like the a1200, is equivalent to a 28MHz 68000. However, the SupraTurbo 28 MZ 68000 accelerator that I have seems to have performance numbers that are much better than the A1200 (14Mz 68EC20). This is using Lamonte Koop's AIBB 6.1. This is accomplished by having a special on-board cache for the 68000. Couldn't beat the price either. <G> For example: AIBBLogFile **************************************************************************** System Static Data —————————————————————————- System CPU: 68000 System FPU: NONE System MMU: NONE CPU Clock Rate: 28.0 MHz FPU Clock Rate: ——– Operating System Version: 37.175 ( 2.x ) ======================= System Memory Information ========================== NODE #1 Node Name: expansion memory Node Size: 4.00 MBytes Bus Port Size: 16 Bit Address Range: $00200000–$00600000 Node Priority: 0 —————————————————————————- NODE #2 Node Name: expansion memory Node Size: 2.00 MBytes Bus Port Size: 16 Bit Address Range: $00600000–$00800000 Node Priority: 0 —————————————————————————- NODE #3 Node Name: chip memory Node Size: 2.00 MBytes Bus Port Size: 16 Bit Address Range: $00000400–$00200000 Node Priority: -10 —————————————————————————- ************************************************************************* ** ========================== Test Information Section ======================== Test Name: EmuTest Parameters: 68000 based standard code Multitasking: DISABLED Test Code Location: Memory Node #1 Test Data Location: Memory Node #1 Test Result: 4.26 simulated MHz (Higher = Better Performance) ============================ Comparison Ratings ============================ This Machine : 3.09 A2000_FU : 1.00 (68000 | SC Math) Base System A1200-NF : 1.82 (68000 | SC Math) A3000-25 : 5.03 (68000 | SC Math) A2000_FT : 3.08 (68000 | SC Math) ————— System Dynamic Information for this test ——————- Instruction Cache: ——– Data Cache: ——– Inst Cache Burst Mode: ——– Data Cache Burst Mode: ——– Data Cache Write Allocate: ——– ————————————————————————- — System Supervisor Stack Address: $00200A08 AIBB Process Stack Address: $002C98B4 Operating System Memory Location: $00F80000 Current OS Dispatch Quantum: 4 *************************************************************************** END OF LOG FILE John G. Ata Sterling, VA
#209804From: Ian J. EinmanAug 27, 1995 1:13 AM
My reply to this: ——————– >consider >a 68020 at 14 MHz, like the a1200, is equivalent to a 28MHz 68000. However, the SupraTurbo 28 MZ 68000 accelerator that I have seems to have performance numbers that are much better than the A1200 (14Mz 68EC20). ——————— You're right and I'm wrong for a couple reasons. The accelerator you have, with a cache, makes the 68000 at 28MHz have MORE than 4 times a stock 68000. I wasn't aware of a cache, this would basically take away the 68020 advantage. The 68030 has two caches, its main advantage over the 68020. The other major mistake I made is that I FORGOT the CPU in a 1200 isn't a 68020, its an EC020 and does not access memory to a full 32 bits in all cases, reducing the "double speed" apparent advantage a 68020 has on a 68000. I still believe my comparison would be true with a full 68020 versus a cacheless 68000, but since neither of those is true, the comparison was inaccurate. I do believe your computer is faster than a 1200 (but not the one next to me, with a 40MHz 68030 :-> I hate those 68EC0x0 chips, they are degraded a lot more than they would have you believe. The 3000 wouldn't even work with a 68EC030. The EC020 isn't missing much – but is 24-bit memory address space, like a 68000. The EC030 and EC040 are missing the MMU, confusing to the user since you will run a Virtual Memory program, and it won't work, even though you have a 68030. But you don't have a 68030… And worse there is a version, of the 68040 without the FPU. Great, how do you add it on later. You can somehow try to add a 68882, but no program will use it right because they think you have a 68040… The EC chips are throw-away processors. You will always upgrade them eventually, it seems.
#209915From: John G. AtaAug 27, 1995 10:42 PM
Ian, You're right about the EC chips. I never understood why anyone would ever upgrade their computer with one, but that seemed the way a lot of accelerators were going a while back. Take care, John G. Ata Sterling, VA
#210017From: Brian BartlettAug 29, 1995 12:16 AM
John, they were cheaper? Given that you have quite a bit less to test on the chip prior to releasing the chip to the retailer/OEM, it saves resources that the manufacturer could spend elsewhere, such as on developing a PPC. Given that savings, and that many users don't seem to care much about having an MMU, after all what good is one on an Amiga?, you charge less, and sell more. Sorry, it makes perfect economic sense here. Now if you want to run VMM, NetBSD, or some other very strange thing, then you need to upgrade. For the average game user, which is characteristic of our community (remember?), who cares. Sell 'em a cheap chip which reduces the end user cost and sells more machines. It's us "power-users" that need the MMU, or indeed, more than a 24-bit address space. Brian Been there…. Done that…. Better.
#210118From: John G. AtaAug 29, 1995 9:45 PM
Brian, You write: >they were cheaper? Yes, but if someone was going to shell out close to 1000 bucks (which is what they seemed to have cost back then), it didn't make sense to me to pay a little less and wind up without an MMU. And then later on, these people would find out they couldn't run Emplant, Gigamem, or like product with the Amiga OS. For example, I think that GVP had an EC30 offering which really wasn't very cheap. Now if the EC accelerators were markedly cheaper (say around 1/5 to 1/2 the cost), it would make sense, IMHO. Take care, John G. Ata Sterling, VA