Article View: comp.sys.ibm.ps2
Article #5Re: ----- MORE GOOD QUESTIONS & INFO ON UPGRADING A PS/2 50Z -----
From: Darius Vaskelis
Date: Tue, 04 Aug 1992 13:49
Date: Tue, 04 Aug 1992 13:49
122 lines
6485 bytes
6485 bytes
Let me preface my remarks that I'm not a definative source of information on Evergreen, Kingston, or Aox. If I get any facts wrong, please correct me. >Some specs on my machine should be in order. I have an IBM PS/2 50Z (10 Mhz) >with 10 Mb RAM and a 60 Mb hard drive, running PC/DOS 4.01 (but I'm planning >to upgrade to 5.0 very soon). > >Up until today, I thought AOX was the only company that provides 486 >upgrades for 286 PS/2s. I found out today that Evergreen Technologies >also have a similar upgrade option called the Evergreen 486 SuperChip, which >by the way sells for $399 ... much cheaper than the MicroMaster (I think it >was in the 4 digit range the last time I've checked). There's gotta be some >major differences then ... anybody knows? > >Unlike the AOX 486 MicroMaster card (which takes up an expansion slot), the >Evergreen chip would replace my 286 chip sitting in the PLCC socket of my >computer. It claims : > > In less than 10 minutes, your 16-bit processor is converted to a 32-bit > processor which can implement the 386's virtual 8086 mode an much > more! Well, my understand is this: The Evergreen 486 CPU uses a Cyrix Cx486SLC. I _think_ it runs at 25MHz. The CPU itself is essentially a 386SX with the handful of added 486 intructions and 1K cache that's not nearly as sophisticated as the cache in the Intel 486, or IBM 386SLC/486SLC2. The chip has no floating point math capabilities by itself (like the Intel 486SX or IBM SLCs.) It's pretty simple, pull out your 80286, and plop in a small board with the Cx486SLC in its place. I'd say the minuses are: (1) Strange math coprocessor options If you buy the model without the math coprocessor, it appears to me that there's no room to add one later. I think you have the spend the money for the math coprocessor model up front. My guess would be that the math coprocessor is a Cx83S87, which is not as fast as the math coprocessor built-in to the Intel 486DX or 487SX. (2) Performance Remember, all you system memory will still be accessed at the same speed. So will the ROM BIOS. As a plus, it's fairly cheap! Based on what I've read, however, I'd feel a lot more comfortable buying the Aox StaX/MC (same concept using an Intel 386SX, with a sophisticated four-way associative cache and much better math coprocessor options) or the Kingston SX/NOW (ditto, except it has a larger cache but it's not as sophisticated... my impression is that it's similar performance to the Aox StaX.) You'd lose the 486 instructions, but I really don't think that's such a big deal with the applications on the market today. >The impression I'm getting so far is that I __CANNOT__ upgrade my 50Z >to a __TRUE__ 386 or a __TRUE__ 486. Is that correct? I'm not >even sure now whether AOX's 486 MicroMaster card is a __TRUE__ 486 >processor, even though I think it's logically possible, as was explained >in their literature. Aox makes three models of the Aox MicroMASTER 486: 25MHz Intel 486SX, 25MHz Intel 486DX, and 33MHz Intel 486DX. I'm not entirely sure of the SX adapter, but the DX models have a 128K cache for RAM and ROM BIOS, and can have up to 16M of RAM on the adapter that is accessed far more quickly than RAM on other expansion adapters and on the motherboard. It's really pretty neat; it's a bus-mastering Micro Channel adapter that takes over CPU tasks of the system. Here are the pluses: (1) Performance The RAM on the adapter is accessed 32 bits at a time. The math coprocessor is built-in to the CPU. (2) Expandability Start with zero RAM on the adapter... go to 16M. It has two SIMM sockets that can take 1M, 2M, 4M or 8M PS/2-type SIMMs. As a side note, I see no reason why the CPU could not be upgraded with an Intel 486DX2-50MHz (for the 25MHz model) or the Intel 486DX2-66MHz (for the 33MHz model.) You'd have to check with Aox to see if this can be done. (3) Backward compatibility You can "turn off" the 486 at boot time and use the native CPU. The RAM on the MicroMASTER is still acessable, although it's not a great memory expansion adapter... The minus is cost. Both in dollars and an expansion slot. (Although the slot isn't so bad when you consider you can load it with RAM.) It's a very nice option, but I couldn't call it cheap. >Are my only upgrade options limited to a 386SX and >a 486SX? Plus, how well would they fare against machines that were >originally a 386SX, 386, 486SX, or 486? Outta curiousity, what is >preventing a common user from purchasing a 386SX, 386, 486SX, or 486 chip >and performing the upgrade by himself/herself? Seems like something >an electrical engineer with microprocessor lab experience can do quite >easily in my opinion. First, don't confuse the 386SX with the 486SX. The 386SX is internally a 32-bit CPU, but is externally a 16-bit CPU with a 16M limit on physical memory. The 486SX in internally a 32-bit CPU with a sophisticated 8K internal cache, and is also externally a 32-bit CPU with a 4G limit on physical memory. The difference between the 486DX and 486SX is that the DX includes a math coprocessor internally. (The 487SX is actually a full CPU plus math coprocessor like the 486DX, so it offers similar performance to a 486DX.) Personally, I don't think a CPU upgrade from a 286 to a 386SX (or similar CPU) upgrade is quite as simple as you describe if any kind of good performance is to be expected. Keep in mind you want the 386SX to run faster than the 286 did, and you'd want an intelligent cache as well. But, I'm not an EE with microprocessor lab experience! - Darius I don not speak for IBM Corp. -- Darius Vaskelis (IBM Marketing/Technical Support) University of Illinois Account Team One IBM Plaza 15-KC9 Telephone: (312) 245-3532 Chicago, IL 60611 Internet: U54345@uicvm.uic.edu
Technical Details
Message-ID:
<92217.084954U54345@uicvm.uic.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!zaphod.mps.ohio-state.edu!news.acns.nwu.edu!uicvm.uic.edu!u54345References:
<1992Aug3.202412.10848@porthos.cc.bellcore.com>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!zaphod.mps.ohio-state.edu!news.acns.nwu.edu!uicvm.uic.edu!u54345 Organization: University of Illinois at Chicago Date: Tuesday, 4 Aug 1992 08:49:54 CDT From: Darius Vaskelis <U54345@uicvm.uic.edu> Message-ID: <92217.084954U54345@uicvm.uic.edu> Newsgroups: comp.sys.ibm.ps2,comp.sys.ibm.ps2.hardware Subject: Re: ----- MORE GOOD QUESTIONS & INFO ON UPGRADING A PS/2 50Z ----- References: <1992Aug3.202412.10848@porthos.cc.bellcore.com> Lines: 122