malpass@VLSI.LL.MIT.EDU (Don Malpass) (12/12/89)
[extract from Gern's msg:] >.... also replace all NEC ICs except the > keyboard controller (and I don't trust that) with non-NEC brands if there > are any in your machine. DO NOT USE a V20 CPU at any rating. Aw c'mon Gern, we've all been through this before. To put it delicately, this is pure prejudice, and as a lot of us have pointed out in the past, not based on technical evidence the rest of us accept. I wouldn't OWN a Z-100 without changing Intel's cpu to a V20. I think NEC has done us all a favor by keeping Intel's feet to the fire, as well as making available an improved product. I feel the same way about Zilog's Z-80 for similar reasons, by the way, and sorry they couldn't make it stick back in the 8-bit days. -- Don Malpass [malpass@LL-vlsi.arpa], [malpass@gandalf.LL.mit.edu] "Why should the TAXPAYERS have to pay for it - let the GOVERNMENT pay instead." [Example of the mentality that got us in this mess.] 12/89
GUBBINS@TOPS20.RADC.AF.MIL (Gern) (12/13/89)
Our Reliablity and Testability Division will differ with that. A NEC V20 10MHz will function in a 10MHz Z-100 and only crash once or twice in a few hours. Z-100 Diagnostics will report TOTAL RAM failure (obviously boarderline as the Diag program is running), pushing the NEC V20, a NEC 8284A, a NEC 74LS240/241/244 Bus driver in a 10.7MHz Z-100, the system will crash usually in under 20 minutes on the average in under 10 minutes. There is no problems at all with Intel/Siemans/AMD 8088-1 and all other non-NEC ICs. This is the only configuration I can recommend. I have done a lot of testing in Z-100 speed-up. I bought the NEC V20 at the time as the fastest Intel 8088 was 8MHz. I have since become biased against them only after seeing their parts consistantly be the problem in Z-100 speedup or Z-100 IC failures (5 or 8MHz machines). Try it: Take a 256Kx1 DRAM 8MHz ZDS factory or HA-108 upgrade new motherboard Z-100 and change the 24MHz crystal to 32MHz. Power it up and the system will be dead without even a Meep. Replace whatever brand of 8MHz 8088 with a non-NEC 8088-1 (10MHz), power up, if there are NEC ICs on the motherboard except perhaps the keyboard controller, the system will have problems and usually fail diagnostics. If there are no NEC ICs or you replace them, the Z-100 will run at 10.7MHz very happily forever and pass diagnostics (except RS-232 and floppy which is a fault of the Diagnostic program), but RAM diagnostics will pass and the system is rock solid. Since you now have a rock solid stable 10.7MHz Z-100, replace the 8088-1 with a 10MHz V20 and you now have a flakey system that will not pass RAM and other diagnostics. I have done this to two Z-110s and a friend to one. I have had the same results with a 32MHz (10.7MHz Z-100) and 30MHz (10.0MHz Z-100, ie. CPUs 'in spec'). The same results for all three Z-100s: A solid 10.7MHz only with non-NEC parts (DRAM too), sorry, but that is the results. As a side note, benchmarking the NEC V20 against an 8088, MHz per MHz, the V20 was faster by 6% at all speeds, the Norton SI showed a 25% increase but this is due to the well known SI bias toward 80186/286 architecture. The benchmarks do not take into account getting a Z-100 with a V20 to not crash long enough to complete a benchmark. As a side-side note, an 8088 (5MHz) run at 8MHz will start off operating for about 30 minutes before it crashes, then each cool-down, repower-up will run about 10 minutes less until it won't run long enough to boot, replacing the 8088 with another at 5MHz is consistant. I toasted five 8088 (5MHz) running at 8MHz, 30 minutes of reliability each before the 8088-2 (8MHz) arrived. Cheers, Gern -------
malpass@VLSI.LL.MIT.EDU (Don Malpass) (12/15/89)
For the record, I received several responses to my reply to Gern's condemnation of NEC chips. There seems to be more basis for such statements than we've been treated to in the past. But I am still puzzled by WHY the experience three of you have had differs so completely from what four of us in this area have had - all of us use'd V20's with only the IMPROVEMENTS noted and NONE of the problems, in my case with Gern's own turbo speedup board. And I think we've all upgraded our mother boards to 768K per the procedures so kindly posted here ages ago. Virtually the only abnormality my Z has EVER given me (which wasn't explained by some dead component or the ubiquitous unseated connector) is trouble making a memory board behave properly as an extended ramdisk. For the hellovit, I'll try it with my old 8086 (if I can find it) sometime. And while the lid's off my Z, I will note the following and post them, as one of you requested. > 1. What rev of the motherboard you have > 2. What rev of the monitor you have > 3. What speed you are running at (and rating of v20) > 4. What you changed other that the v20 > 5. Do you use any graphics intensive software In summary, I will now freely concede that there's room for caution, but I must also add that NEC stuff has NEVER given me any trouble, and I sure can't say that about many of the other dozen or two semiconductor manufacturers I've had to deal with during 3+ decades of digital design. > I'm just trying to figure out what might have > caused my problems (as I have nothing to compare > it to). If you have MASM, maybe I could email > you some code that bombed mine and see if it > sends yours out to lunch..... Scott I know I'll not get to it for a long while (not this year :*) , but I'm genuinely curious, so feel free to fire it off and I'll try to take a look. --- Don Malpass [malpass@LL-vlsi.arpa], [malpass@gandalf.LL.mit.edu] "Why should the TAXPAYERS have to pay for it - let the GOVERNMENT pay instead." [Example of the mentality that got us in this mess.] 12/89
GUBBINS@TOPS20.RADC.AF.MIL (Gern) (12/18/89)
I hope you mean to replace the V20 with an 8088, not a 8086 (same 40-pin IC, ya know) (-: The V20, as noted, seems to act bizarre with RAM, this may be the cause of your Z-205 (which is rated for 5MHz, zero wait states with stock parts and configuration settings) problems. Early versions of Z-100 MULTIPLAN will crash during use with a V20. It was nasty, as it crashed during use (probably after inputting a lot of data) and not straight off. Later versions did not have a problem. Cheers, Gern -------