Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/3/84; site teddy.UUCP Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!whuxl!whuxlm!akgua!gatech!seismo!harvard!talcott!panda!teddy!jpn From: jpn@teddy.UUCP Newsgroups: net.micro,net.micro.pc Subject: Re: Has anyone seen the following advertisement? Message-ID: <1824@teddy.UUCP> Date: Mon, 16-Dec-85 13:46:13 EST Article-I.D.: teddy.1824 Posted: Mon Dec 16 13:46:13 1985 Date-Received: Wed, 18-Dec-85 06:12:55 EST References: <457@eneevax.UUCP> Reply-To: jpn@teddy.UUCP (John P. Nelson) Organization: GenRad, Inc., Concord, Mass. Lines: 82 Keywords: 8088, Intel, IBM PC Xref: watmath net.micro:13132 net.micro.pc:6255 >>Don't miss out on this... IT REALLY WORKS, or YOUR MONEY BACK!!! >> >If anybody has heard of it, please post to usenet since I am sure everyone >is interested. This is the NEC V20/V30 series of chips. The claims made are true, but perhaps a bit misleading. >>No, this is no joke. For $40, plus $1.50 s&h, you get your choice >>of the following This sounds a bit high. I seem to recall prices being closer to $25. I could be wrong - I got mine for free. >> - Fully 8088 compatible - Available in 5 or 8mhz speeds This is the V20. The V30 is 8086 compatible. >>These chips are full 8088/86 clones, except they have been highly >>enhanced internally, resulting in throughput enhancement from 5 to >>100%. Well... PC magazine ran some benchmarks on these chips. (Just the last issue or two - sorry, dont have the issue with me just now!) The V20 (8088 clone) seemed to speed up closer to 5% for any normal instruction mix in existing 8088 code. (i.e. not taking advantage of the new, extra instructions. A few instructions are faster, but most are about the same. PC magazine's explaination is that the 8 bit data bus is still the bottleneck. Apparently the V30 (8086 clone) gets a better average improvement (15% - 30%). Actually, I had a mandelbrot set generator program that sped up significantly. This program was doing LOTS of multiplies. >> It works on Compaqs, IBM's & Clones and have yet to find >>something that doesn't run, and I have tested everything from BASICA, >>to Symphony, to Flight Simulator... Yeah, just about everything I tried worked as well. I had one game program which broke, but it was copy protected, and I suspect the speed difference was the problem more than the V20 instruction set. >> In the Compaq, Norton's SI >>returns a speed index of 1.8, an 80% increase in throughput. Note >>that I say throughput... It processes instructions in less clock >>cycles... there are no changes to the system clock, it stays the same. Oh, it has already been noted that the Norton's SI program must use a benchmark that mostly uses the speeded up instructions. Multiply/Divide, for example. Also, looping is a bit faster. Overall throughput increase is MUCH less than this, however. >>For you techies, it has: >>1) Dedicated effective address calc hardware which does it's job in >> 2 clock cycles, versus 5-12 for Intel. >>2) Dual internal bus, versus single. >>3) Pre-fetch pointer, versus none, speeds calls, jumps, rets, etc. >>4) Dedicated math section, speeds up math intensive applications by >> an average of 60 to 80% >>5) Enhanced instruction set, supports BCD, single bit, and other >> great instructions (new, not in Intel version) >>6) Emulates 8080 (this is amazing, but it does do it!!!) This is a good summary of the enhancements. The other "points" strike me as fluff (WHO CARES if it is CMOS or not!) I recall seeing an ad for a CPM emulator running on top of MSDOS that uses the V20's 8080 instruction mode. I have not seen this in action, however. One thing that WASN'T mentioned is that the V20/V30 supports the 80186 extended instructions. (These are the same as the 80286 unprotected mode extra instructions) If you have a compiler/assembler (Microsoft C 3.0 for instance) that can generate these instructions, you can use them (at the expense of not being able to run the program on an unmodified PC). The new, "extra" instructions are available only to the assembly language programmer. An article was recently posted to net.micro.pc and net.sources (<579@moncol.UUCP>) which supplies a series of macros to take advantage of the new instructions. Using these instructions has the same caveat as above - you cannot run the program on an unmodified machine. John P. Nelson (decvax!genrad!teddy!jpn seismo!harvard!talcott!panda!teddy!jpn)