How the NEC V20 beat Intel Processors in the 1980s!
Вставка
- Опубліковано 5 бер 2022
- We take a look at the NEC V20 CPU and how it had faster instruction timings than Intel CPUs, could emulate the Intel 8080 CPU and how it had additional instructions not in the Intel instruction set.
We also look at an early version of Microsoft BASIC, we emulate the 8080, take a look at my EGA card collection and try to get to the bottom of some corruption in EGA games.
TopBench:
github.com/MobyGamer/TOPBENCH... - Наука та технологія
This is so great. I've always wanted to know more about the 8080 emulation inside the V20!
Thanks. It had been a bit of a mystery to me too until I tried some of this stuff out.
Strange that they didn't emulate the full Z80 instead though. Especially so when NEC was one of the first japanese firms that copied the Z80 in the late 1970s (called NEC μpD780).
@@herrbonk3635 The "emulator" probably just swapped out the instruction decoder. So, it was limited to instructions that exist on the 8086. That architecture can't actually emulate an instruction that's not present.
(as explained in my top-level post, the 8086 is a superset of the 8080, so mechanical translation of existing programs would be possible)
@@JohnDlugosz It's actually the Z80 that is a superset of the 8080. The Z80 can run 8080 code unchanged. These chips even share the same designers, Federico Faggin and M. Shima. The 8086 was losely based on the 8080, but diverged much more than the Z80 in both high level instruction set and internal transistor level implementation. For instance, neither the 8080 nor the Z80 used any microcode, just a PLA, while the 8086 used both (and all three used random logic decoding and sequencing as well, of course).
+1 for NEC optimized stuff. I've always wondered about the full potential of V20/30
Back to this video... now I remember... My search for a video of a demo of the NEC v20 i8080 emulation landed me here, and that's how I found this channel.
Side note, Checkit 3.0 has issues with the EGA ram test. You can ignore those errors.
heck yeah! lets see some V20 optimized demos!
Of course we'd like to see more about the V20 and even the V30!
I will definitely cover the V30. Thanks for the comment.
I would LOVE to see a more in-depth video about the V20 and all these interesting features seemingly no one has actually used before! I'd love to add the video if ever to a blog post about building an XT clone with all kinds of odd parts I'm working on. (Waiting on a CF card adapter)
My XT clone has a 10Mhz equipped V20 made by, and you'd never guess, Sony!
Wow! You mean the actual chip was manufactured by Sony? How amazing is that!
@@PCRetroTech Yeah! Was just as amazed to figure out the mysterious Sony chip was actually a V20! Model is CXQ70108P-8
@@unitedco1904 I've never seen one of those. Maybe it is even a rare part.
So many fond memories, I got my first 386DX40 by soldering hundreds of cables for a guy, payment was a motherboard, CPU and some memory. My god it was fast.
Yeah I'd love to see some NEC V20 specific code/demo as be cool to run on my little XTjr machine which a V20 in it. Great video and well done for appearing in it too, I enjoy videos a bit more when I can see who's behind the hands! I've really enjoyed your content as I'm a bit 8088/EGA obsessed and you've done lots of great videos on that, let us kmow if you ever look at Patreon as I'd be happy to support you in that way. Cheers!
I commented before the end of the video - so to answer your question, yup I like the new format 😁
@@mogwaay Thanks. And a few people have mentioned Patreon. It's not possible yet, but I might think about it for the future. Cheers.
Yes! We (I) would love to see more low level coding and timings. Please :)
I always look forward to your videos, glad to see that you are taking the channel to the next level, so to speak! And yes, I would love to see some V20 programming!
Thanks. It will definitely happen, as soon as I find the time to write the code.
Oh, great, more V20 stuff please! V30 would be nice, too.
And your shelf do look like mine in my basement 😂
Last thing... I thought you would have been a twenty-something with huge ♥️ for retro PC's, twice as old as you.
But great, if you explain stuff personally with face, it feels easier to comprehend and follow.
Thanks for your content, keep it up!!
Thanks for the feedback. I really appreciate it.
What a nice face. Happy you were confident enough to reveal it!
Thanks! 😊
V20 was great. Had one in my XT. Worth about 10% at 4.77mhz
I would also love to see more videos on the V20. Great video, thanks for posting!
Like the segments, and the more personal videos instead of just a voice over. 👍👍
Thanks.
The new format, and your face :) is very fine, do not worry about this. It's always nice to be able to get to know the person behind the videos a bit more. Interestingly though, it always feels to me, that your voice is deeper when you are off-camera. I am not sure though that it's only me, or maybe different microphone is used? No, for sure, _no_ complains here at all, just I noticed, that's all.
Thanks for the observation. I hadn't noticed myself but I should keep an eye on it. A deeper voice is easier to listen to, and I had been working on that. I guess it is going to take some time to get it right. I'm a few videos ahead by now anyway, so it will be a while before you start to notice any change probably. Anyway, this kind of feedback is very helpful. It's something I can change that does make a difference in the long run.
Nice to put a face to voice!!!! Thanks, I am fan of your videos!!!!! And that pic of Stellar 7 for the Apple //e!!!
Great format. Keeps the interest throughout the video.
Glad you enjoyed it!
just a few days ago I assembled an XT compatible based on the NEC V20. Videos like yours help me understand what this machine can do. btw you know I've never seen an EGA card in real life, around here almost everyone had Hercules and then switch to VGA when they moved to 286 or 386
Oh that's interesting. I think I went from CGA clone to VGA so I guess it wasn't that uncommon for people to skip EGA. That makes it interesting though, as we can look at what it was really capable of.
The clone Hercules clone cards could emulate CGA on monochrome.
@@joeturner7959 namely "ATI Graphics Solution 3" for example ;)
EGA was common among home users in the 2nd half of the 80s which is why most PC games released between 1986-1989 targeted it. Hercules was very popular for PCs bought for businesses and schools but CGA was probably more common among European home PC owners where PC clones made by Amstrad and Commodore were a lot more popular.
The new format is great BUT not if it's fatiguing/too much effort. i.e. The effort comes through but beware the burnout monster. The channel is compelling regardless. Keep it fun for you - in the end that's what people will notice most.
It doesn't seem to add that much effort, so I think it is ok. In fact I introduced this format because it lowered effort in some cases.
Good to see you in person! In my experience it's always a positive thing for youtubers to show their face if they make great content like you do. It definitely helps with "viewer loyalty".
Keep up the good work! 👍
Thanks very much. I appreciate it!
@@PCRetroTech Forgot to comment ;-) :
I own a lot of V20 machines and I'm already a big fan of the V20/V30. They were (in a way) the very first upgrade/"overdrive" CPUs if you will so It would be very cool indeed to see some the "special powers" of the V20 in a future video!
That's a cool simple demonstration of the gigantic speedboost in BASIC on the V20 in 8080 hardware emulation mode (interpreter on interpreter vs interpreter on hardware emulation). I was planning to run CP/M on one of my V20s but I have so many hardware and work projects going on that I just never got around to it ;-)
@@pipschannel1222 Thanks! A lot of people seem very interested in that, so it will definitely happen at some point. It's on my future video schedule for sure!
Yesss would be really cool to see V20 optimized stuff!
10:00 What's plan B if you don't get enough thumbs ups to warrent a follow up? tiktok dance videos?
I'm not a very good dancer. Maybe I'll make C++ videos and rant about const correctness. Do you think TikTok could be transformed into a useful pipe if there was content like that on there?
Hell yes, you should definitely write some specialised code for the V20!!! Your channel is awesome!!!
Thanks! It will definitely happen at some point.
I want to see what you can make. One of my favorite 8088 demos is "8088 Domination" where a stock 8088 with CGA and a Sound Blaster displays video full screen, with 16 colours, at up to 60fps using graphic mode (composite output).
Maybe you can ask Trixter for the source code, and optimize it for the V20 to get better quality or performance?
I think Trixter still has ideas for that himself. But I don't think the V20 would make so much difference there. There's just a few instructions used. But who knows.
I also had a 10Mhz XT - but the ram config was different, instead of 1x8 DIPs it used 4x8 DIPs for the upper 128k.
I also added a V20 and an 8087-3. I used Vidspeed to find all the text and graphic modes, which showed slight improvements after I replaced the 8088-3 with the V20-10. Both the NEC and 87 ran hot, so I had heat sinks on them.
That's interesting. I wouldn't have expected too much heat from the V20.
Yes i would like to see more v20 videos.
Yes I want to see a cool demo!
Yeah, I think that style of chip puller is meant for PLCC's like modern BIOS chips, not DIP packages. The puller for those is a much less complicated contraption.
I could see an old school demo, even using tables with precalculated items being a lot more capable than a 8088. And being able to run CP/M on DOS is the first I have ever heard about it with GENCOM.
This was something my mom did when i was a kid in the early 80s. our first PC was an xt clone from around late 84. We had invested a lot in to CPM a few years earlier, and she wanted to switch to some applications that were on PC that would not run on our CPM machine. That and the CPM path was a dead end, the computer we had was at its limits as well. The v20 was just barely available but its 8080 emulation was just perfect. I recall it being quite pricey getting it setup but she was quite pleased with how well her old applications ran. And it allowed her to transition to newer programs that she ran for a good 10 years or more.
I also had this screen memory corruption with a Tseng ET4000 Card when i programmed it in assembly. The solution was to put a NOP between reading Data into the Buffers (of the Card) and actually writing back. It also helped to put the Bus into the 1Waitstate Mode. It seems direct accessing the card is sometimes too fast. You Corruption Pattern is nearly the same i encountered on my experiences.
Oh thanks. That could be a useful comment indeed.
Nice video, please give us more on the V20.
Thanks, will do!
The div and mul operation being faster is actually pretty HUGE! One can get away without div when doing 3D and perspective correction (I once developed texturing with only shifts as hobby task) but generally its hard like hell and div is just used there.
I know someone ported wolf3d to 8088 (search YT for it) and it was going just sloowly... Worth a try if the NEC handles it better because of the divisions.
The bitfield access opcode... I am not so sure about to be honest. maybe slightly faster if need to read multiple bits from random location as a stream, but sounds like some embedded use case mostly. For single / few bits hand optimized regular shift and logic trickery seems faster maybe. I think this chip was used in some more mobile devices that one can carry and if they have any controlling usage it could have helped there. Also likely hugely helped back then with some compilers not yet optimizing that good. I also think effective address calculation can be eliminated from hot code paths with a lot of optimization work in many cases because of enough registers, but surely should help with less hand-optimized codes at least - and make that being feasible to optimization.
I guess in the benchmarks they maybe not using those at all so difference is barely differing while it can be more huge.
Very interesting videos. Subscribed ;-)
PS.: I think this format can also work, but the backside is that it is harder to search back. For example now it will be harder to search back to the EGA card part if someone remembers that and want to show it to someone yet choosing a single topic as title likely helps more first clicks on the video when you choose well. So it gives this interesting dilemma. This format works better for "news" kind of things like Gardiners steam deck news (totally unrelated) and such. I guess maybe you can have a "sister channel" maybe where you upload them separately and with more details or more raw form or something and link back and forth - but I am noob for youtube so don't take my advice seriously.
I've gotta say, you actually don't look like what I imagined.
Don't worry, I don't look like what I imagine either. :-)
@@PCRetroTech None of us do, mate. :D
@@phreapersoonlijk Yeah especially every new decade.
I always thought older cause of the interest in the early x86 :-)
@@Pickle136 I'm old enough.
I recall that the extended instructions on the V20 etc. are taken from the 80186. Compilers can be set to target this, to take advantage of those instructions.
re 8080 emulation: The 8086 is a superset of the 8080. You can map all the 8080 registers and instructions to the 8086. The opcodes are different, but each instruction is present. The translator program goes through and changes each 8080 instruction to the 8086 equivalent. So, I expect the native emulation to be the same speed, not significantly faster.
The translator appears to handle Z80 instructions as well, which needs several instructions to accomplish.
A translator _could_ be faster, by recognizing idioms and replacing that code; e.g. multiply sub-routine calls changed to the native instruction. A fairly simple program could at least recognize pairs of operations done in consecutive instructions, and replace those with a single 16-bit instruction.
My conclusion is that a translator _could_ be much faster than the native code executor.
Whilst it is true that some of the extended instructions were from the 80186, some of the memory, bit precise access instructions are unique to the V20. There was a real mix of additional features and the real purpose of some of the additions is a bit mysterious (to me at least). Presumably they had some application in arcade games or something like that.
The v30 with really optimized code is said to be 286 like in performance. Though I don't know how many applications can practically make use of those instructions. It would be only a certain fraction of them. The v30 also gets a sligtly larger performance boost to the v20 than the 8086 does to the 8088 as well.
That is also my understanding. The data bus is less of a limitation on the V30 than in machines that take a V20.
I like the new format, and I would love to see more coding on V20/V30. I have 2 machines with these NEC CPUs here and coding V20/V30 code is on my (large) TODO list. Thanks for all the interesting information.
It seems to have fired a lot of people's imaginations. So I imagine there will be more to come.
Love the new format mate, well done ✅
Personally I would love to see some demos written and try them on my own
Upgrading a 8086 to v30 made enabled my sharp pc7100 portable to use iomega guest and zip drive over parallel :-)
Guest wont run on a xt by default
Thanks. There is a zip driver for palmzip which actually works on 8088/8086.
Ill keep watching!
Everything in kitchen. Your breakfast must be different. Coffee + chips
More like, do I have time for breakfast today, or do I need to make another video. :-)
I like the show format, with small tidbits on different retro technology. Go for it! Also, you probably know this, but many EGA cards have dip switches to set up for CGA monitors. Just putting that out there in case it wasn't widely known. Thanks for the video!
Thanks. Yes I did know that. It was simply a case of being too lazy to look up the dip switch settings for those cards. I will certainly do that when I show the cards on the channel though, as some of those cards are very interesting indeed!
@@PCRetroTech i was especially surprised that there were EGA cards that could drive a CGA monitor in interlaced mode, meaning higher resolution graphics on something that is really only otherwise locked to the max res of 640x200. Will a 640x400 or 640x350 resolution be possible on a standard CGA monitor? That would be something wouldn't it.
@@molivil Interesting question. I am really not sure at all. Obviously the monitor itself is limited to 200 lines. But it might give the appearance of a higher resolution albeit a bit flickery. I should investigate this in a video!
The glitch in the start screen of Stellar 7 seems to be exactly where the animation of the "star" happens on the Dynamix logo. I don't know anything about how memory works on EGA cards, but maybe this is what you'd see when reading from memory that's not physically present?
Yes, that's what it was. I figured that out in the end.
The chip puller you have is for PLCC chips, it's good for those - not great but that's because PLCCs is a *pain* at the best of times - "sucks a lot less" is as good as you'll ever get with PLCCs!
There's a different type of puller which are intended for DIP chips like these, see the Jonard EX-1 for an example, it's really cheap and there's tons of clones that likely are just as good (it's so simple), it's a bit better than that PLCC puller for this (but you still need to be careful to not grab the socket) but IF you have space on both sides it can often be easier with a thin flat screwdriver instead (but often space can be at a premium).
There *IS* a tool that that's better for this and make removal of 24-40 pin chips simpler and safer, a number of UA-camrs use the Jonard EX-2 for this but it does cost $25/€25. This time there's much less clones available, the only one I can find was the GJ-6. It's available on ebay (search for Chip Extractor GJ-6) for a bit less but is much more plasticy than the EX-2, the price difference isn't massive and depending on shipping and VAT it could actually end up costing more in some countries. They do have a smaller GJ-3 variant that can pull smaller chips (16-22 pin) that is unique to them(!) but usually those small chips are easy with the EX-1 style grabber anyway.
That NEC V20 brings back memory, my first PC was a 8MHz (or possibly 10MHz, don't remember ) Turbo XT clone with a V20 (installed from factory).
Thanks for the extra info. I was aware this was the wrong puller and I have seen the others in various videos. At the time I ordered a chip puller, the other ones didn't seem to be available, and it is alway difficult ordering in German anyway. But with the extra info about model numbers of these pullers I might be able to get the one I was looking for. Thanks!
Even though some of the programming stuff is outside my skill set, it doesn't mean that I don't find this kind of thing absorbing and interesting. It has been decades since I used an XT clone with a V20 containing lots of cards that I didn't understand at all, but now I have been informed about what they were. Better late than never! Just wish I still had the thing to play with. Also, it's nice to see your friendly face on screen and the new format definitely adds interest and detail. Programming that takes full advantage of the V20 would be interesting to see. Unlocking hidden potential perhaps.
Thanks for your kind comments. I am pretty sure I will work on some V20 specific code which I will eventually show on the channel. I'm glad you are enjoying the content.
Nec v series chips were not really designed to be just a cpu/cpu replacement. It’s actually more of a competitor with 6502. Both are general purpose processors primarily made for sbc som applications. This is before there was a lot of microcontrollers or before dsp was really a thing.
LoL - I thought 6 cycle opcodes for memory lookups on 6510 was a lot.
@8:40 Note that while SUB r16,imm16 executes in 4 cycles, as it says in the book, the instruction is 4 bytes long, which takes the 8088 16 cycles to fetch it, assuming 0 waitstate memory (which you usually have on PC/XT machines). When looking at 8088 performance, you always have to worry about bus starvation.
@10:00 As the bus performance of the V20 does not increase compared to the 8088, you will run into bus starvation even more often on the V20 than you do on the 8088, so don't expect too much potential of V20 optimized code. As the V30 has the double bandwidth on instruction fetch and aligned 16-bit data access, it has considerably more potential to outrun the 8086 than the V20 has potential to outrun the 8088.
@20:12 The "parity error" message is the result of a non-maskable interrupt. Some advanced EGA and VGA cards used the NMI to break into compatiblity hacks in the BIOS (e.g. to translate video timings). Possibly your SmartEGA BIOS and your mainboard BIOS are incompatible and the EGA card is unable to intercept the NMI as it expects.
A topic you didn't specifically mention in the video, but which is also a big thing about the V20/V30 processors: They implement the whole 80286 real mode instruction set, and thus are compatible with 286 optimized real mode applications.
Your first two notes I was aware of, but the third is something I didn't know. Thanks for the very helpful information!
Nice to see the face to go with the voice hope it wasn't too daunting😊 I have a couple of Amstrads a 5086 with a 8086 and a ppc640 with a v30. Think the v30 was a drop in for the 8086 so will be interesting to do the same thing with those. Specific code to bring the best out of the v20 would be very cool to see.
I will certainly take a look at the V30. It's a very similar CPU, but for reasons I'll show on a later video, one gets even more speedup with a V30.
@@PCRetroTech cool, look forward to it!
Great move deciding to show yourself on video. I believe it helps viewers connect. Very cool.
Id love to see v20 optimized code. What may save you time is taking one of your 3D tandy or cga 8088 spinning tetrahedron demos which are optimized for intel then optimize it even further for v20 comparing frame rate. I believe those demos use a lot of fills (possibly allowing the v20’s enhanced memory access to shine) and would be somewhat of a practical demo too (and save you time writing new code?)
The EGA results were surprising as well as a “smack the forehead” situation. I was thinking it was a memory speed access issue. For instance the video ram for cga and ega maps into the upper system ram. I was thinking maybe the PC bios settings were set too fast (maybe too few wait states) but your comment on how the card works fine in Cga made that theory mostly bunk and sure enough it was 64k vs 256k. Good to know.
Great video as always
Yeah, the EGA thing was pretty funny when I figured it out. Doing an optimised fill routine for V20 would be an interesting project. Much of the time is not spent doing fill, so the overall speedup would not be huge, but it could be noticeable.
The "Stated" Performance increase on the V20 vs the 8088 was always stated to be 8-12%. I can agree with that on a certain level; but the more RAM the system had (say 256k vs 640k), the speed scaled-up. Also, with games, graphics, or etc, anything that used significant memory access AND calculations, the V20 could show to be up to 35% faster or a bit more. It WAS totally noticeable; especially when the compliler used for an App was using a lot of microcoding with the CPU.
Can't beat the ATI EGA Wonder 800+ card. Btw you can change the mode via software, switching off interlacing, true CGA mode etc, grab yourself the drivers and you'll see what I mean. Great cards.
Great to see you behind the camera.
Thanks. I'll check out those drivers. It is a very interesting card.
@@PCRetroTech Let me know if you can't find them. I can't remember where I found them, but I have them on my IBM 5162 and can copy them over for you easy enough.
I’m loving the multiple subjects in one video!
I’d love to see some more V20 on the channel. Especially what it could eventually be used for.
For instance: Hoe do the numerous additional instructions in the V20/30 compare to, for example, a 286/386 CPU?
It would be interesting to try and find something it could do that is faster than a 286. There might not be anything, but that is an excellent suggestion. Glad you are enjoying the videos!
The problem with most videos on any retro topic, is that they are kids reading wiki pages. They weren’t there and they don’t know what they are really talking about, nor do they know how to write any kind of code.
+1 for Nec
I'd love to see some NEC optimized programs to run on the V30 in my Olivetti M24!
Great! I remember (maybe incorrectly) that certain compiler had switches for V20/V30 CPU just like Basic or Pascal had for 80x87 or emulation. I don't remember what compiler was (may it have been FoxPro or an early version of Pascal?). I remember questioning "what is that V20 CPU?" but I passed the page with 286, 386 and later 486.
That's correct. However I wasn't able to find a compiler which really takes advantage of the other V20 instructions. I'm not even sure anyone has written a modern compiler which really takes full advantage of it. That would be a great fun thing to work on, but way too much for a video on the channel of course.
THE FUTURE!!!
Out of curiosity, did the original 5150 PC have the option of having one of its floppy drives replaced with a 10MB hard drive? I ask because my first computer was purchased at a TRW surplus sale (for $125) in the early 90s and it was a regular old IBM PC, not an XT, but it had said hard drive. And no, it didn't use the 5161 Expansion Unit.
No, I don't believe that was a supported option. For one thing the power supply was not rated for it. The BIOS in the XT was also updated to support hard drives as far as I'm aware.
Great video! Can you tell me which book you were showing when you explaining memory EA? I am hoping to study 8086/8088 real mode programming for fun.
Sure, it is an old book called The 8086 Book by Rector and Alexy.
Thanks! I bought one from eBay.
On motherboard, especially old ones, I recommend to be grounded to avoid static electricity.
I'd love to see a V20 optimized program !
One thing I wonder is : could it be possible to turn a PC with a V20 into a CP/M (non x86) machine with a different BIOS ?
I mean I'd imagine a BIOS who would first put the V20 into 8080 mode and then contain a CP/M bios, letting you use the PC as a "real" CP/M machine :)
I have wondered this myself. It's not clear because for example all the hardware on an actual 8080 computer would be different. They used the S-100 bus for example. So I'm not sure what would actually run.
The PC's BIOS wouldn't matter. All it does is copy the first disk sector into RAM and run it. The BIOS in CP/M is written into CP/M itself by the computer manufacturer and loaded from disk. Digital Research documented this very well with examples. I would think someone somewhere has already done this. I think there is a problem with I/O addresses though. My rusty memory says the 8080 has an 8 bit I/O address space. PC uses 12 bits, so any I/O above 0xFF would inaccessible to the 8080 mode. Maybe someone has come up with a workaround.
XT-IDE has a firmware optimize for the V20, supposed to be 50% faster disk transfert compared to the 8088 firmware on the same machine
Friend of mine had a V20-powered turbo XT back in the day, where I had an 8088-powered turbo XT, both at 9.54 MHz. My experience was quite similar to yours: in specific benchmarks the V20 showed clear advantages, but in most games you'd hardly notice a difference. Perhaps it's also because XTs are severely limited by the poor bandwidth of their video adapters anyway, inserting a lot of waitstates on the bus when you try to read or write video memory.
You would avoid muls and divs like the plague anyway, in 8088-targeted code, or 8-bit platforms in general.
One pretty significant disadvantage of the V20 is that it does not support all the undocumented opcodes. Some software will break on it. 8088 MPH is one such example. It makes use of a common undocumented opcode, which the V20 does not support.
Most of the instructions that you would use in really tight code would be the same as for the 8088. There wasn't much point NEC optimising them because of the fact that the CPU spends almost all of its time accessing the bus (to read the instructions). I'm very convinced one can get some good speedups with carefully written code though.
It is a bit of a shame that the V20 doesn't support undocumented opcodes. I imagine 8088 MPH is also insanely optimised to, and probably even very timing dependent in places, so I could image a V20 doesn't make much of it.
@@PCRetroTech If I'm not mistaken, the V20/V30 actually support the entire 80186 instruction set. Which allows for some decent speedups especially with shifts and rotates, because you can use immediate operands other than just '1'. In fact, if I'm not mistaken, the 186 instructionset is the same as the 286, minus the protected mode. So the V20/V30 should be able to run any code optimized for 286, which does not make use of protected mode.
@@Scalibq Yes, it certainly has optimised shifts and the like, and supports some of the 286 instruction set.
20:22 what are those diods on the right 40 pin IC?
I'm not really sure. I assume it's a mod someone did, but maybe the factory missed something.
Could these extra instructions possibly be related to that .com file that was floating around that would patch the cpu erata or possibly the command.com and speed up the operations of certain things, like I know for sure I remember a friend running that on a machine I had once and it actually worked. Possibly 286,386 era-ish in origin...but maybe I don't remember for sure. Now that I think about it, it could have been a virus or a trick because I only recall the dir command scrolling by about 10x faster than normal after he ran this com file.
It sounds like it might have just modified the timer interrupt or something. I doubt it would be related to the \v20 instructions.
2:15 please show it inside ! I am so curious how an 8086 XT motherboard looks like
When I do the V30 video, I will definitely do that.
V30 was an awesome CPU for the time.
Although I hadn't a PC back then, I remember some arcade machines using them.
Lately, I had a taste from a KCS PowerPC+ board on my A500+.
Better than 8086 and also includes support from 8080 for CP/M 🙂
Interesting. It must have been a popular thing to do at the time. And you have reminded me to mention the arcade games when I make another video on this topic.
Yeah, indeed, I wished it had Z80 not only 8080 mode though ;) But fair enough, strictly speaking 8080 should be OK for CP/M, though some CP/M programs already started to use Z80 extra opcodes over the 8080 ones already.
@ Yes, agreed. It would have been really handy to have a Z80 mode. I don't know why they didn't do that actually.
I just started watching
I almost didnt see you in the video, i kept seeing those cards on the shelf and wondering what they were ;-)
Maybe the v20 can enhance your 3d cube demo?
I' m almost certain it could. There are other things that I want to try as well.
Just cool the CPU that could not handle the 10 Mhz. I used to cool my 286 cpu with wet cotton wads when running on a speed that was not supported, steam was coming off but it kept working till i gave away the old pc :)
Amazing. I guess evaporating water is one way to carry the heat away.
Didn't the V20 also support 80186\80286 opcodes, except protected mode? It should be able to run games like Monkey Island 2 which require a 286. Maybe that's a nice test.
Yes, I believe you are correct.
It might be interesting to test if there is a noticeable performance difference between MBASIC in 8080 compatibility mode and GWBASIC in x86 mode.
Indeed, that could be very interesting. I'd guess MBASIC was very primitive given it had to fit in such a small memory. But speed comparison could still be interesting. Thanks for the suggestion!
Any compilers that can generate for V20/30?
I am not sure. Getting a compiler that does a good job on 8088 is already hard. There is a gcc back end by Reenigne, but it has some limitations I believe. As far as I know it doesn't do something special for V20/30. I would have thought nasm was the best bet, but I don't think even they have done anything special.
Why is memory access so slow? I mean, 4 cycles for register operation and, at least, 23 for the memory one. Its not like there are too many wait states when cpu is running at 4.77 or even 8 Mhz?
Most of it is the CPU itself. It has to do a lot of work to access memory. The memory access itself can happen in theory in just a single bus cycle, which would be four processor cycles. The rest is actual execution time of the CPU instruction, which doesn't even include the bus access itself.
Aaaah! 5 seconds in and he admits to not having a girlfriend!!! This guy rocks! :-D
the 8 mhz cpu nec v20 in the juko motherboard run at 12 mhz....have cpu for much more
Out of curiosity, were you testing those EGA cards on the same machine with the V20 in it?
Oh boy, can you believe I already forgot! I don't think so. I vaguely recall I used a 286, but the videos get made so far in advance that my memory sometimes tricks me.
@@PCRetroTech Huh. That just makes it all the more strange that the Smart EGA card doesn't work, then. I was watching another UA-camr (@Epictronics) who has been working on repairing a prototype IBM PS/2-based workstation, and he has been running into many bizarre errors with it, including a Parity Error Halt on boot, which seemed particularly prevalent when he had a V30 installed (ua-cam.com/video/tmheXQ4sB5s/v-deo.html). It would have been convenient if NEC were the common culprit. Anyway, I sent him over to watch your video. I guess at least knowing that a problem with the video card can cause the machine to throw a parity error could be useful information.
Did you see the new lemmings documentary on exient channel. It offers a technical explanation for why lemmings are the color they are
Didn't see that. Will have to take a look at that one. Thanks for the recommendation!
my compaq portable has one of these, but for me it's a little disappointing because I can't properly run some cool programs like 8088mph, and oddly enough 2048
There's a few things for sure that won't run. My effect in Ara 5150 also won't run, and I think some of the MS Flight simulators. The division instruction and a few others differ ever so slightly.
👍
Nice kitchen
Hardware evolved so fast in 80's and 90's that software developers didn't have time to take full advantage of a given CPU. V20 was doomed from the start, since 80286 was the king of high-end PCs and soon groundbreaking 386 was released.
In late 80's PC market become totally fragmented with 8086, 286, 386 and 486 machines that had to be supported by software. There was simply no reason to spend time and money to make software run better on V20, especially with NEC CPUs being used on lower-end machines that didn't generate a lot of software sales anyway.
However, V20 evolved into a line of embedded SoCs that were used in some palmtops like HP 200LX that could run DOS programs. These palmtops came with built-in software that might use V20 specific instructions.
I agree. It's a shame things evolved so fast that the real benefits were never seen. Interesting comment about the SoCs.
Actually I looked into this, and the LX95 used the V20, but later machines in that line used an Intel SoC called the Hornet. I don't know if I would say they evolved from the V20 per se. Maybe the Hornet was somehow inspired by the V20, I don't know.
parity error, need two chips extra in the ram
Too often manufacturers threw a new product on the market that was not thought true enough. Especially in the world of PC's. They all wanted to be the first and therefore a lot of garbage was on sale. If we look at the screens of personal computers from the past then wel all must admit this was not good enough. Four colours, text mode, screens full of mumbo jumbo. All I see is evenings spoiled on testing this and trying that.
What ought to have done was that all consumers and businesses that bought computer equipment that was not functioning as advertised could bring it back and get the money back, a full refund and something extra for the lost time.
For way too many years the PC business was selling stuff that was not working in combination with other stuff, for years and years the PC market was a rat race in buying ans replacing cards and processors, updating software and everything all over again. Shiploads of money were spilled, time and attention the same. Microsoft could never repay all the mistakes they made, their business model was producing something that was obsolete and unworkable in a few weeks, and then again things had to be bought, technicians had redo the installing of software and hardware. Computer users were held hostage by companies that did not have any ethics, squeezing then out of their money as long as they could.
That would kill innovation. Can you imagine if no one could produce a product that was not 100% compatible with IBM hardware? We'd have no new innovation, because then IBM hardware would not be compatible with it. Anything designed to take advantage of the new hardware would not work on the old.
And companies who wanted to kill their competitors would just have to find some incompatibility (or introduce one) with their competitors. The company that had the resources to fund research into finding incompatibilities (or creating them) would win.
Whilst your idea envisions a nice perfect world, it is a bit dystopian.
I'll agree with your basic point that many products were released too early due to the pace of innovation, but invariably these products failed in the marketplace unless they brought something that really counterbalanced the flaws.
@@PCRetroTech Competitors should have been held responsible for the working and the quality of their product. Everybody would still be allowed to compete and innovate, but making cheap knok-offs and counterfeit cards could be prevented. This would have saved the users a lot of money. And hundreds of thousands computers would never have been made at all. It has no use to produce consumer goods that are almost working and almost just as good as the original product. The real innovators lost revenues on every sale of a cheap Chinese copy of a PC's.
@@vanhetgoor Ah yes, well that part I certainly agree with you on.
It has a face!? UNSUB!!!
Yeah, I didn’t come here for this! ;-D
Ha ha!
So, interesting thing.. the 8080 software emulation on the turbo XT is intended to be somewhat clock cycle accurate to an old 8080 at .91mhz.
Interesting. I wonder what machine that was emulating exactly.
@@PCRetroTech I don't know that it is emulating one in particular. I think that is just how fast it worked out so that is how it was marketed lol. I am not sure but my mom used a program like that back in the day to transition her buisness software from a CPM machine to a newer PC. She found out about the 8080 emulation of the v20 and paid quite a bit to have it all setup. And the performance improvement was as you showed, about 10 fold. Which was amazing given how incapable her CPM machine became over time.
I think this was all in late 1984 so quite early on. She used that setup for probably another nearly 10 years or so... quite amazing really.
@@wishusknight3009 Very interesting. A few people comment on how so many videos are made by people who weren't there at the time. Well I was using PCs at the time, but in Australia. Our experience was quite different there. My father had a Kaypro II at work, but that's about the extent of my exposure to CP/M where I lived.
@@PCRetroTech I honestly don't remember what our CP/M box was. I do know it was a hard transition from manual filing to using a computer for my mom, she got it in 1980 or so, and it didn't take long before it was already too limited. She complained she could do the manual filing much faster.
She did eventually get it working out pretty good and it wasn't long before she swore by it, but still complained the computer had too many limitations. Patches to the software from the company supporting it helped a lot, but 64k of ram on a 1mhz 8080 didn't go very far. If it wasn't for the fact she had $20 or 25k into it, she would have dumped it sooner than she did, but finding something to transition to was just not feasible. And this was in the middle of the reganomics recession.
The V20 was a savior. The person she had doing her IT for her was able to make the transition pretty easy. A bit of a curve in learning a new computer but it wasn't all that dissimilar to CP/M. Of which she was glad to get rid of it.
My first computer was an SX20 386 that i built as a teenager in 1989. I actually inherited that very XT and upgraded it. I was only 13 as well... worked hard as heck trying to get the money together for it all. but it was a flying machine. Still have it and the original v20 and motherboard too!
@@wishusknight3009 Wow, great that you still have it!