Granville Cyrix 6x86-P200 Overview

[@AshtonCoolman]

Wow! You're back. It's great to see a video from you again. I hope that you're doing well

[@joshuareynolds6958]

We're glad to have you back

[@tomaszc4]

We're glad to have you back, Happy new year

[@s3vR3x]

YES!!!!!!!!! YOURE BACK!!! IM SO GLAD YOU DECIDED TO DO ANOTHER RETRO COMPUTER VIDEO!!!

[@Soldoles]

happy new year my dude. all good things to you!

[@BillyBlazeTheKeenest]

Long overdue video. Glad to see your content again!

[@nordicwarriorgaming4953]

Happy New Year bro. Glad to see another main channel video.

[@phillycheesetake]

Happy new year! Welcome back, thanks for the video.

[@cocusar]

happy to see you back dude! happy new year and I hope to see more from you

[@RetroTinkerer]

Happy new year, glad to see a new review from you!

[@RetroTinkerer]

Now that I watched the video to the end... Damn, what a rough year you had with all these troublesome systems and your hard disks. About the 6x86, now I wonder if all the troubles I had on my Pr 200 where related to my lack of criteria selecting a proper motherboard and cooling, but between the 75MHz bus and high power requirements it could had been both LOL I was also pretty puzzled by all the "know it all" telling me how bad of a CPU this was but I kept playing my games pretty satisfied with the performance, the stability issues was what draw me towards a Pentium Pro, maybe I will end up finding a Cyrix again and give it a second chance. Nice to watch you again!

[@HighTreason610]

Stability problems were definitely reported commonly on Cyrix machines and I do think in many cases it was down to under-sized voltage regulators. I'm almost tempted to find a crap motherboard that has such an issue and test replacing the voltregs with beefier ones. Regulators in this time were usually linear, which will net a very stable and quiet supply of power, but only when used within their limits. Being solely current based, biasing a transistor into its linear region, their heat dissipation is proportional to the load, roughly loadA * (Vin - Vout). When these regulators overheat, they usually drop their voltage due to increased resistance, or turn off entirely. In fact, this same tendency made some synthesizers unstable, too. Oddly enough, the PPro wasn't immune as the 2MiB 200MHz model reportedly caused the same problem on boards that weren't designed for such high loads. That would have been around the transitional point of boards going over to PWM buck step-down regulators. These are time based, firing brief pulses to charge an inductor, then letting it freewheel the load (CPU). Because the transistor is saturated, the losses are comparatively low.

[@RetroTinkerer]

@@HighTreason610 I believe that for my general usage and gaming I was super happy with my 6x86 but I had an early 2x CD recorder and each failed burn cost me 10$ I ended up selling my recorder and 50+ CDs and as the amount of failures was really making me feel down. Later I got an Asus board and Pentium Pro 200 (256Kb) believing that Socket 8 might be around for a while like Socket 5/7 but damn I was clueless. That was the first "full upgrade" I did myself. Thanks for your long reply, glad to see you back and hope I can see more of your awesome stuff around here.

[@RediscoveringRetro]

Happy New Year. Glad to see another video from you. Look after yourself mate.

[@Sharkie1717]

So glad to see a new video after the long break. Love watching you tinker with interesting old hardware & software. Really enjoy the classic dance music included sometimes too. Thanks for another quality upload, and for the update included too. Wishing you all the best for the year ahead 🤞 Take care ❤

[@IronicTonic8]

Always good to see a new video

[@nikosuoa]

Awesome, what a way to start 2025 :), best wishes HighTreason610!

[@AncientElectronics]

good timing, I'm currently in editing for my latest video which is a bit inspired by one of your older builds. I mention you being MIA for some time in it. I'll have to edit that out now. glad your still here.

[@HighTreason610]

That's OK, I usually only notice things that need editing _after_ the video goes up. Looking forward to seeing what you've built.

[@nislab_]

Welcome back 😊

[@johnmay4803]

good 2 see you back pal happy new year

[@Daniel-j3m6q]

Welcome back, finally a new video 🙂

[@joshuareynolds6958]

holy sheeeeet! where you been brother?

[@NavJack27gaming]

very glad to see you back at it.

[@georgez8859]

Great Video Thank You and Happy New Year my Friend

[@BeerAndWarcraft]

Welcome back Sir!

[@AngelDemonn]

Glad you are OK mate! And not too shabby of a video! Really interesting how good the 6x86 holds over yhe K5! And Happy New Year! More videos this year pls! 😅

[@ran2wild370]

Oh holly crap!! This is like seeing a ghost of your channel at unexpected moment. Lol.

[@grimreboot]

I'm a bit late to this one, good to see you back, and thank you for the upload! (High treason ... RIP intro!)

[@HighTreason610]

Thanks. The current logo iteration isn't staying, but won't change very much - probably just fixing up the jingle slightly and figuring out why I get stalls in the starfield room's viewport.

[@RoyAntaw]

Happy new year and welcome back.

[@BruceEverett]

Welcome back, High Treason. My understanding is that there was a pipeline feature in the Pentium that allowed an integer operation to get a head start on execution when following a floating point instruction. All the Pentiums had the feature, and some of the late K6s did too. You could take advantage of it if you interleaved instructions manually - say if you had a lot of floating point divides each followed by integer adds - but compilers didn't optimize for it, so a lot of software didn't take advantage. Quake apparently did. A lot of people attribute the better performance of Quake of Pentiums to the better floating point unit, but the floating point unit of the K5 isn't that far behind - IIRC the K5 needed about 23 cycles for a floating point divide where as the Pentium needed about 19. Cyrix FPUs were further behind. The slightly slower FPU of the K5 isn't slow enough to account for the performance relative to the Pentium, though. I've heard it suggested that AMD had considered having the mentioned pipeline feature when designing the K5, but rationalized that because it required manually-written code to optimize for, it wouldn't be taken advantage of by most software, and that the K5's use of out-of-order execution would pick up most of the slack (without requiring any hand-written code) anyway. Quake was an exception to this. There's a video on KZbin somewhere that explains how Quake's instructions pass through the respective pipelines of the K5 and Pentium that explains all this better than I can, although for the life of me I can't remember the name of the 'Tuber. Apparently the out-of-order execution of the K5 doesn't help enough to close the performance gap.

[@HighTreason610]

That explanation definitely makes sense. Admittedly I've never dug super far into how ID's engines work, with my area being Build instead, an engine that does use floats for sloped surfaces and the split screen demo playback sure suggests the FPU isn't necessarily as weak as some make it out to be. At some point, that engine got optimizations made for P6 processors, but those weren't added until somewhere in 1997 and even then, the engine was generally well optimized to begin with.

[@rasz]

You got it almost right. Its FPU instructions being able to run concurrently with other FPU instructions in the Pentium. FDIV is one of the only ones that cant do it tho so doesnt explain Quake Pentium advantage. Its all the other Floating point code doing geometry rotation that takes advantage of it. Duke3D is not a read full 3D game, it doesnt really do 3D rotations because outside of slopes special case (like starting roof) everything else walls and floors/ceilings are at 90 degree like in Doom.

[@HighTreason610]

Depends on how you define "fully 3D". Duke's engine certainly handles many things in three dimensions, mostly using longs and the geometry format itself never uses floats - only the slope _rendering_ code ever does that. I'm not actually sure why floats are necessary for Quake to begin with, as it's generally the uglier of the two titles, leaving one to wonder what is being gained other than, maybe, the ability to look up and down or roll geometry on its axis. Even then, I'm fairly sure other engines managed to do this without relying on floats, especially as there's no shortage of 3D titles on PS1, for example. Yet again, while I know my way around Build's internals and map format fairly well, I admit that I'm generally ignorant of the internals of IDTech, so all I've got is "nothing else seems to run this badly.".

[@rasz]

@@HighTreason610 Duke3D rotates world in 2D - fast and easy, Quake does full rotation matrix multiplication for transformations in Euclidean space. Playstation had special very fast fixed point coprocessor for rotating matrices and the result was fast but jittery triangle points with gaps in geometry :) Descent did fixed point math (VECMAT.ASM) on PC successfully at the cost of order of magnitude less geometry than Quake. Average Quake monster has more polygons than whole world visible on screen in Descent at any moment. The choice was either very blocky sparse world using fixed point math or taking advantage of new tech inside Pentium and screwing non Intel users. Im not arguing Quake Brown on Brown gameplay is superior :) but engine was pushing way more stuff and needed some trick to get there, FPU pipelining was that trick. On Pentium you can have several multiplications (for rotating world) all starting one cycle after another and effectively running in parallel, AMD/Cyrix CPU will wait until first FMUL is finished before issuing next one.

[@JapanPop]

Welcome back🎉

[@screamengine]

Awesome fun vid as always! The best years, for me, for PC 1994-2002. So much hardware and evolution!!!!!!! Midi is still awesome! :)

[@FOIL_FRESH]

i've missed you papa treason

[@RetroRawit]

Happy new year!

[@s3vR3x]

what a great way to cap off 2024!!!!

[@naib_stilgar]

Welcome back. 2025 is already off to a good start then 😎

[@MrKillswitch88]

This is what us poors bought when the cool kids were getting systems that were worth more than a good second hand car.

[@RJRC_105]

Very similar to what I had back in 1998, namely a 200MMX. No 3D card and a recycled Soundblaster 16. Got the job done and ran most things until 1999.

[@jwoody8815]

Never had any Cyrix chips, Mostly AMD chips for some reason, explains why im an AMD user today, first truly new machine I ever owned/built was an AMD K6/2-266, overclocked to 333MHz and ran it like that till I sold it piece by piece. Built it my first year in college just out of High School around late 99, Great machine, 64MB RAM and ran Windows 98SE. It later evolved with a bigger faster HDD and much more RAM and more advanced OSes. (Windows NT, Linux etc) -New subscriber, hello from the USA.

[@HighTreason610]

Much appreciated. Makes total sense at that time, the K6-2 was pretty much uncontested on performance per dollar for quite a while, as far as I know.

[@jasonhaman4670]

VMT... hear the farts... DP... LMAO. I've been missing you, and glad you're back (hopefully the fates leave you alone and you're back for a long time). I'd say 'never change', but I'm sure you need no encouragement to keep being yourself. You're not everyone's cup of tea, but maybe a glass of Islay malt, for those of us who appreciate something unapologetically distinctive, with personality. I've never understood the retro computing people that go upgrading and overclocking and etc. -- seems to completely miss the point of retro computing to me. Refreshing to see someone on the same page.

[@HighTreason610]

It's always worth remembering that most software was written to run on the lowest common denominator anyway, or else it wouldn't have sold well. My policy has long been "default everything" to get a more realistic perspective on how things were at any given time, if only for the fact the average user would never change anything but the desktop wallpaper. The goal is to experience what things were like right out of the box, aside from some of the stranger machines in the collection.

[@Ozfrank]

HighTreason dude - always like))

[@johnsmith-fr3sx]

Great to see you back. I think your cluster of failures is related to your building electricity. I have had bizarre behaviour with power supplies in the past. A relatively new supply would work with vintage motherboards and then magically stop working. The PS was fine itself and not failing. The electrical noise coming into your building can vary and your power company may have had issues. The PS cannot filter all the noise and you were "lucky" to get effects in your components.

[@HighTreason610]

I could blame the electricity supply, but a fair few things weren't even plugged in. I have no solid answers and can only think "some crackhead was trying to talk to aliens" or something. It seems something _was_ going on, because people on the street were having trouble with TVs and such in that time and then it all just stopped. Of course it could simply just be a string of really bad luck.

[@FOIL_FRESH]

apart from the heat, these cyrix cpus are fine! we had the 100mhz (wow 120P rated!!!) version and it ran command and conquer and duke3d just fine back in the day. my first win95 system. i have a real soft spot for it.

[@neuronic85]

As someone who has watched UK TV shows for years, I thought I was used to your myriad accents. I was clearly a cocky bastard. Guess I just have to watch a few more of your videos to get an ear for it. Nevertheless, this is a fantastic video about a neglected CPU. My first computer was a Cyrix, so while I don't think of it in fondness, I respect its place in my history. It played those Apogee & Sierra games beautifully and ran QBASIC like a champ. It came with Windows 95 OSR2, which many people forget about when talking about 95.

[@HighTreason610]

You'll have to look pretty far to find another accent like mine - I don't know where it's from and suspect it's an amalgamation of things. This is on 95B, a version I like to use. People claim it's unreliable and it probably is, but nearly 30 years has made dealing with that something of a reflex. Much like the speed, it doesn't matter very much anyway because the computer isn't doing anything critical.

[@Xan0906]

The stutter in Duke rather looks to me like what the game does when running on a FAT32 partition with no Smartdrive.

[@HighTreason610]

That's definitely similar, but it should be working. For a while I wondered if perhaps the SCSI card didn't support DMA, but apparently it does. I've seen stalls like this when using SCSI to perform small transfers quite a few times over the years, though for some reason it was particularly pronounced on this day.

[@LilMissMurder3409]

Ah, the dulcet tones of phlegm-coloured cases and farty beeps emanating from the PC speaker. Thems were the days.

[@davidp4456]

Do you have an LCD display for your K5 box? I bought the same box new and still have it but I would like to put in the LCD display but don’t know what to get or where to get it. I’m definitely going to bed now.

[@Sabrinahuskydog]

About your newer system: All modern motherboards for desktop are 16X PCI-Express in the top slot and 4x in the bottom slot. This is standard today. It's not this specific motherboard lying to you. What you should try doing is putting the video card down in the bottom slot then you would have a full speed 16x slot up top to use for capture cards or whatever else you need. And yes I also wondered what happened to you and worried about where you were. It's nice to see videos from you again.

[@HighTreason610]

There's no doubt they're all basically the same, but I resent the official spec claiming things that don't exist and the BIOS having the option to enable them. The GPU won't fit in the bottom slot as the heatsink would protrude past the bottom of the chassis - or in this case, the inexplicably placed PSU. For the record, some vendors specify such slots as "PCI--E Xn in PCI-E X16", or use a shorter slot with an open back, which is far more honest. In any case, with the card configuration as it is and the flash drive wasting lanes, all slots apparently run as X1 only. I'm sure this is totally gonna gimp that future GPU upgrade. In any case, thanks.

[@BrassicGamer]

Yep, data loss is probably the worst part of the hobby. Old computers randomly dying is the second worst part. I lost my entire digitised game collection, which included all the levels and patches I had downloaded from places like gameburnworld, which has gone to shit now, and all the images I took of my optical media. Obviously I can get that back but the patches and such - that's years of work. I'm just lucky that the irreplaceable stuff is hosted online and is safe for now. Sorry to hear it's gone to shit but hey, you're here. The irony is that my old hard drives with game saves and such that aren't backed up anywhere are the ones that keep on working. I've got a few Cyrixes that I haven't gotten around to looking at yet (IBM and Cyrix versions of the P166, plus an MII), so I'm now even more intrigued than I was. Oh, and despite the ableism and homophobia, it's likely that "Doom sucks" is the thing that is most likely to see you cancelled.🤣

[@HighTreason610]

When the Xeon drives went, I had actually been working on an episode for Duke 3D. By pure luck I'd just sent the first zip to beta testers a couple of weeks before, so at least managed to get that back. Thus far most things I've recovered are things I just happened to have posted places online. Otherwise, all my data going back to 1994 is gone aside from what was copied onto old PCs. Will attempt sending drives for data recovery once I decide which ones are the best candidates - system drives, backup drives or off-site backup drives. Nobody said the soy breath person was gay, they might be straight (personally I picture one of those "metrosexual" types), it's more that they lack self respect to the extent they'd get down on their knees, even if only metaphorically.

[@jdmcs]

I appreciate department store PCs! 🤓

[@wskinnyodden]

The only CPU in my life that I ever saw after self-destructing itself. Yep, they do run hot as crap, to the point the golden top on the chip will have the solder melt and the have it slide sideways. Yeah, should have taken a photo (I belive that particular PR200+ had previously been mine, but I may be mistaken there were many around at the time)

[@HighTreason610]

Now I'm trying to remember where I got the chip from. It was some years ago now and all I can recall is it being in some random job lot from a closed shop. If you had a PR200, it won't be the same chip as these earlier ones lack the "R" in their rating.

[@wskinnyodden]

@@HighTreason610 Same crap different smell, neither liked to run cool. Honestly these Cyrix would be better used to fry some eggs :P and buy a K5 or K6 and overclock the BUS to match the Cyrix in MHZ (meaning you will need to lower the multiplier) And bang, better option unlocked (assuming it runs with a 75Mhz BUS not all will, the bus speed is also the reason I prefer the DX2 80 to the DX4 100, and prefer the DX4 120 to the DX5 133)

[@HighTreason610]

While I run some fairly weird hardware, I learned years ago that using the common denominator was generally the best option when expecting things to just work, because most software targeted such things. Is the DX2-80 faster than a DX2-66? It is, but everything else probably expects a DX2-66. Can I shove a 486DLC into my 386 board? Probably, but most things expect either a "real" 486 or a 386DX-33, so it's not necessarily the wisest thing to do and usually has some caveat attached to it.

[@wskinnyodden]

@@HighTreason610 True, though my approach was server performance at the time hehe, and a good FSB with a caching storage controller would do wonders when granted such goodies hehe (Still miss the DX50, first CPU to demand a cooler by the way)

[@PROSTO4Tabal]

What city you from ?

[@HighTreason610]

Kingston Upon Hull

[@PROSTO4Tabal]

@@HighTreason610 Oh wow far away I am from Plymouth. I like retro pc hardware too. Thanks for new video

[@danthompsett2894]

got to love those stick the screwdriver in to prize the cpu cooler off, coolers although to be fair ive only accidentally slipped and damaged a motherboard once in my life time.

[@HighTreason610]

What I've seen more is the metal bracket itself gouging the board when pressed down with the screwdriver, or the plastic peg on he socket getting rounded off by it because people were scared to press it far enough.

[@BrassicGamer]

@@HighTreason610 I wonder how many motherboards have been deeply gouged during this process.

[@davidp4456]

Eh up HT. It’s good to see you doing another hardware review. However I’m going to bed now so I’ll watch this tomorrow. Cyrix cpus are interesting beasts that at the time I sadly dismissed as cheap crap and avoided them.

[@davidp4456]

Btw. Happy New Year. The fekkers over the road have quit their fireworks now, so I’m off.

[@danthompsett2894]

i think you need to set up a patroen, pc parts even old used parts arent cheap and dont grow on trees.

[@WaybackTECH]

Dang man! WFT? So the cloud backup didn't have their own redundancy? You might as well have gotten crypto locked. That is terrible you lost all of that vintage hardware. I've thought in the past you had crappy electricity but now I am wondering if you are getting radiated with 5G or some signal in the air they are pushing out.

[@HighTreason610]

This wasn't a cloud service, I just had hard drives stored off site. A fair amount of things should be repairable and in about half, only the NIC got hit, but in such a way that the PC won't even start with it in there. Given how much of it wasn't even plugged in, my suspicion is indeed some outside interference, or "a crackie trying to talk to the aliens", at least as far as NICs, MIOs and video cards. I don't think it was deliberate. Could just as easily be something else. Unknown. I have yet to figure out how the off-site disks failed and also have to decide which ones are best to send off for attempted data recovery. Leaves me unsure of whether I'll bother backing things up in the future. What's lucky is that most of the newer PCs only lost a PCI NIC. The ones that lost other things tend to be way older and therefore, can be repaired indefinitely. The Turbo XT, for example, is generally un-killable because it's all off-the-shelf TH TTL stuff. It'll all be soon to as and when I get to it.