The only problem is, I still don't know how to top this one, not much left to dig up with CPUs aside from ones that are either too elusive, suspected fake, or offer no incentive to be found.
Farscape, hell ya. I like the case. I am surprised you didn't link the cleaning video. Interesting performance. I have never found, yet, a performance difference with the 486 SX and DX chips in these tests, always scoring identical. Must be a 386 thing I guess. Almost like a bug that is being worked around with the Intel chips or something. One of the Cyrix 387 variants would be interesting to throw at this. Late 90's, when the K7 came out, it was rumored that Intel was limiting it's supply of chipsets to certain motherboard makers ( ASUS ABIT ) that were taking their chipset and adding overclocking features. There was also a rumor that ASUS was keeping their initial Slot A motherboard on the "Down Low" , sold unbranded and in a white box, as to not anger Chipzilla, their bread and butter market. No idea if any of this was true, but there was a lot of stories like this, so I would not be a bit surprised if Intel was strong arming the OEM market into not using C&T 386 chips, probably citing this lawsuit as a reason and if used, Intel would cut off supply of their processors.
They shouldn't have cancelled that show, it had at least one good season left in it. Yeah, the 486 is a good baseline because it _should_ be identical if the test isn't using the FPU, so it's definitely limited to this platform as far as I can tell - not like they made a "Pentium SX" or anything to compare it with really and I don't have a 287 - so it definitely seems to be that Intel lose speed somewhere, maybe motherboard dependent though. I'd like to test a Cyrix FPU but don't have one, supposedly they were faster than these by upto 10% according to one or two places. The magazines back then said something about Intel doing that, would not surprise me to be honest, they still don't like overclocking but it seems they're resigned to the fact that it's going to happen now. If memory serves correctly, Intel were found to be paying Dell to use only the Pentium IV at one point and also bribed stores, they got fined a billion or so dollars for anti-trust, seems easy to find on the web anyway. As for the whole advancement of technology thing, it seems the whole world has stagnated a little. We had over a century of innovation but I feel like we're about to go into another few centuries where nothing moves on that much.
This is what I love about old computers. all these years and I can still discover something new and interesting. Thanks for the shout out btw. My Contaq board is also known as the MS-3124. over at Statson.org there is a settings and configuration sheet which lists the C & T 386DX as a compatible processor.
These things are definitely still full of surprises. No worries about the shoutout, you provided information and it was your board I was referencing, so it would be poor manners on my part if I tried to obscure this fact. The extra pins being the same side and the C&T compatibility does definitely suggest some common intended feature between both boards, though it's still not certain exactly what that might be without access to chip datasheets nobody appears to have anymore, if they were ever really available in the first place outside of closed design documents.
@@HighTreason610 I know this is quite a while since you posted this video but I was messing around with this motherboard again this morning and I finally took a good look at the manual which is now available online. in the user manual it does directly state this board is compatible with the Super386 38605DX processor.
Great work dude! Was always curious abt exactly this comparison. Unfortunately I don’t have the C&T 386 in my collection, even I am counting thousands of CPUs in my collection. 👍🏻
Thank you. Oh I know the name CPU Galaxy. The C&T 386/387 are indeed evasive little things, took years of just missing them to finally happen to be in the right place at the right time, it really was down to luck and I hope you manage to find one yourself some day. My collection is comparatively small, probably less than a hundred chips, but there's not much left that I want either, maybe one or two SoCs left and a couple of CPUs I'm not sure even exist. We recently found a rather odd behavior with the C&T CPU getting stuck in some kind of 'unreal' mode when it shouldn't, which triggered when drivers/TSRs were loaded in a certain order. This makes protective mode software act weird or crash, but we have been unable to determine whether this is specific to the configuration of this system in particular or the CPU itself. Was trivial to work around and most users probably wouldn't run into it, but if you ever play with one I'll be curious to know if the same thing happens.
Even with the things I have, there are still things that were thrown away, sold, broken or lost that I wish I still had. Think it happens to all of us.
Great little history and benchmarking on C&T chips! The markings on one of those motherboard chips looks like how Toshiba numbers their chips too (TCxxxx..)
I suppose it's possible Toshiba fabricated some of them, only the CPU is known to be from TI and I'd imagine that outside of that, C&T would probably just use whichever plant was available or cheapest at any given time. Archive.org (from the other comment) is a good resource, but I'm sure they keep deleting stuff which is quite annoying.
They get the job done I guess, it was hard to follow initially which is why the charts were added. In response to your other comment, yes, at that time, console was still king, but only just, the technology was there and people were starting to use it, so the PC was about to overtake them in capabilities. Consoles still sold more though, I'd bet, and the gap probably didn't start closing until the latter part of the 1990s when 3D acceleration got good and gaming on the PC started becoming a larger scene, with some systems even being built for that very task first and foremost instead of work. PC gaming was a fairly big thing in the Athlon days.
Thanks for the answer HighTreason. I totally agree. Half-Life was probably the breaking point, a Playstation couldn't do half of that. I skipped that whole generation, I still had a P2 by the late 90's I think. I remember we still had a 5x86 133 alongside the P2, and Half-Life would run in both! (with very different resolutions xD). Then later, P4 came in, socket 478. But in my mind "PC games" still means something different than console videogames. First, they made you type. Even to start the game. They involved a lot more logic, not such good graphics but solid stories, adventures. I'm thinking about Monkey Island, Leisure Suit Larry, even Alone in the Dark. Long live IBM PC/MS-DOS.
It might be because C&T made use of the EXTRA FPU power, on the other socket, to improve performances overall? Quickly said, the CPU maybe had instructions for using built-in FPU + 2nd FPU if available.
Indeed, that footage is also pushed slightly to the right. It appears to be a timing issue with the AITech capture card and its digital keying circuits, it happens from time to time although this has no effect on speed.
Maybe a FP precision test would reveal the reason for weird difference in intel/c&t combinations? Necroware I think found math accuracy of different FPUs was wildly different around this time period, from memory...
Possibly, but I was never able to find such a test and wasn't too bothered, as I tend to just go with "it seems to work" when tested in applications. The datasheets might have some information on the precision level of each IC. In real world terms I'd hazard a guess that they're at least comparable and the literature seems to suggest as much, although some very specific operations may complete 600% faster on the C&T chip. This is interesting, as it is similar to how some integer operations complete 700% faster on UMC's processors. Of course applications don't tend to be made entirely of those small few instructions, so we see the much smaller gap observed here in practical use.
A short track I slapped together with the CZ-1 for fun, it isn't really complete and has some nasty timing issues when heard on its own. It may be developed further in the future but I'll share the version from the video if you want?
Time and time again we see Intel stifling innovation by nefarious means to make sure no one takes a slice of the PC market. I would have loved to see what C&T could have become.
Seems that every attempt goes the same way, inevitable court case. It's small wonder things stagnated for a while with only AMD left and no Cyrix, C&T or UMC to shake things up. Overall I'd wager UMC would have been a larger threat to Intel if they stuck around, if only because they have their own large fab plants and being located in Taiwan, could probably have undercut the others. Still, Cyrix (TX, USA) and C&T (CA, USA) would have forced innovation. With their efficiency and later their strong presence with VGA chips, I could see C&T having done well in the laptop world if they'd stuck around, or who knows, maybe they would have gone on to build some high-end CPUs later. They did make a SoC at one point, the F8680, though it appears this was only an 8086 core.
They've had quad core procs for years that were fairly decent at the time but for whatever reason they never really made it to market. Intel hogging like 93% of the market didn't leave much for AMD either. :(
Most likely, though I'm actually quite curious as to how this board handles memory and cache because the speed is unusually high, almost to the point where I have to wonder if some kind of interleaving is happening.
Unfortunately (for me), SST 4.78 needs at least 4 MB of XMS to perform bandwidth tests, while I only have 4 MB in total on my 386. Sigh, those numbers you got in the test are impressive.
If you're looking for CD-based PC games, try Rebel Assault, it pretty much run everything from the CD, only having very few and very light files installed on the hard disk. This game also got released for Sega CD but the control suck because there's no analog joystick for the console. You NEED an analog joystick for this game, and if possible, auto-fire option, it might make it much easier.
I'm somewhat familiar with this game, as in, I knew someone who had it on PC years ago. Might give it a shot at some point, just to see how the Sega version holds up. As for your theory on the C&T's FPU usage, I don't know, maybe it does, or maybe it just handles things more efficiently. You have to wonder if the notes I read saying it was slow were perhaps running in a board that wasn't compatible with it, the way it communicates with the rest of the system is clearly different regarding cache and memory, so it stands to reason they probably played with the FPU communications in some way too, and that some boards wouldn't support it.