|
|
Intel P5
Posts: 85
Location: UK | @DXZeff
I've had the Microsoft sound card for a good few years, used with my IBM 386sx for a good while. A pal in America sent it to me after finding it packed away in his office for years, knowing I liked old computer "junk". It works very well in windows 3.1 as you'd expect, and the soundblaster compatibility option for dos works OK letting you select standard IRQ, DMA etc. Nice thing about it is the Phono outputs on the card, nice to hook it straight into an amp with standard leads. I'd say it's a fair choice for a 386 or old 486 system, but I don't know much at all about soundcards in general, and frankly hate messing with the DOS drivers plug & play cards need. That's why I bought that Soundblaster 2.0, it's just a case of plugging in and adding the setblaster line, easy peasey!  A lot of things I really love fiddling with on old PC's but sound card drivers are not one of them! |
|
| |
|
TM Crusoe
Posts: 671
 
| DXZeff - 2017-12-11 11:23 AM
@Brostenen; Somehow forgot to add lto my last post; I guess hooking the CD-ROM drive up to the same card as the hard drive wasn't so common at the time, there are some that can't see them properly or at all. The whole cached controller coming in slower, though, that's not what I'd expect at all, it's not missing a TSR or something is it? Could swear I've run into SCSI controllers like that before, you'd think the card's BIOS would handle it, but perhaps not.
I tried different IDE CD-Rom drivers with the drive, and different drives as well. And though it is an IDE controller, it behaves much like a SCSI in that way. It might be a timing issue between the actual drives, that this controller does not like or it might not support IDE CD Drives at all. Or it might be that controller prevents any drivers from communicating through the BIOS and CPU. What it is, I really don't know.
Regarding the SCSI thing. Yes. I have seen that as well. My Adaptech 1542 shure has this issue of needing a special TSR loaded before the actual CD driver being loaded. This will cause some issues down the road in pure Dos. So I am only loading the external CD-Drive, whenever I need to copy files to the harddrive. This however is a pure DOS issue, and I have no issues in other OS's like Win95, NT or Os/2. So my ISA SCSI setup is installed in my old Unisys 286 machine, alongside a 40mb 3200 Conner SCSI Drive. Wich is a huge upgrade from the original 20mb Miniscribe MFM drive.
I might try and test the Cached controller in a Win95 or Os/2 setup. This way I will see if I can get it to run a bit better.
Anyway....
I have a plan with that board I got from you. I have recieved that rechargeable coincell battery, and I am waiting for a clip-on cooler from USA. I have asked both of my children, if they are up to the challenge of trying to build a computer. My daughter is 9 years old, and my son will turn 7 in jan. They are both up to the task of trying to experience how a computer is being build from the ground up. Installation are of no interrest to them, only the assembling and then pure gaming. So in Januaery, I will be putting them through the process of building a computer the old way. And they will have a basic understanding of how it is put together. And to be honest. They will try something that only a select few children will ever experience today. Yeah... They are lucky. They have a grandfather who raises free range chickens, and my children have seen the slaughter of chickens each year. You know... Axe, boiling and plucking of 120 chickens by hand. They have a father (me) that are into technology. An uncle (my brother) who are into survival alone, in mountains, during the winther on skie in the middle of nowere. He is like a modern crossover, of a budist, liberal, socialist, redneck and hippie. The best of all five things combined. He eats averything eadable, respect everyone, are super calm, practice no religion and like to learn how to do everything on his own. And then a second grandfather who are into gardening, geography and history. Yeah... They are lucky.
It is actually really fun, that people say a 486dx33 can run without cooler. Personally I think that the chip is running too hot for me to trust. It might run without breaking/dying, yet if something gets too hot to touch, then I think it requires cooling. Better be on the safe side and have it running the next 10 years. And stability wise, a heatsink with fan never herd anyone.
Edited by Brostenen 2017-12-11 11:29 PM
|
|
| |
|
TM Crusoe
Posts: 671
| Recieved this little thing in the mail. Was still, is no more, factory sealed....
 (486-Cooler.jpg)
Attachments
----------------
 486-Cooler.jpg (609KB - 494 downloads)
|
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | @CyrixInstead; The phono outputs would bother me if there wasn't a regular jack on it, though in this case there is one as far as I am aware - can't think what else they'd need that many for anyway - but that's only because my existing cabling is set up that way and my amp doesn't use those, it uses DIN.
@Brostenen; I actually meant a TSR for the caching capability, wondering if perhaps it doesn't run as effectively otherwise. No idea, the only 'caching' controllers I've gone near are those large RAID cards with SDRAM slots and have no experience with these older ones at all.
Sounds like a good plan to me, it's a simple enough board to be a good starting point and it should be clunky enough to be resilient to childish handling. Heat seems to vary greatly on which S-Spec is on the chip. SX419 gets extremely hot whereas SX810 does not, they're internally different and even report a different CPUID despite both being an Intel 486DX at 33MHz. Supposedly the older ones dump about 1.5 Watts more (Still barely 5 Watts) power as heat and also don't have the "SL Enhanced" power saving, not that this will make much difference without DOSIDLE installed anyway. Regardless, heatsinks are always good and can't hurt (well, unless they're made of something less thermally conductive than the CPU is, I guess, but that never seems to happen) so are worth installing even if it is just paranoia, just for peace of mind. Personally I always install one, even if it's only passive as is the case with my own DX-33, the U5S also has one despite running barely warmer than room temperature.
I once read that all microchips will run hotter over time due to electron-migration or some such, but I have no idea how true it is and doubt that such an effect would happen very quickly at such low power levels anyway, it would probably take longer than our lifetimes at any rate. That and every so often a P75 or even a P100 will show up with no heatsink on it that is still running, these older chips were quite tough and rated to fairly high temperatures. |
|
| |
|
TM Crusoe
Posts: 671
| The caching stuff is all build into the card it self, acording to the manual. No need for any special TSR on this card.
Though it is recommended by the manual, to disable caching when formatting any harddrive.
It does mention something about the card using something called "WD-1003 Protocol", what that is I do not know.
Regarding the model numbers on dx33's. I did notice that the older model that I have is running way more hot than
my 810 model. So you are right on that. I actually thought that my fingers played a trick on me regarding heat. |
|
| |
|
IDT WinChip
Posts: 237
Location: USA | WD-1003 was the ( my understanding anyway ) original IDE interface, in the 80's was commonly called the "Winchester Disk Interface". Pre- Enhanced IDE standards. |
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | Yeah, or pre-ATAPI, hence the whole CD-ROM problem.
Anyways, I tested the Chips 386 and...
Here it is versus a Cyrix 486DLC-40;
Impressive? Or maybe not, here it is versus an Intel 386DX-33;
As I had suspected, the motherboard appears to be a bottleneck likely due to the older VGA card and the ISA bus - mostly that peculiar VGA card as its weird timings seem to drag it down somewhat. This suits me fine as I still don't know the performance, but I do know the chip is working and besides, I might just have a solution to this that would see the CPU become the bottleneck instead in another board. On the down-side, the Shuttle HOT-307H I used to test it in here only detects the first 640K of RAM with the chip installed and I don't know if the other board I mentioned here will do the same, because it's a weird board, but I don't know that it's weird enough to work with this chip. So yeah, standard pinout for sure, but there's clearly a compatibility issue of some kind. A shame as I planned to leave it in the 307H permanently, namely because I knew the 486DLC was being wasted and the other board is much better for it. Seems this Chips 386 probably won't end up in permanent use then, but it will still be interesting to test properly and I still, honestly, feel quite honored to actually own one given its historical relevance.
BIOS updates for the 307H aren't really an option, if any even exist, because being a fairly late 386 board (around 1993) Shuttle cut costs by soldering the EPROM (yeah, so not even something I can flash in board, if the circuits existed in the first place) right onto the board. Here's hoping the AOpen has better luck with it, though as I said, that will be the new home for the 486DLC, so the HOT-307H will probably be scaled back to an Am386DX or something. I doubt that there are any updates for the VI9 if it also does not support the chip. Still, nothing blew up and that was my main concern here. Plus, at least for now, I like the actual performance of the chip remaining a mystery for a little while. |
|
| |
|
IDT WinChip
Posts: 237
Location: USA | I don't think a bios update would really make much, if any difference anyway.
Edited by waybacktech 2017-12-16 11:45 PM
|
|
| |
|
Intel P5
Posts: 85
Location: UK | Great news that the Chips CPU is working, must be something to do with the Chipset on the motherboard and the way the CPU handles the RAM addressing I guess? I suppose it could be some BIOS settings even? I know my 386DX has a VERY complex BIOS setup whereas the SX is simple as anything. |
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | The AOpen board does not boot with the Chips 386 installed, it just locks with the keyboard leds all on.
The chipset was what I was blaming on the Shuttle, because the memory detection issue also happens if the Cx486DLC (mostly its cache) is wrongly configured, which is why I pondered updates, but I don't think there ever were any. Either way, looks like all tests will have to be done with the Shuttle because it's the only one I have which boots, so unfortunately we'll be stuck with only 640K of RAM, limiting what can be tested and demonstrated.
|
|
| |
|
IDT WinChip
Posts: 237
Location: USA | You know Paul, you should keep an eye out for a CHIPS chipset 386 motherboard. Seem to me if anything would be compatible with that chip, it would be their own chipset. I found this one on ebay, ships world wide, is new with cpu btw... might want to have a look at it and see what you think. https://www.ebay.com/itm/New-Bulk-Peak-DM-386DX-Motherboard-from-Tok... lkAAOSwAvJW-YLf
Edited by waybacktech 2017-12-17 6:58 PM
|
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | Haha, they say great minds think alike and all that. I actually bought that exact same model of board through that same thought process, but this board came from elsewhere. It has a C&T BIOS where that one seems to be AMI. I looked around for a while and tried to find one where the chip bin numbers were from late on, the DM386 boards seem to average between '92 and '93, so might just be late enough to have proper support. If that doesn't work, I don't know what will. But I may have a backup plan if it doesn't.
The only negative here so far is that it's ISA only, but then I doubt the CPU is as fast as a 486DLC-40, heck it might even be really slow and the "10%" claim dotted around the web may refer to the J38605 version (Definitely wouldn't fancy the odds of getting one of those going) instead. I did do tests, but will share them another time and would rather wait to see if they can be done better if this Peak/DM-386 works. NSSI actually detects the chip though, as an SX model, but it does detect it. Haven't tried PCCheck, but would not be surprised if it detected it as well, that program won't reliably detect anything else but less common CPUs, it'll detect those without fail so far.
If that motherboard doesn't work, I might take a gamble on one with a UMC chipset in the future, purely because every 486 UMC board (8498 / 8881) has always had very good compatibility and that might extend to the 386 platform too. They show up at reasonable prices from time to time. I suppose on the up-side, this adventure sure is netting a good pile of 30-Pin SIMM RAM which I was short on before, so that SX can finally be started up and some sampling gear is going to get a long overdue upgrade if nothing else... Except that bloody SampleCell... 4MB modules only... to save you money they said (actually, it kinda does as doing the same with 1MB sticks would be more) but that's already taken care of. |
|
| |
|
IDT WinChip
Posts: 237
Location: USA | I imagine the odds of that happening are quite astronomical given the staggering number of 386 motherboards that were produced. Of course, narrowing it down to CHIPS chipset would increase those odds. Sure looks like a pretty BOSS motherboard, as far as 386 boards go. Only thing I don't like is the clock gen is soldered on. I hope it's operating at 33Mhz. Is that connector above the ram for ram expansion card? I thought it was an interesting looking connector that might house something fun.
I did a test with my ET4000 ISA video card, the Diamond Speedstar 24, at one time with my K5 and was still pushing in the 20's with quake so I figured the ISA bus probably wasn't a factor in holding the card back on, say 486 or 386 boards. I personally haven't "felt" a difference between a good ISA card and a VLB card up to, say maybe a 486 DX50 or even 66. Sure some of the benchmark results, especially Topbench, say a different story but when it comes to how a game feels, nothing there really. Windows might be a different story though in some cases.
|
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | In the case of the J38605, it uses a different pinout and I've never seen a board or even a CPU, so likely odds of nil in that case.
As for this one, yeah, that connector looks to be "32-Bit external memory" from what the internet says. Every one of them I've seen is 33MHz but there may be a 25Mhz version out there going by this - http://www.uncreativelabs.de/th99/m/S-T/30937.htm - crystals aren't hard to swap over though.
With ISA, there may be something else crippling the Shuttle board, given it really choked on Nitemare 3D in some later levels. Possibly that weird BIOS bug it ran into a while back has flipped a setting over that has gone unnoticed, not sure, didn't look that way but who knows what kind of nonsense is going on in the CMOS RAM. Could be worth just wiping it out and starting again to make sure, likely to happen when everything gets switched around anyway as the system will not be staying in its current configuration or case. |
|
| |
|
Intel P5
Posts: 85
Location: UK | You could always message the person on OS/2 museum to find out what motherboard they used to test the Chips 386?
http://www.os2museum.com/wp/the-forgotten-386/
Seems a few of us are on a 386 "kick" at the moment! I'm really enjoying playing with my two new ones  |
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | Suppose that's an option, maybe if things don't work out.
As it stands, I personally fancy the chances of the Peak DM-386 for one reason, something that I couldn't quite figure out;
Notice the CPU socket? There's an extra row of pins that aren't used. I've seen stuff like that before, sometimes unused pads for a 486, sometimes the FPU socket has unused pads for a Weitek, but there are not enough unused pins around that CPU for anything commonly encountered.
On the other hand, I have found a picture of the J38605DX chip and, well;
(I hope CPU-World don't mind me borrowing their images)
It seems suspiciously similar to that footprint around the CPU socket there. It might be possible that this board was designed to work with the 38605 at some point, a chip which was supposed to have a different pinout, I even do recall seeing 144-Pin instead of the regular 132-Pin somewhere. It is possible that the extra pins were shaved off later in its lifespan, that some of the documentation I read is wrong, or that the chip was cancelled entirely. Either way, this would leave little reason to fit unusual (also likely more costly) sockets to the motherboards built to work with these CPUs. Whilst I don't have a 38605, if my suspicions about that socket are true, then it would seem support for the 38600 is far more likely.
No deliberate 386 kick here really, had work to do with mine and just happened to find some neat stuff in the process. Probably spent more time with the K5 overall. |
|
| |
|
IDT WinChip
Posts: 237
Location: USA | So what explains the size difference? Hard to believe the 38605DX only used more pins because of the L1 cache. I count 29 unused solder pads there around the chip. It makes sense to me for some boards to have been designed, at the time, for the 38605DX, but never ended up being used because the processor ended up not being manufactured beyond some samples. |
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | Your guess is as good as mine. CPUs always seem to have loads of power and ground pins, several per row usually, so it's probably safe to assume some of them do that. The Cyrix DLC has an issue where certain instructions (largely related to DMA) will flush the cache, perhaps this was Chips' way around it, those pins could simply bypass whatever causes this. The pins at the top left seem to head off towards the L2 Cache and Memory, this board actually has an unusual L2 cache implementation because it is spread over eleven SRAM chips. The only thing I've seen like that is the AMI Baby Voyager, that similarly aged 486 board I have, that one has its cache spread over twelve chips in some configurations, or eleven in my case. Towards the bottom, some of those traces might be heading for the FPU socket, but it's really hard to tell what is going on until my board arrives.
Skimming about the web, there are other boards with extra pins, the MS-3124 / Contaq-386 might have them and is another late board but they only seem to be on one side of the CPU.
Of course it's possible this is just a red herring, the board maker just did something weird and it lines up by pure coincidence.
In the case of Cyrix's 486DLC it's genuinely quite amazing that they ever got those to work properly, yet somehow they did, in their case with a bunch of extra signals on existing pins. Cyrix were capable of some truly astonishing feats when their designers/engineers put their minds to it. |
|
| |
|
Intel P5
Posts: 85
Location: UK | That motherboard sure looks like it was made for the J38605DX. Were any other CPU's made with the same number of pins, perhaps some of the more exotic Cyrix parts? I can't imagine that socket being that way for a trivial reason. I do think your right that that is a VERY good candidate for your Chip's 386 as well.
Look forward to hearing your results when the parts arrive. |
|
| |
|
IDT WinChip
Posts: 237
Location: USA | I noticed that interesting cache layout as well. Strange it has those 2 just like, there in the middle top of the motherboard. Given the placement, it's like whatever is going on between the chipset there is being cached by those 2 sram's or something. Placement of the other cache is where it should be. I hate those motherboards that place the cache like miles away from the cpu and ram, clear on the other side of the motherboard down where VLB slots usually are. I might get a bit of a laugh behind the computer screen if that board didn't work with the CHIPS 386. Ya know, it would just figure... cause it's life, the REAL four letter word...
Outside of ES's for the J38605DX chip, I don't think there were any x86 chips that ever used that particular pin layout. Sparc comes to mind as being one that might have used that number of pins. or somewhere close to that. I have a 64 bit Sparc cpu that is in between a 386 and 486 in size |
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | This server keeps crashing due to cold getting to the disks. Unfortunately there's not much I can do about it, maybe I'll shut the CPU fan off as the Atom can run on passive cooling and it might just get a little heat into things that way.
I'm not sure about CPUs with that pin count, there are Cyrix processors that have non-standard pinouts but they seem to have the same pin count as a regular 386, I want to say the Ti SXL but that might have had a 486 footprint now I think about it. There are definitely other CPUs that have that kind of footprint, but none of them are x86 and so wouldn't really be candidates for that board.
The cache does have another oddity about it, the board has silkscreen for other sizes but the jumpers aren't populated and all the documentation suggests 64K is the only option. Have yet to see any of them with those jumpers in place and you have to wonder why they'd take them off. Interestingly, it seems there's a board on the same chipset, minus the extra pins, with a similar cache setup here with which the owner is having issues upgrading his cache, seems to be a different version of the chipset though. There's an ECS PD386 which also uses the other chipset version but uses a more common cache configuration, only five SRAM chips. I forgot to link it last time, so that Contaq board is on this page and seems to belong to 0blivi0n100. Seems it might have only one extra pin row, but it is on the same side as the Peak DM board, oh, it has a less common cache layout too, 10 SRAMs this time.
The 64K restriction of the Peak DM-386 isn't exactly a problem for me, 64K is enough for a 386, it should be good up to 16MB of RAM anyway due to WT mode and I can't really see many situations where a 386 would need to go further. Hell, the T3200SX handled all of its music work in ~5MB of RAM with no cache.
No, I'd have to laugh if it didn't work, not much else you can do in those situations really. Always like to have a spare to hand, currently have no spare 386 boards and I'm sure the AOpen will break some day anyway, so it's fine.
What an interesting rabbit hole this has lead down. Delivery should be quite quick, likely the end of this week unless the truck falls into a black hole or something.
Edited by DXZeff 2017-12-19 6:39 AM
|
|
| |
|
IDT WinChip
Posts: 237
Location: USA | I bet those are tag's. Statson's diagram of the peak/DM386 shows 2 tag chips, but is a different layout from the one in the photo there. It's strange to me to have the tag's so far away from the cache banks, but that seems to be the way a lot of them were for a time. Speaking of 0blivi0n, I actually bought a computer from him a couple summers ago. I haven't done much with it other than test it.
Edited by waybacktech 2017-12-19 8:23 PM
|
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | Try this one - http://www.uncreativelabs.de/th99/m/S-T/30937.htm - there are a few very similarly named boards out there. Yeah, they do look to be TAG SRAM, but it's still an unusual way of laying it out. It's certainly not unique in doing so, but I may have to research why at some stage.
Meanwhile I'm about to drag my whole block's internet down again. Also my StarTech might be busted, but it could just be a software bug as Windows has updated - annoying in itself because it keeps setting my keyboard back to UK layout every time - but even then, I might still change it in the future.
Edited by DXZeff 2017-12-19 8:33 PM
|
|
| |
|
TM Crusoe
Posts: 671
| Found an ET4000-ISA, that were not listed as an ET4000. So the price was really fair. Snatched it right up and paid for it like instantly. |
|
| |
|
TM Crusoe
Posts: 618
Location: Hull, UK | Nice find, they're getting harder to get hold of now, you'll almost never find a VLB one and the prices are just stupid now.
Anyway, I'll just leave these here before I fall asleep for the day;
Board needs cleaning due to battery leakage.
Edited by DXZeff 2017-12-21 7:15 PM
|
|
| |
Recently Played/Bought/Messed with
