The official COAST standard (from a certain stage on) has 3.3V AND 5V supply voltages, one at each side of the stick.
There were sticks with 5V TAG and 3.3V PB-SRAMS (very common) and other combinations.
Question is - as always - did ZEOS abide to the COAST standard?

snufkin wrote on 2022-12-02, 17:58:

Sphere478 wrote on 2022-12-02, 16:43:

These are the only other possibilities that I own. I’m pretty sure that top one is what you are talking about

Looks like the memory you've ordered is suitably overkill. For reference, the sticks in that photo look like a couple of 4MB and a 16MB, I think they're all 5V FPM (417400 datasheet calls it Enhanced Page Mode, which from a quick skim looks like FPM). The pair of 4MB are probably useful for testing. Just noticed that the manual from that RetroWeb link in the first post says the maximum RAM is 384MB. So it's possible that board doesn't support 128MB sticks and might just see them as 64MB.

Manual also says the cache needs to be 3V. Datasheet (https://datasheetspdf.com/pdf-file/537700/All … uctor/AS7C512/1) for the chips on the stick in the photo gives recommended supply voltage as 5V. But it also says that data is retained down to 2V, and any signal above 2.2V is read as '1'. So it might be a good idea to check what voltage the Vdd pins on the socket are. When the manual says 3V SIMM it might just mean the signal levels rather than supply voltage. But as dionb suggested, probably best to get the board started without cache to start with.

Okay, so these should work for test?

Sounds like my coast stick is 3.3/5v tolerant you are saying?

Another possibility, 128mb simms may not have existed back then. I hope they routed the address lines for 512mb. Gonna suck if not.

I suppose if they detect and work as 64mb I can just order two more.🤔

Whatcha think about these configuration resistors next to coast socket?

You need to find out if there´s some proprietary pinout at work here or if it´s COAST standard.
It´s a good idea to check if the Vdd (3.3V) and Vcc (5V) voltages are at the pins like in the official pinout. Maybe there´s Vcc only but the raster must match.

Most proprietary layouts I´ve seen had massive divergences here.

Sphere478 wrote on 2022-12-02, 18:24:

Okay, so these should work for test?

Think so, assuming they test ok in a known working board.

Sounds like my coast stick is 3.3/5v tolerant you are saying?

Not really. The datasheet for the chips on the stick says their supply voltage must be 5V (actually between 4.5 and 5.5), but that signals must be <0.8V to be read as '0' and >2.2V to be read as '1'. Signal voltage mustn't ever be more than 1V above supply voltage. So if the socket supplies 5V then the stick should work as long as the signals get under 0.8V and over 2V. If the socket only supplies 3.3V then the stick won't be guaranteed to work, particularly at full speed. If (and this would be weird) the socket only supplies 3.3V and the signals get above 4.3V then there's a chance of damaging the chips. A thing I don't know is what the manual means by '3V SIMM'. If it means the supply voltage is only 3V (or 3.3V) then that stick probably won't work. If it means the supply is 5V and the signalling is 3V then it might work.

Dionb mentioned about possible issues with async or burst mode, but I don't know enough about that. Looks like Majestyk just answered a question I had as I don't know if there was a standard COAST pinout or not, and it sounds like there wasn't. I think I'd ignore the cache for now, then if the board POSTs I'd find out where the power pins are on the connector and compare that with the power on the stick, and compare both of those with online pinouts, then see if anything matched up.

Thread here where mpe found out the exciting way: Pentium COAST / CELP L2 cache sticks standards

Okay, so that pinout doesn’t appear to match, I have 5v here

On a COAST slot this would be a data line.
Have you checked with an ohmmeter between that pin and 5V psu connector?

Per continuity check of vss, vcc5 vcc3 I have determined that this is a match to the zeos board. Through measurement of pins listed as such for these three power signals

How do you suggest I check the cache stick for compatibility?

I guess I will need a datasheet for one of the chips to attach to for measuring pins

This is definitely not COAST, but some proprietary (or preliminary / predecessor) standard.
Be careful what you insert in that CELP slot!

Now that you have the Vdd and Vcc (and Vss) contacts of the slot you can investigate the stick if ground and supply voltages are routed from these very contacts to the Vss and Vcc pins of the single chips.

AS7C512 SRAM chips: https://datasheetspdf.com/pdf-file/537700/All … uctor/AS7C512/1 , GND pin 16, Vcc (5V) pin 32
FCT373 Latches: https://pdf1.alldatasheet.com/datasheet-pdf/v … CT373ATQCT.html , GND pin pin 10, Vcc (5V) pin 20
PI5C3245 Switch: https://www.diodes.com/assets/Datasheets/PI5C3245.pdf , GND pin 10, Vcc (5V) pin 20

No idea about those option resistors, that'd involve tracing their connections to try to work out what they might do. Might be for configuring chipset on power up.

I don't know exactly when Intel came out with the COAST spec, but I've found a datasheet for a COAST 1.1 stick from Dec '94. So I wonder why this board is non-standard. Looks like even the pin 1 mark doesn't match.

Eyeballing the photos of your cache stick and it looks like power traces (slightly thicker) match up with the pinout that you found that matches the socket. So it might be the right stick.

Sphere478 wrote on 2022-12-02, 18:24:

[...]

Okay, so these should work for test?

Yes, they are 16MB 5V FP SIMMs that definitely should work

[...]
Another possibility, 128mb simms may not have existed back then. I hope they routed the address lines for 512mb. Gonna suck if not.

I suppose if they detect and work as 64mb I can just order two more.🤔

To be honest, you're getting well ahead of yourself if you're already spending significant money on new components before you've verified that the board isn't completely FUBAR... I'd recommend getting it minimally running with two SIMMs and no cache, and only when that works start investing in things to pimp it.

Assuming the board does work, the question is not whether 128MB SIMMs existed when this board was made, but rather whether the board - or more specifically the chipset - supports them. According to spec, i430NX does NOT support 128MB SIMMs - it supports max 256MB, but that's using a lot more than two SIMMs. I've not seen it do more than 32MB per module, so depending on what your BIOS thinks of out-of-spec stuff, those big boys will either cause POST to fail with a RAM error, or will POST, but with significantly less capacity available, most likely 1/4 of their nominal value (as only able to address the first 16Mb of every 64Mb chip).
Slight caveat: Intel was repeatedly more conservative than necessary with their RAM support specs. In the late 1990s multiple chipsets were listed as supporting max 64Mb chips when in practice they could handle 128Mb. It wouldn't be unprecedented for other chipsets to alos do that. But I would be rather surprised in this case.

Darn, I hope these sticks work.

To get to 512mb there is no other option but to use 128mb in at least two slots. And 64 in remaining 4

Manual seems to say 384 is max according to other poster.

So that is 64x6 or 128x2 64x2 or 128x 2 32 x4

When I ordered them I didn’t realize that I had those 16mb fp sticks.

Didn’t test all the gold fingers but those that I did test seemed to be a good match to the diagram for socket that I posted earlier.

Probably safe to say that the stick is a match

Okay, i started out with a p75

Result

- -

🙁

Hunch: p200

Result, post codes!

scrolls then halts on 49, 47

Thoughts? Gonna try the “not a coast” next and see if I get same codes. After that, another video card

Edit:

Now with not a coast I have halt at c0, 91

It lives!

Different video card did it (radeon 7000)

‘Not a coast’ is installed btw

Same post code hault. So co must mean failed boot?

It seems to have a phoenix bios.

I definitely want to try and run a k6-3+ and dual pentium mmx 233 in this. Any suggestions for a bios update?

I think Jan can only patch award bioses.

Bios:

Okay, so there is a laundry list to do now..

-Install the second socket.
-Update the bios.
-Convert to socket 7 dual plane (socket 5/7/ss7 voltage interposer tweaker)
-Get scsi working.
-Investigate max bus clock.
-Investigate max ram/max cachable.
-Confirm ‘not a coast’ is working.

Great news! But er... maybe re-order this list...

Sphere478 wrote on 2022-12-03, 18:12:

Okay, so there is a laundry list to do now.. […]
Show full quote

Okay, so there is a laundry list to do now..

-Install the second socket.
-Update the bios.
-Convert to socket 7 dual plane (socket 5/7/ss7 voltage interposer tweaker)
-Get scsi working.
-Investigate max bus clock.
-Investigate max ram/max cachable.
-Confirm ‘not a coast’ is working.

How about:
1) get a CMOS battery hooked up one way or another so anything you do in BIOS will persist.
2) investigate max bus clock (with absolute minimum of stuff attached, hence so early) - don't get hopes up, i430NX is very early and 66MHz was challenging.
3) check the cache with CACHECHK under DOS (fact no magic smoke comes out proves it's correct for your board).
4)update BIOS
5) investigate max RAM and max cacheable once big SIMMs come in. My money is on max 32MB per SIMM. Supposedly i430NX is able to cache all 256MB of max RAM, so that should be good.

You should be able to do all of this right now. Then as and when parts come in:

6) get SCSI working
7) install second socket
8 ) confirm regular P54 CPUs will run in SMP
9) try fancy stuff with voltage (and multiplier) interposers
10) try fancy stuff with even higher clocked non-SMP stuff

Dual AMD K6 computer

This is also on the list 😀

I have to admire your chutzpah - you treat known limitations as challenges, which could be described as foolish or worse, but you just chip away at it and discover stuff like the K6-2+ cache unlock by ignoring other people telling you what's not possible. That one alone earns you a place in the annals 😀

Here again I'd say you're bordering on insane with what you want to do, but if anyone can do it, you can. That said, I'll eat my hat if you get two K6 CPUs w0rking on this board. But afterwards I will get myself a similar board and reproduce it 😉