Iirc they (at least the Nehemiahs) had bridges on top of the CPU package similar to the Thunderbirds.
I also read in the datasheet (will edit this post to provide link later) the multi could be changed vy software

I had a look at my C3 CPU's last night to check on those bridges.
It turned out that 4 CPU's I have, have those bridges:
The 866Mhz one has 12 bridges and the 2 1000's and the 1200 have 11.
I cross referenced them with the BF pins in the datasheet but couldn't deduct from the differences which bridges are the BF ones, even though they give me enough info to get started.

The datasheet can be found here: http://datasheets.chipdb.org/VIA/Nehemiah/VIA … heet%20R113.pdf
If you do a search for "MSR 0x2A [25:22]" you'll land at the relevant section right away 😉

The bridges on the bottom of the CPU's are like this:
From left to right they go (0 is Open(the bridge is cut) and 1 means closed (the bridge is not cut)) :
866(6.5x multi): 111011111100
1000(7.5x multi): 00010010011
1200(9.0x multi): 00111110011

Apparently to change the multi from 7.5 to 9, the blue bridges here need to be closed (with a pencil or something) : 00111110011

Sorry if my posts is a bit a mess, but I just woke up after a too short a night 😜

Edit: Just read the datasheet again quickly...apparently 9.0 is listed twice? 😵
The bottom part of the column is in another font, dunno why though.

My question to whoever reads this:
Does anyone else here have a C3 here which has bridges?
If so, could you please list your part and the corresponding bridges here so we can improve the amount of data we can work with?

Tetrium wrote:

I wouldn't bother replacing caps on a Slot 1 board, it's easier (and usually cheaper) to simply get another Slot 1 board. […]
Show full quote

I wouldn't bother replacing caps on a Slot 1 board, it's easier (and usually cheaper) to simply get another Slot 1 board.

The Katmai 550 may not be dead, I've found cleaning the contacts with a paper cloth and some rubbing alcohol turns "dead" Slot 1 CPU's back to life on more then 1 occasion 😉
Also with Slot 1 it's important to make sure you put it in the slot deep enough. I've found many of the newer "half open" Slot 1 CPU's are not hold down by the 2 plastic CPU thingies and basically gets all it's support by the Slot 1 motherboard connector alone, so be sure to push it all the way down 😉

I wouldn't bother using any board with even 1 bulged cap, it's not worth the trouble as trouble is all you're guaranteed to get.
(Super) Socket 7 boards are relatively common on Ebay as long as you don't buy from those sellers who hope to sell some crappy board for 200 bucks.
Btw, I presume you're looking for an AT Socket 7? Or would an ATX Socket 7 board also be alright for you?

Anyway, good luck! 😉

But when you buy another Slot 1 board, the caps on it are likely to be just as old and prone to failure as the ones on the board you are replacing. So you buy a new board and maybe 3 months later the caps on that board fail, too. Is it better to keep buying new (old) boards every time the caps fail, or just replace them with new ones and have a board you know is going to last a while? You can ask anyone who owns a Mac SE/30. Replacing the caps when they go bad is still a money making industry for those with the skill in that community. SE/30 motherboards are never allowed to go to waste due to bad caps.

sliderider wrote:

Tetrium wrote:

I wouldn't bother replacing caps on a Slot 1 board, it's easier (and usually cheaper) to simply get another Slot 1 board. […]
Show full quote

I wouldn't bother replacing caps on a Slot 1 board, it's easier (and usually cheaper) to simply get another Slot 1 board.

The Katmai 550 may not be dead, I've found cleaning the contacts with a paper cloth and some rubbing alcohol turns "dead" Slot 1 CPU's back to life on more then 1 occasion 😉
Also with Slot 1 it's important to make sure you put it in the slot deep enough. I've found many of the newer "half open" Slot 1 CPU's are not hold down by the 2 plastic CPU thingies and basically gets all it's support by the Slot 1 motherboard connector alone, so be sure to push it all the way down 😉

I wouldn't bother using any board with even 1 bulged cap, it's not worth the trouble as trouble is all you're guaranteed to get.
(Super) Socket 7 boards are relatively common on Ebay as long as you don't buy from those sellers who hope to sell some crappy board for 200 bucks.
Btw, I presume you're looking for an AT Socket 7? Or would an ATX Socket 7 board also be alright for you?

Anyway, good luck! 😉

But when you buy another Slot 1 board, the caps on it are likely to be just as old and prone to failure as the ones on the board you are replacing. So you buy a new board and maybe 3 months later the caps on that board fail, too. Is it better to keep buying new (old) boards every time the caps fail, or just replace them with new ones and have a board you know is going to last a while? You can ask anyone who owns a Mac SE/30. Replacing the caps when they go bad is still a money making industry for those with the skill in that community. SE/30 motherboards are never allowed to go to waste due to bad caps.

My personal experience is that mobo's without bulging caps tend to just go and go.
The capacitor plague made caps go *POP* in a MUCH shorter timeframe then was usual.
Either the board would develop bulging caps, or it would most likely not.
Another thing about boards with bulging caps is, even if you replace the caps, the defective caps may have already damaged other components on the board and you may end up still having to buy another board, in addition to already have waisted money, many hours recapping + troubleshooting the still malfunctioning board.

If a board hasn't developed bulging caps in the last 10+ years, chances are very good that they won't suddenly develop within 3 months after you bought it.

Tetrium wrote:

My question to whoever reads this:
Does anyone else here have a C3 here which has bridges?
If so, could you please list your part and the corresponding bridges here so we can improve the amount of data we can work with?

My C3 has those bridges on the underside. I'll remove it from the slotket and take a picture of the CPU's underside.

Ace wrote:

Tetrium wrote:

My question to whoever reads this:
Does anyone else here have a C3 here which has bridges?
If so, could you please list your part and the corresponding bridges here so we can improve the amount of data we can work with?

My C3 has those bridges on the underside. I'll remove it from the slotket and take a picture of the CPU's underside.

Cheers!

Could you please also upload a pic of the topside?

I've found these C3's are kinda a pain to identify as theres so little information available about them on the net.
I'm kinda wading in unknown territory here...with mah machete 😁

I actually just found a 933MHz Cyrix C3 lying in a bag of CPUs along with some Coppermine Pentium IIIs and Celerons. The BIOS actually runs this CPU properly. The multiplier, unfortunately, is locked to 7x, so the slowest it goes right now is 466MHz. I'd either need this CPU to be faster than 933MHz without cache(at that rate, its performance according to SpeedSys is in between a 50MHz 486DX2 and a 133MHz AMD 5x86, but a little closer to the 50MHz 486DX2, so I'd assume the performance I'm getting should be close to that of a 66MHz 486DX2). Instead of speeding up the CPU to make it faster without cache, I'd like to unlock the multiplier and set it as low at it can get, then work my way back up until I get smooth gameplay in X-Wing and Doom. Right now, with this other C3 running at full speed without cache, X-Wing has a little bit of slowdown and Doom is a bit choppy, but still playable.

EDIT: When running my Soyo SY-6BA+ motherboard with a 133MHz FSB, the AGP slot runs at no less than 88MHz. What graphics cards are known to handle that speed? Does an nVidia Riva TNT2 work well on an 88MHz AGP slot?

EDIT 2: Seems my 933MHz Cyrix C3 is an Ezra model of the C3. It does seem to set the multiplier the same way as my 1GHz Nehemiah Cyrix C3, though the placement of the jumpers is different.

Does your Ezra have the little bridges on the bottom of the chip?
I know Nehemiah "should" be able to change it's multiplier with a program (or at least, so sais the almighty datasheet 😜 ) but I have no idea if this'll work for the earlier models as well.
I don't think I have an Ezra (nor it's related Ezra-T), only a couple Nehemiah's and a couple Samuels.

Ace wrote:

EDIT: When running my Soyo SY-6BA+ motherboard with a 133MHz FSB, the AGP slot runs at no less than 88MHz. What graphics cards are known to handle that speed? Does an nVidia Riva TNT2 work well on an 88MHz AGP slot?

People usually refer to that as the 89MHz AGP issue, we are missing the proper AGP bus divider on the i440BX. There is no way of telling exactly what card is best, because I had differing results with exactly the same Nvidia chipsets. Most tolerant card I have here is a certain Geforce 2MX which could stand 133MHz AGP for 40 minutes. A Voodoo3 was less tolerant but never gave me a problem at 89MHz.

On your board you can use the SoftFSB windows program to easily change the FSB from 66 to 133MHz (or beyond).

I found that Speedsys gives very mysterious CPU identifications when underclocking CPU's. When my Pentium II SL2QF was running at 100MHz (2.0x50), it was calling it a 633MHz Celeron?

I enjoy reading about underclocking and the C3's here. Could you maybe add a little more specification codes of the CPU's you talk about. And when talking about cache is it the L1 or the L2?

Tetrium wrote:

I know Nehemiah "should" be able to change it's multiplier with a program (or at least, so sais the almighty datasheet 😜 ) but I have no idea if this'll work for the earlier models as well.

The sheet shows a minimum multiplier of 5.0x.. that is rather high. so 333MHz@66FSB or 250MHz@50FSB.

I saw on mini-itx.com that wcpuid could vary the c3 multiplier on an epia board, from 3x to 13x:

http://www.mini-itx.com/hardware/overclocking/

Needs windows but I wondered if it also depended on the motherboard.

gerwin wrote:

I enjoy reading about underclocking and the C3's here. Could you maybe add a little more specification codes of the CPU's you talk about. And when talking about cache is it the L1 or the L2?

The thing is, I don't know.

Info on these chips is sketchy and theres little written about C3 on forums.
All I know is VIA released the Samuel which is supposed to lack any L2 cache.
Then Samuel 2 was released which had L2 cache and a die shrink.
Then came Ezra and Ezra-T which were minor tweaks of Samuel 2
And finally came Nehemiah which featured a full speed FPU instead of the half speed FPU the Ezra's and Samuel's had.

It's very possible that within each CPU type, VIA kept adding minor tweaks resulting in 1 Nehemiah not being the same as another Nehemiah, etc.

Add to this that it's very hard to see from the CPU itself what kind of chip exactly you're holding in your hand.

Darn.. Like you said. I tried to distinguish the different C3 types (Samuel,Samuel2,Ezra,Nehemiah) But CPU world is incomplete, and on wikipedia there are 2 lists, but they differ from oneanother. That, or I am failing to understand it well.

It seems the first three types (Samuel,Samuel2,Ezra) differ in process size. then the Nehemiah is the same process as Ezra but with full speed FPU.

gerwin wrote:

Darn.. Like you said. I tried to distinguish the different C3 types (Samuel,Samuel2,Ezra,Nehemiah) But CPU world is incomplete, and on wikipedia there are 2 lists, but they differ from oneanother. That, or I am failing to understand it well.

It seems the first three types (Samuel,Samuel2,Ezra) differ in process size. then the Nehemiah is the same process as Ezra but with full speed FPU.

That's it basically. Theres Samuel with no L2 cache, then Samuel2, Ezra and Ezra-T with a half-speed FPU and finally theres Nehemiah with a full-speed FPU.
And if it weren't complicated enough, theres different subtypes as well -_-
And some of the info posted on the web is darn right false. I read on a site that Nehemiah will only work in Tualatin boards but the first time I had a crack at it, my Nehemiah's all posted successfully in the ASUS CUSL2, which is a Coppermine board. It worked perfectly fine with a 1.2Ghz Nehemiah and provided it's correct 1.45v vcore.

Some sites list Nehemiah starting at 1Ghz but there seem to also be 800Mhz and 866Mhz variants and noone on cpu-world has ever seen a 1.3Ghz or 1.4Ghz versions.

Theres also this chip, which apparently is one of the other cores with just a different name on it.

So far I haven't figured out a single bulletproof way to distinguish between the different models. One of the better ones is to check at what voltage the chip runs.
And if it has lots of those little resistors on the top, it's probably a Nehemiah...or an Ezra -_-

As I don't own an Ezra, I'm not in the position to check visually for subtle differences between it and Nehemiah.
Perhaps Nehemiah is the only one with bridges on the underside?
And so far I haven't seen any Samuels with those bridges. Those look, at first glance, somewhat similar to the old gold-topped Pentiums

Tetrium wrote:

My question to whoever reads this:
Does anyone else here have a C3 here which has bridges?
If so, could you please list your part and the corresponding bridges here so we can improve the amount of data we can work with?

Here's some pics of my C3, which shows as a VIA C3 900A on the boot screen.

Has anyone tried a Nehemiah in a dual socket 370 yet? What was the result?

feipoa wrote:

Has anyone tried a Nehemiah in a dual socket 370 yet? What was the result?

Iirc after reading the datasheet of the C3, the C3 isn't multiprocessor capable. You'd be able to use only 1 socket.
Correct me if I'm wrong

Ace wrote:

Well, it looks like my Soyo motherboard's BIOS is full of s***, as SpeedSys reports the CPU running at 667MHz. With cache, SpeedSys ranks the C3 in my computer as equivalent to a 300MHz Pentium II. Without internal cache, SpeedSys ranks the C3 a little lower than a 50MHz 486DX2.

By the way, does anyone have the latest BIOS for the Soyo SY-6BA+ motherboard? I keep finding things for the Soyo SY-6BA+ III or IV, but not the plain Soyo SY-6BA+.

Here is the latest bios for the Soyo 6BA+
http://www.elhvb.com/mobokive/Archive/Soyo/bi … 86/6bap2da2.bin
And here for the 6BA+ III
http://www.elhvb.com/mobokive/Archive/Soyo/bi … 86/6ba32ba6.bin
And if anyone wants here the bios for 6VBA 133
http://www.elhvb.com/mobokive/Archive/Soyo/bi … 86/vba-2bab.bin

By the way, if anyone is interested in buying 3x1GB ECC Registred SDRAM DIMM memories (IBM certified), then i have those, and i dont need them, because I dont have any motherboards that support that kind of a desity.

Tetrium wrote:

Iirc after reading the datasheet of the C3, the C3 isn't multiprocessor capable. You'd be able to use only 1 socket.
Correct me if I'm wrong

I haven't poked around at the data sheet yet, but I do recall shooting e-mails back and forth with VIA at the time the Nehemiah's were coming out, and I was told that they were going to be SMP-capable. Perhaps they failed?

I couldn't get VIA to sell me any. I even offered to perform stability tests them for free in an SMP system, but was declined on the grounds that these chips were not intended for solo consumers.

I grabbed this line from the spec sheet,

The VIA C3 Nehemiah processor currently does not support multiple processors; however future versions may implement dual processors support.

My guess is that they never implemented SMP because they didn't renew their socket 370 lincense with Intel. Even for their current EBGA package, they have no mention of SMP. SMP would have only really been practical in the socket 370 form factor.

Oh well, it would have been fun to use a dual VIA C3 1.2 GHz for the sake of novelty in a retro rig.