mockingbird wrote on 2023-08-31, 13:29:

Gmlb256 wrote on 2023-08-31, 13:18:

This is one of the reasons why I set the FSB/PCI to 124/31 MHz on mine instead of 133/33 MHz. With 133 MHz FSB (aside from the issue with the AGP @ 89 MHz), I had to set the VIO jumper setting to Test for stability and that wasn't great for the northbridge's lifespan.

Interesting... Maybe the 440ZX was less capable in that regard. It wasn't only the heat coming out of the northbridge, the CPU @ 1.6V was an order of magnitude hotter than it was when it was running at 1Ghz. 0.15V shouldn't have that effect. This silicon just wasn't meant for that. The clockgen is also hot -- I might put a heatsink on it. See the second edit of my post above about Prime95, btw.

Finally managed to get 133 MHz FSB stable without toggling the VIO jumper by replacing the PC133 SDRAM sticks with higher quality ones from Infineon that I got during week. The previous ones used Micron chips and apparently, they weren't up to the task regardless of the CAS latency settings.

To handle the AGP @ 89 MHz, I'm now settling for a GeForce2 GTS card which turns out that the VBIOS it has doesn't support sideband addressing.

I have now added a signature that shows the specs of my current and now only retro computer. Wish that the vintage edition of CPU-Z let me validate it.

Were they micron modules or 3rd party? Might have been remarked from 8ns... (Though I've had plenty of 8ns do 133 on Cas3)

BitWrangler wrote on 2023-10-14, 23:19:

Were they micron modules or 3rd party? Might have been remarked from 8ns... (Though I've had plenty of 8ns do 133 on Cas3)

Third-party with 4 Micron chips on each side and those weren't stable at CAS3 settings when running the FSB at 133 MHz.

Gmlb256 wrote on 2023-10-14, 23:02:

Finally managed to get 133 MHz FSB stable without toggling the VIO jumper by replacing the PC133 SDRAM sticks with higher quality ones from Infineon that I got during week. The previous ones used Micron chips and apparently, they weren't up to the task regardless of the CAS latency settings.

To handle the AGP @ 89 MHz, I'm now settling for a GeForce2 GTS card which turns out that the VBIOS it has doesn't support sideband addressing.

I have now added a signature that shows the specs of my current and now only retro computer. Wish that the vintage edition of CPU-Z let me validate it.

Very nice, good work... I had no luck at all with Infineon chips for some reason, but most other stuff worked fine... Keep an eye on the northbridge temperature... I'd put a heatsink on it at the very least at 133Mhz if there isn't one already.

That's what gobbled up most of my original stock of 486 and early pentium heatsinks back in the day, 1998-2003 northbridge cooling.

What is a safe voltage range for Neh.'s? I got my 1.0 to 1.26 but anything over that it becomes unstable.

ifrit05 wrote on 2023-11-03, 14:59:

What is a safe voltage range for Neh.'s? I got my 1.0 to 1.26 but anything over that it becomes unstable.

The rated CPU core voltage is around 1.4v - 1.45v, but it can be bumped to 1.5v safely if the overclock is unstable.

Edit: According to the "Recommended Operating Conditions" in the VIA C3 Nehemiah Processor Datasheet, the safest voltage range is 1.365v - 1.635v on Socket 370 motherboards and slotket adapters that doesn't support Tualatin.

Edit #2: That was the CPU voltage termination range what I mentioned in the first edit. 😜

Gmlb256 wrote on 2023-11-03, 16:20:

The rated CPU core voltage is around 1.4v - 1.45v, but it can be bumped to 1.5v safely if the overclock is unstable. […]
Show full quote

ifrit05 wrote on 2023-11-03, 14:59:

What is a safe voltage range for Neh.'s? I got my 1.0 to 1.26 but anything over that it becomes unstable.

The rated CPU core voltage is around 1.4v - 1.45v, but it can be bumped to 1.5v safely if the overclock is unstable.

Edit: According to the "Recommended Operating Conditions" in the VIA C3 Nehemiah Processor Datasheet, the safest voltage range is 1.365v - 1.635v on Socket 370 motherboards and slotket adapters that doesn't support Tualatin.

Edit #2: That was the CPU voltage termination range what I mentioned in the first edit. 😜

Thanks.

I tried 1.33@1.5v, 3DMark 2001SE crashed then 99/00/01 said I didn't have a DX6/7/8.1 GPU... So I went back down to 1.26@1.45v.

EDIT: Seems like 1.7v is the absolute maximum core voltage. I guess I can try 1.55v.

ifrit05 wrote on 2023-11-03, 14:59:

What is a safe voltage range for Neh.'s? I got my 1.0 to 1.26 but anything over that it becomes unstable.

My two best Nehemiah CPUs (which are 1.2 GHz SKUs), can reach a stable 1.53 GHz @ 1.55V. That's a 27% overclock.
The other five can only go up to 1.46 GHz (21% overclock).

So, if your 1 GHz Nehemiah is stable at 1.26 GHz (26% overclock), that's actually quite good and in line with what my best chips can do.
Going beyond that is probably very difficult to achieve with a 1 GHz part (i.e.: you'd have to win the silicon lottery big time!).

bloodem wrote on 2023-11-04, 17:17:

My two best Nehemiah CPUs (which are 1.2 GHz SKUs), can reach a stable 1.53 GHz @ 1.55V. That's a 27% overclock. The other five […]
Show full quote

ifrit05 wrote on 2023-11-03, 14:59:

What is a safe voltage range for Neh.'s? I got my 1.0 to 1.26 but anything over that it becomes unstable.

My two best Nehemiah CPUs (which are 1.2 GHz SKUs), can reach a stable 1.53 GHz @ 1.55V. That's a 27% overclock.
The other five can only go up to 1.46 GHz (21% overclock).

So, if your 1 GHz Nehemiah is stable at 1.26 GHz (26% overclock), that's actually quite good and in line with what my best chips can do.
Going beyond that is probably very difficult to achieve with a 1 GHz part (i.e.: you'd have to win the silicon lottery big time!).

Well it seems my P3B-F doesn't quite like booting reliably at 133MHz, so I lowered it to 100MHz with a 13x multi @ 1.45. Seems to be pretty stable now, although not sure how much the lower FSB is a bottleneck now, but the whole point was to have a scalable CPU for older DOS games.

ifrit05 wrote on 2023-11-05, 15:13:

Well it seems my P3B-F doesn't quite like booting reliably at 133MHz, so I lowered it to 100MHz with a 13x multi @ 1.45.

It depends on the quality of the SDRAM sticks, previously I used two cheap PC133 ones that weren't stable at 133 MHz FSB, no matter how the CAS latency settings were set in the BIOS.

Seems to be pretty stable now, although not sure how much the lower FSB is a bottleneck now, but the whole point was to have a scalable CPU for older DOS games.

There will be some performance left on the table due to lower RAM bandwidth. It is still plenty for many DOS games though.

Any tips on finding a good brand?

ifrit05 wrote on 2023-11-05, 15:34:

Any tips on finding a good brand?

At the time of this writing, there is currently a guy on eBay selling pairs of Infineon 128MB PC133 SDRAM CL2 sticks.

16x64 is good? Or do I need to look for 16x32 for a 440BX board?

Those 128MB Infineon SDRAM sticks are single-sided with 8 chips (128 Mbit density), which are good enough for a motherboard based on the Intel 440BX/ZX chipset.

Ok, so I'm finishing up (assuming the only glitch is a whiny bios that tells me to set the right multi) an Ezra-T based "time machine"/ early 3D gaming rig with a Voodoo 3 and I'd like to know which MiniGL works best with and takes advantage of the C3's 3Dnow implementation. So far it looks like the stock opengl in the driver package works best which kind of confounds me.

Edit: Nevermind, noticed the board was overvolting the Ezra-T to an uncomfortable degree (1.5 volts!) so I swapped in a 1.2 GHZ Nehemiah.

Can I get a sanity check please? Is Quake3 1.32 (or 1.32C) incompatible with the Ezra-T (or similar?). I get constant crashing while trying to run timedemo, but it works fine with the earlier release. Benchmark should be 1024x768, everything maxxed out and both color depth and texture depth set to 32-bit. Tried with several different nVidia GPUs. Running DirectX 7a with 45.23 Detonator drivers.

Thanks

EDIT: hmmm, no, it's not the CPU. I just popped in a P3 and the same thing happened.

mockingbird wrote on 2024-02-08, 04:51:

Can I get a sanity check please? Is Quake3 1.32 (or 1.32C) incompatible with the Ezra-T (or similar?). I get constant crashing while trying to run timedemo, but it works fine with the earlier release. Benchmark should be 1024x768, everything maxxed out and both color depth and texture depth set to 32-bit. Tried with several different nVidia GPUs. Running DirectX 7a with 45.23 Detonator drivers.

Thanks

EDIT: hmmm, no, it's not the CPU. I just popped in a P3 and the same thing happened.

Sound card? Any other device?
Remove everything that you don't need and try again. 😀

Also, is this a 440BX motherboard or VIA Apollo Pro 133?

bloodem wrote on 2024-02-08, 15:35:

Sound card? Any other device?
Remove everything that you don't need and try again. 😀

Also, is this a 440BX motherboard or VIA Apollo Pro 133?

Nothing -- bone stock 440BX system. I think I have it narrowed down -- I'll report back soon.

Got any USB cards in it? That's a frequent bugbear.