Nemo1985 wrote on 2022-02-16, 10:03:

I can confirm that the rpm issue is solved on the CUBX

Very good, thank you!

Nemo1985 wrote on 2022-02-16, 10:03:

I also tested the via C3 cpus I have (just 2 unlucky: a C3 800 rated for v1.25 800 mhz which seems a Nehemiah cpu despite it should start from 1ghz and the 1.2 ghz cpu v1.45):
The first one isn't detected, debug card shows the infamous ----

It's likely that the power controller on the board only goes down to 1.3 V. If the CPU really requests 1.25 V via its VID pins, the controller will disable its outputs.

Nemo1985 wrote on 2022-02-16, 10:03:

while the second is able to boot but there are 2 quirks: it gets identified as pentium 3 from the bios and despite I change the voltage setting in the bios it always provide v1.5 (I tried to go down to 1.3)

Are you sure the voltage does not get applied correctly? The HW Monitor output can be misleading, see this post for details.

Nemo1985 wrote on 2022-02-16, 10:03:

Is there a chance to fix those issues? I will gladly do any testing if needed.

Well, it depends on how deep this rabbit hole turns out to be. 😉 If it's more or less about the cosmetics, it should be feasible. Please check out the attached BIOS.

[Edit: removed obsolete attachment]

DenizOezmen wrote on 2022-02-16, 20:23:

The HW Monitor output can be misleading, see this post for details.

Nemo1985 wrote on 2022-02-16, 10:03:

Is there a chance to fix those issues? I will gladly do any testing if needed.

Well, it depends on how deep this rabbit hole turns out to be. 😉 If it's more or less about the cosmetics, it should be feasible. Please check out the attached BIOS.

Thank you very much for the quick answer and the dedication.
I didn't know about the bios hw monitor output, I will check with hwinfo.

Bad news:

This problem was also present on the previous bios version

The cpu is still identified as p3 with that "bios update data incorrect" message during the post.

No it got identified as Pentium II.

Yes right, sorry

Nemo1985 wrote on 2022-02-17, 09:30:

Thank you very much for the quick answer and the dedication.
I didn't know about the bios hw monitor output, I will check with hwinfo.

Thanks for the continued testing!

Let's try and do this step by step. Please check whether the attached BIOS fixes the processor identification and the microcode message. I don't have an idea yet what might cause the missing frequency settings.

[Edit: removed obsolete attachment]

I just tried the newest bios, nothing changed. Recognized as Pentium II and the same error as before.
If it may help troubleshooting the problem I can try the cpu with the other 2 intel 440bx I have: Abit BE6-II and Msi BX Master or the QDI Advance 10T (VIA Apollo Pro133T) let me know if it could be useful.
Thank you for the help

Nemo1985 wrote on 2022-02-17, 19:21:

I just tried the newest bios, nothing changed. Recognized as Pentium II and the same error as before.

Now that is confusing. Could you post a CPU-Z report in .txt format? That should contain the results for the various CPUID levels.

Edit: I guess I spotted a bug. Could you try the new version?

[Edit: removed obsolete attachment]

Could this work too?
I'd have to install windows 98 from scratch

Nemo1985 wrote on 2022-02-17, 21:04:

Could this work too?
I'd have to install windows 98 from scratch

Yes, thank you! I needed to verify the vendor and raw family/model/stepping values.

DenizOezmen wrote on 2022-01-23, 12:18:

Thanks for the feedback and the offer! Since you volunteeered, would you be willing to test the fan mod on your CUBX-L? 😀

Sure and sorry for the delayed answer. I got hands on a CUBX-E in the meantime. I'll test the CUBX-L and CUBX-E BIOS next weekend.

I just found this:

LSS10999 wrote on 2016-03-16, 13:07:

It seems some boards have issues with cache initialization for Nehemiah CPUs. I was able to boot a C3 Nehemiah 1.2AGHz on my ASU […]
Show full quote

It seems some boards have issues with cache initialization for Nehemiah CPUs. I was able to boot a C3 Nehemiah 1.2AGHz on my ASUS CUV4X out of box, but the board recognized it as "Pentium II 1200MHz" with no cache memory (Cache Memory: None). Without the cache memory the processor is fine under DOS but extremely sluggish under Windows. I'm currently on the latest BIOS (1010.003).

Also, the BIOS additionally complained about unable to update microcode for CPUID=0000069A (which stands for Nehemiah) during the boot process. Given Nehemiah doesn't have a microcode update mechanism, it seems to be just a warning that it couldn't properly identify the processor. The board (CUV4X) is VIA C3 Verified, but so far only C3 Samuel support was mentioned in one of the early official BIOSes.

So it seems that pre-Tualatin boards may boot with the C3 Nehemiah, but in certain cases like mine, without internal cache, making it too slow to be practically used.

EDIT: Just tried a Tualatin adapter over the 1.2A Nehemiah on CUV4X. The result is the same as without it, the cache is still not initialized (None). Guess it seriously depends on whether the BIOS of a given C3-capable board can properly init its cache or not for a Nehemiah to work at proper speed, despite it can boot regardless.

So maybe any effort to properly support the nehemiah it won't work :\

Nemo1985 wrote on 2022-02-17, 23:08:

I just found this:

LSS10999 wrote on 2016-03-16, 13:07:

It seems some boards have issues with cache initialization for Nehemiah CPUs. I was able to boot a C3 Nehemiah 1.2AGHz on my ASU […]
Show full quote

It seems some boards have issues with cache initialization for Nehemiah CPUs. I was able to boot a C3 Nehemiah 1.2AGHz on my ASUS CUV4X out of box, but the board recognized it as "Pentium II 1200MHz" with no cache memory (Cache Memory: None). Without the cache memory the processor is fine under DOS but extremely sluggish under Windows. I'm currently on the latest BIOS (1010.003).

Also, the BIOS additionally complained about unable to update microcode for CPUID=0000069A (which stands for Nehemiah) during the boot process. Given Nehemiah doesn't have a microcode update mechanism, it seems to be just a warning that it couldn't properly identify the processor. The board (CUV4X) is VIA C3 Verified, but so far only C3 Samuel support was mentioned in one of the early official BIOSes.

So it seems that pre-Tualatin boards may boot with the C3 Nehemiah, but in certain cases like mine, without internal cache, making it too slow to be practically used.

EDIT: Just tried a Tualatin adapter over the 1.2A Nehemiah on CUV4X. The result is the same as without it, the cache is still not initialized (None). Guess it seriously depends on whether the BIOS of a given C3-capable board can properly init its cache or not for a Nehemiah to work at proper speed, despite it can boot regardless.

So maybe any effort to properly support the nehemiah it won't work :\

Thanks for the information. It's entirely possible that the L2 cache will be a problem, though the BIOS showing "none" doesn't necessarily mean it's not enabled. From a quick look at the CUV4X BIOS it seems like the board would at least not heed the "L2 Cache Enabled/Disabled" setting for Nehemiah cores. Whether or not this leads to a disabled cache is not completely clear.

Could you try the BIOS I attached a few posts above?

Hirsch wrote on 2022-02-17, 22:19:

DenizOezmen wrote on 2022-01-23, 12:18:

Thanks for the feedback and the offer! Since you volunteeered, would you be willing to test the fan mod on your CUBX-L? 😀

Sure and sorry for the delayed answer. I got hands on a CUBX-E in the meantime. I'll test the CUBX-L and CUBX-E BIOS next weekend.

No problem, thank you! 😀

DenizOezmen wrote on 2022-02-18, 00:20:

Could you try the BIOS I attached a few posts above?

Ups sorry I didn't notice you uploaded a new version.

It's a huge success!

If it can be useful, the other C3 cpu is identified as this (I used the QDI Advance 10T because it still doesn't work on the CUBX), it has a different CPU id-signature:

Nemo1985 wrote on 2022-02-18, 10:10:

It's a huge success!

Progress! 😀

Regarding the results:

• The cache display is indeed a visual bug.
• It should be possible to add the 133 MHz options, but that's something for another day. 😉
• I don't know enough yet about the CPU temperature readings. I guess they are done through the HW monitor chip, but I'm not sure how to influence the mechanism.
• There's a bug regarding the CPU voltage monitoring. The mod checks for the wrong voltage range for 069xh processors. That's why VCore is displayed in red; 1.34 V should be well inside the defined range.

Nemo1985 wrote on 2022-02-18, 11:37:

If it can be useful, the other C3 cpu is identified as this (I used the QDI Advance 10T because it still doesn't work on the CUBX), it has a different CPU id-signature:

Thanks! It's the same model, but given the default voltage, it will never boot in the CUBX without some sort of hardware mod. The QDI board supports lower voltages.

I've attached an updated mod. It should fix the voltage range issue and can hopefully display the L2 cache size. Could you check whether disabling the "CPU Level 2 Cache" option actually disables the cache?

[Edit: removed obsolete attachment, uploaded to first post]

DenizOezmen wrote on 2022-02-18, 16:31:

Progress! :-) […]
Show full quote

Nemo1985 wrote on 2022-02-18, 10:10:

It's a huge success!

Progress! 😀

Regarding the results:

• The cache display is indeed a visual bug.
• It should be possible to add the 133 MHz options, but that's something for another day. 😉
• I don't know enough yet about the CPU temperature readings. I guess they are done through the HW monitor chip, but I'm not sure how to influence the mechanism.
• There's a bug regarding the CPU voltage monitoring. The mod checks for the wrong voltage range for 069xh processors. That's why VCore is displayed in red; 1.34 V should be well inside the defined range.

Nemo1985 wrote on 2022-02-18, 11:37:

If it can be useful, the other C3 cpu is identified as this (I used the QDI Advance 10T because it still doesn't work on the CUBX), it has a different CPU id-signature:

Thanks! It's the same model, but given the default voltage, it will never boot in the CUBX without some sort of hardware mod. The QDI board supports lower voltages.

I've attached an updated mod. It should fix the voltage range issue and can hopefully display the L2 cache size. Could you check whether disabling the "CPU Level 2 Cache" option actually disables the cache?

Now the it shows the correct l2 cache (64k) during post or none when disabled in the bios.
I also confirm that it is correctly enabled or disabled according to the bios option. I also tested the l1 cache option and it also works as intended.
I confirm the other C3 doesn't work, the cpu id code is slightly different 800 mhz is 698, while 1,2 ghz is 69A, the issue could be that the slower one is an hybrid, in fact I can't find any info about a nehemiah rated for 800 mhz anywhere else:
https://imgur.com/a/RxtyAnC

Out of curiosity, do you know what the bios update option means in the bios? I also tried to write the latest bios from the motherboard instead of using the eeprom writer but it didn't work, is there any write protection option in the bios? I checked the jumpers and I didn't find it.

If you need anything else tested I will gladly do it.

Nemo1985 wrote on 2022-02-18, 17:31:

I confirm the other C3 doesn't work, the cpu id code is slightly different 800 mhz is 698, while 1,2 ghz is 69A, the issue could be that the slower one is an hybrid, in fact I can't find any info about a nehemiah rated for 800 mhz anywhere else:
https://imgur.com/a/RxtyAnC

C3s seem to be kind of exotic. I haven't been able to find a definitive list of models and their specs so far. There are some lists around, but all of them are incomplete in some respect. Maybe Via could afford to lower the voltage significantly for the lower-clocked model and went for it?

Nemo1985 wrote on 2022-02-18, 17:31:

Out of curiosity, do you know what the bios update option means in the bios?

IIRC the "BIOS Update" option controls whether or not the BIOS loads microcode updates into the CPU. It shouldn't have any effect on a C3.

Nemo1985 wrote on 2022-02-18, 17:31:

I also tried to write the latest bios from the motherboard instead of using the eeprom writer but it didn't work, is there any write protection option in the bios? I checked the jumpers and I didn't find it.

I'm not aware of a write-protect mechanism on the board or within the BIOS. Was there any specific error message?

Nemo1985 wrote on 2022-02-18, 17:31:

If you need anything else tested I will gladly do it.

Thank you. That should be it for now, but who knows what might come up. 😀 Thanks a lot for all the swift tests!

DenizOezmen wrote on 2022-02-18, 18:38:

C3s seem to be kind of exotic. I haven't been able to find a definitive list of models and their specs so far. There are some lists around, but all of them are incomplete in some respect. Maybe Via could afford to lower the voltage significantly for the lower-clocked model and went for it?

It's probably that way, a real mess, one hypotesis is that they tried to introduce the newer version to upgrade systems with older c3.

DenizOezmen wrote on 2022-02-18, 18:38:

I'm not aware of a write-protect mechanism on the board or within the BIOS. Was there any specific error message?

Not much, uniflash was unable to identify the rom chip and complained about beint write protected, it was also unable to do a backup.

If you plan to keep working on bios, count me in for testing 😁

DenizOezmen wrote on 2019-09-23, 18:57:

1008cu004_20220218.2.zip
1008cul004_20220218.zip
p3bf1084_20220218.zip

I successfully tested the three files quoted with a Lin-Lin + Tualatin Celeron 1400 on CUBX and CUBX-L and with a Pentium II 450 on the P3B-F. (Something seems to be wrong with my Upgradeware SLOT-T + Tualatin Celeron 1400 on it. I'll check this later.) . The speed of a Noctua NF-8A slowed down to 1300 RPM is displayed correctly and no hardware error is reported during POST.