adalbert wrote on 2020-11-14, 00:57:

I have a dual-cpu Slot-1 Tyan S1834D Tiger 133 mainboard with Via 694X chipset. I just sucessfully modded MS6905 Master 2.3 slotket and booted that mainboard with one SL5XL Pentium IIIS 1400MHz CPU (didn't try dual-cpu setup yet, I need to get second slotket). I set voltage jumpers to 1.45V, I measured 1.54V on MOSFET tab next to the CPU slot, but I guess that's fine. (edit: that was next to the empty slot, working slot gets 1.47V under load, so that's perfect)

It POSTs just fine, but displays wrong multiplier. CPU is detected as PENTIUM III 1400MHz (215x6.5). There is no -S suffix.
Should I care about this? Do I lose anything by not having proper microcode in BIOS? I can't find a BIOS with Tualatin update for this motherboard. Are there any tools I can use to add the microcode by myself? Or does it require hex-editing the BIOS by hand? I have a programmer so i'm not worried about bricking the board.

There is Bios Patcher 4.23, which works for Award BIOSes . See Help needed with booting slot1 boards with unsupported CPUs
Be careful, though. It can create a non bootable BIOS. Make sure you can recover from a bad flash (by hot flashing or preferably with an EEPROM programmer) .

Some useful tools and microcodes can be found on www.rom.by (you might want to use Google Translate...).

Additionally, I found the Award BIOS Editor to work very well for this purpose. I can post some more detailed instructions later... Once you have figured out the process it becomes really easy.

adalbert wrote on 2020-11-14, 00:57:

It POSTs just fine, but displays wrong multiplier. CPU is detected as PENTIUM III 1400MHz (215x6.5). There is no -S suffix.

I expect the raw multiplier in MSR 0x002A is 0x1F which is x10.5, but the BIOS is only looking at the low 4 bits as x6.5 is 0x0F. I further suspect the BIOS does not have code that knows about the CPUID 2 cache type 0x83 which is L2 512KB . I wonder what CPUID 80000002/3/4 return, it should be "Intel(R) Pentium(R) III CPU - S 1400MHz".

I could comment further if you posted the Menu->File->Save Local files from my SIV utility, ideally run SIV32X on Windows XP or later.

Thanks for the answers,

darry wrote on 2020-11-14, 04:55:

There is Bios Patcher 4.23, which works for Award BIOSes . See Help needed with booting slot1 boards with unsupported CPUs
Be careful, though. It can create a non bootable BIOS. Make sure you can recover from a bad flash (by hot flashing or preferably with an EEPROM programmer) .

gex85 wrote on 2020-11-14, 08:33:

Some useful tools and microcodes can be found on www.rom.by (you might want to use Google Translate...).

Additionally, I found the Award BIOS Editor to work very well for this purpose. I can post some more detailed instructions later... Once you have figured out the process it becomes really easy.

i will check these tools. I have TL866II programmer, so I don't even have to program BIOS in-system.

red-ray wrote on 2020-11-14, 15:24:

I could comment further if you posted the Menu->File->Save Local files from my SIV utility, ideally run SIV32X on Windows XP or later.

I attached saved files. I ran SIV32X on Windows 7.

adalbert wrote on 2020-11-14, 19:20:

I attached saved files. I ran SIV32X on Windows 7.

Windows 7 is good. My initial thoughts were correct and as I suspected the name in CPUID 80000002/3/4 did not have the - S, I don't think this really matters.

The BIOS only detected 2 DIMMs, but looking at Menu->System->SMB Bus there look to be three DIMMs present. How many DIMMs are actually installed? I think there is a Golden Empire DIMM with some sort of issue.

SIV did not detect any SIO chip so no motherboard temperatures, fan speeds or voltages were reported. Do you know if the board has one/does the BIOS report these?

Thanks for checking the logs.

Yes, I had three DIMMs installed, and they were not detected correctly. First I tried populating 4 banks with 256MB modules (8 chips on each side of stick). With four modules BIOS detected 640MB? or so. With three it detected 512MB, and in such configuration I saved the logs. After that I removed all the memory, installed 2 x 256MB modules with 4 chips on each side and they are correctly detected as 512MB. Memory problems were mentioned in BIOS changelog, I should update to the newest BIOS and see how it behaves. But it still won't have Tualatin support.

BIOS reports both fan speed and temperature. It has W83977EF-AW chip.

adalbert wrote on 2020-11-14, 21:44:

BIOS reports both fan speed and temperature. It has W83977EF-AW chip.

OK about the DIMMs, let us know how things transpire.

Looking at the attached W83977EF-AW datasheet it does not do hardware monitoring, is there another Winbond chip?

Sure, I will try to use whole gigabyte of RAM with this board. Currently I am trying to inject Tualatin microcodes with CTMC and Award Bios Editor. If I get everything to work I will post the results.

There are no other Winbond chips. There is a programmed GAL next to it and some logic chips, there is also a Myson MTP008P28 chip in 28 pin case, I couldn't find a datasheet. I don't know how motherboard reads the temperature and fan speed.

//EDIT: i found this document https://www.fnal.gov/projects/ckm/ckm_cluster … sors.conf.ckmxx

chip "mtp008-*" # The values below should be correct if you own a Tyan S1834D motherboard. # If not, please contact us, so […]
Show full quote

chip "mtp008-*"

# The values below should be correct if you own a Tyan S1834D motherboard.
# If not, please contact us, so we can figure out better readings. (...)

so it looks like that Myson chip might be used for the sensors, but still I can't find datasheets. Edit: more info is available when looking just for MTP008

Update:

Mainboard was able to detect 1024MB of RAM after trying to use different sticks, but when I used more than 2 sticks i got black screen after changing any BIOS settings and board refused to boot until CMOS was cleared. I will try to get 4 identical sticks, or two 512MB sticks, maybe it will be better.

I added microcodes (up to 6B1) with CBROM utility. I tried to use Award BIOS Editor, but it caused BIOS checksum error displayed on the screen and the board refused to boot. I am still getting exactly the same message on boot (PENTIUM III 1400MHz (215x6.5)). I don't know if I did it properly, i just merged all microcodes with copy /B *.bin cpucode.bin and replaced them with CBROM.

By the way, USB isn't working under Windows 7 (Code 12 on USB controller), i need to check if using older OS would fix it. The same problem existed on stock BIOS.

maybe I will try BIOS Patcher 4.23 linked earlier and see if it fixes anything

adalbert wrote on 2020-11-14, 22:35:

I don't know how motherboard reads the temperature and fan speed.

Thank you for the information and I suspect they are read via SMBus device [ 0_2D ], see Menu->System->SMB Bus. If you press [ 0_2D ] this should dump all the data and pressing [Timer] should update the data every few seconds. Do any values change as you plug/unplug fans and if so which ones?

I may be able to deduce what is what if you post a screen shot of what the BIOS reports so please do this.

I have adjusted the attached SIV32X 5.54 Tyan-01 test SIV so it should report Tyan S1834D (Tiger-133) and the BIOS version as V1.05, please may I have new save files so I can validate the changes.

It looks like Windows 7 loaded the CPU µCode 001C from mcupdate_GenuineIntel.dll so if not being in the BIOS is not a real issue. I suspect WXP would also do this.

Looking in the manual it specifies Supports up to 2 GB at 100 MHz and Supports up to 1.5 GB at 133 MHz so I suspect only three DIMM slots can be used with PC133 memory.

Update: I just found a PII CPU CARD VER1.5 + SL36C that I have not used for at least 15 years and I am wondering if it will work with a Tualatin CPU. The two jumpers are JP1 + JP2 and I have no idea what they are for.

adalbert wrote on 2020-11-14, 23:56:

By the way, USB isn't working under Windows 7 (Code 12 on USB controller), i need to check if using older OS would fix it. The same problem existed on stock BIOS.

red-ray wrote on 2020-11-15, 10:47:

Do any values change as you plug/unplug fans and if so which ones?

I may be able to deduce what is what if you post a screen shot of what the BIOS reports so please do this.

I put that system aside for a moment, I ordered new PSU and heatsinks for the CPU because I was running it with a tiny heatsink and underpowered Pico PSU inside a cardboard box. I will mount it in a case, do a fresh install of OS and upload logs when I reassemble it.

USB: maybe OS reinstall will fix it, because i just took a HDD with preinstalled Windows 7 out of different machine with VIA chipset. BTW USB keyboard works fine in BIOS, with or without Legacy USB keyboard setting enabled, but as soon as OS boots it stops working.

red-ray wrote on 2020-11-15, 10:47:

Update: I just found a PII CPU CARD VER1.5 + SL36C that I have not used for at least 15 years and I am wondering if it will work with a Tualatin CPU. The two jumpers are JP1 + JP2 and I have no idea what they are for.

That slotket is built for Celeron Mendocino CPUs.

I found info about jumpers here: https://de.comp.hardware.cpu-mainboard.misc.n … d-ver1-5-jumper (google translate)

JP1 stands for the FSB, open 66MHz, closed overclock,
JP2 stands for dualprocessing, open none, closed dual.

It will not work out of box with Tualatin CPUs, also probably not with regular P3. S370 had 3 variants: PPGA, FC-PGA, FC-PGA2. Maybe it could be modded for Coppermine CPUs with PPGA -> FC-PGA mod. But the problem is that this adapter doesn't have voltage clamp chip. Coppermine and Tualatins use lower I/O voltage. I don't know if this is a real problem, there are mixed opinions. But your adapter also doesn't have jumpers for setting core voltage, and that would require additional mod not to fry the CPU.

I am using MS6905 Master slotket with voltage clamp chip. I insulated 3 pins in the socket: DYN_OE(AN3), RESET2#(AJ3), VttPWRGD(AK4). I connected VttPWRGD(AK4) on the CPU to VTT (AN11) with kynar wire. Similar mod should work with other slotkets which support Coppermine CPUs (they have that rectangural voltage clamp chip).

Here is some info about modding Mendocino slotket (like yours) for Coppermine, but I didn't try it: Re: Simple passive slotket on 694X dual cpu board for Coppermine/Tualatin?

adalbert wrote on 2020-11-15, 15:29:

That slotket is built for Celeron Mendocino CPUs.

Thank you for the info, I don't think I will bother as I have no real need to use it.

OK about the rebuild, good luck and do you plan to add the 2nd CPU as well?

red-ray wrote on 2020-11-15, 16:41:

OK about the rebuild, good luck and do you plan to add the 2nd CPU as well?

Yes, i just hope I will get another slot adapter soon. I am now using SL5XL P3-S 1.4GHz, i have second SL6BY P3-S 1.4GHz, so that's a different stepping but the same speed and cache. I wonder if it will work, or would I have to buy another matching CPU. I found some Compaq Xeon server manual that said you need to put the CPU with lower stepping in first slot when there is a mismatch, don't know if it will always work.

adalbert wrote on 2020-11-15, 17:54:

don't know if it will always work.

The only way to know for sure is to try it. Have a look at Re: In an SMP system how does DOS/98 choose the CPU to utilise?, all of the Speeds, L2 Cache and Stepping are different!

adalbert wrote on 2020-11-14, 23:56:

I added microcodes (up to 6B1) with CBROM utility. I tried to use Award BIOS Editor, but it caused BIOS checksum error displayed on the screen and the board refused to boot. I am still getting exactly the same message on boot (PENTIUM III 1400MHz (215x6.5)). I don't know if I did it properly, i just merged all microcodes with copy /B *.bin cpucode.bin and replaced them with CBROM.

Yeah that should work. When I did this mod the last time for my Diamond C400, I used the Award BIOS Editor to replace the microcodes and it worked pretty well. Just used CBROM in the end to verify that it correctly detects all the microcodes.

I have attached a modded version of the S1834 BIOS version 1.05 where I included the microcodes for pretty much everything from early Pentium II all the way up to Tualatin. Maybe you want to give it a shot (use at your own risk, but if you have an external programmer, you should be OK in case something goes wrong).

Here's the output of CBROM v1.55 for this file (the mc-final.bin is my assembled collection of microcode updates):

1C:\biosmod\CBROM155.EXE V1.55 [10/12/05 Release] (C)Phoenix Technologies 2001-2005
2
3              ******** C:\biosmod\1834v105\1834v105-mod.bin BIOS component ********
4
5 No. Item-Name         Original-Size   Compressed-Size Original-File-Name
6========================================1381Bh(78.03K)==1834W105.BIN============
7  0. System BIOS       20000h(128.00K)
8  1. XGROUP CODE       09D30h(39.30K)   068A3h(26.16K)  awardext.rom
9  2. CPU micro code    13000h(76.00K)   0A1BBh(40.43K)  mc-final.bin
10  3. ACPI table        026FBh(9.75K)    00F50h(3.83K)   ACPITBL.BIN
11  4. EPA LOGO          0168Ch(5.64K)    002AAh(0.67K)   AwardBmp.bmp
12  5. YGROUP ROM        02760h(9.84K)    01B74h(6.86K)   awardeyt.rom
13
14  Total compress code space  = 36000h(216.00K)
15  Total compressed code size = 26FE7h(155.98K)
16  Remain compress code space = 0F019h(60.02K)
17
18                          ** Micro Code Information **
19Update ID  CPUID  |  Update ID  CPUID  |  Update ID  CPUID  |  Update ID  CPUID
20------------------|  PPGA   04   06B0  |  PPGA   1C   06B1  |  PPGA   01   068A
21SLOT1  13   0630  |  PPGA   11   0681  |  PPGA   14   0683  |  PPGA   08   0686
22SLOT1  0D   0681  |  SLOT1  0C   0683  |  SLOT1  20   0632  |  SLOT1  34   0633
23SLOT1  35   0634  |  SLOT1  32   0650  |  SLOT1  30   0651  |  SLOT1  14   0652
24SLOT1  01   0653  |  SLOT1  04   0660  |  SLOT1  06   0670  |  SLOT1  03   0671
25SLOT1  04   0672  |  SLOT1  05   0673  |  SLOT1  14   0680  |  PPGA   10   0683
26PPGA   01   06B4  |  SLOT1  2A   0652  |  SLOT1  0A   0660  |  SLOT1  0E   0673
27SLOT1  40   0650  |  SLOT1  40   0651  |  SLOT1  10   0653  |  PPGA   03   0665
28SLOT1  10   0672  |  SLOT1  07   0686  |  SLOT1  02   1632  |  SLOT1  32   0633
29SLOT1  33   0634  |  SLOT1  27   0650  |  SLOT1  29   0651  |

Here's the output for the original v1.05 BIOS file, still having the CPUCODE.BIN microcode file, which is only half the size of the updated mc-final.bin:

1C:\biosmod\CBROM155.EXE V1.55 [10/12/05 Release] (C)Phoenix Technologies 2001-2005
2
3              ******** C:\biosmod\1834v105\1834v105.bin BIOS component ********
4
5 No. Item-Name         Original-Size   Compressed-Size Original-File-Name
6========================================13812h(78.02K)==1834W105.BIN============
7  0. System BIOS       20000h(128.00K)
8  1. XGROUP CODE       09D30h(39.30K)   068A3h(26.16K)  awardext.rom
9  2. CPU micro code    09000h(36.00K)   05163h(20.35K)  CPUCODE.BIN
10  3. ACPI table        026FBh(9.75K)    00F50h(3.83K)   ACPITBL.BIN
11  4. EPA LOGO          0168Ch(5.64K)    002AAh(0.67K)   AwardBmp.bmp
12  5. YGROUP ROM        02760h(9.84K)    01B74h(6.86K)   awardeyt.rom
13
14  Total compress code space  = 36000h(216.00K)
15  Total compressed code size = 21F86h(135.88K)
16  Remain compress code space = 1407Ah(80.12K)
17
18                          ** Micro Code Information **
19Update ID  CPUID  |  Update ID  CPUID  |  Update ID  CPUID  |  Update ID  CPUID
20------------------|  PPGA   11   0681  |  PPGA   10   0683  |  PPGA   03   0665
21SLOT1  20   0632  |  SLOT1  34   0633  |  SLOT1  35   0634  |  SLOT1  40   0650
22SLOT1  40   0651  |  SLOT1  2A   0652  |  SLOT1  10   0653  |  SLOT1  0A   0660
23SLOT1  06   0670  |  SLOT1  03   0671  |  SLOT1  10   0672  |  SLOT1  0E   0673
24SLOT1  14   0680  |  SLOT1  0D   0681  |  SLOT1  0C   0683  |

gex85 wrote on 2020-11-16, 10:40:

1834v105-mod.zip

Thanks 😀 unfortunately it didn't boot with that BIOS, I got checksum error and beeping (but that's no problem since TL866II is sitting right next to me). I had exact the same error when I was using Award BIOS editor. It looks like any changes in Award BIOS editor cause checksum error, but i'm not able to recalculate the checksum with that software.

BUT i extracted your microcodes to CPUCODE.bin and installed it into original BIOS file using CBROM software:

1cbrom 1834v105.bin /cpucode release.
2cbrom 1834v105.bin /cpucode CPUCODE.BIN

I used this tutorial http://www.geocities.ws/tuacubx/reads/MBMG.htm

I attach fixed file. It works, but doesn't fix wrong multiplier 😀

Multiplier is OK with BIOS patched by Bios patcher 4.23, it also recognized Tualatin CPU, but I looked inside generated file and it has typos in CPU identification like Pentium 2 Deshutes, I also don't like the fact that it modifies boot screen message. I will try to fix that with hex editor, test it and upload it here. I also noticed that it doesn't show the number of CPUs, so I wonder if it will correctly recognize both CPUs when I install second one.

I'm still waiting for my power supply to arrive so I can install CD drive and reinstall the OS (i have only one molex now), then I will test SIV32X

adalbert wrote on 2020-11-17, 20:20:

I attach fixed file. It works, but doesn't fix wrong multiplier 😀

I am not surprised. Looking at Menu->Hardware->BIOS Tables->BIOS Tables the only µCode was for Intel (Klamath) [A0], so not even for Coppermine CPUs. I wonder what will get reported for the new BIOS.

To fix the ratio the code that interprets MSR 0x0000002A would need to change, the change is simple enough, but I suspect patching it into the BIOS would be tricky.

mult[3:0] come from MSR 0x0000002A[25:22] and for ( model >= 8 && stepping >= 3 ) || ( model >= 11 ) then mult [4] needs to be set from MSR 0x0000002A[27]. I expect there is also a 16 entry table that converts the raw value to the multiplier that will need to be extended to have 2 x 32 entry ones as below.

1static CONST  BYTE p3c_ratios[ 32 ] =                               // P-3 Coppermine Ratios
2{                                                                   //
3 10,   6,   8,   4,  11,   7,   9,   5,                             // multipliers x 2 
4  7,  14,  16,  12,   4,  15,   3,  13,                             //
5 18,  22,  24,   4,  19,  23,  17,   5,                             // Coppermine 83
6  7,  14,  16,  20,   4,  15,   3,  21                              // and later only
7};                                                                  //
8
9static CONST  BYTE p3t_ratios[ 32 ] =                               // P-3 Tualatin Ratios
10{                                                                   //
11 24,   7,   7,   8,  11,   7,  23,   7,                             // multipliers x 2 
12 32,  14,  16,  12,   8,  15,   7,  13,                             //
13 18,  22,  24,  32,  19,  23,  17,   7,                             //
14  7,   7,  26,  20,  28,   7,  30,  21                              //
15};                                                                  //
16
17LONG cpu_mul_p32(                       // get CPU Multiplier for P-2 + P-3 
18           ULONG        msr,            // raw MSR value
19  CONST    CPUB        *cpu )           // CPU Context
20{          ULONG        val;            // working value
21
22  val = ( msr >> 22 ) & 0x0F;                                       // get the ratio
23
24  if( ( cpu->xmodel   >= 0x08 ) &&                                  // CPU a Coppermine and
25      ( cpu->stepping >= 0x03 ) )                                   // newer Coppermine ?
26    val |= ( msr >> 23 ) & 0x10;                                    // yes, merge in the extra bit
27
28  return p3c_ratios[ val ];                                         // return the scaled ratio
29}
30
31LONG cpu_mul_p3t(                       // get CPU Multiplier for P-3 Tualatin
32           ULONG        msr,            // raw MSR value
33  CONST    CPUB        *cpu )           // CPU Context
34{
35  return p3t_ratios[ ( ( msr >> 22 ) & 0x0F ) |                     // return the scaled ratio
36                     ( ( msr >> 23 ) & 0x10 ) ];                    //
37
38}

All right, so gex85's microcodes work fine with Tualatin, but after I installed Dual P3-Katmai 450 PC unfortunately froze on POST screen (before memory count).
I decided to use BIOS patched with BP 4.23. Looks like it doesn't modify original BIOS contents (except microcode), but adds another module which bypassess some settings? Anyway, I removed some stuff that I didn't like (custom BIOS message, typo in CPU name, disabled BIOS recovery mode, un-deleted CGA font, removed comma before CPU count because it is in wrong line now) and added TA suffix to BIOS version. I got 1GB of memory with 2x512MB right now.

Here is updated BIOS file:

POST screen and sensors page:

I still couldn't get USB to work in current Windows 7 install and have no time to reinstall, but I just installed PCI network card, so I can easily share the files and we are back in the game 😀 I downloaded SIV32X Tyan-01 tool and ran it with current modded BIOS. Main window screenshot:

This is how SMBus looks like with fan plugged and unplugged (i have a tiny 486 heatsink and fast fan right now, so it gets hot very fast):

and full log dump: