No, it's not that. I connected VCCIOPLL and VCCA together for more current. The bus is 100 MHz anyway.
The last decision is to check out multipliers for RAM and external bus, exactly AGP, PCI. Maybe the cause is hiding in BIOS and the motherboard calculates multiplier(s) not correct. Then the system starts at a stable external frequencies ie at FSB 100 MHz.

It is necessary to delve in BIOS.

Because it is very difficult to find a man who can disassemble the BIOS and look for issues I want to try to decide this problem by using program clock generator setPLL. Exactly setPLL because this program is reading info in LUT-files which have been written in simple TXT format. Thus it is more simple to add supporting any needed generator than in other similar programs (setFSB etc). But these setPLL's LUT-files have some 16-bit codes and it is necessary to place them in correct sections and correct sequences.

My clock generator chip is ICS952617BFLF. Today I have a file named G7S620NG.LUT which consist of only FSB commands. Here is:

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: PLL: ICS952617BFLF
:: SYSTEM: G7S620NG
:: BCLKs: 100-200Mhz
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

[INIT]
set bytecount=0x0F
set local0=0x[BCLK_B1]7F00FF47D70711
set local1=0x0F07008D9AE62F00

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

[BCLK]
100, 87
108, 88
114, 84
117, C4
120, 8C
123, CC
126, 82
130, C2
133, C7
148, C6
152, 8E
156, CE
160, 81
164, C1
166, CF
170, C9
175, 85
180, C5
185, 8D
190, CD
200, 8F

Now I need to add a right line(s) in [INIT] section and "00--FF" codes in [BCLK] section which will set PCI/AGP frequencies. It should be:
[INIT]
set bytecount=0x0F
set local0=value
set local1=value
set local(2,3,etc)=value

[BCLK]
FSB FREQ number, FSB FREQ command value, PCI FREQ command value, etc (00 --- FF)

If somebody knows how to use it tell please. For example I place here LUT for MSI-GT627:

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: PLL: ics9lpr113aklf
:: SYSTEM: MSI GT627, SP9400-2.40 (266Mhz BCLK)
:: BCLKs: 266-300Mhz
:: AUTHOR: Moral Hazard@NBR
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
::
:: Create two sections INIT, BCLK.
:: Add the (optional) PCIE section for pcie overclock as well
::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: [INIT] section
:: 1. set the PLL as when setfsb does a "SET".
:: 2. Note the bytes that change when alter the BCLK, placing
:: [BCLK_B1]..[BCLKB5] in those positions.
:: 3. If creating a PCIE section then note the bytes that change when
:: alter the PCIE clock, placing [PCIE_B1]..[PCIE_B5] in those positions.
::
:: Eg: the setFSB diagnosis window OR r-w "Clock Generator" data appears as
::
:: 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F
:: 10 11 12 13 14 15 16 17 [ local3 ]
:: [ local4 ] [ local5 ]
::
:: Converts to be in the following format below
::
:: set local0=0x0001020304050607
:: set local1=0x08090A0B0C0D0E0F
:: set local2=0x1011121314151617
:: set local3=..................
:: set local4=..................
:: set local5=..................
::
:: The byte(s) that change during BCLK changes is substituted with
:: the strings [BCLK_B1], [BCLK_B2] .. [BCLK_B5]. (Optional) The byte(s)
:: that change during PCIE changes is substituted with the strings
:: [PCIE_B1], [PCIE_B2] .. [PCIE_B5] from the LUT below. Ensure localx
:: is 64-bit (16 chars), pad with 00 if needed.
::
:: bytecount = number of bytes to be send to the PLL
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

[INIT]
set bytecount=0x16
set local0=0x48EFFFFFFFA64A15
set local1=0x1F07E8[BCLK_B1][BCLK_B2]F22318
set local2=0x7DFF3B00005F80[BCLK_B3]
set local3=0x[BCLK_B4][BCLK_B5]7D7D7D003D00

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:: [BCLK] section. Found by placing the two PLL byte that change in setFSB
:: when move the slider noting the BCLK they correspond to.
::
::[BCLK], [BCLK_B1] (byte 0xB), [BCLK_B2] (byte 0xC), [BCLK_B3] (byte 0x17),
:: [BCLK_B4] (byte 0x18), [BCLK_B5] byte 0x19)
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

[BCLK]
266, 98, 42, 42, 42, 42
267, D8, 42, 42, 42, 42
268, 18, 43, 43, 43, 43
269, 58, 43, 43, 43, 43
270, 98, 43, 43, 43, 43
271, D8, 43, 43, 43, 43
272, 18, 44, 44, 44, 44
273, 58, 44, 44, 44, 44
274, 98, 44, 44, 44, 44
275, D8, 44, 44, 44, 44
276, 18, 45, 45, 45, 45
277, 58, 45, 45, 45, 45
278, 98, 45, 45, 45, 45
295, D8, 49, 49, 49, 49
296, 18, 49, 49, 49, 49
297, 58, 49, 49, 49, 49
298, 98, 49, 49, 49, 49
299, D8, 49, 49, 49, 49
300, 18, 4B, 4B, 4B, 4B

- if it can help. Thank you very much!

P.S.: I tried to seek datasheet for ICS952617BFLF but not found it.

PARUS wrote:

P.S.: I tried to seek datasheet for ICS952617BFLF but not found it.

Took a quick look, could not find at my usual datasheet haunts. 🙁

mockingbird wrote:

Coincidentally, I have an Asus P5PE-VM here, which is similar to your board in the respect that it's an 865G Socket775 board that also handles up to 65nm Core2 chips.

Among others motherboards, I also have this one. However, even if it runs stable with a QX6800 CPU, I am not able to change to multiplier above 11x or raise the FSB via Clockgen (even a 1-2 MHz increase freezes Windows).

Did you manage to overclock the board?

PasatoiuTD wrote:

Among others motherboards, I also have this one. However, even if it runs stable with a QX6800 CPU, I am not able to change to multiplier above 11x or raise the FSB via Clockgen (even a 1-2 MHz increase freezes Windows).

Did you manage to overclock the board?

Does QX run with all 4 cores activated? Or only 2 active cores?

To change up multiplier use a program ThrottleStop 6.00 (or 8.00). But mind you that if startup frequency is 2,2 GHz and you're switching it up the gameplay time in many games will run faster because perfomance timer value is set during starting Windows.

/\ In what version of Windows? AFAIK Vista uses HPET for the performance timers. And don't Core2 processors have constant TSC anyway, or else SpeedStep would mess it up?

Yes, I didn't say. I mean Windows XP. And I don't know how to change Perf timer in Windows XP without rebooting and changing real CPU clock.

PARUS wrote on 2014-11-26, 19:39:

I just sent it in private message. The changes in this BIOS are: - edited for supporting all CPUs 65nm Conroe / 45nm Wolfdale; - […]
Show full quote

I just sent it in private message. The changes in this BIOS are:
- edited for supporting all CPUs 65nm Conroe / 45nm Wolfdale;
- ACPI moved from IRQ9 to IRQ10 (for use SoftMPU in DOS by IRQ2/9, otherwise with original BIOS there is a conflict);
- edited ISA startup diagnose, detecting ISA PnP devices (for use AV310, otherwise with original BIOS this sound card won't work in DOS).

OK. Please see it. I'm waiting. Thank you very much again.

Hello, I've got the DFI G7S620-N-G. Would you share the modded BIOS with me with all the patches and fixes you mentioned including microcodes for newer CPUs? I haven't tested the board yet because I have only 775 CPUs with Core 2 Duo and newer. I think I will buy a Pentium D 820 first. And I have found mentioned a bios mod called LLC Edition in the modlabs.net forums. Do you have that bios? Thank you.

Also what I found is a pdf for I think official CPU compatibility for the board, see attached file.

gmipf, the final BIOS version supports properly 65nm Conroe CPUs. But you can't just plug it and enjoy. It is necessary to do some isolations/groundings and one signal reroute, on the CPU, via foil and adhesive tape. 45nm Wolfdale CPUs can't work properly not because BIOS but hardware limitation - sometimes freeze, sometimes no starting, sometimes power off at any moment. OK? Only 65nm Conroe! The best choice is limited to two models - E4700 (x13) and X6800 (x31) because their multipliers as max as they exist on Conroe family. The high multiplier is very important because FSB800 max.

BIOS which is on previous page is not final and has bugs.
Core 2 CPU with ISA bus is one of the best machines for "old"/"new" DOS+Win98+WinXP gaming. It can do 80286-Core2 speeds without lags. Multiplier range is x1-nominal.
The final BIOS is available for owners only. Please PM me if you want to do CPU mod and get Conroe working properly. Advice: buy X6800 for this board. Only X6800. This unique board must work with most widely scalable CPU.

P.S.: you should disable C1E in BIOS setup and use Conroe at fixed VID.
If you still want it PM me.

PARUS, thank you for the information. Soldering a wire is no problem to me. If you can tell me how the shielding is done in detail, I can do it too. The PM function isn't available to me at this forum, can I PM you at old-games.ru or modlabs.net? I have accounts their too. Thank you.

I have PMd you at the old-games.ru board. For the start I have bought a Pentium 4 660 for 10€ to be able to even start the board, I have no compatbile CPUs at the moment. I think the Pentium will arrive in 3 days. I will buy the X6800 CPU, if its the best, what the board can do. It is a little expensive though. Do you know where to get it cheap?

You also mentioned 80286 speeds, can it also disable L1 and L2 cache? If yes, can it also disable only L2 without L1? This flexibility would be awesome. I also wanted to buy a Super Socket 7 Asus P5A as a second machine for older stuff.

X6800? On ebay, I did so. You have motherboard whichs costs about $300-$1000 today and is very very rare. Is $40-$60 for X6800 really expensive? This CPU is available now, it is not necessary to seek it during months-years. And this CPU is much better than many popular "retro gaming" CPUs like Tualatin-1400S, Barton-2500+, Coppermine-1000 FSB100 etc. You can use CPU X6800 on i865 chipset at 200x1=200MHz (perfect for DOS and without error 200), at 200x15=3000MHz (perfect for WinXP), at 200x10=2000MHz (perfect for Win98). You can use it even at 100x1=100MHz and L1+L2 disabled and it will be ~80386SX speed; if you add throttle.exe you'll get about XT-80286. With i865+Core2 machine you really need no other slower machines for old stuff. 386-486 speeds are really without problems, without lags. 286 speeds are possible at FSB400 (100MHz) - it needs just reboot.
Yes, it can also disable L2 separately but in DOS it is not much useful (almost no slowing down) because L1 volume is absolutely enough for most or all DOS games on such machine. In DOS we usually disable L1+L2 totally to take effect on Core 2 systems.

Earlier all these operations were made by programs:
- SETMUL
- THROTTLE.EXE
- MSR Editor
- Cache CTL
Now we can use CPUSPD only and with the best usability, efficiency, many thanks to kalohimal!!!

My RAM recommendations for this motherboard (read it carefully and don't ignore please, believe me I have this mobo for 10 years and spent many many hours, weeks and months for its testing):
- use four single-sided 512MB memory modules, 2GB in total (don't use double-sided 512MB or 1GB modules).
- use BURNMEM.SYS driver, 64MB limitation for DOS, 512MB limitation for Win98, system.cb [386Enh] section limitation for Safe Mode.
- Samsung UCCC, NCP are better. Hynix is worse for this board.
- one more time - don't use 2x1GB, 2x1GB+2x512MB. Use please 4x512MB identical modules. Don't use double-sided modules.

I received your message. Will find final BIOS. And ask you please donate retro replica project if you wish. Also I must tell you how to modify CPU contacts.

Thanks for the advises. Then I have to find new RAM modules too. I still wait for the cheap CPU to arrive and to test the board. Of course you are right, this mainboard is much more expensive and rare compared to the CPU. I'm in fact pretty lucky after for long waiting and searching this mainboard, I have got it very cheap. I will order the cpu after validating its functionality.

Remember please Presscott at 3,6GHz in closed case can kill your motherboard. The most cold NetBurst CPU on LGA775 of those I saw is Presler up to 3,2GHz.

Are you serious? OK, I will only start it then shortly on an open bench.

It is working! First it wouldn't start, it was a combination of dirty socket pins and wrong RAM. As you said, the board is finicky about RAM, I have then found a 256MB single bank (8x soldered chips) RAM and cleaned the pins and it booted to the bios.

The next step is to buy an X6800, but only at the end of this month. Is it also OK if I send the donation next month?

Hi gmpif!

I know this thread is old and like you at the time of this thread i can not send PM. I am really interested to buy a DFI like yours in the pic and i read about Parus stuff.
you are still online on this forum but not Parus (for what i see) and i would like to know what to solder and how and to hae the final bios you are talking with Parus or where to find it (if possible!)

Obviously we can talk on the thread or in other forum if you can give me the link or wherever you want!

in the while thanks in advance!

Hi breakz,

he never shared his final bios with me because he isn't reacting anymore since then. But I have at least cleaned up his half broken bios by injecting all microcodes beside prototypes and 45nm cpus. I was able to confirm that at least one of his mods are working:

PARUS wrote on 2014-11-26, 19:39:

- ACPI moved from IRQ9 to IRQ10 (for use SoftMPU in DOS by IRQ2/9, otherwise with original BIOS there is a conflict);

I also already got an X6800 but not yet tested it since then. I have attached the bios for you.

EDIT: Regarding the soldering, I think it is just the generic BSEL mod, you can google that. I haven't done it.

HI gmipf! really thanks bout replying after all this time and sharing your bios

About the "final versioN" could be that it is rady and published somewhere in "modlabs.net"? on "oldgames.ru"there they were 　talking of "sending the bios to the owner of the board" by the rules of "modlabs.net"
I found some of the people that Palrus is thanking on the "oldgames.ru" thread but still do not know where to search about bioses.do you know any thread abou this "misterious bios" on modlanbs.net?

i really need to find a mobo like yours!

thanks again"

I already tried to ask some mods at modlabs and oldgames.ru. But they couldn't contact PARUS either anymore. Never asked forum users who already had received the final bios. Maybe you could have better luck?