I was able to get the QX6800 up and running. Installing the X6800 first then configuring the BIOS options did the trick. It still throws a uCode error on bootup, but it runs with all 4 cores in Prime95. It doesn't perform nearly as well though. The best I could do was 285 FSB with the 5:4 divider (definitely seems to be the best option so far with this P5PE-VM modded BIOS) before it started acting up in either Prime95 or SuperPi. I threw in the PQI TCCD ram and was able to get the timings down to 2.5-3-3-5 (this board also doesn't seem to like CL2 at all).

With the X6800, I was able to reach the same FSB speeds as with the E7600, but the E7600 runs cooler and seems to slightly outperform the X6800 with the higher bus/clock speeds.

I have a feeling the QX6800 will do better with the Asrock 4CoreDual-VSTA once I start messing with that. I've read on other forums people using QX9650 on that board, but I don't see how it's possible to reach 1333MHz bus speed with that VIA chipset, even with DDR2 ram.

FYI, I do not recommend the E6500K with the P5PE-VM.

I thought I was clever and that I would run it at 15 x 200 with the stock BIOS (so that I can keep the FSB at 800 with the stock BIOS). Well, the BIOS doesn't allow multiplier adjustment past 11. Also, I don't know if my Chinese E6500K was actually an E6500K or maybe a rebadged E6500, because it acted funny with the board.

Either way, I ended up getting an Core2 E7500 and the Russian BIOS (after updating 067A to the 2010 version with MMTool). Everything is perfectly fine, running at 266 x 11 (1066FSB). So this seems like the most reasonable "fastest" CPU for the P5PE-VM. You just need the Russian BIOS for 45nm support.

Detection is not automatic with the BIOS. You need to disable automatic multiplier selection and set it to 11, and enable overclocking and set the FSB to 266 and the divider to 5:4, otherwise it'll boot to sub 1Ghz speeds.

Gigabyte GA-8I865GME-775 , it even has drivers for windows 98.

mockingbird wrote on 2021-08-04, 00:38:

FYI, I do not recommend the E6500K with the P5PE-VM.

I didn't have great results with the E6500K either. The E7x00 or X6x00 CPUs are definitely the best performance and best for overclocking.

I had issues with the BIOS auto-detecting CPUs too. If I remember correctly, you need to fiddle with the multiplier. Like the E7600 has a multiplier of 11.5, so when switching to another CPU, that multiplier is out of range.

Gigabyte GA-8I865GME-775 , it even has drivers for windows 98.

No support for Core2 though 🙁

Personally I wouldn't bother procuring a motherboard with only NetBurst support. Contemporary MB with AGP and Socket 939/754 would be much cheaper (back then) and more energy efficient.

Just did a quick search on auction sites; MB with AGP supporting Core 2 cost roughly between US$20-50 only.

If Win9x support is needed then one can rule out VIA PT880 Pro / Ultra.

dormcat wrote on 2021-08-04, 19:03:

Personally I wouldn't bother procuring a motherboard with only NetBurst support. Contemporary MB with AGP and Socket 939/754 would be much cheaper (back then) and more energy efficient.

Just did a quick search on auction sites; MB with AGP supporting Core 2 cost roughly between US$20-50 only.

If Win9x support is needed then one can rule out VIA PT880 Pro / Ultra.

I like to run XP on these setups. I have both the Asus P5PE-VM and the Asrock 4CoreDual-VSTA. I haven't even messed with the Asrock yet and it's been a while (again) since I was running the P5PE-VM.

bnice7 wrote on 2021-08-04, 16:45:

mockingbird wrote on 2021-08-04, 00:38:

Gigabyte GA-8I865GME-775 , it even has drivers for windows 98.

No support for Core2 though 🙁

I run a dual core 3.4Ghz and its pretty decent. Tops out at 2GB ram though.

bnice7 wrote on 2021-08-04, 19:43:

I like to run XP on these setups. I have both the Asus P5PE-VM and the Asrock 4CoreDual-VSTA. I haven't even messed with the Asrock yet and it's been a while (again) since I was running the P5PE-VM.

Those are very good MB under WinXP. However, IMHO the main purpose of keeping AGP is to maximize Win9x compatibility, creating a 3-in-1: an overkill build for Win9x, a very nice build for WinXP, and still a good build for Win7.

If no Win9x combability is needed then Core 2 / Core i / Sandy / Ivy CPU on PCIe MB are very easy to find and cost very little: a friend retiring an old machine, at a yard/garage/trunk sale, at a local recycling center, you name it.

You are seeing Ivybridge junked already? I thought I was doing well scoring a C2Q for free last month... I think we need to revive Limewire though, I was getting 3-5 year old machines for free once they were clogged up with viruses. Nuke and repave and you'd got a decent machine.

BitWrangler wrote on 2021-08-05, 03:04:

You are seeing Ivybridge junked already? I thought I was doing well scoring a C2Q for free last month... I think we need to revive Limewire though, I was getting 3-5 year old machines for free once they were clogged up with viruses. Nuke and repave and you'd got a decent machine.

Not "junked" but being sold for ~US$50 for an entire working system, usually retired from big companies, schools, or government offices. Most of them have weak PSU and no independent video card but for that price you get MB + Sandy/Ivy CPU + 4-8GB DDR3 RAM + 500G-1T HDD + DVD burner + legal Win7, and you can replace any undesirable component with a better one. That's far cheaper than many sought-after individual retro components e.g. ISA Sound Blaster, Gravis UltraSound, SS7 MB with K6-III+, Socket 370 MB with Tualatin, Voodoo3/4, etc.

C2D/C2Q and corresponding AMD builds like Athlon64 X2 do appear in junkyards from time to time; I was really lucky to have picked up an Athlon64 X2 5000+ on MSI K9A2 Platinum in the trash dump of my neighborhood.

bnice7 wrote on 2021-08-04, 16:45:

No support for Core2 though 🙁

There are some revisions, which support only P4 and Celeron, another with dualcore support and yet another with C2 support.

bnice7 wrote on 2021-08-04, 16:45:

I had issues with the BIOS auto-detecting CPUs too. If I remember correctly, you need to fiddle with the multiplier. Like the E7600 has a multiplier of 11.5, so when switching to another CPU, that multiplier is out of range.

Are you using 5:4 "low latency mode"? It crashes after Windows loads in this mode, but works fine with the high latency setting. Keep in mind, the RAM is still running 20Mhz over spec with 5:4 and a 266 FSB.

Well, the time came to ditch the Russian BIOS and the E7400.

It was kind of stable with the aforementioned settings and a 5:4 divider, but only if I set the FSB to 250Mhz, and then the RAM ran at spec, but the CPU ran at only 2.75Ghz.

But that must have messed something up because when UDMA was enabled (I was wondering why my HDD was so slow - it was running at PIO Mode 4 - maybe intentionally enabled by default by our Russian author), it wouldn't boot to Windows.

So I went back to the vanilla BIOS and put the E6700 in... Well, the BIOS uses a 3:2 divider for this CPU, and RAM ran at 354Mhz... Fine and everything, but the system was unstable. I don't know why there's instability at odd FSB and RAM speeds, maybe it has something to do with the PCI or AGP divider.

I settled on my E4600, which has an FSB of 800Mhz, which means the RAM runs at 1:1, and everything is fine and dandy at 2.4Ghz.

Not to mention that the northbridge was a lot cooler. The 865G really doesn't like more than an 800Mhz FSB.

Conclusion: The E4600 or the E4700 is the best bet for this board. Totally stable, no headaches, nada... The only mod I have to the BIOS is just newer OROMs and microcodes. This board is a revision 1.02, maybe 1.03 is better. I had considered doing the memory voltage mod, but I also tried the 2.65V option in the Russian BIOS and that didn't help enough, even at stock speed.

mockingbird wrote on 2021-08-17, 03:37:

Well, the time came to ditch the Russian BIOS and the E7400. […]
Show full quote

Well, the time came to ditch the Russian BIOS and the E7400.

It was kind of stable with the aforementioned settings and a 5:4 divider, but only if I set the FSB to 250Mhz, and then the RAM ran at spec, but the CPU ran at only 2.75Ghz.

But that must have messed something up because when UDMA was enabled (I was wondering why my HDD was so slow - it was running at PIO Mode 4 - maybe intentionally enabled by default by our Russian author), it wouldn't boot to Windows.

So I went back to the vanilla BIOS and put the E6700 in... Well, the BIOS uses a 3:2 divider for this CPU, and RAM ran at 354Mhz... Fine and everything, but the system was unstable. I don't know why there's instability at odd FSB and RAM speeds, maybe it has something to do with the PCI or AGP divider.

I settled on my E4600, which has an FSB of 800Mhz, which means the RAM runs at 1:1, and everything is fine and dandy at 2.4Ghz.

Not to mention that the northbridge was a lot cooler. The 865G really doesn't like more than an 800Mhz FSB.

Conclusion: The E4600 or the E4700 is the best bet for this board. Totally stable, no headaches, nada... The only mod I have to the BIOS is just newer OROMs and microcodes. This board is a revision 1.02, maybe 1.03 is better. I had considered doing the memory voltage mod, but I also tried the 2.65V option in the Russian BIOS and that didn't help enough, even at stock speed.

Mockingbird, just a few things:

For hard drives, it's imperative to use IDE and not SATA for anything other than stock FSB speeds. There's no lock in place for the SATA PCI bus, so when you up the FSB speed, you're also overclocking the SATA ports. I use a SATA SSD with an IDE to SATA adapter, plugged into the IDE port on the motherboard.

I have noticed weird issues with the locks between my rev. 1.02 board vs. my rev. 1.03 board, but I can't remember exactly what they are (probably posted here earlier in this thread). I just remember the 1.03 board I have performs much better for me. Pretty sure if I remember correctly, clockgen showed the PCI lock not working at all on my 1.02 board.

The 2.55 / 2.65v DDR Reference Voltage setting in the hacked bios won't do anything at all for overclocking. DDR Reference Voltage is a different setting vs. vDDR supply voltage, which is not adjustable by the hacked bios at all.

Use whatever setup works best for you and that your happy with, but just wanted to clarify those few things.

bnice7 wrote on 2021-08-17, 04:22:

Mockingbird, just a few things: […]
Show full quote

Mockingbird, just a few things:

For hard drives, it's imperative to use IDE and not SATA for anything other than stock FSB speeds. There's no lock in place for the SATA PCI bus, so when you up the FSB speed, you're also overclocking the SATA ports. I use a SATA SSD with an IDE to SATA adapter, plugged into the IDE port on the motherboard.

I have noticed weird issues with the locks between my rev. 1.02 board vs. my rev. 1.03 board, but I can't remember exactly what they are (probably posted here earlier in this thread). I just remember the 1.03 board I have performs much better for me. Pretty sure if I remember correctly, clockgen showed the PCI lock not working at all on my 1.02 board.

The 2.55 / 2.65v DDR Reference Voltage setting in the hacked bios won't do anything at all for overclocking. DDR Reference Voltage is a different setting vs. vDDR supply voltage, which is not adjustable by the hacked bios at all.

Use whatever setup works best for you and that your happy with, but just wanted to clarify those few things.

Ah, ok, that explains it...

When you say "stock" FSB speeds, at 1066, you're operating the memory at 425.6Mhz instead of 400Mhz, so you're saying your voltage mod could fix instability issues with that... But is the PCI running out of spec with that? If not, then perhaps instability could have indeed been solved with your memory voltage mod (and I have a 20K pot, I was just too lazy to take the motherboard out of the case and solder sense and control wires).

Also, I take issue with what you're saying about the SATA/IDE bus... They're both a function of the southbridge... Why would one not be affected by an overclocked PCI bus and not the other? I think what you're accomplishing with the adapter is that you're running the HDD in PIO or MWDMA mode. Ostensibly, you could limit a native sata connection in this way in the BIOS. By the way, there is also an option for "32-bit data transfer" in each individual HDD listed that is not enabled by default and significantly slows down performance when disabled. I highly recommend enabling that.

Either way, there was a very noticeable increase in heat at 1066Mhz FSB, and I also have an FX5800 in there so I didn't want to heat things up too much. Active cooling for the northbridge is an option, but this just isn't good in this cramped case for the long term.

I did however order a 45nm Pentium E5800, which gives a significant clock boost over the Core2 E4600 (3.2Ghz vs 2.4 Ghz), and the E5800 is very comparable otherwise (2MB cache, and more importantly 800Mhz FSB) so I'll report back how that goes with the modded BIOS. Perhaps it is a better choice.

bnice7 wrote on 2021-08-17, 04:22:

Mockingbird, just a few things: […]
Show full quote

Mockingbird, just a few things:

For hard drives, it's imperative to use IDE and not SATA for anything other than stock FSB speeds. There's no lock in place for the SATA PCI bus, so when you up the FSB speed, you're also overclocking the SATA ports. I use a SATA SSD with an IDE to SATA adapter, plugged into the IDE port on the motherboard.

I have noticed weird issues with the locks between my rev. 1.02 board vs. my rev. 1.03 board, but I can't remember exactly what they are (probably posted here earlier in this thread). I just remember the 1.03 board I have performs much better for me. Pretty sure if I remember correctly, clockgen showed the PCI lock not working at all on my 1.02 board.

The 2.55 / 2.65v DDR Reference Voltage setting in the hacked bios won't do anything at all for overclocking. DDR Reference Voltage is a different setting vs. vDDR supply voltage, which is not adjustable by the hacked bios at all.

Use whatever setup works best for you and that your happy with, but just wanted to clarify those few things.

Ah, ok, that explains it...

When you say "stock" FSB speeds, at 1066, you're operating the memory at 425.6Mhz instead of 400Mhz, so you're saying your voltage mod could fix instability issues with that... But is the PCI running out of spec with that? If not, then perhaps instability could have indeed been solved with your memory voltage mod (and I have a 20K pot, I was just too lazy to take the motherboard out of the case and solder sense and control wires -- though I realize it's just a matter of the correct resistor after fiddling with the pot).

Also, I take issue with what you're saying about the SATA/IDE bus... They're both a function of the southbridge... Why would one not be affected by an overclocked PCI bus and not the other? I think what you're accomplishing with the adapter is that you're running the HDD in PIO or MWDMA mode. Ostensibly, you could limit a native sata connection in this way in the BIOS. By the way, there is also an option for "32-bit data transfer" in each individual HDD listed that is not enabled by default and significantly slows down performance when disabled. I highly recommend enabling that.

Either way, there was a very noticeable increase in heat at 1066Mhz FSB, and I also have an FX5800 in there so I didn't want to heat things up too much. Active cooling for the northbridge is an option, but this just isn't good in this cramped case for the long term.

I did however order a 45nm Pentium E5800, which gives a significant clock boost over the Core2 E4600 (3.2Ghz vs 2.4 Ghz), and the E5800 is very comparable otherwise (2MB cache, and more importantly 800Mhz FSB) so I'll report back how that goes with the modded BIOS. Perhaps it is a better choice.

mockingbird wrote on 2021-08-17, 14:50:

Ah, ok, that explains it... […]
Show full quote

Ah, ok, that explains it...

When you say "stock" FSB speeds, at 1066, you're operating the memory at 425.6Mhz instead of 400Mhz, so you're saying your voltage mod could fix instability issues with that... But is the PCI running out of spec with that? If not, then perhaps instability could have indeed been solved with your memory voltage mod (and I have a 20K pot, I was just too lazy to take the motherboard out of the case and solder sense and control wires).

Also, I take issue with what you're saying about the SATA/IDE bus... They're both a function of the southbridge... Why would one not be affected by an overclocked PCI bus and not the other? I think what you're accomplishing with the adapter is that you're running the HDD in PIO or MWDMA mode. Ostensibly, you could limit a native sata connection in this way in the BIOS. By the way, there is also an option for "32-bit data transfer" in each individual HDD listed that is not enabled by default and significantly slows down performance when disabled. I highly recommend enabling that.

Either way, there was a very noticeable increase in heat at 1066Mhz FSB, and I also have an FX5800 in there so I didn't want to heat things up too much. Active cooling for the northbridge is an option, but this just isn't good in this cramped case for the long term.

I did however order a 45nm Pentium E5800, which gives a significant clock boost over the Core2 E4600 (3.2Ghz vs 2.4 Ghz), and the E5800 is very comparable otherwise (2MB cache, and more importantly 800Mhz FSB) so I'll report back how that goes with the modded BIOS. Perhaps it is a better choice.

About the FSB lock and using SATA ports, I'm just going by the notes from the guy who created the hacked BIOS. I'm not sure why it's an issue.

https://docs.google.com/document/d/1cQkYVhTn2 … dit?usp=sharing

Yes, lots more heat on the NB when run at 1066MHz. I did have a small fan on top of the heatsink at one point. It helped a little bit.

The vddr mod could potentially help the memory hit 425MHz, but it also depends a lot on the RAM sticks.

bnice7 wrote on 2021-08-17, 15:04:

About the FSB lock and using SATA ports, I'm just going by the notes from the guy who created the hacked BIOS. I'm not sure why […]
Show full quote

About the FSB lock and using SATA ports, I'm just going by the notes from the guy who created the hacked BIOS. I'm not sure why it's an issue.

https://docs.google.com/document/d/1cQkYVhTn2 … dit?usp=sharing

Yes, lots more heat on the NB when run at 1066MHz. I did have a small fan on top of the heatsink at one point. It helped a little bit.

The vddr mod could potentially help the memory hit 425MHz, but it also depends a lot on the RAM sticks.

Yes, I see it now, thanks:

"CPU External Frequency (MHz) - sets the value of the system bus frequency, the values are changed by pressing the "+" and "-" buttons. For owners of SATA drives, only the standard series 100,133,166,200,266,333 is recommended."

"PCI Frequency (MHz) - 33 / Auto - selection of a fixed frequency or fsb dependent (SATA lock on 33 does not help, on the contrary)"

So the 250Mhz FSB I was using to keep the memory in spec was indeed wreaking havoc with the PCI bus speed.

SATA ports are not tied to PCI bus. All integrated stuff in south bridge use proprietary bus since early 2000s.

I'm back on the Russian BIOS 😀

Hopefully this will be the conclusion to this epic.

The Pentium E5800 arrived today and it turns out it is a good compromise between the lower clock speed of the Core2 E4600 and the larger cache of the E7x00. Plus, it is 45nm vs the E4600 (65nm) so it runs cooler, eventhough it's 3.2Ghz as opposed to 2.4Ghz.

But more importantly - it is stable! Same setup as before - you need to manually set the multiplier to 16, enable overclocking and manually set the FSB to 200Mhz (I think I left the ratio on Auto, but set it to 1:1 if you want), and the system is perfectly stable. No weirdness like with 1066Mhz (266) FSB.

Now, if only Intel hadn't disabled SSE4 for the E5800 in the CPU microcode... That's asking too much.

So -
Core 2 E4600 for the ordinary BIOS
Pentium E5800 if you feel like flashing the chip

Make sure to use the newest 067A microcode as the one included by the author isn't the latest.