TPU's Nostalgic Hardware Club

[Retrorockit]

Some OEM market MB have a locked PLL chip and can't raise the fsb. You could try the E7600 2 core and see if you can get 3.83GHz (11.5x333fsb). It has SSE4 and would be better for old single thread gaming.

 

[Greenslade]

I hope you guys are having a good weekend. I have another problem.i am trying to dual boot Windows 7 and XP.I have downloaded EasyBCD but it fails to install.it nearly gets to the end, but it says*Installation Aborted set up was not completed successfully.Show more says *Testing internet connection installation cancelled?*The version is 2.4.

 

[MachineLearning]

No, you're right. 8GB is max, I got something wrong & corrected my previous post.

By the way, I gave it a try with different CPU configuration(s) and no, it doesn't work. All I can squeeze out of Q6600 is approx. 3.00GHz, 333MHz. Regardless of the voltage, it simply hangs on POST & reverts back to fail-safe config. Same thing actually as my other Q6600 which ALSO happens to be SLACR and is (oddly enough) also paired with MSI motherboard. This one however is much older model (P43 NEO) and uses DDR2.

Interesting... Both G41 and P43 have FSB walls so I wonder if that's the problem. Apparently it was intentional to limit competition with P45.

From that G41 overclock.net link:

Steps:

1) Turn off the usual bits (C1E, HPET, etc.)
2) Set mems to 1:1 ratio
3) Set PCI-E frequency to 104 Mhz
4) Lower multiplier to x8
5) Start increasing fsb to desired level!

...

That's it, originally I couldn't even get past 400 Mhz without getting a BSOD or my PC crashing. Now I'm at 425 fsb rock stable @prime95 using 425x8 for my q6600 but I believe this is memory limited since I'm using Kingston Value Rams on my rig

edit: also, strictly for testing, using 1 DIMM may help you at least boot higher frequencies, if DRAM is a limiter (which I frankly doubt)

 

[Retrorockit]

I got a nice bump on my G41 from Crucial Ballistix Sport 2x4GB DDR3 Ram. The CAS9@1600 modules run 6-6-6-16 at 1066 speed in my Dell.

 

[Mr.Scott]

I got a nice bump on my G41 from Crucial Ballistix Sport 2x4GB DDR3 Ram. The CAS9@1600 modules run 6-6-6-16 at 1066 speed in my Dell.

That's no bump. Cl9@1600 is faster than Cl6@1066.

 

[Retrorockit]

According to this it's actually the same. 11.3ns latency..

RAM Speed Calculator | XBitLabs

This is a calculator to determine an absolute RAM latency or RAM speed.

www.xbitlabs.com

The "bump" is going from the typical 7-7-7-20 @ 1066 to 6-6-6-16 timing at the same speed.
If you have an old computer that can't run RAM at 1600 speed it's a big bump

 

[Mr.Scott]

According to this it's actually the same. 11.3ns latency..

RAM Speed Calculator | XBitLabs

This is a calculator to determine an absolute RAM latency or RAM speed.

www.xbitlabs.com

The "bump" is going from the typical 7-7-7-20 @ 1066 to 6-6-6-16 timing at the same speed.
If you have an old computer that can't run RAM at 1600 speed it's a big bump

Ahhh.............didn't know you couldn't run @ 1600.
My bad.

 

[Trekkie4]

Some OEM market MB have a locked PLL chip and can't raise the fsb. You could try the E7600 2 core and see if you can get 3.83GHz (11.5x333fsb). It has SSE4 and would be better for old single thread gaming.

Right, OK. But these are not OEM boards... Especially not P43 Neo, it was hand-picked for a gaming rig back in 2009.

Interesting... Both G41 and P43 have FSB walls so I wonder if that's the problem. Apparently it was intentional to limit competition with P45.

From that G41 overclock.net link:


edit: also, strictly for testing, using 1 DIMM may help you at least boot higher frequencies, if DRAM is a limiter (which I frankly doubt)

Interesting... Yes, this could certainly explain a thing or two. I'm not going to remove RAM module because I'm already running under full load all the time (it would seem that 4GB isn't what it used to be, 10 years ago) but I may try disabling all these other options & reducing the multiplier.

Edit
Nothing... Even with C13, HPET disabled, PCIe frequecy at 104MHz, multiplier at 8X and memory to (as close as possible) to 1 : 1, I can't even get to 3.00GHz mark any longer.

 

[MachineLearning]

Interesting... Yes, this could certainly explain a thing or two. I'm not going to remove RAM module because I'm already running under full load all the time (it would seem that 4GB isn't what it used to be, 10 years ago) but I may try disabling all these other options & reducing the multiplier.

Edit
Nothing... Even with C13, HPET disabled, PCIe frequecy at 104MHz, multiplier at 8X and memory to (as close as possible) to 1 : 1, I can't even get to 3.00GHz mark any longer.

Hmm, what are your CPU VTT, vCore and NB voltages at? And how much vDroop is there for vCore under full load, is LLC a BIOS option?

 

[Trekkie4]

Hmm, what are your CPU VTT, vCore and NB voltages at? And how much vDroop is there for vCore under full load, is LLC a BIOS option?

They were all default during that last test. There was no need to increase voltage because I could barely even get to factory clock of 2.4GHz.

 

[Retrorockit]

Ahhh.............didn't know you couldn't run @ 1600.
My bad.

I can't adjust my RAM timing either. Those are the timings those run at 1066 speed as is. No one publishes those specs.

They were all default during that last test. There was no need to increase voltage because I could barely even get to factory clock of 2.4GHz.

You might just try a Q6700. They're all G0 and if you have a little extra Voltage it should go 3.33GHz.

G41 is famous for the FSB not moving. It's tied to some bus that crashes everything.

 

[MachineLearning]

They were all default during that last test. There was no need to increase voltage because I could barely even get to factory clock of 2.4GHz.

If you feel like taking another crack at it, this is what I'd try:
- PCIe frequency 100
- multiplier 9x
- LLC enabled if possible
- northbridge voltage @ 1.3-1.35v
- CPU VTT @ 1.4-1.45v
- vCore @ 1.50v, check vDroop. Will probably be a couple dozen mv lower from what you set in BIOS, dropping even further under load. Might end up around 1.4v under load.
- DRAM to spec'd timings for primaries (manually)
- all other voltages (SB, etc.) to their default setting (again manually)

This configuration should alleviate any bottleneck sans:
- vrm overheating / inadequate (vDroop over 200mv for instance)
- chipset limitations not affected by voltage (i.e. intel meddling)

edit: not suggesting 1.50v for daily use necessarily, but at least to eliminate that variable from causing an issue.

I can't adjust my RAM timing either. Those are the timings those run at 1066 speed as is. No one publishes those specs.


You might just try a Q6700. They're all G0 and if you have a little extra Voltage it should go 3.33GHz.

Q6700 is not worth the trouble, they have 2 SLACR CPUs and I refuse to believe that both are of such low quality that 3GHz is their max. Especially when the one in their photo was made in 2008, that's a mature 65nm right there

 

[Retrorockit]

They were all default during that last test. There was no need to increase voltage because I could barely even get to factory clock of 2.4GHz.

How are those capacitors looking?

Q6700 is not worth the trouble, they have 2 SLACR CPUs and I refuse to believe that both are of such low quality that 3GHz is their max. Especially when the one in their photo was made in 2008, that's a mature 65nm right there

G41 raises the PCIe bus with the fsb. If you can adjust that down as you go you might get somewhere. It's not the CPU it's the G41 chipset. I work around it other ways because I odn't have other options.

 

[MachineLearning]

G41 raises the PCIe bus with the fsb. If you can adjust that down as you go you might get somewhere. It's not the CPU it's the G41 chipset. I work around it other ways because I odn't have other options.

I've not been able to find another mention of this, only people saying they were trying different PCIe bus frequencies to get a higher FSB. Are you sure it's not something specific to your board or am I missing something? Also they're having issues with P43 OC as well so I bet it's either Intel artificially limiting both chipsets or the OC configuration

 

[DR4G00N]

They were all default during that last test. There was no need to increase voltage because I could barely even get to factory clock of 2.4GHz.

From searching around pretty much all of MSI's 775 boards are pretty meh at best, even the high-end ones.

The P43 Neo is also pretty low end so I think running a quad is a bit much for it. Just because it's on the support list doesn't mean it'll work well.

 

[Retrorockit]

P43 I have no idea.
G41 is known to be hard to overclock. Most people give up on it. I'm so used to locked BIOS Dells I do other things already. So I can't say what will work if you insist on fsb overclocking a G41.
But I do see it in the SetFSB posts to lower the bus 1 for very 3-4 FSB. The furthest I got with SetFSB was 337fsb. I didn't know the 3:1 PCIe trick then. Unfortunately SetFSB isn't working for me right now or I would be trying it myself. I would love to see how far my 65W Q9505s will go on a MB that supports a 120W Xeon.

 

[Robert B]

Volunteer - 657c 123 r

NEW HW FOUND!!!

1. ELSA VICTORY Erazor-4 - nVIDIA RIVA 128 4MB PCI *** VICTORY Erazor-PCI-4
2. Intel Pentium ES - SX963 - 100MHz - ES is written on the back.
3. AMD DX-2 80MHz *** A80486DX2-80 *** Am486DX-2 80
4. Team Xtreem 1GB DDR500 - TXDR1024M500HC3 *** 3-4-4-8 - just one stick
5. Intel Pentium III 550MHz - SL3F7 - Katmai - Slot 1
6. A couple of copper CPU coolers - Titan CU5TB ( a very very very whisper silent cooler not ) and a Spire one. You musn't leave a slab of copper to go to waste.
7. ELSA Winner 1000PRO - in bad shape. Repair in progress.
8. Gigabyte 7N400 Pro 2 - nForce 2 Ultra 400 - Sk.A + AXP 2200+ with all the bells and whistles. One memory slot is missing the clip and piece of plastic from the slot. The memory slot works well without it. Not ideal but for the price I got it I can't complain.
9. Gigabyte GA-586ATV Rev. 4B - Sk.7 + IBM 686L PR166+ *** 6x86L-2VAP166GB 133MHz 2.8V
10. ASUS A8V sk. 939 AGP8x R2.01 + A64 3200+ - needs 15 caps. The original Nippon Chemi-Con 820uf 6.3V are bulged. I hope that the board will POST after repairs. Minor damage on some plastic bits.

 

[erek]

Just ordered this for $160 shipped

 

[Mr.Scott]

Nice price.

 

[NoiseBox]

This hardly qualifies for "antique" but just got a free system, a Socket 775 (Xeon e3110) Asus P5QL-VM DO (uATX) system with 8GB DDR2 in some kind of cube case that I've not seen before. Sadly, the NIC card seemed damaged (code 10 under windows 10) and the front USB cables, when plugged into the mobo give me the "overvoltage detected, shutting down" even when nothing is plugged into them.
I'm a rather big fan of the old Shuttle XPC boxes, and had considered picking one up to make the ultimet C2Q system, but this box will do for now.

 

[Ruru]

This hardly qualifies for "antique" but just got a free system, a Socket 775 (Xeon e3110) Asus P5QL-VM DO (uATX) system with 8GB DDR2 in some kind of cube case that I've not seen before. Sadly, the NIC card seemed damaged (code 10 under windows 10) and the front USB cables, when plugged into the mobo give me the "overvoltage detected, shutting down" even when nothing is plugged into them.
I'm a rather big fan of the old Shuttle XPC boxes, and had considered picking one up to make the ultimet C2Q system, but this box will do for now.

775, 1366 and AM2/AM3 are totally suitable for this thread.

 

[MachineLearning]

This hardly qualifies for "antique" but just got a free system, a Socket 775 (Xeon e3110) Asus P5QL-VM DO (uATX) system with 8GB DDR2 in some kind of cube case that I've not seen before. Sadly, the NIC card seemed damaged (code 10 under windows 10) and the front USB cables, when plugged into the mobo give me the "overvoltage detected, shutting down" even when nothing is plugged into them.
I'm a rather big fan of the old Shuttle XPC boxes, and had considered picking one up to make the ultimet C2Q system, but this box will do for now.

ASUS sometimes has jumpers next to the internal USB ports to select between them using 5VSB or 5V for power. If yours has those, give that a shot

 

[NoiseBox]

ASUS sometimes has jumpers next to the internal USB ports to select between them using 5VSB or 5V for power. If yours has those, give that a shot

I found the manual, and indeed it does, I'll swap it over and see what happens, thanks for the heads up!

 

[Trekkie4]

How are those capacitors looking?

You mean, on G41 board? Last time I checked, looking good (all original). This is a fairly modern board, so I don't expect them to go bulging up any time soon.

 

[Retrorockit]

The thing about G41 is with only 2 RAM slots it would have been a 4GB system back then. With an onboard GPU using some of that it would have been for 32 bit OS office computers. Mine they even left off an expansion slot so they could use the same MB# for Desktop and Mini Tower. G41 was all about cutting corners. Many of them had a 95W CPU limit, even though the chipset can go higher.
You started overvolting an already overclocked system and it crashed. Did they cut corners on the capacitors too? You may be already pushing 120W through a 95W VRM. The truth is any computer that can run the Q6600 at 3Ghz can probably also run the 95W Q9650 or in the case of G41 the E5450 80W Xeon as a starting point. That doesn't mean it won't crash when you move the fsb about 4 clicks.They're known for this. Try a 65W 2 core and see if it crashes when you bump the fsb. If it does, think about the E7600 at 3.83GHz with SSE4. If it doesn't think about a more modern CPU to match the more modern MB.
I got away with an X5470 which is equal to 2x E8600 CPUs. OC never happened.
Averaged 60fps at Superposition 1080P Medium setting. It took a couple tries to break the 8000 point barrier. I didn't see any other LGA775 systems around me there. i3 and i5 mostly. Some AMD stuff too.

UNIGINE Superposition benchmark score

UNIGINE Superpsition detailed score page

benchmark.unigine.com