4PCA3+ OC help

[dragsteracer]

OK first off the system specs:

4PCA3+ rev 2.2
2.6c batch Q337A275 (Think its a malay) CPUZ reports stepping 9 rev D1
2X256 Kingston HX3200K2
2 80 gig Hitachi SATA drives in Raid 0
Gainward GF3 Ti200 video card
Sound Blaster Audigy Gamer
Win XP home
Home made watercooling rig, temps never go above 40 deg c under load

Now the problems:

Best stable overclock is 250 fsb 5:4 divider with aggressive timings 2 2 2 6 with the Vcore at 1.6 Vagp at 1.6 and the vdimm at 2.9.No matter what I do with the memory timings and voltages 250 stable is pretty much it.If I disable HT I can get into windows at 260 or so but prime errors immediately. Memtested the memory overnight and found no errors. I think its a CPU/Mobo voltage problem as no matter what I set the vdimm to in bio it will not go above 2.9_. If I upp the Vcore and vagp it seems to draw the vdimm down even more.....Tried different PSU and they read close to the same.....The main reason I wanted to try a higher Vdimm is that I was under the impression that the KHX3200K2 memory used BH5 and it likes 3.0+ volts but now that I read some posts in ohter furums I am not so sure.........Another clue its not memory related it I can switch to 6:4 and OC will not get any better....Can anyone help me out with this? I am supicious that the mobo is doing something funky with the voltages also because in CPUZ the cpu is actually the vcore voltage I just figured out last night.....What should I do? RMA mobo? Bad CPU for Overclocking? I am at a loss for what to do next.
I do know I really like this mobo but 250 fsb with the 5:4 divider on water is very dissapointing......TIA for any help!

[keithwalton]

How is your northbridge cooling ?
without a cpu fan the air around there could get pretty stagnent.
rev 2.2 i just got a replacement for my now dead rev1.0 and it was a 2.0 though they dont seem to have changed much different northbridge heatsink and different colour connectors / cables. that and the different vcore chip thingy.

250fsb is still quite good though :-)

K

[dragsteracer]

Before it was kinda toastly lol.I lapped it and switched to AS3.... As soon as I get the chance I am going to make a waterblock for it but for now I have a 92mm case fan wired up @ 7 volts moving air across it. (once you make to switch to water you just HATE fan noise lol)It also moves some air over the heat spreaders on the kingston and around the capacitors on the other side of the cpu, and over my heat sinks on the ti200 memory....

SOS help! SOS need some advice from you 280-290 FSBers!!!

[Goat Fajitas]

Originally Posted by dragsteracer

Another clue its not memory related it I can switch to 6:4 and OC will not get any better....!

My first guess, after reading what you have tried so far is that your CPU has hit its limit.

How far have you upped the voltage on it ?

[dragsteracer]

Originally Posted by Goat Fajitas

My first guess, after reading what you have tried so far is that your CPU has hit its limit.

How far have you upped the voltage on it ?

Goat I swapped the 2.6 SL6WS out for a SL6Z3 2.4 today and am rocking on quite nicely in prime at 285 fsb 5:4 divider and 2 2 2 5 timings....

When it comes to voltage settings on this board I have no idea what readings to belive. I get different readings in MBM the bios and when I take a volt meter and measure it on the board....Reading some forums they say the vcore undervolts badly but when I check the mosfets beside the CPU that change with the vcore setting, they read +.05 of what I set the vcore to in the bios.....I am unsure weather this is the correct place to measure them, but I can tell you the figures change to the T with every vcore change in the bios........right now as I type this running prime priming I am "supposedly" at 1.54 in the bios and that shows up on the mosfets at 1.59 - 1.60....How high can you go on water??? Anything above 1.70 or so would scare the pants off me since I seen the "sudden northwood death syndrome" back in the days of the 1.6A clockmonsters......

[Goat Fajitas]

Believe the voltmeter measurement , nothing else. Delete MBM and all programs like it... worthless junk.

[soloflow]

where are the measuring points on the 4pca?

thanks

[EPoX Tech]

I really don't like it when people suggest RMA because their board doesn't overclock as they wish it would. All that does is continue to drive up the relative cost of mainboards.

[Clevor]

The 2.60C-2.80Cs can be pretty obstinate and while they can do 3.3 gig at below default voltage, they hit a wall fast and won't budge.

As for your 285, 5:4, 2-2-2-5, try benching 3D . Not just on the Epox, but almost every other board I've tried ('cept the Asus) you can be 2D and even SuperPI stable over 280 at 5:4, but 3DMark2001 will kick out or reboot the computer. All you can do is drop the fsb until both 3DMarks pass, and throw a lot of other 3D at it, particularly the SS2nd Encounter and IL-2 Sturmovik demo loop. I have passed all three 3DMarks but you may encounter SS dumping to the menu screen or get an error box halt IL-2. That or just plain lockups. IL-2 tends to reboot the computer when running or exiting.

Also, on the KHX3000 (BH-5) I tried on the 4PCA3+, it would not pass 3DMark2001 at 5:4 above 265 fsb, even though it did DDR490, 2-2-2-5 at 3.3 volts at 1:1. Figure that one out. The SD/CW boards runs best at 1:1, then 3:2, then 5:4.

I have three 2.40Cs that max out at 287-288 on other boards. I can only run them up to 279, 5:4, 2-2-2-5 on the 4PCA3+.

[Goat Fajitas]

Hmmm ... Mine is 3d stable 24/7 at 288, of course my NB is watercooled

[-CTPAX-]

Clevor, the 4PCA3+ doesnt top 280 fsb. I can bench 3dmark2001se at 292 on air 5:4, cpu limited. It may clock lower than p4c800, but theres no such thing like "280 fsb limit".

[Clevor]

Originally Posted by -CTPAX-

Clevor, the 4PCA3+ doesnt top 280 fsb. I can bench 3dmark2001se at 292 on air 5:4, cpu limited. It may clock lower than p4c800, but theres no such thing like "280 fsb limit".

But how come my 2.40C can run 287, 5:4, 2-3-2-5 on the P4P800 and P4P800 and only do 279, 5:4 on the Epox boards? And I've tried two 4PDA2+s and four 4PCA3+s. I did get up to 282, 5:4 on one 4PDA2+ board.

Also, the 280 fsb limit is a well-known issue with every motherboard out there except the Asus boards (I know as I've tested over a dozen boards by Abit, DFI, MSI, Gigabyte, AOpen, Asus). Hipro over at Extreme can do 371 fsb but he still sees the 5:4 cap at around 270-280 fsb on his Abit board. It does seem the people on phase change are immune to it.

[Goat Fajitas]

Originally Posted by Clevor

But how come my 2.40C can run 287, 5:4, 2-3-2-5 on the P4P800 and P4P800 and only do 279, 5:4 on the Epox boards? And I've tried two 4PDA2+s and four 4PCA3+s. I did get up to 282, 5:4 on one 4PDA2+ board.

Also, the 280 fsb limit is a well-known issue with every motherboard out there except the Asus boards (I know as I've tested over a dozen boards by Abit, DFI, MSI, Gigabyte, AOpen, Asus). Hipro over at Extreme can do 371 fsb but he still sees the 5:4 cap at around 270-280 fsb on his Abit board. It does seem the people on phase change are immune to it.

You just got lucky with your Asus boards...

There IS a limit issue that occurs between 270 and 290 on MANY i865/875 boards, but it effects them all.

Some Asus cant hit it, some can
Some Abit cant hit it some can
Some Epox cant hit it, some can.
Most of those other brands wont ever hit it.

On the other hand, can a default 200mhz bus not running stable at 280 be called as "issue" ? I'd say it is quite a bit over spec already.

My advice to you... Although I do understand your frustration, stop trying all these motherboards looking for the "holy grail" and use what you have. 279*12= more then plenty fast enough to run any game or app out there.

[EPoX Tech]

We had 302 here on the 4PCA3+ when we released it - I'll try to find an image out if I can - was quite a while ago.

It is true that there seems to be a hitchy gap between a certain area and then things smooth out a bit. There are reasons relating to sine lengths etc., but really a tad too outlandish to try to explain fully other than to state that something does happen!

Nice o/c there Goat Fajitas - and......

On the other hand, can a default 200mhz bus not running stable at 280 be called as "issue" ? I'd say it is quite a bit over spec already.

Quite in agreement here in context of what I'd earlier stated in relation to RMA'ing etc. and hiking everyone elses prices up.

[OC_Newbee]

Mine is also above 280fsb with 5:4 on the memory at 2-2-2-5 3.1v. 100% stable at anything you throw at it, including 3dmark and games. All on AIR and tiger1 on the northbridge.

[Goat Fajitas]

Also noteworthy... Someone at Vrzone modded the VCORE, VDIMM, and Chipset and got their 4pc3a up to 330mhz.

[Goat Fajitas]

Originally Posted by EPoX Tech

Nice o/c there Goat Fajitas - and... Quite in agreement here in context of what I'd earlier stated in relation to RMA'ing etc. and hiking everyone elses prices up.

Thanks, I actually have the 12/25 BIOS now, and I upded my CPU voltage... Currently testing at 291@5:4 mhz , with the same config in my sig. Stable so far But I am worried about my CPU at 1.8 volts.

Is the 12/25 an official BIOS ? Its available via magicbios, but not on the website. If it is official, Is there a changelog ?

[althes]

Originally Posted by Goat Fajitas

Thanks, I actually have the 12/25 BIOS now, and I upded my CPU voltage... Currently testing at 291@5:4 mhz , with the same config in my sig. Stable so far But I am worried about my CPU at 1.8 volts.

Is the 12/25 an official BIOS ? Its available via magicbios, but not on the website. If it is official, Is there a changelog ?

I got my rev1,2 up to 292 stable. It can do 300fsb but not really stable.

[Goat Fajitas]

Originally Posted by althes

I got my rev1,2 up to 292 stable. It can do 300fsb but not really stable.

Cool ! 292@ 5:4 or 3:2 ?

[Clevor]

Since I can't get above 279 at 5:4 I just run run my XMS3500 at 1:1 and switch to a high multi CPU. I was able to run a 2.80C at 250, 1:1, 2-2-2-5 at 3.2 volts. I got 3.5 gig and 3600+ unbuffered . My 3DMark2001 points went up by 1400 over a little 2.40C. I have a 3.0C in store for this board, which I can run at 3.72 gig on air at 1.58 volts. Will have to drop a tad on the fsb to 248, but still in the sweet spot of my XMS3500.

I have even more in store, as the 4PDA2+, Rev. 2.0 with full PAT got me 3700+ unbuffered at just 240, 1:1, 2-2-2-5 at 3.2 volts with the XMS3500. The memory tweaks to get full PAT stresses out the ram so I can't bench 3D at 250, 1:1, 2-2-2-5 like I normally can, but I get higher unbuffered anyway at 240 with the tweaks.

Normally, to get this kind of bandwidth/CPU power I would have to run a 2.40C at 310 fsb, 5:4 (DDR496), 2-2-2-5 to net me the same numbers. I'd need an exceptional CPU, board, and probably phase change. So there is more than one way to skin a cat .

The key to getting this kind of unbuffered bandwidth is the Epox 4PCA3+/4PDA2+ boards and good BH-5. I have tested six boards so far, and every one of them can run my XMS3500 at 250, 1:1, 2-2-2-5 at 3.1-3.2 volts, even though some boards undervolt as low as 3.14 and as high as 3.24 (when set to 3.2).