ASUS P5N32-E Sli Memory timing not changing after changing it in BIOS

[Widjaja]

[SOLVED] ASUS P5N32-E Sli Memory timing not changing after changing it in BIOS

Not sure where to put this but I think it maybe a motherboard problem/limitation rather than a memory issue.

The RAM timings according to the sticks of RAM are 5-5-5-12 but the timings the motherboard keeps the RAM at are 5-5-5-18 regardless of manually setting the timings in BIOS.

There is a Sli Ready Memory option which is greyed out due to having 4 sticks of RAM installed which is supposed to allow the motherboard to run the RAM at manufacturer specifications so that's out of the option as well.

Voltages, multipliers FSB are all functional.

So I am guessing with more than 2 sticks of RAM this motherboard refuses alter timings?

I've searched the net and there is no info on the subject.

 

[Bundy]

Not sure where to put this but I think it maybe a motherboard problem/limitation rather than a memory issue.

The RAM timings according to the sticks of RAM are 5-5-5-12 but the timings the motherboard keeps the RAM at are 5-5-5-18 regardless of manually setting the timings in BIOS.

There is a Sli Ready Memory option which is greyed out due to having 4 sticks of RAM installed which is supposed to allow the motherboard to run the RAM at manufacturer specifications so that's out of the option as well.

Voltages, multipliers FSB are all functional.

So I am guessing with more than 2 sticks of RAM this motherboard refuses alter timings?

I've searched the net and there is no info on the subject.

Wow, I'm amazed you still have one of those still working, I have blown up 2...

Anyway, as I recall, Sli ready is the same terminology as EPP now. It shouldn't matter.

I didn't think having 4 sticks should matter but to provide some ideas, can you post the memory profiles and SPP info from CPU-Z or memset? Also, what voltage are you running on the NB? Is the rest of the rig stable? The BIOS will revert to default if it thinks you are not stable.

 

[Widjaja]

This machine was left to run at defaults since the day it was bought so it was never stressed in anyway so I guess this is why it is still alive.

I recently upgraded the BIOS to 1903 as well although I forgot to set everything at defaults and the machine would not post until I cleared CMOS.

I'd say the RAM is less stressed than before at the moment since I have st the RAM to 1.86v (which is the lowest this motherboard allows) instead of 2.1v since the SPD suggests 1.8v

Otherwise I have left everything at default including the NB voltage.

Here's the memory read outs in CPU-Z

When I tried to overclock the CPU is would tell me it failed and would revert to stock when it entered windows.
I red this motherboard is terrible for overclocking quads so I have left it at stock.

Yes the rig is absolutely rock solid BTW.

 

[Bundy]

I think you will need to run higher volts on the NB and the RAM. I also had a thermalright cooler plus fan on the NB because I needed 1.5V going through it to keep eveything on track.

The issue of Quad cooling is true, it was difficult but possible. The problem was no ability to adjust the GTL volts and they were set "wrong" at default. The answer was to pump the vtt volts to high levels and that seemed to work ok for FSB up to about 350Mhz.

Even at stock though, I think you need more NB volts for 4 sticks.

 

[Widjaja]

Should there have been a BIOS post telling me the timings failed?

I would have expected to see the BIOS back at stock timings after I had found the clocks are still at default in CPUZ.

Also after the BIOS flash and CMOS reset the Q6600 was set at X6 multiplier which I found odd.
But it may have been done to get the system to boot or something.

It all seems pretty strange to me.
Not a big deal though as I am happy with the current performance.

 

[John Doe]

1.35v NB is alright for stock, let it be 2 or 4 sticks. But if you're going to OC, bump it up to 1.45. Oh and, SLi-ready mem. only works with 2 sticks due to chipset limitation.

That said though, you're better off selling it for a cheap P35/P45 or even a P965. nVidia chipset Intel boards are a nightmare.

 

[Widjaja]

Don't know how accurate HWmonitor is as the temp readings are pretty off.
I read something at -54degC and my CPU temps are always exactly 10degC higher than in Real temp.

Anyway it appears my NB voltage is at 1.25v stock according to HWmonitor.

 

[Bundy]

Should there have been a BIOS post telling me the timings failed?

I would have expected to see the BIOS back at stock timings after I had found the clocks are still at default in CPUZ.

Also after the BIOS flash and CMOS reset the Q6600 was set at X6 multiplier which I found odd.
But it may have been done to get the system to boot or something.

It all seems pretty strange to me.
Not a big deal though as I am happy with the current performance.

I agree that you should have got messages for a failed boot so maybe it was not that. I still suggest volts to be your answer for this chipset. BTW, as you might have noticed, they run very tight NB timings so will appear to have higher performance/higher memory bandwidth than equivalent boards. IMO they were too tight.

1.35v NB is alright for stock, let it be 2 or 4 sticks. But if you're going to OC, bump it up to 1.45. Oh and, SLi-ready mem. only works with 2 sticks due to chipset limitation.

That said though, you're better off selling it for a cheap P35/P45 or even a P965. nVidia chipset Intel boards are a nightmare.

I found the boards to be ok, and faster. Agreed on the difficulty but that didn't stop me from running an easy 400FSB with a core duo and 350FSB on a Quad. I never got any better on an Intel board anyway, due to heat (Aussie is hot) I am sure if I put a WC loop on my CPU that I could have done better.

 

[John Doe]

It's energy saving, you can bump it up to 1.35-1.45v. I've had a 680i and a 780i, both of which use the same chipset.

"I found the boards to be ok, and faster. Agreed on the difficulty but that didn't stop me from running an easy 400FSB with a core duo and 350FSB on a Quad. I never got any better on an Intel board anyway, due to heat (Aussie is hot) I am sure if I put a WC loop on my CPU that I could have done better."

Uhm, what are you on about? nVidia chipsets run much hotter, have FSB holes and can't clock up properly. 400 FSB is easy as pie. You can bump a C2D from 333 to 400 just by changing the value. And 350 FSB on a quad is disapointing at best. A solid P35 can get up to 400-450. Intel chipset boards are better in every single way.

 

[Widjaja]

I'll bump up the voltage and set the timings to see if the RAM timings show in CPUZ when I get around to it.

I don't know whats meant by tight NB timings though.
If anything, all I'd want is to be able to bump the FSB to 333 so I can get 3Ghz on the Quad stable ut have not seen any threads where someone has the board stable at 3Ghz or more.

 

[Bundy]

I'll bump up the voltage and set the timings to see if the RAM timings show in CPUZ when I get around to it.

I don't know whats meant by tight NB timings though.
If anything, all I'd want is to be able to bump the FSB to 333 so I can get 3Ghz on the Quad stable ut have not seen any threads where someone has the board stable at 3Ghz or more.

Tight NB timings - The NB has timings like your RAM but you cannot adjust them on that board. BIOS sets them and they were tight as compared to other boards, this giving better performance but increased unreliability. thats why more volts helps these chips. Back in the day, this is where nvidia pwned.

On Intel X48, NB timing can be set, it's called performance level on Gigabyte boards. It's a number, the number of clocks to wait for the NB to pass an instruction to or from the RAM.

Now that the memory controller is on the CPU, this number is largely redundant.

EDIT: to get 333, you will need active NB cooling and 1.5 V on it. Vtt (As I faintly recall), needs also to be at the max. setting of 1.5V as well. This is to compensate for the shitty GTL voltage. An inelegant approach I admit. Also note though that this will be well outside of spec and might degrade the CPU a bit, unless the cooling is good.

 

[John Doe]

It's called performance level (from 1 to 10) on DFI boards as well. You don't need to adjust it unless you're going for a heavy OC. The problem with nVidia boards is the chipset itself. It has many issues such as wrong sensors, vdroop and so on.

1.5v VTT/NB just for 333 FSB? Joking, right? That's way too high.

 

[Bundy]

You don't need to adjust it unless you're going for a heavy OC.

It may be adjusted for any OC, or you may leave it on auto.

The problem with nVidia boards is the chipset itself.

thats what I said

It has many issues such as wrong sensors, vdroop and so on..

They all worked on mine

1.5v VTT/NB just for 333 FSB? Joking, right? That's way too high.

thats what I implied but not so forcefully. It was to help stability. The "holes" you refer to were mainly due to tight NB timing, as mentioned before. Increasing NB volts could make them largely disappear. I agree the vtt is extreme and if Widjaja can get the CPU up to those speeds without a high vtt, then that is best. I still think the NB will need 1.5 V.

When I quote speeds, they are fully stable speeds, run on air in an environment of 30C plus. I had no problems posting a lots of higher speeds, but they were not stable.

 

[Widjaja]

Oh well.
Looks like I'll keep the system at stock.

I can't place any active cooling on the NB due to the GPU being triple slot.

My Brother had an Intel chipset and had no problems with overclocking and did tell me in the past about the 680i chipset at one stage frying memory.

My future build will unlikely be another nVidia chipset anyway.
Didn't have much luck with either of my nForce4 boards.
I only bought this setup due to being a good deal on the combo.

 

[Bundy]

You should be able to get some extra out of it.

lol - I burned my finger on the NB heatsink though, that was when I first got the boards and did an auto OC.

 

[John Doe]

It may be adjusted for any OC, or you may leave it on auto.

You don't need to mess with GTL or performance levels for a mild OC. If anything, you'd make things unstable at mild OC's. Those're only needed for high OC's.

thats what I said

No, you didn't. You said the boards sucked because of the "NB timing", which doesn't exist.

They all worked on mine

You can't know whether software is accurate or not. You need a multimeter, and I've had a 790i as well. On all 3, BIOS voltage and software readings were different. If I put 1.35v in BIOS, software showed 1.30.

thats what I implied but not so forcefully. It was to help stability. The "holes" you refer to were mainly due to tight NB timing, as mentioned before. Increasing NB volts could make them largely disappear. I agree the vtt is extreme and if Widjaja can get the CPU up to those speeds without a high vtt, then that is best. I still think the NB will need 1.5 V.

Look, there's no such as "NB timing". Those are called performance levels. And his NB does not need 1.5v. It's the combination of board with a weak, old built (coil chokes) and the problematic chipset that can't OC. You're giving poor advice.

When I quote speeds, they are fully stable speeds, run on air in an environment of 30C plus. I had no problems posting a lots of higher speeds, but they were not stable.

Yeah, and you do realize 350 FSB on a quad is miserable, right? I don't think you know what you're talking about at all.

 

[Widjaja]

Today I change the NB voltage to 1.35v and changed the ram timings a bit lower than standard, increased the RAM voltage to 2.1v, changed to manual and standard OC profile but still no change mentioned in CPUZ even though the NB voltages changed fine.

It's almost as if memory timings have been disabled for some reason.

Very strange.

When I get around to it, if I can be bothered I'll remove a couple of the sticks to get it to run in dual channel to see what happens.

 

[Bundy]

You don't need to mess with GTL or performance levels for a mild OC. If anything, you'd make things unstable at mild OC's. Those're only needed for high OC's.



No, you didn't. You said the boards sucked because of the "NB timing", which doesn't exist.



You can't know whether software is accurate or not. You need a multimeter, and I've had a 790i as well. On all 3, BIOS voltage and software readings were different. If I put 1.35v in BIOS, software showed 1.30.



Look, there's no such as "NB timing". Those are called performance levels. And his NB does not need 1.5v. It's the combination of board with a weak, old built (coil chokes) and the problematic chipset that can't OC. You're giving poor advice.



Yeah, and you do realize 350 FSB on a quad is miserable, right? I don't think you know what you're talking about at all.

There is a term for people like you, anyway read this here

Today I change the NB voltage to 1.35v and changed the ram timings a bit lower than standard, increased the RAM voltage to 2.1v, changed to manual and standard OC profile but still no change mentioned in CPUZ even though the NB voltages changed fine.

It's almost as if memory timings have been disabled for some reason.

Very strange.

When I get around to it, if I can be bothered I'll remove a couple of the sticks to get it to run in dual channel to see what happens.

Yea darn mate, sorry I can't help more, it's been ages since I used that board. I don't even have the manual any more. I do think it would be a good idea to try 2 sticks as you suggest.

 

[Widjaja]

No problem, thanks for offering help though.

It is rather strange this is happening.
I have never overclocked a system with all slots loaded and this scenario pops up.

No big deal, will get around to it when I do.
As for now, I'll just run the rig as is.

 

[Widjaja]

[SOLVED]

When I flashed the BIOS the system would not boot as mentioned.
Flicked the battery out and it resolved that.

It appears removing the battery did not clear the BIOS completely the first time round.
Removed the CMOS battery again and now I am able to adjust everything as I should.

As you can see SPD still says 5-5-5-18 1.8v when my sticks say 5-5-5-12 so going to Corsair for answers on that question.

I found the solution after googling ASUS Striker Extreme boards to see if any of them had the same problem since the ASUS P5N32-E SLi is pretty much the same board.
Turned out someone with an ASUS Striker Extreme did indeed have an identical problem.

Case closed.