X48 Rampage Formula Preview.

[greg.m]

error @ 71%

give atry @ 1.61nbv and see..........

[kuebk]

greg.m why you are not using memtest86+? And did you try to adjust tREF?

error @ 71%

give atry @ 1.61nbv and see..........

Maybe it's your ram man.

My ram requires 2.4v to work stable at 600mhz, I wouldn't be surprised if your ram would want more power.

[Zucker2k]

Yeap, until you run 2.1 vdimm and fail, you don't have much of an argument. Besides,other factors like high fsb and improper NBV and timings could also be factors. Give a little higher cpu volts too, just to eliminate the possibility of that being an issue.

[greg.m]

greg.m why you are not using memtest86+? And did you try to adjust tREF?



Maybe it's your ram man.

My ram requires 2.4v to work stable at 600mhz, I wouldn't be surprised if your ram would want more power.

Ive already done that with success and with using much lower nb,ram volts like A-Grey. Thats the prob...

[kuebk]

Yeap, until you run 2.1 vdimm and fail, you don't have much of an argument. Besides,other factors like high fsb and improper NBV and timings could also be factors. Give a little higher cpu volts too, just to eliminate the possibility of that being an issue.

I highly doubt that PL <= 7 is reachable above 600mhz on ram without much volting MCH/RAM.

Ive already done that with success and with using much lower nb,ram volts like A-Grey. Thats the prob...

You can't compare volts, some sticks require less - some more. I would start with 2.3v on ram.

[Alien Grey]

Yeap, until you run 2.1 vdimm and fail, you don't have much of an argument. Besides,other factors like high fsb and improper NBV and timings could also be factors.

I've been running MemTest86+ V2.11 without problems with DRAM Voltage 1.90V and NB Voltage 1.41V.

I'm going to try it Windows with MemTest to see if it's really the memory that fails.

Why should I use DRAM Voltage 2.10V and 1.55-1.60V to run at FSB 465MHz if I can have FSB 450MHz with DRAM Voltage 1.80V and NB Voltage 1.33V?

This doesn't make sense.

[neo_rtr]

just to say that with 0902 bios and same settings as before i fail in 2-3 Prime pass.
I took it up +1 on CPU, PLL,NB,FSB (x63 and x65) and i fail almost all the time.
qx6700 and 8x450, 333 strap 1018 mem.

[Alien Grey]

I highly doubt that PL <= 7 is reachable above 600mhz on ram without much volting MCH/RAM.



You can't compare volts, some sticks require less - some more. I would start with 2.3v on ram.

I shoudn't try that with these sticks without active cooling.

[kuebk]

I've been running MemTest86+ V2.11 without problems with DRAM Voltage 1.90V and NB Voltage 1.41V.

I'm going to try it Windows with MemTest to see if it's really the memory that fails.

Why should I use DRAM Voltage 2.10V and 1.55-1.60V to run at FSB 465MHz if I can have FSB 450MHz with DRAM Voltage 1.80V and NB Voltage 1.33V?

This doesn't make sense.

Yes it does, after some point the amount of volts is disproportionate to the amount of mhz you can earn.

[greg.m]

I highly doubt that PL <= 7 is reachable above 600mhz on ram without much volting MCH/RAM.



You can't compare volts, some sticks require less - some more. I would start with 2.3v on ram.

Do you know about which mem are we talking about?
If not here it is: G.SKILL F2-9600CL5D-4GBPI. There is also a thread here in xs.
This mem have done crazy things. And there are other mobos with which it works like a charm..

Anyway i wouldnt have problem to raise vram beyond 2.1(rated) but how far can i go. Wahts the vlimit before.....?

Edit: i updated my sig with right comps

[Alien Grey]

Ive already done that with success and with using much lower nb,ram volts like A-Grey. Thats the prob...

Do you agree if I say that CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps are the settings that we should use but that it can't stay stable with these settings?

I think that with some small adjustments in the BIOS on the CPU and NB Clock Skew it would be possible to keep it stable.

It has nothing to do with tRD, Ai Clock Twister moderate, DRAM Static Read Enabled or DRAM CLK Skew on Channel A/B.

I tried everything and nothing helps to keep it stable. From the moment you go above DDR 1200MHz it isn't possible anymore to keep it stable.

[Alien Grey]

Do you know about which mem are we talking about?
If not here it is: G.SKILL F2-9600CL5D-4GBPI. There is also a thread here in xs.
This mem have done crazy things. And there are other mobos with which it works like a charm..

Anyway i wouldnt have problem to raise vram beyond 2.1(rated) but how far can i go. Wahts the vlimit before.....?

Edit: i updated my sig with right comps

You can try 1600MHz with 2.6V and dry ice like the guy on that thread that you are saying.

[greg.m]

Do you agree if I say that CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps are the settings that we should use but that it can't stay stable with these settings?

I think that with some small adjustments in the BIOS on the CPU and NB Clock Skew it would be possible to keep it stable.

It has nothing to do with tRD, Ai Clock Twister moderate, DRAM Static Read Enabled or DRAM CLK Skew on Channel A/B.

I tried everything and nothing helps to keep it stable. From the moment you go above DDR 1200MHz it isn't possible anymore to keep it stable.

Except the first test iv done all the others is with cpuskew100 and nbskewnormal

and all these : tRD, Ai Clock Twister moderate, DRAM Static Read Enabled or DRAM CLK Skew on Channel A/B set to auto or as it was.

But ill be back with my exact bios settings. Right now im testing with 1.61nbv.....

[greg.m]

With 1.61nbv error came much earlier @ 18% .........f@kc

[kuebk]

Do you know about which mem are we talking about?
If not here it is: G.SKILL F2-9600CL5D-4GBPI. There is also a thread here in xs.
This mem have done crazy things. And there are other mobos with which it works like a charm..

Anyway i wouldnt have problem to raise vram beyond 2.1(rated) but how far can i go. Wahts the vlimit before.....?

Edit: i updated my sig with right comps

Yes I know, but don't forget that every stick is different. As I said some are better some are worst. Maybe quad is the problem - its used to put much stress to the fsb.

[Alien Grey]

Except the first test iv done all the others is with cpuskew100 and nbskewnormal

and all these : tRD, Ai Clock Twister moderate, DRAM Static Read Enabled or DRAM CLK Skew on Channel A/B set to auto or as it was.

But ill be back with my exact bios settings. Right now im testing with 1.61nbv.....

You didn't understand me. What I'm saying is that if you would use CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps it could have been stable without having to force it. But with this BIOS it isn't and that's why I think that corrections should be done in the BIOS to make it stable with CPU Clock Skew Delay 100ps and NB Clock Skew Delay 100ps.

[greg.m]

Yes I know, but don't forget that every stick is different. As I said some are better some are worst. Maybe quad is the problem - its used to put much stress to the fsb.

The think that both A-Grey and Fritz the germ have almost the same prob with running that mem above 1200 stable, points either to the mobo or to the mem. But i think its the rampage that limits mems performance. Enough for today, time to play crysis and relax

[greg.m]

You didn't understand me. What I'm saying is that if you would use CPU Clock Skew Delay 200ps and NB Clock Skew Delay 100ps it could have been stable without having to force it. But with this BIOS it isn't and that's why I think that corrections should be done in the BIOS to make it stable with CPU Clock Skew Delay 100ps and NB Clock Skew Delay 100ps.

I totally agree that asus should do some corrections/fixes because its unacceptable to stop caring for a "top" mobo after a year. They made a new bios which supports "new cpus"(who cares but if so which ones), but i havent seen since a long time ago a new bios which "supports new ram" or "improve ram compatibility" or "improve ram performance" or whatsoever.

[Alien Grey]

The think that both A-Grey and Fritz the germ have almost the same prob with running that mem above 1200 stable, points either to the mobo or to the mem. But i think its the rampage that limits mems performance. Enough for today, time to play crysis and relax

Oh... and don't forget Roger_D25. He is also running with this fine memory on his ASUS Rampage Formula board. He is also saving on his electricity bill like I do and is running at 8 X 450MHz and DDR 1200MHz with tRD 6.

Why are we having so much trouble with DDR speeds above 1200MHz on our ASUS rampage Formula board that OFFICIALLY has no DDR 1200MHz limitation.

That's something that I just found out.

[Alien Grey]

I totally agree that asus should do some corrections/fixes because its unacceptable to stop caring for a "top" mobo after a year. They made a new bios which supports "new cpus"(who cares but if so which ones), but i havent seen since a long time ago a new bios which "supports new ram" or "improve ram compatibility" or "improve ram performance" or whatsoever.

Visit the ASUS rampage Formula forum and add your thoughts to the ASUS Rampage Formula BIOS 0902 thread.

Thank you for your help sorting things out. We didn't get any wiser but we know enough now.

[Zucker2k]

Okay those of you having problems with DDR2 1200 try this: Move rams into white slots, set cpu skew to delay 100 (I agree finer adjustments eg. 50ps would do wonders), set nb skew to normal. Set your ram to max volts (2x2GB skills don't suffer from over-voltage even up tp 2.80v) ie. 2.100v - make sure this is actual. Set NB to 1.4.5 - 1.5 (for FSB 450). Advnace both DRAM CLCK SKEW Channels A/B to 300. Set Ai Clock Twister to moderate.

I would test this myself, but I no longer own this board. Please give me some feedback; I'm more interested in MCI Memtest (Windows) than Memtest 86+ under DOS mode. Thanks.

[greg.m]

Visit the ASUS rampage Formula forum and add your thoughts to the ASUS Rampage Formula BIOS 0902 thread.

Thank you for your help sorting things out. We didn't get any wiser but we know enough now.

ASUS Rampage Formula BIOS 0902 UPDATED

[Mikefra2008]

para fsb strap 266 que lvl perfomance usan ?

[Alien Grey]

Okay those of you having problems with DDR2 1200 try this: Move rams into white slots, set cpu skew to delay 100 (I agree finer adjustments eg. 50ps would do wonders), set nb skew to normal. Set your ram to max volts (2x2GB skills don't suffer from over-voltage even up tp 2.80v) ie. 2.100v - make sure this is actual. Set NB to 1.4.5 - 1.5 (for FSB 450). Advnace both DRAM CLCK SKEW Channels A/B to 300. Set Ai Clock Twister to moderate.

I would test this myself, but I no longer own this board. Please give me some feedback; I'm more interested in MCI Memtest (Windows) than Memtest 86+ under DOS mode. Thanks.

Not possible. ASUS already optimized the DRAM CLK Skew on Channel A/B for running these sticks without problems on the ASUS Rampage Formula. They have fix this issue in BIOS 0902.

Changing the DRAM CLK Skew on Channel A/B isn't going to work. The DRAM Clock Fine Delays are on 11-9-11-9 and that's where they should be.

I'm going to run MemTest in Windows XP with the maximum available free memory. This means that I have to use two copies of MemTest with both half the free memory.

[Alien Grey]

This is a copy of my post at the ASUS Rampage Formula forum.

Just like I expected. Running 2 copies of MemTest to use all free memory available in Windows gave ZERO errors.

What does that tell me?

There are no Memory problems with BIOS 0902. I've been running MemTest86+ V2.11 Test 5 for more than an hour and it didn't find any errors on the Memory.

To see that there are no problems with the Memory in our OS I have been running 2 copies of MemTest for Windows to use all free memory available to test.

Most users prefer to use MemTest for Windows because it's more stress full and is going to pick up errors much quicker than MemTest86+ V2.11 because it stresses the NB more.

To look if DDR 1239MHz, DRAM Static Read Control Enabled, Ai Clock Twister Moderate and tRD 6 are a problem, I did the test with MemTest.

Here's is a screenshot of my result. No errors after 100%. Some users are going to say that 100% isn't enough. I tell you I can run MemTest86+ V2.11 or MemTest for Windows all day long and it will pick up ZERO errors.



How come that Linx already fails after the second pass as you can see in this screenshot?



The answer is very simple. It isn't the Memory that makes Linx fail it's something else.

If it isn't the Memory it must be the CPU.
Well it isn't the CPU that makes Linx fail because my CPU can be perfectly stable at 3.72GHz.

What can be wrong that it randomly fails in Linx or Prime95 after some time?

The answer is more easy than I expected. It's been there for quiet some time but we didn't look at it. Let's have a look at my settings for 8 X 465MHz and DDR 1117MHz with tRD7.

What are the differences between these settings and the ones I've been using for DDR 1239MHz?

A different Strap, faster Memory speed, tRD 6 and CPU Clock Skew Delay 200ps.

MemTest86+ V2.11 and MemTest for Windows already showed us that there is no problem with the Memory so we can rule out the different strap, faster Memory speed and tRD 6.

These settings seem to be bullet proof.

When all these settings seem to be bullet proof there is only one that remains and that is the CPU Clock Skew Delay 200ps.

Brilliant! Here's our problem. The difference between the CPU Clock Skew Delay and The NB Clock Skew Delay is making these random failures.

As I don't know the exact amount of delayed time I can only speculate what the problem is. It could be that the CPU Clock Skew is to far or to short delayed. It can be that the difference between the CPU and NB Clock Skew delay is to short or to large.

Anyway if what I'm thinking is true the random failures can be removed with some adjustments on the CPU and NB Clock Skew.

It's possible that we need smaller delay times of 50ps like we have on the DRAM CLK Skew on Channel A/B.