TWIN3X2048-1333C9DHX + Striker II Extreme

[cje]

Well I ran blend all day. Temps at highest were 58 degrees. (around 24 degrees ambient room temp).

 

I was experimenting with the RAM this evening. I clocked them up to 1600 mhz to match the fsb, timings: 9-9-9-24-2T. I gave them 1.6v (1.66v actual).

 

They passed 16 passes of memtest over 5 hours, but froze in vista after 5 minutes. I know it was an optimistic clock. Am I going to be realistically able to clock these to 1600mhz? Or should I just save my self the hassle and buy the TW3X4G1600C9DHX kit instead?

 

Also, can I take this opportunity to thank DerekT, which has been a massive help. Thanks mate. Really really appreciate it.

[DerekT]

I clocked them up to 1600 mhz to match the fsb, timings: 9-9-9-24-2T. I gave them 1.6v (1.66v actual).

 

They passed 16 passes of memtest over 5 hours, but froze in vista after 5 minutes. I know it was an optimistic clock. Am I going to be realistically able to clock these to 1600mhz? Or should I just save my self the hassle and buy the TW3X4G1600C9DHX kit instead?

Set to 1.85v Set your Northbridge to 1.50v. Retest. You are warranted up to and including 2.0v on your memory but I have not seen >1.85 really do any benefit other than heating the modules up. You don't need to do more than a few passes of memtest. You're just driving the heat and electronic discharge up when you do that.

 

Also, can I take this opportunity to thank DerekT, which has been a massive help. Thanks mate. Really really appreciate it.

 

No problem.

[cje]

Are you sure 1.85v will be safe? The sticker on the side says 1.5v. I just want to check first, before i try it. I had set VDIMM to 1.72 before, and 1 of my sticks developed a fault, and have been RMA'd. Mind you, only 1 chip developed a fault. the other 3 were fine.

[DerekT]

Better not do it. Forgot you had four modules.

 

Are you sure 1.85v will be safe? The sticker on the side says 1.5v. I just want to check first, before i try it. I had set VDIMM to 1.72 before, and 1 of my sticks developed a fault, and have been RMA'd. Mind you, only 1 chip developed a fault. the other 3 were fine.

[cje]

I'm running 2 modules at the moment, as I'm still waiting for RMA on the other 2. Been 2 weeks now.

 

So, I know my overclock runs sweet with RAM @ 1066 8-8-8-20-1T @1.6v

 

So, naturally, being itchy fingered, I had to see how fast I can get these 2 modules. Currently got them stable so far. (Need more memtest runs, only did 5). No bsods or errors in vista.

 

Current settings are:

fsb: 1600

ram: 1600

 

Vcore:1.25

Loadline Calibration:Enabled

PLL:1.52

VDIMM: 1.62

VTT: 1.32

NB:1.44

 

Timings: 9-9-9-30-1T

trc:50

trFc:100

 

 

It used to freeze after 20 minutes idling with 9-9-9-24, so raised it to 30 and seems a lot better.

 

Well it runs sweet for an hour or two like that, but when left alone it had restarted. Any suggestions on getting stability?

[DerekT]

Higher Vcore. Come on. You need to play with it and find out what your machine will and will not do. This is merely hand holding.

 

I'm running 2 modules at the moment, as I'm still waiting for RMA on the other 2. Been 2 weeks now.

 

So, I know my overclock runs sweet with RAM @ 1066 8-8-8-20-1T @1.6v

 

So, naturally, being itchy fingered, I had to see how fast I can get these 2 modules. Currently got them stable so far. (Need more memtest runs, only did 5). No bsods or errors in vista.

 

Current settings are:

fsb: 1600

ram: 1600

 

Vcore:1.25

Loadline Calibration:Enabled

PLL:1.52

VDIMM: 1.62

VTT: 1.32

NB:1.44

 

Timings: 9-9-9-30-1T

trc:50

trFc:100

 

 

It used to freeze after 20 minutes idling with 9-9-9-24, so raised it to 30 and seems a lot better.

 

Well it runs sweet for an hour or two like that, but when left alone it had restarted. Any suggestions on getting stability?

[cje]

I am not expecting hand holding. My cpu is stable with that amount of vcore. Because with a RAM speed of 1066 it is totally stable. but when I raise frequency to 1600 mhz, that's when problems occur.

 

Sorry! Just noticed that vcore is mis typed. Vcore is actually 1.35v! I don't think it would even boot with 1.25.

[DerekT]

You have to overclock the memory controller to reach the high speeds of the memory. When you populate all four DRAM slots, you lose the ability to clock as highly. You have to see what you can get with the two modules and then accept what you can make. That's what all enthusiast overclockers do.

 

Your CPU is a 266Mhz (266 X 4 = 1066 Quad Pumped) FSB part. 1600Mhz DRAM is an 800Mhz (800 X 2 Double Pumped) part. That's the issue. Your CPU even when overclocked highly has trouble reaching a stability with the memory controller when it is being driven up. You would do better with at 333/1333Mhz FSB CPU part.

 

At any rate, I doubt if I can help you so I will just leave this thread, and yes, you are expecting hand holding. You are overclocking your machine highly and asking for help with gaining stability at that high overclock. Sorry, you won't get the answers from me as I am not going to advise you to raise voltages higher than midline and your overclock of both the CPU and memory is higher that that already. Overclocking the memory to a bin higher is not a simple thing and most have to access higher voltages than I would advise or anyone working on this site because you forgo warranty and DRAM that shows the characteristics of overvoltage will lose RMA.

 

Here is a 106 page thread dedicated to your motherboard. Ask them about your Q6600 and trying to make 1333Mhz memory run at 1600Mhz.

 

http://www.xtremesystems.org/forums/showthread.php?t=180445

[cje]

But I thought you said my ram is warranted up to 2v? I am no way going up to that, I'm just confused you say they are warranted up to 2v, then you say taht 1.7v invalidates my warranty. Which is correct?

[DerekT]

Show me where I say that 1.7 invalidates your warranty? I am saying that you will have to do operations on voltages to gain such stability and I am not going to give you that advice.

 

This is NOT an overclocking site. So, my advise to you was to go to the site I told you and where you will get the advice you need.

 

I gave you a site where you can get that advice. Your board is already known for overvolting memory so unless you find the Vdimm points on the motherboard and test them, you can not be certain where your voltage really is on the memory. You are not going to find a stability in the longer term with what you are doing currently which is why you are not finding stability. You are looking for higher end speeds and for this reason I gave you a site to find your information.

 

But I thought you said my ram is warranted up to 2v? I am no way going up to that, I'm just confused you say they are warranted up to 2v, then you say taht 1.7v invalidates my warranty. Which is correct?

[cje]

OK, thanks mate. I will run them 1066 mhz for now.

 

Provided my board is stable, would this kit work ok? TW3X4G1600C9DHX?

 

I have been told that the Striker Ii will only be stable with RAM ratio of 1:1, 1:2, 3:2. This seems true as 1333mhz (with the only 2 sticks in) is unstable.

 

So, since I'm sure y've realised I am new to overclocking, I will buy a rated 1600 mhz 4gb kit. Then I won't fill all 4 slots.

[DerekT]

Thanks for understanding. There is a line where advice can lead to damage of components and I will not cross that line for others although I might well for myself.

 

I am not certain if it is unstable or if it is the CPU <--> Memory Controller<--> Stream

 

Let me explain this. Your CPU is a 266Mhz FSB Processor. The Memory controller is set to Intel SPec for 266/333/400Mhz. So 267 - 332, 334 - 399 MHz is out of Intel SPec. The CPU when run in SPec can handle an overclock stream in SPec (CPU <--> Memory Controller <--> DRAM) better than it can when out of spec. So, your CPU overclock of 333Mhz should more easily give 1333Mhz for your memory and a CPU overclock of 400Mhz should give more easily 1600Mhz.

 

Think of the overclock as a stream of the three components rather than separate components. Intel has predefined specific latencies at specific NB FSB speeds. They are referred to as straps. There is a strap for when the NB FSB is 1066Mhz and under, 1333Mhz FSB and under, 1600Mhz FSB and under, ect. When you go from the 1066Mhz FSB strap to the 1333Mhz FSB strap, the north bridge's internal latencies loosen to allow for greater stability.

 

So, set low to find out if it is stable. Set your CPU FSB to 333Mhz and your multiplier to 6. Set your memory at 1333Mhz with 9-9-9-24 2T with 1.7v. Memtest both single channel and dual channel. If you are memtest stable then the memory produces correctly. Then unlink the memory, set the FSB to 400 and the memory to 1333Mhz and retest both singly and in dual channel.

 

If you get a failure, then the stream is the issue, not the memory or the CPU. The memory controller is issuing problems with stability of the stream.

 

See?

Download Memtest86+ V2.11 from--->

Here

and extract the ISO image. Burn the ISO image to an CD-ROM disk. Enter your BIOS. Load Setup Defaults. Save Setup Defaults. Set to these values:

Extreme Tweaker 

CPU Level Up = Auto
Memory Level Up = Auto
CPU Multiplier = 6
FSB - Memory Clock Mode = UnLinked
FSB (QDR) Frequency = 400 (1600)
Mem (DDR) Frequency = 667 (1333)
LDT Frequency = 5x

Memory Timing Section 

P1 = Auto
P2 = Auto
tCL = 9
tRCD = 9
tRP = 9
tRAS = 24
Command Per Clock = 2T
tRRD = Auto
tRC = Auto
tWR = Auto
tWTR = Auto
tFAW = Auto
tRD = Auto
tRTP = Auto
tRFC = 80
tREF = Auto

Over Voltage

CPU Voltage = Auto
Loadline Calibration = Disabled
CPU PLL Voltage = Autov 
CPU VTT Voltage = Autov
Memory Voltage + 1.7v
NB Core Voltage = 1.32v - 1.45v
SB Core Voltage = 1.50v
CPU GTL_REF0 Ratio = Auto
CPU GTL_REF1 Ratio = Auto
CPU GTL_REF2 Ratio = Auto
CPU GTL_REF3 Ratio = Auto
DDR3 CHA1 Ref Voltage = Auto
DDR3 CHA2 Ref Voltage = Auto
DDR3 CHB1 Ref Voltage = Auto
DDR3 CHB2 Ref Voltage = Auto

CPU Configuration

CPU Internal Thermal Control = Auto
Limit CPUID MaxVal = Disabled
Enhanced C1 (C1E) = Disabled
Execute Disable Bit = Enabled
Virtualization Technology = Disabled (Unless you are using Virtual Machines)
Enhanced SpeedStep Tech = Disabled

All CPU Cores = Enabled

Spread Spectrum Control = All to disabled

 

Boot to the Memtest Version 2.11 CD and allow for two full passes. If stable, enter Windows run CPU-z. Post a screenshot of CPU-z's CPU, Memory and SPD tabs.

 

 

OK, thanks mate. I will run them 1066 mhz for now.

 

Provided my board is stable, would this kit work ok? TW3X4G1600C9DHX?

 

I have been told that the Striker Ii will only be stable with RAM ratio of 1:1, 1:2, 3:2. This seems true as 1333mhz (with the only 2 sticks in) is unstable.

 

So, since I'm sure y've realised I am new to overclocking, I will buy a rated 1600 mhz 4gb kit. Then I won't fill all 4 slots.

[cje]

Alright, I finally understand what you are saying. Thanks for your patience. Will try tonight, not home till 6pm GMT, so will test ou tonight.

 

BTW, how do i test single channel and dual channel in memtest?

[DerekT]

Single Channel is one module tested at a time. Dual Channel is two modules tested in your first and third slots.

 

Alright, I finally understand what you are saying. Thanks for your patience. Will try tonight, not home till 6pm GMT, so will test ou tonight.

 

BTW, how do i test single channel and dual channel in memtest?

[cje]

Tested at 1333 mhz fsb and ram. passed 4 in memtest in both single and dual channel mode. Also passed 3 with 1600 fsb and 1333 ram with the settings you posted. here is the cpu-z screenshot:

 

http://img16.imageshack.us/img16/2355/1600fsb.th.jpg

[DerekT]

Well done! Now test in Windows with Linpack.

 

Download LinX - Linpack clone.

 

LinX Version 0.5.7

 

Allow for a twenty iteration pass and set the memory to "All". Then choose start.

 

Results?

 

Tested at 1333 mhz fsb and ram. passed 4 in memtest in both single and dual channel mode. Also passed 3 with 1600 fsb and 1333 ram with the settings you posted. here is the cpu-z screenshot:

 

http://img16.imageshack.us/img16/2355/1600fsb.th.jpg

[cje]

Results are... PASS! So I guess that cpu is causing instability?

 

I also went ahead of myself and tested 1600 fsb and RAM linked and synced, but with multiplier to 6x. And passed 5 memtest, and 20 linpack.

 

So, if I'm not mistaken, Loadline Calibration seems to be the common factor that causes instability, I have read through the thread in Xtremesystems, and Loadline Calibration is known to cause idle voltage to drop.

 

Am I correct in thinking this? Or have I got the wrong end of the stick?

 

OR, seeming that my settings haven;t changed, bar loadline, is the fact that before testing, I loaded defaults, saved defaults, POST again, then enter BIOS and enter settings directly.

 

I noticed that when loading and saving defaults, the LEDs on the board cut out for a split second, if I just change a voltage value or something minor, the light are continually on. Don't know if this is an issue.

[DerekT]

Loadline Calibration allows to create less of a voltage droop of the Core. Here's the thing. You need to learn this stuff on your own. I told you about the multiplier. Raise it to seven and retest. etc. etc.

 

Google loadline calibration.

 

http://www.google.ca/search?pz=1&ned=ca&hl=en&q=loadline+calibration&btnmeta%3Dsearch%3Dsearch=Search+the+Web

 

52 thousand hits.

 

That site will give you all the information you need. It's time now for you to start learning through research.

[cje]

Ok, no matter where I search, people say set your command rate to 2T with this board.

 

I currently have 2x 1gb sticks, and it freezes in vista with 2T. With 1T however, I've been playing games and messing about and linpack testing 50 passes with no problems. Not a single freeze, blue screen or anything.

 

But, sometimes freezes on "USB INIT" on POST. Found out that it was my RAM causing this. Set back to 1066 and no POST freezes. Guess I need more VDIMM. I also read that the striker II is notorius for drooping VDIMM under load. So need to experiment a bit more.

[RAM GUY]

I have heard that before with this MB and some CPU's, I would suspect the BIOS may need to mature a bit more but for sure if the system is stable at 1T I would run the memory controller at 1T Command Rate it will not hurt the system.

[DerekT]

And this is why you need to play with your system to find what works. There are always exceptions to the rule. The best rules of thumb do not work consistently with computer systems and I can not count the times when my advice on one system works poorly but which works well with many of others of the same system hardware. That's why it is a "Your Results May Vary" thing.

 

I'm happy to see that you are going to experiment. :):

 

Ok, no matter where I search, people say set your command rate to 2T with this board.

 

So need to experiment a bit more.

[cje]

Yes, was experimenting now. I've set multiplier to 6x (2.4ghz) To rule out cpu. RAM is at 1066 mhz. As you said, and other people with this board say the only stable ratio's they had was 1:1, 1:2 and 3:2. 1066 is 3:2.

 

2T resulted in a bsod but no restart after 7 minutes. Now testing 1T. I had gone into the habit of changing more than one BIOS option. Which I know is stupid as I can't nail down what's causing these bsods.

 

Will try experimenting with NB GTL. Some people say leave on auto, some say +60mv with 1600 ram and fsb. We'll see what my board likes.

[DerekT]

Here's the method to be certain that the memory controller is or isn't the issue with regards to the blue screens.

 

Set up and testing proves stability. In other words, LinX 20 iteration and Memtest proves stable but the blue screens still occur.

 

Download LinX - Set to All RAM and 20 iterations

 

Download LinX from-->

Here

Remove two sticks and test for repeats of the blue screen. If it does not appear, then you have to tweak the memory controller, but be clear, this does not mean that you will find stabilty with the memory controller (as that often depends on the controllers quality), just that you have isolated the issue. You can lower the performance (Latencies and Skews) of the memory controller but you may find that tightening the controllers latencies and dropping the bandwidth gives you better results.

 

Download Everest Trial Version from here -->

Here

Install Everest and right click on the (i) icon in the system tray. Choose Tools --> Cache and Memory Benchmark. Run the Benchmarks for testing your results.

 

 

Yes, was experimenting now. I've set multiplier to 6x (2.4ghz) To rule out cpu. RAM is at 1066 mhz. As you said, and other people with this board say the only stable ratio's they had was 1:1, 1:2 and 3:2. 1066 is 3:2.

 

2T resulted in a bsod but no restart after 7 minutes. Now testing 1T. I had gone into the habit of changing more than one BIOS option. Which I know is stupid as I can't nail down what's causing these bsods.

 

Will try experimenting with NB GTL. Some people say leave on auto, some say +60mv with 1600 ram and fsb. We'll see what my board likes.

[cje]

I was experimenting as I said, and it bsod again. So, being tired and a little annoyed, I just went into BIOS loaded setup defaults and started windows.

 

BUT, it froze on the windows logo. So I switched it off, cleared CMOS and all is fine. No BSODS with defaults, hopefully.

 

Since this happened, I am going to run it on stock like this for a while now. Just to make sure that my OS isn't corrupt or something.

[DerekT]

Good idea. I still think your issue is the memory controller and a four DRAM slot population. The reactions are such at any rate.

 

I was experimenting as I said, and it bsod again. So, being tired and a little annoyed, I just went into BIOS loaded setup defaults and started windows.

 

BUT, it froze on the windows logo. So I switched it off, cleared CMOS and all is fine. No BSODS with defaults, hopefully.

 

Since this happened, I am going to run it on stock like this for a while now. Just to make sure that my OS isn't corrupt or something.