"Callin' n7" Please post your 4-up Mushkin 4x2GB setup on Asus P5Q-Dlx

[GundamF91]

n7, you posted a list of parameters on the 4x2GB Mushkin setup with Asus P5Q-Dlx. Now that I have the same setup, I can't find your post. This should save me a lot of initial trial and error. Thanks!

 

[Blazer7]

I believe that this is what you've been looking for : http://forums.anandtech.com/me...AR_FORUMVIEWTMP=Linear

 

[n7]

These are my present 24/7 settings, but you need to be aware they may not all work for you.

First thing, don't use such high vcore; you have a 45nm chip...i'd try to stay at 1.35v or lower to be safer.

Also, i don't suggest using manual subtimings.
Leave the values in 1st, 2nd, & 3rd @ auto, excepting the first four; set those to 5-5-5-15 (or whatever your RAM is rated for).
Do set Row Refresh Cycle Time to 55 or higher.
I'd say leave Common Performance Level @ auto as well at first.

Basically, don't tweak the RAM till you figure out your CPU's max.

BTW which Mushkin kit do you have?
I'd say try running lower speeds initially while you work with the CPU...worry about RAM speed/timings afterwards.





AI Overclock tuner: Manual
CPU Ratio Setting: 8
FSB Strap to North Bridge: 333
FSB Frequency: 442
PCI-E Frequency: 101
DRAM Frequency: 1061
DRAM CLK Skew on Channel A1: AUTO
DRAM CLK Skew on Channel A2: AUTO
DRAM CLK Skew on Channel B1: AUTO
DRAM CLK Skew on Channel B2: AUTO
DRAM Timing Control: Manual

1st Information:
CAS# Latency: 5
DRAM RAS# to CAS# Delay: 5
DRAM RAS# Precharge: 5
DRAM RAS# Activate to Precharge: 15
RAS# to RAS# Delay : 3
Row Refresh Cycle Time: 55
Write Recovery Time: 6
Read to Precharge Time: 3

2nd Information :
READ to WRITE Delay (S/D): 9
Write to Read Delay (S): 4
WRITE to READ Delay (D): 6
READ to READ Delay (S): 5
READ to READ Delay (D): 7
WRITE to WRITE Delay (S): 5
WRITE to WRITE Delay (D): 7

3rd Information :
WRITE to PRE Delay: 15
READ to PRE Delay: 6
PRE to PRE Delay: 2
ALL PRE to ACT Delay: 7
ALL PRE to REF Delay: 7

DRAM Static Read Control: Disabled
DRAM Read Training: Disabled
MEM. OC Charger: Enabled
AI Clock Twister: Moderate
AI Transaction Booster: Manual
Common Performance Level: 08
All Pull-Ins: Enabled

CPU Voltage: 1.4
CPU GTL Voltage Reference (0/2): AUTO
CPU GTL Voltage Reference (1/3): AUTO
CPU PLL Voltage: 1.52
FSB Termination Voltage: 1.22
DRAM Voltage: 1.96
NB Voltage: 1.24
NB GTL Reference: AUTO
SBridge Voltage: 1.1
PCIE SATA Voltage: 1.5

Load Line Calibration: Enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew : AUTO
NB Clock Skew : AUTO

 

[GundamF91]

Thanks for the info. I have the Mushkin kit that's 5-4-4-15 @ 800mhz @ 1.8v. I plan on running at most 425mhz x 8multi = 3.4Ghz. I think I'll just run FSB to memory at 1:1, so the Mushkins just need to operate at 850mhz.

There's one confusion I still have. "FSB Strap to North Bridge: 333"

Why is FSB Strap to NB set at 333, even if I run 400mhz FSB and 400mhz memory? Or should I leave it to auto?

 

[n7]

Just put FSB strap to auto, & choose the same speed RAM as FSB.

Lower strap usually = better latency though, if you were benching for that sorta thing.

 

[GundamF91]

I never had an Asus board, so I wasn't sure about the "FSB Strap to NB" thing. On Abit IP35 Pro that I had, it just has ratios that I get to pick. So by "lower strap", do you mean 266? or do you mean 400? I thought higher would be better latency.

 

[n7]

No.

Try POSTing using the 200 strap...likely won't even be able to.

The difference is usually made up for with the higher speed, etc. but lower NB strap is generally better latency, at least in theory.

Obviously, you can usually more than make up for that with the higher speeds you'll get with the higher straps, so it's not such a big deal.

In my case, 8x442 5:6 (333 strap) PL 8 + pullin is actually faster overall in Everest than 10x350 2:3 (266 strap) PL 6 + pullin.

Anyway, there are a ton of factors involved...you could try testing for yourself; always best way to determine what settings to run.

Keep in mind tRD (aka Performance Level - PL) has a big effect on things for RAM benching, so once you get your overclock all figured out, work on tightening that.
Pull-ins enabled tend to equate to roughly the same as one lower timings, but will often work where one lower doesn't.

E.g., i cannot POST @ tRD 7 with my FSB @ 442 as it is currently.
But i can do tRD 8 + all pull-ins enabled stably, which brings scores up a bit.

AI Clock Twister also affects numbers.
Lighter/Light are easy to do, & good to start out with.
I am on Moderate right now though; boosts scores a bit.
Any higher & again, no POST...so these things can be picky.

Especially with 8 GB.

I'd suggest posting on the Mushkin forums once you figure out your CPU, & Greg will set you up with guidelines for all the timings/subtimings if you want to play with them.

 

[n7]

Get in the habit of using Memtest86+ (especially test 5) for basic testing.

And HCI Memtest is a must for testing 8 GB...nothing else actually stresses it as much, aside from LinPack.

P95 is good for CPU obviously, but pretty much useless for 8 GB...it only will stress a tiny amount of it.

 

[GundamF91]

Thanks, I have it stablized at 1.225v vCore, and 1.8v memory. Running at 400x8=3.2Ghz, and 1:1 memory ratio. Everything worked out nicely. I had one problem with booting when I had left everything at AUTO (just to see what happens), so I changed timing to manual. Row Refresh Cycle Time to 55 in particular is probably where it helped when using 4 banks of memory. I still haven't pushed for better timing yet. Right now I need a stable rig to finish my class in a couple of weeks, then I have more time to push it to the edge.

As for the "FSB Strap to NB", I have it set at 333. I'm still confused why it's 333. When I have FSB at 400, and memory at 400. What makes the strap 333? Wouldn't this make FSB strap to NB at 400 also? Or is this value independent of what FSB and memory settings are?