Quick RAM questions.

[bront]

The first one, which I can't seem to find :?, stock voltage for DDR is 2.7V?

The second one: How much of a performance hit is 2T over 1T? I've tried setting my memory to 1T, but the system fails within 30 minutes in windows (though it passed memtest for a few hours). Am I better off using 2-3-3-6 2T or relaxing the timings to get 1T?

I'm doing my first true foray into Overclocking, which has been slow due to the nature of quality testing. I just want to make sure I get it right.

 

[bront]

that mentiones the normal latencies (2-3-3-6 is what I run at at 200, was stable at 216, 2.5-3-3-7 was stable up to 250, at least so far in my testing), but I didn't see them mention 1T vs 2T.

I'm running Prime95 on the system now, with the system at 8x250 and 1:1 on the memory, with 2.5-3-3-7 timings. It passed Memtest for 3+ hours on stock voltage.

 

[TabrisDarkPeace]

Stock voltage for DDR is: 2.5 to 2.6 V

2 CMD RATE is more a way around having a large number of ranks of memory (not to be confused with banks, or sides, or # of sticks).

Once you go over 4 ranks of installed memory (per AMD CPU Socket, and thus per memory controller) which may be only 2 DIMMs, or it may be 4 DIMMs depending on your RAM, 2 CMD RATE will need to be used for stability regardless of the other timings.

MemTest86 alone is not enough

I suggest combining MemTest86+ (not MemTest86, unless it has a more recent timestamp, which is highly uncommon) with Prime95 (in Torture Test, Blend Mode, or you are just wasting your time with Prime95 aswell).

Funnily enough many 'so called' premium DIMMs are actually 2 ranks per stick, so if you install 2 of them you may need 2 CMD RATE. If people don't understand it and believe the 'bullshot' advertising that becomes their problem. - ie: The sticks 'work' so you won't get a warranty replacement.


Search http://www.amd.com for CMD RATE, memory controller, Opteron, etc (the 1st AMD64 CPUs where Opterons, and much of the documention reflects this) as they invented the memory controller on the Athlon 64 / Opteron processor, and you'll find 100's of pages of whitepapers and techdocs beyond want you'll ever 'need' to know.

 

[bront]

Ok, that explains the 1T vs 2T. I'll just stick to what's stable, and I'm not unhappy with running DDR400 at DDR500 speeds with 2.5-3-3-7 at auto voltage (which seems to come up at 2.75, so maybe my mem asks for more voltage).

I'm running the custom test twice (I have an X2) with 2048 - 4096 with my physical memory split between the two runs, which, as I read, should stress the RAM out. I'll check it out in about 10 hours (it's been running for 3 already) and see if it's clear. SuperPI ran clear for a while on 32M.

 

[TabrisDarkPeace]

You can run Prime95 from two different folders (or more, up to 32 should be OK) at once btw, so it hits both CPUs and memory at once.

It will cut your testing time in half if you have a dual-core system.... if you are not already doing it. Even on single-core with HyperThreading you'll still gain a fair bit of time. On quad-core it helps heaps

Tabris - Recommended Prime95 settings.txt:

Min FFT size (in K): 128
Max FFT size (in K): 4096 (or larger if supported)
Run FFTs in place: Disabled / Off
Memory to use (in MB): 1/n *
Time to run each FFT size (in minutes): As low as 15 and as high as 60 minutes to personal taste. (15 min recommended though, even for todays systems)

* - Where n is the number of cores/threads the system can run in parallel, and how many Prime95 instances are being run, each from their own folder at the same time.

eg: 4 threads at once = 1/4 memory = eg: 4 GB total, so 1024 MB each

Prime95 can detect errors in calculations (usually caused by faulty memory, or overclocking CPU to far), I don't know if SuperPI can though.

 

[bront]

I'm running 2 copies of it simultaniously at 2048 to 4096 per the Wiki entry I found, using enough to max out my physical ram.

 

[bront]

BTW, I found this article to be quite enlightening on the timings issue.

Of course, now I need to figure out if I should OC the memory further by relaxing timings even more. I'm not unhappy with DDR500 2.5-3-3-7 though, and will do some benchmarking once I am asured of it's stability.

 

[bront]

2-3-3-6 is the "default"

Had to set that manualy.

http://www.newegg.com/product/product.asp?item=N82E16820145450

That's what I have.