Can I believe these DDr2 timings?

[Idontcare]

OK so I am playing around with ram timings down in the low speed range, 533MHz stuff, but looking at what timings my ram will support.

Ram = Mushkin mushkin REDLINE (2 x 1GB) DDR2 1000 (PC2 8000) (set Vdimm to 2.3V)

Mobo = ASUS Striker Extreme

BIOS = 0402 (as delivered)

I've got a QX6700 loaded, stock FSB for this discussion. 1067MHz FSB. Ram is set to synch, so 2x266 = 533MHz.

Here are the timing I can set in the BIOS which in turn have proven memtest86+ stable as well as windows/orthos stable:

• tCL = 2
  tRCD = 2
  tRP = 2
  tRAS = 2
  tRC = 1
  tRFC = 11
  tRRD = 1
  tWR = 1
  tWTR = 4
  tREF = 1
  tRD = 1
  Async Latency = 1.0 ns

Anything lower in any timing parameter results in either no-post or failed memtest/windows/orthos to some degree.

I just can't beleive these sub-timings are real, even at 533MHz speeds. For example, tRC is usually a double-digit value, as is tRAS usually. But to have a tRC of only 1 and tRFC a mere 12!?

Anyone else have similiar experience or can give me a reality check on this? Is this Striker BIOS blowing smoke up my tail or is it actually possible/typical/common for such tight sub-timings at 533MHz speeds?

Edit: added tREF and tRD

 

[n7]

LOL CAS 2 DDR2-533?

I think not. :laugh:

Your settings are either completely f*cked, or maybe your RAM is actually running @ DDR2-266 (133) speeds if that's even possible?

Why don't you try Sandra or Everest to show timings in Windows?

Wait, doesn't nTune show the timings?

 

[myocardia]

Why not just download CPU-Z, and see what timings it says your RAM is running? If CPU-Z says it's running at those timings, I'd personally believe it. Oh, it will also tell you the speed that it's running.

 

[n7]

Ah i see CPU-Z has added support for nF 680i in the latest version.

If i'd known that, my first post would read a little differently.

 

[myocardia]

Originally posted by: n7
Ah i see CPU-Z has added support for nF 680i in the latest version.

If i'd known that, my first post would read a little differently.

The never-ending story of trying to keep your software updated. I've been doing it since dirt was discovered (it seems), and I always thought it was like trying to keep a funnel filled up.:laugh:

 

[Idontcare]

OK, I grabbed a screenshot, let me know if this provides the data you guys need:

http://img400.imageshack.us/my.php?image=533222221ttm3.jpg

CPU-z says Freq = 266.7MHz, and DDR2 runs at 2X the CPU-z frequency, so that should be DDR2-533 (which the mobo reports as well during the post).

These timings just seem ridiculously tight, even for DDR2-533. If I can't trust the timings reported by the board then I can't really overclock with confidence either, which is my current concern.

 

[Avalon]

I almost tend to think those are your actual timings :Q
Looks like you're going to have to do some sandra comparisons between "normal" timings and yours to see if there is a difference, which ought to give you a definite answer.

 

[The-Noid]

I can get to around 580 mhz at CAS 2. That is definetly true timings . I showed some results with them on Superpi. 3-3-3-8-1T @ 820-840 or 4-4-4-8-1T @ 1000 mhz is still faster.

It wont boot CAS 1 regardless of timings though, I really wanted to see how that would do.

 

[Idontcare]

Thanks Yoxxy for the reality check, at least I can proceed knowing my striker BIOS timings are plausible.

My goal was to definitely get a decent FSB speed to maximize the bandwidth. This asynch mode, at least on striker, is pretty much useless from what I can tell.

Whether I clock my ram as 1:1 with FSB (DDR2-1067) or down to synch-mode with FSB (DDR2-533) I get nearly the exact same bandwidth per Sandra.

Latency increases, per Sandra, as the DDR2 speeds decrease, unless I correspondingly tighten the timings, so it does make some difference.

 

[n7]

Sweet jesus CAS 2 is for real :Q

Insane...

 

[twjr]

Originally posted by: n7
Sweet jesus CAS 2 is for real :Q

Insane...

I didn't think it was possible................... :Q:shocked:

 

[Idontcare]

What blows me away is that not only is the CAS a mere 2, but all the sub-timings are so freaken tight as well.

Too bad my quad-core doesn't have on-die MCH to take serious advantage of such tight timings though.