DDR2 and T1....

[Pantlegz]

So, this may be a stupid question but I can't for the life of me figure it out... So I figured I'd toss it out there for input.

I have 1gb of OCZ DDR2-6400 that according to OCZ should have timings of 5-5-5-10, I am assuming T1. but my crappy MoBo(EpoX) defaulted it to 5-5-5-15(T2) and I can't for whatever reason drop it down to T1 and get the system to boot into windows (Vista x64 Ultimate if it matters) I have tried kicking the voltage all the way up to 2.5v which is where my mobo maxes out and I can get it to start to boot but then I have to spend 30 min trying to boot into windows becase it really doesn't like it. I have also tried to reduce the speed of the ram to 533(I think) with the FSB at 200Mhz and still nothing. Does anyone have any suggestions??

Also, if I could get it to boot at say 6-6-6-10 T1 would it be worth the CAS increase for T1 timings or would I be hurting my performance that way? I'm not sure it would work, just thought of it as I was tpying.


I can run the RAM without issue at default speed, 800 Mhz, with the FSB jacked up to 230Mhz - which is all the further I've moved it so far and had no issues with Prime95 running for 12-15 hours, and I keep moving(10Mhz incrimints) it up untill under load I get a temp of over 40c or Prime95 catches an error.

 

[Yellowbeard]

You are MUCH better off running your system at 2T with the CPU running at stock or OCed. Don't slow the CPU just to get some nearly worthless chipset setting. Command rate has a very neglible effect on actual performance. 1T is pretty for benchmarks but that's about it.

 

[lopri]

What is your platform? Knowing that you have an 1T option in the BIOS, it's probably nForce board? 1T isn't always guaranteed. It depends on chipsets, boards, BIOS, and memory. It is also possible that different BIOS revisions will work differently and BIOS programmers can't test every memory sticks out in the market. Like Yellowbeard says, most of the times the difference 1T makes is negligible compared to CPU frequencies. That said, 1T does make a small difference and it's great if it's attainable without too much hassle (voltages, stability, etc.). But again it's not worth chasing after with the cost of system stability and lifespan, IMO.

 

[lopri]

And no, I wouldn't run DDR2 with CL6. As a matter of fact I don't even like CL5 and you'll be better off with high frequency/CL4/2T, than low frequency/CL6/1T

 

[n7]

Originally posted by: lopri
And no, I wouldn't run DDR2 with CL6. As a matter of fact I don't even like CL5 and you'll be better off with high frequency/CL4/2T, than low frequency/CL6/1T

Have you benched that or is that just speculation?

I mean, for the OP'd case: CAS 5 2T vs. CAS 6 1T.

I have this feeling it might be rather close, if not better at 1T.


OP, i'd try it & run some benches for us

Be aware you may screw up your OS install by trying to run w/ unstable RAM, if you haven't already.

 

[PCTC2]

Yeah. I ran my memory, just as a comparison (Corsair XMS2 Dominator PC2-6400C4D) at 667 1T and 2T in SiSoft Sandra XIb (Int/Float)

3-3-3-3-2T: 6967/6995
4-4-4-12-1T: 6987/7005

So, CL3 2T is about the same as CL4 1T. I'm guess it will be the same for CL5 2T vs. CL6 1T.

EDIT: I did not run it at 800 b/c it lacks stability. For both tests, my E6400 was at 400*8 (3.2). Subtimings were tightened for the CL3 Run.

 

[Pantlegz]

Thanks for all the advice. I guess I'll update my BIOS and hope that might help something, otherwise I'll try to drop the CAS down a lil. My current setup, not that it matters anymore, is a EPoX MF4 Ultra - Nforce4 - AM2 3800+ - OCZ Gold something memory...

PCTC2- Yea, I forgot to drop the FSB down when I was playing with the CAS and had to reinstall Vista, lucklyI have 3 HHD's and all the stuff I had I could easily drag over... for the most part.

If the BIOS doesn't help, would running dual channel help anything? or not really, I kinda figured it wouldn't but I would much rather get T1 then mess with CAS timing anymore. It doesn't seem to like me much

 

[Acanthus]

I wouldnt get too many headaches over this, youre talking a 1% performance window or less in everything but synthetic memory specific benchmarks.

 

[Pantlegz]

yea, but my mistake slowed my computer down a ton... I guess single channel ftl I moved the ram around and got it to dual and it runs so much faster and got the T1 timings w/o dropping anything but the FSB which I'll slowly bump as it passes Prime95

thanks for all the help