Fast & Flashy 2GB from Corsair ---> TWINX2048-4400PRO OCing report!!!

[PRJ]

Hmm..... did u try the 5:6 divider?
If you ran it with a 5:6 div, you should be able to get the RAMs at 275MHz at 330HTT.

That's what i did, I used divider 5:6 alias 166



PRJ

[pcmoddingmy]

Sigh... 280MHz seems unstable with my setup at 1T. Sometimes would pass Super PI 32M, but it will BSOD if I left the system Priming throughout the night.

IMO, 275MHz at 3-3-2-5 is still the best for these modules...

[Deathspawner]

Anything higher with the 4400PRO's and you would be lucky. I haven't overclocked mine for a few weeks, but I've been happier with the results from the 3500LL PRO tbh.

If you really, really want to get higher then 275MHz, you are going to have to play around a lot with drive strengths and secondary timings, if you have not already been.

[Slammin]

Too bad I spent my refund money elsewhere, because this does look like fun.

[Deathspawner]

Ya, what were you thinkin'

[pcmoddingmy]

Just realised that Super PI 32M doesn't prove total stability. My PC4400PROs completed Super PI 32M at 275MHz 3-3-2-5-1T 2.7V, but will immediatly error out when I fired up Windows Memtest.

I had to get it to 2.75V to run Windows Memtest at 275Mhz 3-3-2-5-1T. It is running for more than 8 hours now, about 6000% Coverage. IMO, 2000% Coverage is already good enough with Windows Memtest.

[Deathspawner]

6000% is overkill.. usually I go to 300% and that's it. It's more brutal than MemTest for DOS, so if there are errors, it will usually find them right away.

For total stability, I'd run Prime95 Small FFT for 8 hours.

[high5]

small FFT isn't very good stability test, that's been discussed several times. stick with large FFT test.

as for Win Memtest, 1000% is IMHO good enough.

[pcmoddingmy]

I see...... so any idea why my PC4400PROs at 3-3-2-5-1T can pass all Super PI 32M, Prime Blend Stress (24 hours) and 3D stable but only to fail Windows Memtest??

If so, does this make Windows Memtest a better memory testing tool than the others??

[Deathspawner]

SuperPI is more CPU dependent than memory, obviously. So even if your memory is unstable, that doesn't mean that it will cancel SuperPI. Sometimes it may, but not always.

Prime Blend should be run twice on a dual core for the best stress. You can manually set each instance to a specific affinity. This is my opinion, I am sure others will vary.

The test that's most important though, is whether or not it's 3D Stable. Try running 3D Mark 01 all the way through, or a quick 10 minute run of HL2 and see if it's stable. If it's not, it may crash entirely, or halt at unexpected times.

I had an overclock just like yours, that even proved MemTest stable, but loading up a game proved otherwise.

[PRJ]

Finally after completing 24hour Prime blend Stress test, I decided to tighten the timings and ended up with 3-3-2-5-1T at 275MHz. It passed Super PI 32M and it was 3D stable...

Here's a screenshot of the kit at 275MHz | 3-3-2-5-1T | 2.7V running SP2004 Blend Test for 1 hour.



Will let it Prime until it reaches 8 to 10 hours...

Edit: Prime just breached the 11th hour now...

Hi again

I just got hold of one more set of these memory, and want to see if its any good with 4 GB. I changed to 2T and used you timings @ DDR526

Can I tighten some of the other timings now that I am running 2T, any idéers of what to try?

Thanks

PRJ

[Deathspawner]

If it's anything like the 3500LL PRO, you shouldn't have to change any timings except the 2T. I have 4GB of 4400PRO here though, so I may try to see how that acts also.

If you wanted to tighten anything, tRCD or tRP may be ok to try with some extra voltage.

[PRJ]

If it's anything like the 3500LL PRO, you shouldn't have to change any timings except the 2T. I have 4GB of 4400PRO here though, so I may try to see how that acts also.

If you wanted to tighten anything, tRCD or tRP may be ok to try with some extra voltage.

Hi

"so I may try to see how that acts also" That would be real nice

This is at 2T with timings from earlier...

http://lp.pcmoddingmy.com/albums/use...r_3Dstable.jpg

PRJ

[pcmoddingmy]

Guys, I'm done testing these 2GB 4400PRO kit and my final result was at 280MHz... impressive?
The below screenshot shows the modules running 280MHz at 3-4-4-8-1T, 2.84V. It was Super PI 32M, Windows Memtest and 3D stable.



You can read about of this and more in my review at THIS LINK. Hope you guys have a good reading coz this is basically the "fastest" 2GB DDR memory kit that glows!

[Deathspawner]

The below screenshot shows the modules running 280MHz at 3-4-4-8-1T, 2.84V. It was Super PI 32M, Windows Memtest and 3D stable.

Was that with 3-3-2 timings or 3-4-4?

[pcmoddingmy]

Was that with 3-3-2 timings or 3-4-4?

280MHz was done with 3-4-4

[PRJ]

Hi again

I am running with 4 Gig and can only read 2.5 Gig in XP.

I read about "MMIO hole in Bios" and tryed to enable this, XP still shows 2.5 Gig.

Should I keep "MMIO hole " enabled?

Still love to see some timings for DDR526 at 2T with 4 GB

Thanks

Greetings PRJ

[celemine1Gig]

AFAIK, it's a XP SP2 problem. SP1 is said to be able to use over 3Gb of RAM. XP X64 and others (I guess Server 2003 X64; Server 2003 ?) should be OK, too. Nice to see a good 4GB combionation BTW.

[PRJ]

Thanks..

yes, its a good 4GB combionation, and it looks real good.

I'm running XP SP2 for now hoping to see Vista soon.

But I still am confused if I should enable this "MMIO hole " or not ?

What can i tighten in memory timings after slowing down to 2T @ DDR526 ?

PRJ

[PRJ]

I am back with 2 x 1 gig, never found optimal 4 gig timings..

I found these settings and I got one pass at 3DMark03 with those timings at 2.87 Vdim, seems like the higher voltage I give them the futher i get. But I only gave them up to 2.87 Volts because the next step is 2.92 ( and I don't dare go this high )

Timings are shown here, but from this site...

http://lp.pcmoddingmy.com/albums/use...r_3Dstable.jpg


DRAM Frequency (Mhz)= 200
Command per clock (CPC)= Enabled
CAS Latency control (TCL) = 3.0
RAS# TO CAS# delay (TRCD)= 3
MinRAS# active time (TRAS)= 5
Row precharge time (TRP)= 2
Row cycle time (TRC)= 7
Row refresh cycle time (TRFC)= 14
Row to row delay (TRRD)= 3
Write recovery time (TWR)= 2
Write to read delay (TWTR)= 1
Read to write delay (TRWT)= 3
Refresh period (TREF)= 3120
Odd Divisor Correct= Disable
DRAM bank enterleave= Enable


DQS skew control= Auto
DQS skew value= 0
DRAM drive strength= Auto
DRAM data drive strength= Level 1 ( reduce 50% )
Max async latency= 08.0 ns
DRAM responce time= Normal
Read preamble time= 05.5 ns
Idlecycle limit= 016 Cycles
Dynamic counter= Disable
R/W Queue bypass= 16 X
Bypass max= 7 X
32 byte granalarity= Disable ( 4 Bursts )


Do I dare give them more than 2.87 Volts ?

What timings could I loosen, to get them stable?

Thanks