Recommend RAM for NF7-S rev 2.0

[serlv]

I've been using TwinMOS PC3200 in my rev 1.2 boards, but have heard that the TwinMOS PC3700 may not be all that good. Not sure if that info is correct, but until I hear more I'll lay off.

What RAM, rated PC3500 or better, do folks suggest for the rev 2.0 Abit. I'd like to run at at least 225 MHz.

TIA
Scott

[smokinjryan]

I would either go twinmos 3200 or twinmos 3700

http://search.store.yahoo.com/cgi-bi...w.memoryx.net/

[deathstar13]

ive been seeing the twinmos 3700 will only run above 200fsb at cas 2.5 and thats where alot of people are having rough times forcing cas2

i use xms3500 and can do 228 stable 2.0-3-3-7 2.9v
also 225 stable 2.0-2-2-5 2.9v

i have a stick of kingston hyperX 3500 coming this week to add and ill test it fully then also.

[Sh@rky]

I would recommend OCZ PC3500 EL - my 512MB-module reached 236MHz@2.0-2-2-6@2.8V (ASUS P4PE, FSB177, FSB:RAM=3:4)

[serlv]

Guess I'll do a little experimenting. I have the OCZ EL 3500, 3200 XMS and some TwinMOS 3200.

Having a little bit of a tough time dialing an ABit rev 1.2 in with the OCZ and a Tbred 1700 DUT3C, though. The RAM seems very touchy.

[Calcu]

With my TwinMOS PC3700 and a 8RGA+, the best I could do even at 200FSB, was 5,3,2,2.5 and the highest I could go was 225FSB @ 7,3,3,2.5. Now on my NF7-S, I get 233FSB @ 3,3,2,2 @ 2.92v(set to 2.8v in BIOS). So I actually think that the 8RGA(v1.0 A3 NB) has some serious problems that need to be worked out still,

Calcu

[serlv]

Thanks for your input. As far as NF7-S rev 2.0 owners go, that is good news indeed. You are talking 'bout the 2.0, or the 1.2?

For this system, I've already gone w another two sticks of the TM/WB 3200. But I'm sure I'll be getting RAM for another system soon...

What DRAM is on that 3700? Ya know?

edit - I see from you're sig, you're on the 2.0

[Liquid3D]

Originally posted by Calcu
...So I actually think that the 8RGA (v1.0 A3 NB) has some serious problems that need to be worked out still...Calcu

I would definately concur with that theory. For a NB that was sold on it's ability to reach 400MHz memory speeds (notwithstanding the memory multiplier, and that is but one of it's attributes) it certainly struggled to do so. Many believed the NB was merely undervolted, but were now realizing the problems run deeper. They (mobo manufacturers) obviously expected the A3 stepping NB to perform well under default voltages, and probably knew allowing end-users to have access to more voltage would exponentiate the already troublesome thermal issues, and increase the failure rate hence RMA costs.

My criticisms of the nForce2 A3 NB are not meant to accuse nVidia of a mephistophelean plot. Just as Via's KT266 chipset underwent it's growing pains, and turned out being one of the best Socket-A chipsets (KT333/KT400), so the nForce2 should be allotted the same opportunity for growth. Of course as enthusiasts we push most devices to their limits, where the average user would have no problem with the nForce2 chipset, as it performs flawlessly at 133/166MHz. I have nothing but praise for nVidia's Dual DDR (128-bit memory bus), not to mention it's DASP pre-processor, which I feel it's potentiality was completely overlooked/unappreciated. I don't know if it is the "StreamThru" technology, "DASP," and/or "Twinbank" memory architecture which conspire to offer such stunning visual quality. I do know when I took my Tyan Tachyon G9700 Pro, from my KT400 based 8K9A (which was an excellent overclocker, just see my website) and dropped it in the nForce2 based 8RDA, the difference was truly dramatic. We usually only experience small incremental changes if one is a frequent "upgrader" like myself, but this difference was truly tangible. My 3DMark2001SE scores jumped an immediate 3,4000 points between my 8K9A and 8RDA, the only variable being the board themselves (graphics Tyan G9700 Pro, memory, CPU and PSU remained unchanged). My 3DMark2003 shot up by over 1500 points as well. The only dissapoinment I've experienced between board's is in memory clock-speed. While the bandwidth difference is approximately 30% in favor of nForce2, the clock speed lags. Even using the memory multiplier (which results in a no post) there's no comparison, my KT400 overclocked OCZ PC3200 to 441MHz (220FSB) with only 2.63VDIMM. While I can't attain 205FSB with the nForce2 no matter the timings, or voltage. Even reverting to 64-bit bus made no difference (except to negate bandwidth).

The gist of this (semi-short story) being; my next board will most likely be the Abit NF7 rev.2, and as an advocate for OCZ Technology's leadership in the RAM market (they've shrunk the die process again a few months ago, which I believe will be implemented in their quadband memory) I do know Bhuilider Sethi (VP Marketing an an excellent designer in his own right) stated the Abit board is the only board to support their new PC4000, (500MHz utilizing in-house OCZ 4ns IC's) ergo, I believe it's going to be a wonderful match for their "newer" memory. OCZ's in house IC's being PC3700 series and above (to my knowledge). I've been using OCZ PC3200, PC3500, and PC3700 for over a year through all their incarnations. I'm steadfast in my conviction OCZ is the best overclocking memory out there (consistently). Sure there are intermittent IC's w/exceptional stepping's from Elixer, Hynix, Micron, Nanya, Samsung, and Winbond and of course manufacturers such as A-Data, Corsair, Mushkin, Transcend and Twinmos bid on these stepping's, but their short lived. The only company I've seen come close to OCZ Technology's consistency is Corsair XMS (which followed OCZ's EL, and their Twin-X followed OCZ's Dual Optimized).