TwinX2048-3500LLPRO Kit

Hi, I'm trying to nail down whether its a timing issue with the BIOS27 or if the Abit NF7-S V2.0 motherboard is not capable of running these sticks at CAS2.0

My original problem from the old thread was getting it to run stable. http://www.houseofhelp.com/forums/showthread.php?t=53425

I can't find anyone running this memory and motherboard combo to find out for sure if it can run the default stock 2.0-3-2-6 latencies.

I'm currently running which is the only way I can get windows to not blue screen:

Memory Timings [Expert] = Manual

Row-active delay [6]

RAS-to-CAS delay [3]

Row-precharge delay [2]

CAS Latency Time[2.5]

With DDR SDRAM Voltage 2.8v

If I put it to CAS2.0, blue screen right away more or less.

I have been running Windows MemTest for the past 4 hours now and has only picked up 1 error.

Some people have said nForce2 boards are extremely picky with 1GB sticks. I was wondering if any experts can clarify if its just a board problem or a memory problem.

Somebody had mentioned on nforcershq.com that setting the Row-active delay to 10 or 11 works better for nForce2.

Can you please advise.

Thanks.

As I'm dreading the likes of an RMA from the UK to the States.

The motherboard is the Abit NF7-S version 2.0 with the latest BIOS version 27 thats been installed for the past 2 years.

The CPU is the AMD Athlon XP (Barton) 3200+ 2.2Ghz. (200x11 is the 3200+ speed) and 220x10 to get the XMS 3500 to show up as 438Mhz upon BIOS post.

Power supply should you need to know in the system is Enermax Liberty 620W ELT620AWT ATX2.2 Modular SLI Compliant PSU

BIOS settings below, all other settings are default values from clearing the CMOS. No performance settings enabled as its set for stability.

CPU Operating Speed 3200+

External Clock 200Mhz

Multiplier Factor x11.0

AGP Frequency [66Mhz]

CPU FSB/DRAM Ratio [Auto]

CPU Interface [Disabled]

 

Power Supply Controller - System default

CPU Core Voltage 1.65v

DDR SDRAM Voltage 2.6v

Chipset Voltage 1.6v

AGP Voltage 1.5v

 

Memory Timings [Expert = Manual]

Row-active delay [6]

RAS-to-CAS delay [3]

Row-precharge delay [2]

CAS Latency Time [2.5]

System BIOS Cachable [Disabled]

Video RAM Cachable [Disabled]

AGP Aperture Size [256MB]

FSB Spread Spectrum [Disabled]

AGP Spread Spectrum [Disabled]

CPU Thermal Throttling [50.0%]

Enhance PCI Performance [Disabled]

CPU Disconnect Function [Enabled]

AGP Data Transfer Rate [8x]

AGP Fast Write Capability [Enabled]

I have cleared the CMOS several times while testing, reseated the memory. Tested the XMS 2GB kit 3500LL Pro in banks 3 and 2 and 3 and 1.

It fails memtest at the end of test #5 at CAS2 with 158 errors, with 1 stick and 2 sticks together. Soon as I test the memory at CAS2.5 3-2-6 with 2.6v, 5 complete passes.

Just to clarify I've had the same results at 200x11 and 220x10 and fails at CAS2 with 2.6v, 2.7v and 2.8v but passes at CAS2.5 with 2.6v.

Increasing the voltage does not solve a thing, so I have the system stable at CAS2.5-3-2-6 with stock 2.6v.

I definitely can't see it being faulty sticks as it would error no matter what latency surely...? as I would think another replacement set would show the same results.

I'm quite convinced the nForce2 chipset cannot handle 1GIG sticks at CAS2 but can handle 512 and 256 CAS2 sticks. All previous Corsair memory I've had before (TWINX512-3200LLPT (2x256) and TWINX1024-3200XLPro (2x512)) that has worked at CAS2 on the Abit NF7-S.

One other thing from reading a lot of reviews of the TwinX2048-3500LLPRO Kit, theirs had said v1.2 as mine is v1.3

I had used the memory compatibility configurator to make sure and double check the memory was compatible with this board before purchasing.

And sure enough it is, but I am pretty positive the nForce2 chipset does not like 1GIG sticks at CAS2. Perhaps it's too agressive for it?

Is there anyway Corsair can test a setup like this in their labs to know for sure?

Thanks.

 

It fails memtest at the end of test #5 at CAS2 with 158 errors, with 1 stick at a time in the system and 2 sticks together. Soon as I test the memory at CAS2.5 3-2-6 with 2.6v, 5 complete passes.

Just to clarify I've had the same results at 200x11 and 220x10 and fails at CAS2 with 2.6v, 2.7v and 2.8v but passes at CAS2.5 with 2.6v.

Sorry, miscommunication there. I've corrected the above. I have tested both sticks with 1 stick at a time in all 3 slots and they fail at CAS2, but do pass at CAS2.5. I do not have another motherboard/system to try sadly.

A replacement set will most likely show the same results.

So I am leaving the system at 2.5-3-2-6 with DDR voltage 2.6. Which passes memtest.

As I have been told...

You will 'see' no performance difference between CAS 2 and CAS 2.5 unless you are running some memory bandwidth tester in which case you will see a 1-2% decrease in theoretical performance in CAS 2.5 over CAS 2. In real world application the difference will be less then 1%. As for NForce chipsets, 2G of memory and CAS 2 it's a known issue. If you look closely most vendors will even say somewhere CAS 2 for Intel platforms and CAS 2.5 for AMD platforms.