Enough RAM testing to warrant RMA?

[wraith84j]

For quite a while now, I've been dealing with random in-game and out-of-game crashing (BF2, Firefox, plain Windows XP). Due to my lack of dedicated time, it's gone on for several months, but during that time, I've been slowly trying to narrow down the cause of the problem. For the record, I have had the video card and motherboard replaced (with the same model) to no avail.

 

Running stock settings (all G.Skill's RAM timings on their website - except for RAM frequency, which is 200 MHz), I've run Memtest86+ for about 160 hours. During that time, it completed 181 passes and produced 19 errors at 3 distinct memory locations for tests 2 & 7. Seeing this, I find myself asking the question: Does RAM tend to produce different 'patterns' of errors depending on whether it's bad or just improperly configured in BIOS? Will improperly tuned RAM tend to produce large blocks of errors? On the other side of the coin, are sparse, disjoint memory errors an indication of bad RAM?

 

Without knowing the answers to those questions, I've tried to move forward and retest the RAM with more relaxed settings in hopes of removing the errors. I've gone through some memory related threads here trying to learn some more about all of the different ram settings to facilitate this. I can provide additional information about RAM settings, but loosening parameters such as CPC, Tras, Trc, DRAM Drive Strength, DRAM Data Drive Strength, and R/W Queue Bypass in the direction of stability have not removed the errors (In these later tests I focused Memtest86+ on the suspected tests 2 & 7 for small memory ranges around the 3 memory locations - actually producing a larger error/pass ratio).

 

I'm trying not to initiate another RMA needlessly, so I would like to get some input to determine whether or not I've tested my RAM properly/enough, if the current errors are an indicator of bad RAM, and if not, what are some recommended courses of action for further testing? Please don't hesitate to ask for more information.

 

Thanks!

[CrAsHnBuRnXp]

Usually only 12-24 hours of testing in Memtest is required. If you produce any sort of errors in that time, you can be sure you have bad ram. Since you have tested for ~160 hours and have produced 19 errors, I say RMA it or buy new RAM. Its bad.

 

I recently had RAM go bad on me. (Tested for a bit over 2 hours and soon got over 25k errors in about 2 minutes) It was my second RMA on the same RAM. Called up the company and they are upgrading me to the next line of their RAM for free. Im still awaiting its arrival. So if this is your second RMA, make a fuss if you have to and tell them you either want a refund or that you want to be upgraded to something better.

[MalibuSS]

Usually only 12-24 hours of testing in Memtest is required. If you produce any sort of errors in that time, you can be sure you have bad ram. Since you have tested for ~160 hours and have produced 19 errors, I say RMA it or buy new RAM. Its bad.

 

I recently had RAM go bad on me. (Tested for a bit over 2 hours and soon got over 25k errors in about 2 minutes) It was my second RMA on the same RAM. Called up the company and they are upgrading me to the next line of their RAM for free. Im still awaiting its arrival. So if this is your second RMA, make a fuss if you have to and tell them you either want a refund or that you want to be upgraded to something better.

 

It does not necessarily mean that the ram is bad...It means one of two things:

 

Your secondary timings are wrong.

You have bad ram.

 

Good deal on your second RMA though.

[loggan26]

Have you tried settings like this for your ram

 

 

FSB Bus Frequency.............................  -  241
LDT/FSB Frequency Ratio.......................  -  x3.0
CPU/FSB Frequency Ratio.......................  -  11.0
PCI eXpress Frequency.........................  -  100Mhz

CPU VID StartUp Value.........................  -  1.475v

CPU VID Control...............................  -  1.325v
CPU VID Special Control.......................  -  Above VID *113%
LDT Voltage Control...........................  -  1.30v 
Chip Set Voltage Control......................  -  1.50v
DRAM Voltage Control..........................  -  2.60v

DRAM Configuration Settings:

DRAM Frequency Set............................  -  200=RAM/FSB:01/01
Command Per Clock (CPC).......................  -  Enable 
CAS Latency Control (Tcl).....................  -  03 Bus Clocks
RAS# to CAS# delay (Trcd).....................  -  04 Bus Clocks
Min RAS# active time (Tras)...................  -  08 Bus Clocks
Row precharge time (Trp)......................  -  04 Bus Clocks
Row Cycle time (Trc)..........................  -  07 Bus Clocks
Row refresh cyc time (Trfc)...................  -  14 Bus Clocks
Row to Row delay (Trrd).......................  -  02 Bus Clocks
Write recovery time (Twr).....................  -  02 Bus Clocks
Write to Read delay (Twtr)....................  -  02 Bus Clocks
Read to Write delay (Trwt)....................  -  04 Bus Clocks
Refresh Period (Tref).........................  -  3120 Cycles
Write CAS Latency (Twcl)......................  -  Auto
DRAM Bank Interleave..........................  -  Enabled

DQS Skew Control..............................  -  Auto
DQS Skew Value................................  -  0
DRAM Drive Strength...........................  -  Level 5
DRAM Data Drive Strength......................  -  Level 1
Max Async Latency.............................  -  08.0 Nano Seconds
DRAM Response Time............................  -  Normal
Read Preamble Time............................  -  06.0 Nano Seconds
IdleCycle Limit...............................  -  256 Cycles
Dynamic Counter...............................  -  Disable
R/W Queue Bypass..............................  -  16 x
Bypass Max....................................  -  07 x
32 Byte Granularity...........................  -  Disable(4 Bursts)

 

Borrowed from Tasr OCDB listing

 

Logan

[Guest Kobalt]

Have you tried settings like this for your ram

 

 

FSB Bus Frequency............................. - 241

LDT/FSB Frequency Ratio....................... - x3.0

CPU/FSB Frequency Ratio....................... - 11.0

PCI eXpress Frequency......................... - 100Mhz

DRAM Configuration Settings:

 

DRAM Frequency Set............................ - 200=RAM/FSB:01/01

...

Borrowed from Tasr OCDB listing

 

Logan

Wait, if FSB is 241, and your running 1:1 RAM, then RAM is also running that speed, correct? So set your FSB to 200, or set RAM ratio to something lower to get it as close to whatever the rated speed of the RAM is.

[calvintang]

well if it failed ANY memtest test according to spec, it will be RMA'ed, thats corsair's policy though,should be the same.

[wraith84j]

Logan & Kobalt,

 

The gskill ram is rated for 250 Mhz, but I'm only running it at 200 Mhz. I'm not overclocking at all, just trying to get a stable system. I've also taken a look at the stock settings database to get a feel for some of the timings for my ram.

 

However, running timings that are 250Mhz-stable for others (using the same brand/model ram) @ 200Mhz for me 'should' produce something stable for me I would think (correct me if I'm wrong).

 

The settings reposted from Tasr aren't too off from what I've been testing, but I'll give them a go and report back in a bit.

[wevsspot]

wraith,

 

try these settings;

 

[color=blue][b]Genie BIOS Settings:
FSB Bus Frequency - 244
LDT/FSB Frequency Ratio - 3
LDT Bus Transfer - 16/16
CPU/FSB Frequency Ratio - 9
PCI eXpress Frequency - 100Mhz

CPU VID StartUp Value - Startup

CPU VID Control - 1.40v
CPU VID Special Control - Above VID * 104%
LDT Voltage Control - 1.30v 
Chip Set Voltage Control - 1.60v
DRAM Voltage Control - 2.60v
DRAM + .03 - Enable

DRAM Configuration Settings:

DRAM Frequency Set - 200 = RAM/FSB 1:1
Command Per Clock (CPC) - Enable
CAS Latency Control (Tcl) - 3.0
RAS# to CAS# delay (Trcd) - 04 Bus Clocks
Min RAS# active time (Tras) - 08 Bus Clocks
Row precharge time (Trp) - 04 Bus Clocks
Row Cycle time (Trc) - 07 Bus Clocks
Row refresh cyc time (Trfc) - 14 Bus Clocks
Row to Row delay (Trrd) - 02 Bus Clocks
Write recovery time (Twr) - 02 Bus Clocks
Write to Read delay (Twtr) - 02 Bus Clocks
Read to Write delay (Trwt) - 03 Bus Clocks
Refresh Period (Tref) - 3684 Cycles
DRAM Bank Interleave - Enabled

DQS Skew Control - Auto
DQS Skew Value - 0
DRAM Drive Strength - Normal 4
DRAM Data Drive Strength - Level 3
Max Async Latency - 8.0 Nano Seconds
DRAM Response Time - Normal
Read Preamble Time - 5 Nano Seconds
IdleCycle Limit - 256 Cycles
Dynamic Counter - Disable
R/W Queue Bypass - 16 x
Bypass Max - 07 x
32 Byte Granularity - Disable(4 Bursts)[/color][/b]

 

This will put your cpu running at 2.196ghz and your RAM running at DDR 488. These settings should work just fine as a starting point. If you continue to get failures I'd suspect you have a bad kit of RAM that needs RMA'd. These settings are using the 0704-2bta BIOS.

[wraith84j]

Wevsspot,

 

I tried your recommended settings and found two things:

 

1. Testing the entire memory for all tests for ~8 hours has produced 0 errors for 12 passes.

2. Testing a select memory range (encompassing a location that produced errors before) for test #7 for ~1 min has produced ~30 errors for ~30 passes. (roughly 1 to 1).

 

The errors are infrequent enough that when I test the entire memory range, I'd have to keep Memtest86+ running for such a long time to actually see anything pop up (see first post concerning ~160 hour test). Yet when I concentrate the test on that location, errors pop up immediately.

 

I'm beginning to think an RMA is inevitable.

[wevsspot]

That is very interesting. I never focus my Memtest on a specific memory range, alternative testing the entire range during an overnight Memtest of all eight tests. If the overall testing passes for eight or more hours I consider my rig Memtest stable. I'm not really enough of an expert to comment on why specifying a certain memory address or addresses for test would result in failures so I better leave that up to the experts. However, here is a good explanation of what each Memtest test does. Please see the attached link.

 

http://www.memtest86.com/#details