Kingston PC1066

[rambus]

I'm seriously wonder if all Kingston PC1066 modules were 100% tested as Kingston claimed. I have 2 128M modules with part number KVR1066X16-4/128. Runing at 4*133MHz on a ASUS P4T533-C m/b makes the system very unstable. They caused the system crashes or even reboots randomly.

I checked them with MemTest 86. Both modules failed! They could only pass the tests at PC800 mode, sigh..

I have filed a RMA request but got no response yet. It was a really frustrating memory for me. And I've ordered Samsung PC1066 modules (Intel Certified). I will post the results whenever available

 

[crontab]

I did not know that the 128mg version was also having problems. I currently have an RMA in to replace 2 RIMMS of KVR1066X16-8/256. The pair failed memtest-86 with a total of 67,676 single bit errors and 3 double bit errors in 17 different address locations. In all fairness to Kingston, they were very understanding and helpful with the RMA but waiting 4 to 6 weeks for the replacements to arrive (and yes I am doing a cross ship with a CC) is just a little too long for my taste. When I purchased the Gigabyte GA-8IHXP ver 2.1, I never expected these sorts of problems. The first issue was with OCZ memory that was purchased when I first decided to build this 2.53 P4 system. When the system would not boot and gave 3 beeps, I knew the first 64K could not be allocated or read. Oh great, memory problems! I was really looking forward to trying out my ATI 9700 Pro card with this config but OCZ memory killed that thrill! SO I sent it back and purchased the Kingston (at a PREMIUM PRICE). I figured Kingston would be a the MOST RELIABLE BRAND! But I guess today, even a quality brand is not immune to problems of mass production.
From what I could get out of the Kingston folks, this problem is only with the KVR1066X16-8/256 RIMM and that they are recalling all of the memory and offering free replacements when you call. Do yourself a favor, download and run memtest-86 before you call them and have the model number of your memory handy. One more thing, they only give out RMA numbers during the weekdays/standard business hours. Nights or weekends will require you to call them back during normal business hours or for you to give them your number to be called back. I hope this info helps others who are having the same problems that almost drove me off the deep end for three weeks! Feel free to post questions and I will do my best to respond!

Cheers, crontab -l

Go ahead,,,make my day! root@sys02 /root> kill -9 1<P ID="edit"><FONT SIZE=-1><EM>Edited by crontab on 10/29/02 02:49 PM.</EM></FONT></P>

 

[Crashman]

Hmm, they did have a few bad chips at the beginning of production, maybe these are early production units? Kingston will replace those with high quality units.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[crontab]

I checked with Kingston and they said it seems that quite a few RIMMS got past final inspection during July-August in their U.S. plant only. And yes, you are correct, Kingston will do their best to replace the units with a top quality product. So far they are telling me my RMA will be processed during the first or second week of November. We'll see if they keep the ETA! What have you heard about this memory issue Crashman?

Cheers, crontab -l

Go ahead,,,make my day! root@sys02 /root> kill -9 1

 

[rambus]

I sent the RMA request (Cross Ship with credit card) 2 days ago. I had to call them this noon after checking my RMA status many times. The status was always shown as "Technical Support Review". They promised me to give me a response tomorrow. They did not give me a specific date of shipment.

I should have checked Intel System Validation page, before I ordered the PC1066 modules. Here it is:
http://www.intel.com/technology/memory/rdram/valid/rmvalid.htm

 

[Crashman]

I've heard that only the 256MB modules are on backorder. Such delays would indicate either a rework of the chip design or a revamping of the production process. Because only the 256MB modules seem to be affected by this delay, I hypothesised that the larger modules might use different chips than the smaller modules (perhaps the same number of chips at twice the density).

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[rambus]

The Kingston people in RMA department are really responsive. They approved the RMA request yesterday. However I was told to wait 4-6 weeks for the shipment. I cannot wait that long. So they agreed to refund. I have to wait for the Samsung modules however, which are going to arrive next week. I wish I have some good luck with PC1066 memory this time.

 

[srscbc]

The Kingston 256MB PC1066 recalled the whole batch and is re-engineering/re-processing. The 2 sticks of RDRAM I ordered from Kingston in mid-September are due 11/8/02!

Scott
"Champagne Taste on a Beer Budget"<P ID="edit"><FONT SIZE=-1><EM>Edited by srscbc on 10/31/02 06:42 PM.</EM></FONT></P>

 

[rambus]

After agreeing to refund, Kingston people informed me that the replacements had been already shipped. And I received the 2 modules today. The new modules passed MemTest-86 tests at 4*133 MHz. And I continued the the tests at 4*145MHz for 20 minutes. One module failed this time at Test Number 4 at one address. It seems that the modules can run at 4*143MHz without too much risk. Obviously they meet PC1066 specifications though their overclockability is not so satisfactory.

During the test I had enough time to take a close look at the old defective modules. To my surprisement, they are very likely Samsung's PC800 modules! On the left edge of the PCB is the Samsung part number: MR16R1624AF0. On the aluminium spreader is Samsung's website address. Yes, of course, the Kingston label on the plastic package says "ASSY IN South Korea". These modules did a perfect job at 4*100MHz during the tests. However they can not live at 4*133Mhz.

The Kingston memory was the last thing I suspected in the frustrating process of identifying the source of crashes. I replaced the stock PSU with a Antec True380. And bought a MCX4000 heatsink.

For those who considering buying PC1066 modules, this article maybe helpful:
http://www.gamepc.com/labs/view_content.asp?id=stateofpc1066&page=1&bhcp=1<BR>

 

[crontab]

I see you got yours sooner than expected. The same thing happened to me. I called Monday morning (Oct 28th) to check the RMA and they said 1st or 2nd week of November. Then Wednesday morning (Oct. 30th), the door bell rang and it was FedEx with the replacement RIMM's. I installed the memory and tested with memtest-86. Over clocked the memory up to 150MHz and it was clean with no errors reported. The faulty RIMM's that were replaced were showing a "Made in USA" manufacturing origin lable and the replacements are from "South Korea". How strange that my memory manufacturing origin is exactly opposite of yours. Thanks go to Kingston for getting this problem resolved quickly! I think I will send the guy who wrote memtest-86 a few sheckels for this great tool.
Crashman, this may not be a bad test tool for Tom's Hardware to use when testing the the "Overclockability" of memory. I know it REALLY helped me in this situation and I would suggest for everyone to use it when chasing hardware bugs or stability problems.

Cheers, crontab -l

Go ahead,,,make my day! root@sys02 /root> kill -9 1

 

[rambus]

You got some good luck with Kingston this time. If you check with the Intel RDRAM Validation page I mentioned in a previous post in the same thread, you will find out that only _some_ products of Samsung have been validated under PC1066-32P catagory. You can not find Kingston PC1066 there.

MemTest-86 is really a great tool of doing memory tests. However, using it to test memory overclockablity means a lot of work. You have to test many modules for many hours.

For me, it is already a good indiction that KVR1066X16-4/128 is of low overclockability. the two defective sticks failed at 4*133MHz and one of the replacements failed at 4*145MHz.

Two Samsung modules will be shipped to me in a few days. I wish they could run at 4*150MHz as your modules do.

 

[alcdotcom]

So does overclocking RDRAM cause more long-term wear to the modules? Should I expect a shorter life of the product if I set it to 4*150 as opposed to 4*133? (assuming it will go that high) How much better is the performance of 4*150 compared to 4*133? Is it really noticable? Just curious.


<font color=blue> <i> There are two major products that came out of Berkeley: LSD and UNIX. We don't believe this to be a coincidence. </i> </font color=blue> --J.S. Anderson

 

[rambus]

I think overclocking RAM does no harm to it. You system gains quite a lot by increasing bus speed. You could notice the difference if you run CPU and memory intensive applications.

 

[rambus]

Received Samsung MR16R1624DF0-CT9 modules today. They ran Memtest86 under 4*150Mhz for 2 hours without any errors. It seems to me that Intel Certification really means something.