Found the source of my unstable system

**Bubba** · 02-09-2008, 10:21 AM

Recently I upgraded to 3GB while on Vista. Instant blue screen.
So then I install XP and it's ok for awhile. Then InstallShield keeps telling me CABs are corrupt on CD's I've only used once. So I alter the memory timings and everything seems to be ok. Recently I started having applications throw exceptions that have never thrown before. Finally I broke down and burned an ISO of Windows Memory Diagnostic.

Tested the full 3GB and every test failed.
Tested each stick independently and 1 in particular always failed.

The third stick would do this:

Expected: fe-fe-fe-fe::fe-fe-fe-fe
Actual: fe-fe-fe-fe::fe-fe-ee-fe

Every time on every test the third byte in the high order word was wrong. So if you had a pointer that relied on some object at an address the pointer would be wrong and of course would throw an exception.

Problem solved.

In short when you buy memory...test it. Woulda saved me a lot of headaches.

SlyMaelstrom · 02-09-2008, 10:38 AM

RAM from my experience has a fairly high failure rate... over the years I've bought 20 or so sticks of RAM and have had to replace, I think, four of them for one reason or another. I'm talking very reliable brands, as well... you just have to get used to the fact that they have such a high failure rate.

Bajanine · 02-09-2008, 03:57 PM

I have never had new RAM fail until just recently! I agree it can be a real pain in the rear. Thorough testing is a must! I always use MemTest and Orthos to verify my RAM.

Lately, it seems like getting good RAM is just pure chance. I'll tell a friend how easy it is to upgrade their RAM by directing them to a online memory configurator and if the don't get the wrong RAM then they get faulty RAM! So much for easy.

Mario F. · 02-09-2008, 04:16 PM

Still... quiet annoying the fact RAM is being sold that a simple diagnostic can show to be faulty.

**Bubba** · 02-09-2008, 08:03 PM

Yes, very. From my experience with quality control I'd say they test a sample size of each production run. This means that not every module would be tested so I'm sure some bad ones get through.

AloneInTheDark · 02-09-2008, 08:32 PM

Which brand did you buy?

zacs7 · 02-09-2008, 11:49 PM

> This means that not every module would be tested so I'm sure some bad ones get through.
Not to mention it *could* get damaged in transport. After the diagnostic perhaps?

AloneInTheDark · 02-10-2008, 06:36 AM

Quote:

Originally Posted by zacs7

> This means that not every module would be tested so I'm sure some bad ones get through.
Not to mention it *could* get damaged in transport. After the diagnostic perhaps?

Somehow, I don't see that as the true cause. I mean, RAM has no moving parts and the rate of error is way too high. I almost doubt that many drives would get messed up in the transport.

Elysia · 02-10-2008, 06:47 AM

What diagnostic? Things such as RAM is not tested before it's shipped since it would take way too much time.

SMurf · 02-10-2008, 10:14 AM

I'd say it's almost completely luck of the draw, although fate seems to smile on me as far as memory goes, never had a failure yet (although admittedly I only buy new memory for new equipment, that's every 4 years on average).

Same with hard disk drives. My stuff works fine, then I pass it on and I hear a few months down the line that it's seized up.

**Bubba** · 02-10-2008, 10:34 AM

Quote:

What diagnostic? Things such as RAM is not tested before it's shipped since it would take way too much time

I've never been in any production environment that did not test a small sample size from each production run. I've been involved with machines, glass, engines, parts, etc. Everything that is made has some type of QA analysis setup so that the factory can monitor quality. You can't just produce, forget, and ship. It doesn't work like that.

Elysia · 02-10-2008, 10:36 AM

Samples, sure. But that's just a small subset of all the production.
That's why warranty exists, just in case something should break.

Mario F. · 02-10-2008, 10:41 AM

Quote:

Originally Posted by Elysia

Samples, sure. But that's just a small subset of all the production

These "samples" are used to calibrate the production line. Under high quality control requirements, once calibrated a production line produces basically the exact same product (there's a well known failure rate).

What it seems is that, quality requirements have been lowered or production lines are not being adequately calibrated or whatever.

**Bubba** · 02-10-2008, 10:52 AM

Quote:

What it seems is that, quality requirements have been lowered or production lines are not being adequately calibrated or whatever.

Hehe. Sounds like a fellow QA guy. Rememeber it's always an auditor's fault but never the line's.

Elysia · 02-10-2008, 10:53 AM

Quote:

Originally Posted by Mario F.

What it seems is that, quality requirements have been lowered or production lines are not being adequately calibrated or whatever.

I'm thinking you might be right.

02-09-2008, 10:21 AM	#1
Bubba Super Moderator Join Date: Aug 2001 Posts: 7,812	Found the source of my unstable system Recently I upgraded to 3GB while on Vista. Instant blue screen. So then I install XP and it's ok for awhile. Then InstallShield keeps telling me CABs are corrupt on CD's I've only used once. So I alter the memory timings and everything seems to be ok. Recently I started having applications throw exceptions that have never thrown before. Finally I broke down and burned an ISO of Windows Memory Diagnostic. Tested the full 3GB and every test failed. Tested each stick independently and 1 in particular always failed. The third stick would do this: Expected: fe-fe-fe-fe::fe-fe-fe-fe Actual: fe-fe-fe-fe::fe-fe-ee-fe Every time on every test the third byte in the high order word was wrong. So if you had a pointer that relied on some object at an address the pointer would be wrong and of course would throw an exception. Problem solved. In short when you buy memory...test it. Woulda saved me a lot of headaches.
Bubba is offline

02-09-2008, 10:38 AM	#2
SlyMaelstrom Devil's Advocate Join Date: May 2004 Location: Out of scope Posts: 3,776	RAM from my experience has a fairly high failure rate... over the years I've bought 20 or so sticks of RAM and have had to replace, I think, four of them for one reason or another. I'm talking very reliable brands, as well... you just have to get used to the fact that they have such a high failure rate. __________________ Terms of Service By quoting or replying directly to this post, you consent to the fact that all of the information in the post above is completely accurate and highly intelligent and no comments will be made towards its validity, thoughtlessness, and/or grammatical structure. Violators will be prosecuted to the fullest extent of the law.
SlyMaelstrom is offline

02-09-2008, 03:57 PM	#3
Bajanine Registered User Join Date: Dec 2001 Location: The most peaks over 10,000 feet! Posts: 388	I have never had new RAM fail until just recently! I agree it can be a real pain in the rear. Thorough testing is a must! I always use MemTest and Orthos to verify my RAM. Lately, it seems like getting good RAM is just pure chance. I'll tell a friend how easy it is to upgrade their RAM by directing them to a online memory configurator and if the don't get the wrong RAM then they get faulty RAM! So much for easy. __________________ *"A government big enough to give you everything you want, is big enough to take away everything you have." - Thomas Jefferson* MSVS 2008 Pro / DevPartner / CB NightlyBuilds / MinGW / Cygwin
Bajanine is offline

02-09-2008, 04:16 PM	#4
Mario F. (?<!re)tired Join Date: May 2006 Location: Portugal Posts: 5,645	Still... quiet annoying the fact RAM is being sold that a simple diagnostic can show to be faulty. __________________ Originally Posted by brewbuck: Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
Mario F. is offline

02-09-2008, 08:03 PM	#5
Bubba Super Moderator Join Date: Aug 2001 Posts: 7,812	Yes, very. From my experience with quality control I'd say they test a sample size of each production run. This means that not every module would be tested so I'm sure some bad ones get through.
Bubba is offline

02-09-2008, 08:32 PM	#6
AloneInTheDark Registered User Join Date: Feb 2008 Posts: 74	Which brand did you buy?
AloneInTheDark is offline

02-09-2008, 11:49 PM	#7
zacs7 Woof, woof! Join Date: Mar 2007 Location: Australia Posts: 3,292	> This means that not every module would be tested so I'm sure some bad ones get through. Not to mention it could get damaged in transport. After the diagnostic perhaps? __________________ "I.T. gets the chicky-babes" - M. Kelly bakefile \| vim
zacs7 is offline

02-10-2008, 10:14 AM	#10
SMurf Registered User Join Date: Aug 2001 Location: Newport, South Wales, UK Posts: 1,094	I'd say it's almost completely luck of the draw, although fate seems to smile on me as far as memory goes, never had a failure yet (although admittedly I only buy new memory for new equipment, that's every 4 years on average). Same with hard disk drives. My stuff works fine, then I pass it on and I hear a few months down the line that it's seized up.
SMurf is offline

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Line Counting	00Sven	C Programming	26	04-02-2006 08:59 PM
fopen();	GanglyLamb	C Programming	8	11-03-2002 12:39 PM
found source! i be tha thrilla killa'!!!	doubleanti	A Brief History of Cprogramming.com	6	10-31-2001 09:15 AM