The Corsair User Forums  

Go Back   The Corsair User Forums > Corsair Product Discussion > Memory

Reply
 
Thread Tools Search this Thread Rate Thread Display Modes
  #46  
Old 10-27-2004, 02:57 PM
Zoltan Zoltan is offline
Registered User
 
Join Date: Oct 2004
Posts: 5
POST ID # = 151040
Zoltan Reputation: 10
Default .

sorry, misspelling.. it is 2.8v

latest bios v1.3

back to my question dough.. how much does the 2-2-2-5 / 2-3-2-5 difference do ?
Should I get it replaced ?

The mb should be able to handle it, as everybody elses does.. shouldnt it ?

Last edited by Zoltan; 10-27-2004 at 03:01 PM.
Reply With Quote


  #47  
Old 10-27-2004, 03:01 PM
RAM GUY's Avatar
RAM GUY RAM GUY is offline
Corsair Product Guru
 
Join Date: Apr 2001
Location: Fremont, CA
Posts: 118,268
POST ID # = 151043
RAM GUY Reputation: 10
Default

It's nominal 50-200 Mbs difference. I would test it both ways and what test are you getting errors on?
__________________
Support accounts and tickets can be created at https://support.corsair.com.
Reply With Quote


  #48  
Old 10-28-2004, 07:11 AM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151171
herman Reputation: 10
Default

Its not much, but thats not what you paid for.

If you are going to do that, then just get LL which are MUCH more stable.

Either way don't let them get away with that.

----

In my oppinion, these sticks get too hot.

The problem with tesing one at a time is that they are not getting as hot as they would sitting together. I have to run more passes to get mine to fail when they are individual.

The neo2 tried to do more of a BTX layout with the CPU in the middle and the sticks on the right. To get everything close enough they alternate the channels for the slots, and also put all the termination is one spot. This means that the sockets are right next to each other and the sticks are closer together for two dimms each in a channel.

No matter what the ~$280 dollar memory should just be able to deal with it. I have a HUGE fan blowing ot the back and plenty of circulation.
Reply With Quote


  #49  
Old 10-28-2004, 07:25 AM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151173
herman Reputation: 10
Default

And by the way, I got my replacements from my RMA for my 2x512 XL. I'm not totally certain, but I might have just gotten the same memory back. If I didn't, then this means that over three models, four motherboards, and five different dual sets, that so far corsair is *2 for 5*.

1) 2x256 XMS C2 2700 - fry's - always worked
2) 2x512 XMS LL 3200 - fry's - failed
3) 2x512 XMS LL 3200 - from RMA - always worked
4) 2x512 XMS XL 3200 - newegg - failed
5) 2x512 XMS XL 3200 - from RMA - failed, very hot

I moved the little spreader retaining wires so they sort of alternate which means four points on the spreaders between the two sticks touch, and this seems to help some. If it is heat, then corsair is not tesin in a way that accounts for heat in a normal persons case. Open air conditioned testing != inside case. For example AMD tests processors at a *very* elevated test room temprature for this very reason.

--------

Needless to say this is very frustrating.

If I got the same ones back then what next? Do I need to send you AMD64 parts to to test with? Do I take advantage of some poor retailers return policy and get fresh ones that work? I'm sure you know what I mean.

Last edited by herman; 10-28-2004 at 07:45 AM.
Reply With Quote


  #50  
Old 10-28-2004, 10:33 AM
Zoltan Zoltan is offline
Registered User
 
Join Date: Oct 2004
Posts: 5
POST ID # = 151182
Zoltan Reputation: 10
Default quick question

its not impossible you got the same one back, that happens a lot. I used to work in support for a pc company and I saw that a lot. Or they switched 2 ppls memory so that the user wouldnt be able to check the serial number...


Can anybody tell me what results they get (memory speed, mb/s) with the 2*512 - 3200XL memory with memtest86+ so I can compare.
And if anybody is overclocking, what results do you get with the 2-2-2-5 or 2-3-2-5 timings ?

(damn its really anoying that I cant run it at 2-2-2-5, only 2-3-2-5)
as herman said its not much, but I still paid a lot of money for it.

Last edited by Zoltan; 10-28-2004 at 10:50 AM.
Reply With Quote


  #51  
Old 10-28-2004, 01:30 PM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151235
herman Reputation: 10
Default

I just tried mine at 1T 2325, 2.7V "aggressive" and it passed with flying colors like 7 full passes.

I used 2.7 V this time just to see. I dont think it can do 2225 consistantly no matter what I do with the voltage. First pass with test 5 only always works, but once they get hot....

Haven't tried just test 5 by itself (can run many more passes), but I will later. I dont trust it till its run like 50 passes on test 5. Its sad, but I think you have to pound it on test 5 long enough to get get really hot. You would think that it would only fail maybe in memtest, but I watched it crash windows after only failing test 5 like this.
---

My results using memtest86+ were like:

3500+ and neo2 @ stock:
18113 CPU, 4500 Cache
2698 Mem - XL, Agressive, 1T, 2225
2618 Mem - XL, Agressive, 1T, 2325
2399 Mem - XL, Agressive, 2T, 2225
2333 Mem - LL, Agressive, 2T, 2326 (16:03 to complete one full pass)
2158 Mem - LL, A Disabled, 2T, 2326 (16:51 to complete one full pass)

XP 2500+ and asus A7N8X v2.0 @ 2.0 GHz and 400 DDR:
12296 CPU, 3914 Cache
1423 Mem - XL, Agressive, 2T, 2225 (26:11 to complete one full pass)
1373 Mem - LL, Agressive, 2T, 2326 (26:33 to complete one full pass)

I wrote them down weeks ago because I had same question, is XL worth it if I dont overclock. For us we can forget overclocking anyway, i.e. just want it to run stable at the Corsair rated speed at all.

Also I tested way many more option combos than this, but these are the only ones I wrote down.

Lastly, you can begin to see why, in theory at least, a fast Athlon 64 might cause it to fail and an XP wouldn't. The lower latency on the A64 causes the reads and writes to be spaced on average much closer together, and either results in higher heat or yet a different type of timing failure for a particular memory, or combination of both, saying there isn't some other tolerance issue with Athon 64's on die mem controller. I dont think there is any other generally available/affordable platform that pounds the mem like this. And Corsair doesn't, to my knowedge, test with Athon 64's, dual or otherwise, and not with case temps like we would have (few degrees difference is all it takes). Since they qualify the XMS (out of spec) memory by testing in real boards, it would make sense that by not testing in a more agressive platform it would cause nothing but pain for us. "Live by the sword, die by the sword", the new XMS slogan.

Last edited by herman; 10-28-2004 at 02:23 PM.
Reply With Quote


  #52  
Old 10-28-2004, 02:49 PM
RAM GUY's Avatar
RAM GUY RAM GUY is offline
Corsair Product Guru
 
Join Date: Apr 2001
Location: Fremont, CA
Posts: 118,268
POST ID # = 151250
RAM GUY Reputation: 10
Default

Well there are ways of testing in another platform and doing the same thing, by increasing the memory frequency you are testing at which we do. So I would have to speculate based on what you have posted that maybe the CPU you have is either not being properly seen by the bios or the CPU has weak memory controller! Also do you have a 90nm CPU, if so you might need to get the latest beta bios from MSI to properly support that CPU, and or just give it some time for the MB/Bios to mature?
__________________
Support accounts and tickets can be created at https://support.corsair.com.
Reply With Quote


  #53  
Old 10-28-2004, 04:18 PM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151294
herman Reputation: 10
Default

Are you saying you dont guarentee XLs at 2225 with A64s?

There are plenty of people who have XL working fine on this model cpu and board as far as I can tell. I was only arguing that maybe, just maybe, the QA may not really be validating the mem well enough for dual with Athlon 64 and/or the extra heat.

I certainly wouldn't want to say that I know absolutely the reason we are having trouble, or how valid my conjecture is entirely. However, I will say cranking the bus speed from 400 to say 420 at 2225, boosting transfer and potential command frequency a little, is not the same as how fast *on average* a particular platform might/can actually do a read after a write. Boosting mem clock only changes the potential (in this case a tiny 5%amount), not if something is actually going to always test the 2 clocks vs 3. Thats the beauty of test 5 and the cacheless block transfer to begin with, it increases the potential for seeing timing problems.

Last edited by herman; 10-28-2004 at 04:23 PM.
Reply With Quote


  #54  
Old 10-28-2004, 04:46 PM
RAM GUY's Avatar
RAM GUY RAM GUY is offline
Corsair Product Guru
 
Join Date: Apr 2001
Location: Fremont, CA
Posts: 118,268
POST ID # = 151300
RAM GUY Reputation: 10
Default

No that's not what I am saying at all, I am asking to check for newer bios and give the MB some time to mature and see if the problem goes away. In the mean time I have just got off of the phone with MSI and they will check and get back to me with any info. All of our modules are tested on a MB at their rated speed so it would hard to get a bad module let alone 2 sets of them. I think some prudence would be suggested as this is not a normal case of a bad module, from what you have posted. In addition, in the read me file for memtest86.com it states that errors in test 5 and or 8 with AMD Platform are not uncommon and do not always suggest a bad or failing module!
__________________
Support accounts and tickets can be created at https://support.corsair.com.

Last edited by RAM GUY; 10-28-2004 at 04:49 PM.
Reply With Quote


  #55  
Old 10-28-2004, 05:22 PM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151309
herman Reputation: 10
Default

OK, I understand.

For me when it fails in memtest it crashes windows eventually, like 10-15 mins. If it passes its fine.

I know there are people who have this stuff and works fine. And there isn't alot about Corsair specific problem with neo2 boards on the MSI forum. Maybe the 90nm procs, which I dont have, need bios help. Like last time, one of the two sticks works fine.

Also the "AMD platform" thing is too broad now. The arch has changed dramatically over the years. Even the old nForce 1 vs VIA KT333/400 thing has become irrelevant, where alot of this came from.

The memtest site actually says:
There have been numerous reports of errors in only tests 5 and 8 on Athlon systems. Often the memory works in a different system or the vendor insists that it is good. In these cases the memory is not necessarily bad but is not able to operate reliably at Athlon speeds. Sometimes more conservative memory timings on the motherboard will correct these errors. In other cases the only option is to replace the memory with better quality, higher speed memory. Don't buy cheap memory and expect it to work with an Athlon! On occasion test 5/8 errors will occur even with name brand memory and a quality motherboard. These errors are legitimate and should be corrected.
----

I hope I bought good quality name brand memory. :-)

This snippet highlights the argument we are having, where I am suggesting the possiblity that your ram as tested is just not consistantly module to module able to operate at "Athon 64 speeds", just like what happened before a few years ago. So far no one has been willing to consider the possiblity, which makes me sad.

It is one of the main tenants of memtest, that if you cant read it or write it here, then it just doesnt work. That windows, linux, whatever, will read and write just the same way. Its not like you get alot of choice either, load, store, sse, etc..., built into proc x86.

Like I said, test 5 highlights timing issues. I'm willing to believe that it could be the proc or the MB, but I'm not willing to believe that the mem doesn't work with 3500+ and neo2 at all yet.

I say this stuff with all the love in the world. I know you guys have to deal with alot.

Last edited by herman; 10-28-2004 at 05:58 PM.
Reply With Quote


  #56  
Old 10-28-2004, 06:14 PM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151324
herman Reputation: 10
Default

Also keepin mind that ther aren't alot of different 939 (dual channel) boards out there right now, where the neo2 is probably the most popular among people that would pay this much for memory.

It makes sense that you would get alot of reports for problems with neo2 with XL memory which in some cases might just be the memory. If all neo2 were having problems with it, you might easily have 10x-100x the reports.
Reply With Quote


  #57  
Old 10-28-2004, 06:26 PM
RAM GUY's Avatar
RAM GUY RAM GUY is offline
Corsair Product Guru
 
Join Date: Apr 2001
Location: Fremont, CA
Posts: 118,268
POST ID # = 151328
RAM GUY Reputation: 10
Default

Granted, and I would agree with you on most counts. However, there is no reason for me to believe that we are making a memory that will just not run in your system. I don't buy that! I would like some time to set this MB and CPU up and test it here and get some feedback from MSI. I just do not want to put you through the RMA process again and have you right back saying it's our modules. So I understand your frustration, but please give me a few days to do some research. We will get it solved if it's or modules. And we might be able to get the MB maker to take another look at it if it’s not the memory!
So just too be clear, my suggestion is too run the system at 2.8 Volts with the timings set to 2-3-2-5 for a short time so I can get with MSI and AMD and test the configuration you have.
What would really help me to help you is a detailed spec of your configuration and the # from your CPU so I can try and match it!
__________________
Support accounts and tickets can be created at https://support.corsair.com.
Reply With Quote


  #58  
Old 10-28-2004, 07:27 PM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151343
herman Reputation: 10
Default

I was suggesting that it should work normally too.

I believe that only a relatively few people are having problems, but many would have a neo2 and that would account for things. I also believe that you would need marginal memory, like what I had and have for it to fail. Many many people have working systems with 2x512 XL 3200, 3500+ and neo2 running 2225 stock, and quite a few fairly well overclocked. I just want mine to work at all.

Main:
Corsair TwinX 1024 (2x512) 3200 XL (set #1 new / set #2 from RMA)
...tried in 12 (inside toward CPU) and 34 (outside)
Antec TruePower 380s - comes with Sonata case (new)
...all rails under load are within %1
New AMD Athlon 64 3500+ 939 130nm Retail - ADA3500AWBOX (new)
...Serial# 1010064G40685
New MSI K8N Neo2 Platinum - 601-7025-010 (new)
...Serial# K0409017183

Other:
Antec Sonata case (new)
Zalman 7000A Cu Heatsink (new)
Western Digital Raptor 73GB (new)
BFG nVidia ti4600 (older AGP 4X version)

Software:
memtest86+ v1.26
neo2 bios v1.3

Basic Settings:
clear cmos then set "failsafe" defaults
...*should* be the same thing
set mem to 2.8V
(should now have 2T 2225 @ 400 DDR and 2.8V)

Procedure:
run test 5
...should fail once gets hot enough
...might need simulate normal in case temp

Faster Settings (all should be supportable on this board):
core cell bios menu:
- mem to 1T
- aggressive
- spread spectrum off

For mine both sets seems to have one marginal stick. Much harder to get to fail by itself, but if you use "Faster Settings" above, it will fail after a several passes. Faster setting is still 2225 but gets the requests flying. My theory is that when they sit together, they get much hotter and fail more easily. Slots on MSI are very close together. Maybe need better thermal material for spreader for XL.

I know its supposed to be rare, but since I dont think you have systems that could catch this, and as these build up that come back I could totally see getting someone else's bad set.

Last edited by herman; 10-28-2004 at 07:38 PM.
Reply With Quote


  #59  
Old 10-28-2004, 07:35 PM
Zoltan Zoltan is offline
Registered User
 
Join Date: Oct 2004
Posts: 5
POST ID # = 151344
Zoltan Reputation: 10
Default

This is what I am useing (when stable):

Cpu AMD64 3500+ NewCastle 0.13
3200XL 2*512 at bank 1+2
MSI Neo2 Platinum Motherboard, latest bios version.
Running with agressive setting
Mem volt 2.8
Agp volt 1.6
Memory settings 2-3-2-5 1T 200Mhz

The speed I get in memtest86+ is 2333MB/s with 200 FSB setting.
Memory works just fine with higher FSB settings. Tried up to 209 and it is still stable as long as I use the 2-3-2-5 1T settings.

Need any more info?

Thank you RAM GUY and herman for good information and the patience to see this trough for everyone that is reading the forum and having the same problems.

ohh, and just to look cool like herman above, Im useing the same Raptor HDD :)

Last edited by Zoltan; 10-28-2004 at 07:41 PM.
Reply With Quote


  #60  
Old 10-28-2004, 07:41 PM
herman's Avatar
herman herman is offline
Registered User
 
Join Date: Dec 2003
Posts: 60
POST ID # = 151345
herman Reputation: 10
Default

NP Thats why I am being so verbose, to help anyone else.

This stuff sucks. I guarentee though that if it was the whole world, there would be a total flood, not just the 10-15 or so in the last weeks. We either got bad mem, procs or MBs (or outside chance on the PS for some).

I still think its most likely the mem. They have no way to test an A64 at all, unless very recent.

Last edited by herman; 10-28-2004 at 07:45 PM.
Reply With Quote


Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -4. The time now is 08:35 AM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2019, vBulletin Solutions, Inc.