Jump to content
Corsair Community

Twinx1024 3200c2 Memtest86


CrowTech

Recommended Posts

So, before I get going on my issue I just wanted to say thank you to Ram_Guy and all the other active members posting here. A lot of great info went into my solutions directly from here, 'preciated!

 

What I've got...

 

Intel 3.2 HT 512Cache

Abit 1c7-g (BIOS 28 Jan2005)

2 Gig (2 x TWINX1024 3200C2) Corsair (3-3-3-8 Default By SPD)

 

I was having the odd nasty crash in Max Payne 2 and Punisher - symptoms of crash near identical so I decided to get testing stuff. Funny though, games like Doom3, Half Life 2, CS Source - no problemo - occasional *rare* crash to desktop (which I give hats off to game code going awry). I mean even Vampires runs fine, and it's a piece of coding **** ;) It came down to the RAM cause everything else appears okay...and well...here we go!

 

Ran MEMTEST86 at the default SPD and things test out just fine. So I start a scratchin' the noggin thinking perhaps I can bring the ram up to its stress level and reveal something...so...up we go to 2-3-3-6 (rate speed mind you).

 

MEMTEST86 ran fine until TEST 7. Ouch. Nasty...lots o errors, and I'm like...uh oh. So I set it back to default SPD and things go back to normal, no errors. Now, after reading a lot of the great stuff in here I get the idea to go ahead and yank that voltage up from 2.6 default setting to 2.7. Back to MEMTEST86....and, well - nope.....much less errors, but STILL errors. Also, side note - normal speeds ran prime95 for 15hrs no errors. Have yet to attempt that with the lower timings...

 

I guess my question is where to go now? More voltage (obviously no more than 2.9)? Reduced timings (I didn't buy this RAM to run at lower speed did I)? Now, what I'm really worried about!

 

Playing in the ic7-g ram slots ain't fun...there's hardly ANY room up there and well - with full banks (2 gigs loaded) it ain't a whole lot of fun removing and replacing sticks in single channels to find out which of the sets has gone bad...if at all and while I'm taking RAM in and out I stress the hell outta the MOBO. I also do wonder of the reliability of MEMTEST86 and in my case - why JUST TEST 7??. Further, I find it a bit erroneous that Corsair doesn't at least provide a *trusted* memory test (it seems lacking in the entire industry so hey, no harm no foul, just odd). Also, these pairs are tested pretty good at factory, no? So I'm kinda feeling sick having paid a whole bunch for ram I have to test for bad, damn that sucks. So what if I get it replaced, the time and energy that's been wasted here cannot be compensated for. Sorry, it's just the hurtful truth.

 

Alas, I want the machine to stop being "weak" and according to what I have in front of me now, the memory may be the "weak" spot...maybe not...but that's why I'm asking for help now.

 

What do I do now? Help, help...and thanks :)

Link to comment
Share on other sites

So, before I get going on my issue I just wanted to say thank you to Ram_Guy and all the other active members posting here. A lot of great info went into my solutions directly from here, 'preciated!

 

What I've got...

 

Intel 3.2 HT 512Cache

Abit 1c7-g (BIOS 28 Jan2005)

2 Gig (2 x TWINX1024 3200C2) Corsair (3-3-3-8 Default By SPD)

 

I was having the odd nasty crash in Max Payne 2 and Punisher - symptoms of crash near identical so I decided to get testing stuff. Funny though, games like Doom3, Half Life 2, CS Source - no problemo - occasional *rare* crash to desktop (which I give hats off to game code going awry). I mean even Vampires runs fine, and it's a piece of coding **** ;) It came down to the RAM cause everything else appears okay...and well...here we go!

 

Ran MEMTEST86 at the default SPD and things test out just fine. So I start a scratchin' the noggin thinking perhaps I can bring the ram up to its stress level and reveal something...so...up we go to 2-3-3-6 (rate speed mind you).

 

MEMTEST86 ran fine until TEST 7. Ouch. Nasty...lots o errors, and I'm like...uh oh. So I set it back to default SPD and things go back to normal, no errors. Now, after reading a lot of the great stuff in here I get the idea to go ahead and yank that voltage up from 2.6 default setting to 2.7. Back to MEMTEST86....and, well - nope.....much less errors, but STILL errors. Also, side note - normal speeds ran prime95 for 15hrs no errors. Have yet to attempt that with the lower timings...

 

I guess my question is where to go now? More voltage (obviously no more than 2.9)? Reduced timings (I didn't buy this RAM to run at lower speed did I)? Now, what I'm really worried about!

 

Playing in the ic7-g ram slots ain't fun...there's hardly ANY room up there and well - with full banks (2 gigs loaded) it ain't a whole lot of fun removing and replacing sticks in single channels to find out which of the sets has gone bad...if at all and while I'm taking RAM in and out I stress the hell outta the MOBO. I also do wonder of the reliability of MEMTEST86 and in my case - why JUST TEST 7??. Further, I find it a bit erroneous that Corsair doesn't at least provide a *trusted* memory test (it seems lacking in the entire industry so hey, no harm no foul, just odd). Also, these pairs are tested pretty good at factory, no? So I'm kinda feeling sick having paid a whole bunch for ram I have to test for bad, damn that sucks. So what if I get it replaced, the time and energy that's been wasted here cannot be compensated for. Sorry, it's just the hurtful truth.

 

Alas, I want the machine to stop being "weak" and according to what I have in front of me now, the memory may be the "weak" spot...maybe not...but that's why I'm asking for help now.

 

What do I do now? Help, help...and thanks :)

 

 

I know how you feel; read my post. We've spent lots of energy and time on this, and what now? And most importantly, what about out loss?

 

Interestingly, I've the same model as you, but mine is Twinx 512Mb.

 

Corsair never more! Time for moving to a HONEST/COMPENTENT memory brand.

Link to comment
Share on other sites

Nice to know I'm not alone out here, although - that doesn't make it any nicer for either of us (or the, hmmm - others maybe?? Speak up for God sakes...they'll need our help to fix it). Of course - I'm still not convinved the RAM is to blame here though it's *very* conspicuous at this point. I'd like to give Corsair the chance to help me outta this, umm - expensive venture into RAM in some amicable way. Try this, try that has got me this far, now I need their (your) help to see if I'm truely umm, in trouble or not...lol.

 

So, anyone....anyone....wanna fish me outta this (RAM_GUY, please help)? At least help me get this heap of RAM to do what it says it was going to do, that, or keep replacing it until it does....though, how long am I supposed to be exchanging parts to find the *holy grail* set that gives a clean 2GB set of RAM....at least it may seem like that to me...anyway...I'm still open to testing stuff out...regardless, I ain't givin' up that easy!

 

More updates for those who wanna help...

 

I've since tried uping voltage to max on board setting of 2.8 with the 2-3-3-6 settins. It ran memtest86 much better but STILL errors, damn - damn - damn. Then, bravely - I loaded the OS up to try prime95 with these settings, ya know - another app...can't be two bad apps type thinking (and up to now I had *never* attempted an OS boot). Well, OS started fine no worries there, but...not two bad apps for sure. Prime95 failed pretty quick, third test of the default stress test "blend" mode. Settings have been returned to stable (*cough* mostly) settins of 2.7 voltage, 3-3-3-8 by SPD. Things seems fine again...but I'm sure the afore mentioned games still puke, though, I have yet to test them (been more relying on Prime95 for the bench here). Last night, I had run Prime95 at the lower timings (though voltage was set at 2.6 then) and succesfully ran the app without errors for a reported 15 hours, but I understand it can take longer to fail in *weird* cases (possibly mine). Memtest86 runs fine at the 2.7 voltage with 3-3-3-8 timings. Running Prime95 as this is typed...no problems yet (about an hour run time by now).

 

Thanks to LOPAN for directing me to his posts as well, like I said - there is *some* comfort in not being alone here and being back in Windows for now (can't believe I just said that!). *Sigh* - why is it always the favorite games that just won't stay going...damn, it's the worst part of this whole situation!!

Link to comment
Share on other sites

What is the revision of each TwinX pack? Keep in mind that while each TwinX pack is rated @ 2-3-3-6, they neither test nor guarantee the timings for 2 TwinX packs together. In fact, I've never heard of any memory manufacturer guaranteeing 4 sticks together @ a certain timing setup.

 

Try this: Take one TwinX pack out, test the remaining one @ 2-3-3-6 with memtest. After that, take the other pack and swap it into the same slots as the first pack. Test them. If all of the tests run fine, no errors, no crashes, etc, then it isn't the memory. If the packs are different revisions, it may be the memory controller not liking to play with 2 different sets of ICs, albeit if that is the case, I'd imagine you'd see much more frequent crashing. If they're the same revision, try 3-4-4-8 with all 4 sticks in.

 

Also, run all of the tests @ 2.75 or 2.8 volts. BTW, don't forget that the memory controller on the chipset may be causing the problem with 4 sticks as well. Trying all 4 sticks on a friend's setup couldn't hurt either, but I'd leave that pain in the *** as a last resort! :)

Link to comment
Share on other sites

What is the revision of each TwinX pack? Keep in mind that while each TwinX pack is rated @ 2-3-3-6, they neither test nor guarantee the timings for 2 TwinX packs together. In fact, I've never heard of any memory manufacturer guaranteeing 4 sticks together @ a certain timing setup.

 

Try this: Take one TwinX pack out, test the remaining one @ 2-3-3-6 with memtest. After that, take the other pack and swap it into the same slots as the first pack. Test them. If all of the tests run fine, no errors, no crashes, etc, then it isn't the memory. If the packs are different revisions, it may be the memory controller not liking to play with 2 different sets of ICs, albeit if that is the case, I'd imagine you'd see much more frequent crashing. If they're the same revision, try 3-4-4-8 with all 4 sticks in.

 

Also, run all of the tests @ 2.75 or 2.8 volts. BTW, don't forget that the memory controller on the chipset may be causing the problem with 4 sticks as well. Trying all 4 sticks on a friend's setup couldn't hurt either, but I'd leave that pain in the *** as a last resort! :)

 

Will try this - however, I'm not pleased to hear about modules not working together as such, very dissappointed - not like it's something I've heard before even though it's likely true. Then...what am I supposed to do with the other sticks and why on Earth can I load 4 modules if they don't *like* to run together??

 

I'll look into revisions, etc - and try the single set test(s), however, the issue with Max Payne 2 was on lower timings with only one set at the time...alas I think I may be wasting my time...though I will waste it just to be on the safe side of things. Thanks for the ideas :)

 

Anyone else got some thoughts....I'd really like to hear what you have to say Ram_Guy, even if you're like "try what he said, get back to us" type thing...

Link to comment
Share on other sites

  • Corsair Employees

Please put the memory module on slot 1 & 3; make sure that you have the latest BIOS for your MB, and set the following settings on your BIOS;

TwinX512 or TwinX1024-C3200C2

Softmenu Setup

CPU Operating Speed: User Define

CPU Clock/AGP/PCI: 250/66/33

Memory Clock: 400MHz

CPU/DRAM ratio: 1:1

AGP ratio: Fixed

AGP/PCI: 66/33

Spread Spectrum: Enabled

DDR Voltage: 2.8/2.85 Volts

AGP Voltage: Default *unless you have ATI (9200 or 9600) or NVIDIA (5200 or 5700) then 1.6 Volts suggested!*

Advanced Chipset:

SDRAM Timing Settings: User Define

SDRAM CAS Latency: 2T

SDRAM RAS to CAS Delay (tRCD): 3T

SDRAM Active to Precharge Delay (tRAS): 6T

SDRAM Row Precharge (tRP): 3T

All other settings should be set to default settings!

Then please test them one at a time with http://www.memtest.org and let’s make sure it's not some other issue! I would run the test for at least 2-3 passes to be sure! If you have additional PCI card installed on the system, please make sure you do not put it on PCI slot# 4; since PCI slot# 4 is sharing IRQ with the AGP card, and this will cause conflict to your system. Please let me know how you make out!

Link to comment
Share on other sites

Sorry RAM_GUY, I was in the midst of this as you posted...please read, this is how I made out...and how I feel right this second about testing this RAM anymore than I have. BUT, BUT, BUT - thank you so much for the config info - that's awesome of you!!

 

I've never heard of any memory manufacturer guaranteeing 4 sticks together @ a certain timing setup.

 

I'm sorry, but that's ridiculous to me!

 

Why?

 

It seems to me, that if I have read the details correctly, the only issue that running more sticks raises is the potential for one incompatible stick (or sticks) to not run "high speed"...i.e. One pair is capable of speeds the other is not. That the RAM was tested as a pair, not a quad set. So, someone actually sifts through all this RAM to find pairs that "work" in DUAL channel mode at rated speeds and I guess this entitles mem. companies to say "High Speed DUAL Channel", where that plainly is an overstatement. There is no such thing as DUAL channel. It's just "tested" pairs of RAM, tested for DUAL mode operation. This is an awful nice garuntee, I don't deny it, but it comes across to me as a marketing campaign made out of your own saftey net. Especially if you're ME....looking for DUAL channel with 2GB of RAM (which doesn't include fat hulking 1GB modules either). I had figured Corsair was good enough, there wouldn't be a problem - it's excellent RAM, that's why I bought it...I had plenty of other choices (hmmmm).

 

What has been asked of ME to find out and try...

 

Quickly looking with software, I believe it correct that the pairs are NOT the same as one pair will list support for lower FSB speeds in the utility and the other pair only lists 200MHZ speeds (Sandra if you were curious). The pair with more timings listed (133, 166, 200) is the new set, the old set only lists one timing, 200MHZ (now I'm really not happy). But, is this enough evidence to support "not the same revision"? I really want to avoid taking RAM in and out unless it's getting replaced (if there's nothing wrong with the MOBO now, re-installing RAM endlessly sure will screw it up eventually). One pair was purchased in Oct 2004, the other a couple weeks ago. I think the likelyhood is high that they're not the same revision, but, who cares whether they are or not??!! I don't, and now I want to go back to what I was thinking in this whole mess...(BTW, I'll not be lower the timing any more than the default, it's bad RAM if I have to do that...no other explination, actually I'm already convinced it's bad).

 

Now, the way I interpret what you've said to me....

 

Well dude, sorry we didn't test your RAM in a quad set - we haven't done that because the market isn't there for these types of RAM requirements and as you have first hand experience - testing quads is a major butt puffer, we can't make money that way! It's much easier for us to just say, well - it's probably that fancy chipset not liking the RAM being *different*, even in a subtle and hard to reveal way. That, or your MOBO is junk. It's just NOT our RAM, no way. Besides, ridiculous performance isn't THAT big a deal man.

 

The simple fact is, by Corsair's admission and predetermined choice - this second set of RAM is just a waste, or was it both sets...lemme go on a bit more, more gently...(I feel like I got stuck with a Lada, when all I really wanted was a Ferrari). Actually, the fact that the RAM does DUAL at all in this config in a somewhat stable manner is proof you guys can do this...

 

What I want is 4x512 MB sticks running in DUAL channel mode at 2-3-3-6, no more, no less. No!!? Not garunteed? So, NO!?? You're saying NO to money, now and in the future (since I'm in the hole here already)?!

 

Why!?

 

My take is that it is because Corsair can't actually make a profit making RAM that works like that, or worse - it's beyond Corsair's capability to make this stuff *that* realiable. Really - if you made RAM that was I wouldn't be here at all, further to that, it wouldn't be too hard to find 4 sticks that run together this way, would it?

 

What I was hoping could be done - and I think Corsair would win major points if you did this...exchange some RAM with me. Get me the "holy grail", though you didn't promise it I guess, you should be delivering it or PLAINLY STATING OTHERWISE!!!!! You can't be an industry leader in second, third - anywhere but first place, and it ain't a easy trip getting AND staying there. Besides, I paid a lot of money already - I almost expect it ya know. The RAM I have does not do what I want it to and I paid for RAM that does what I want it to. I guess I'd rather just through it in the garbage and get sets that will. Harsh?? Well, I'd rather that then either...

 

a) Drop down to single channel mode (which probably won't help anyway)

b) Use timing other than the ones I want (why, just - why?!!)

c) Lose 1GB of RAM (of which, what am I supposed to do with anyway)

d) See an itch of any error in any memory test, at least for a good long while

 

I also came to a conclusion in that, lol - if one stick goes bad on me, you replace both?? You have to, don't you. Or at least get them both back so you can FIND a stick that works with the original, that or just give me a pre-tested new pair and recycle. It'd be pretty grim to get a brand new stick back and find that, because of the replacment, I can't run the config I had set before hand....or does that happen too??

 

Well, as I said - this is what I want you to do to fix my 4x512 needs. And before you say, no way - not a chance, realize that I still believe Corsair can do this. I'd like to continue to praise Corsair and have others go, HOLY COW...I want another set of RAM now too! Maybe then you can make some money at this quad config, besides - other companies would go this mile, don't have me mention names now do I (that's just hideous to see in your forum isn't it)?

 

*sigh*

 

All this, or I throw all the RAM in the trash, *never* recommend Corsair to anyone, ever...probably get it removed as product from at least one local retailer and replace it with RAM that DOES THE JOB! I am sorry, but it's just that easy to me. Though, believe me, the fact remains in this conclusion I may have just sent several hundred dollar bills to the "bit bucket" - and that's a VERY hard lesson, OUCH.

Link to comment
Share on other sites

Okay, some settings I couldn't make..

 

First, did NOT set 250/66/33 - left at default 200/66/33...umm, I have a 3.2GHZ CPU, is that not accurate?? Did other settings manually though, yep...except...

 

Can't set memory clock to 400, system hangs. By SPD (default) it boots up and posts stating it is in fact at 400MHZ (as always). All other settings, besides spread spectrum (unavailable or not named thusly) were changed and booted to memtest86, of course first without removing RAM...yep, still errors...again only test 7, I don't understand that, but that's okay - errors, I understand enough.

 

I will now attempt to switch the sticks, dreaded switch the sticks - play around with the pairs and see if that isolates something at least. I hope so, for the effort and for the hopes that I can finally be rid of the issue during a nice RMA (which, I'm thinking will happen according to Wired's other posts around and other such posts, resolutions). Thing is, I hadn't read yet (and I'm still having the creepy feeling) that Corsair won't support the 4 stick config I want...oh please, why is that too much to ask after I've gone to this much trouble...ya know I wouldn't if it wasnt that important, so I'd hopefully thank you in advanced for sweet support like that, hopefully.

 

Also, fyi - I've only lowered the timings to 3.0-3-3-8, all other settings are being maintained...though I haven't really hammered a few tests yet, she seems as stable as normal (maybe even more so, I dunno yet - don't care, I want that gap closed)...so....I'm off to switch sticks and test that with the lower 2-3-3-6....(pray for me) :confused:

Link to comment
Share on other sites

I know I said it before, but I'm saying it again - thank you for the info on the config changes, suggestions and such. Save for a few settings, it all helped me get through all this. I should have been more careful and not blindly bought into the new set without dilegently checking revisions, etc, etc....anyway...I *thought* it'd be okay since Corsair is so good!...right??

 

I have finished testing each pair. First, the pairs are in fact two different revisions, confirmed. The early set is revision 4.1, the older is 5.1. Testing each pair revealed no errors, hear that guys - NO ERRORS!! LOPAN and others...well, I got it...maybe with a little faith, you will get your troubles too... :p:

 

Okay, but I'm still damned - I can't get all four sticks to go at speed. So, I ask in favor - can I replace the revision 4.1 for 5.1 or even the 5.1 for 4.1 as I'm fairly confident, and you'll probably agree - this will give me what I want - 2-3-3-6, 4x512 - and a big freakin' smile on my face?? &&|| Do I get store credit (if I freakin' can) since either pair actually works and still (I would think with the latest set) may just fall in the 30 day return time (please don't make me ask these nice guys, come on...they've been great...unless of course Corsair is gonna back them up on the other end of that deal).

 

What are the options now - both pairs are fine on their own, not together the way I want, need them to run. Please help me, I know this RAM will do it...just a matter of getting the right set....come on now, I've done my job to the letter. Lets see you do yours beyond the call, yeah Corsair!

 

EDIT...thought this would do too....

 

WORKING SETTINGS

 

CPU 200/66/33 (3.2 GHZ)

MEMORY CLOCK BY SPD (400 MHZ DEFAULT)

CPU/DRAM RATIO 1:1

AGP RATIO FIXED

AGP/PCI 66/33

SPREAD SPECTRUM (UNKNOWN - NOT FOUND)

DDR VOLTAGE 2.85

AGP VOLTAGE 1.55 (DEFAULT - RADEON XT800 PE)

TIMINGS 2-3-3-6

 

Again, thanks to RAM_GUY and Wired for their responses and help through this. Lets hope I can get 4x512?! :bigeyes:

Link to comment
Share on other sites

OK, couple of things to clarify:

 

It seems to me, that if I have read the details correctly, the only issue that running more sticks raises is the potential for one incompatible stick (or sticks) to not run "high speed"...i.e. One pair is capable of speeds the other is not. That the RAM was tested as a pair, not a quad set. So, someone actually sifts through all this RAM to find pairs that "work" in DUAL channel mode at rated speeds and I guess this entitles mem. companies to say "High Speed DUAL Channel", where that plainly is an overstatement.

 

Running more sticks sort of raises the potential. Let's say 1 / 100,000 sticks "passes" the tests internally, but in the real world, fails. The more sticks you buy, the higher chance you have of getting a bad stick, yes, however it's still slim. If a stick does everything it's supposed to do within normal, guaranteed specs (whatever that may be), it won't suddenly go bad or anything when run with more sticks. Now, there are some memory controllers out there that when 4 sticks are run, they will lower the FSB for example, and this is by design.

 

Abit IC7-G uses the Intel 875P chipset. That chipset's setting off some bells in my head, can't remember why though. Gotta research it.

 

 

It's just "tested" pairs of RAM, tested for DUAL mode operation. This is an awful nice garuntee, I don't deny it, but it comes across to me as a marketing campaign made out of your own saftey net.

 

Quickly looking with software, I believe it correct that the pairs are NOT the same as one pair will list support for lower FSB speeds in the utility and the other pair only lists 200MHZ speeds (Sandra if you were curious). The pair with more timings listed (133, 166, 200) is the new set, the old set only lists one timing, 200MHZ (now I'm really not happy). But, is this enough evidence to support "not the same revision"? I really want to avoid taking RAM in and out unless it's getting replaced (if there's nothing wrong with the MOBO now, re-installing RAM endlessly sure will screw it up eventually). One pair was purchased in Oct 2004, the other a couple weeks ago. I think the likelyhood is high that they're not the same revision, but, who cares whether they are or not??!! I don't, and now I want to go back to what I was thinking in this whole mess...(BTW, I'll not be lower the timing any more than the default, it's bad RAM if I have to do that...no other explination, actually I'm already convinced it's bad).

 

Correct, the memory isn't dual channel, the memory controller is. It does sound like they are different revisions, but the only way to confirm that is to check the sticker. Also, ignore what the SPD chips are reporting to Sandra. Don't know why you're not happy about one pair listing only 200 Mhz, when that's the speed they were purchased for, but it doesn't matter. They will underclock if you so choose to do so, and yes, I understand you ultimately want 4 sticks running @ 200 Mhz, 2-3-3-6. Just clearing up the SPD issue.

 

 

Now, the way I interpret what you've said to me....

 

Well dude, sorry we didn't test your RAM in a quad set - we haven't done that because the market isn't there for these types of RAM requirements and as you have first hand experience - testing quads is a major butt puffer, we can't make money that way! It's much easier for us to just say, well - it's probably that fancy chipset not liking the RAM being *different*, even in a subtle and hard to reveal way. That, or your MOBO is junk. It's just NOT our RAM, no way. Besides, ridiculous performance isn't THAT big a deal man.

 

1. I don't work for Corsair at all. The only official Reps here are Ram Guy and Cool Guy (water cooling stuff).

 

2. You're reading WAAY too much into my post. Yes, your interpretation on why they don't test 4 sticks at a time is probably correct, in that it's not really asked for much. The ICs being different isn't a subtle difference, it's potentially very significant. Without knowing the revision and lot #s of the sticks though, the difference can't even be guessed.

 

 

The simple fact is, by Corsair's admission and predetermined choice - this second set of RAM is just a waste, or was it both sets...lemme go on a bit more, more gently...(I feel like I got stuck with a Lada, when all I really wanted was a Ferrari). Actually, the fact that the RAM does DUAL at all in this config in a somewhat stable manner is proof you guys can do this...

 

What I want is 4x512 MB sticks running in DUAL channel mode at 2-3-3-6, no more, no less. No!!? Not garunteed? So, NO!?? You're saying NO to money, now and in the future (since I'm in the hole here already)?!

 

Why!?

 

My take is that it is because Corsair can't actually make a profit making RAM that works like that, or worse - it's beyond Corsair's capability to make this stuff *that* realiable. Really - if you made RAM that was I wouldn't be here at all, further to that, it wouldn't be too hard to find 4 sticks that run together this way, would it?

 

Where are you getting that corsair admitted and decided in advance that 4 sticks is not worth it? Again, NO ONE said it's not possible to run 4 sticks, just that it's not pre-tested! THAT IS IT. Assuming they are different versions, they have different ICs. Running multiple ICs is kind of like talking between 2 people that both speak English, but different variations, like Britain English vs. Australian English. They're mainly the same, but there's JUST enough of a difference that for them to work together they have to slow down their speech rate just a tad. Now is it possible that they can run at the same timings, together? SURE. But again, IT HAS NOT BEEN TESTED. It is OUT OF CORSAIR's HANDS! They do NOT make ICs. In fact, I only know of one memory company that does.

 

Let's take it up a notch. Let's say you're putting something mechanical together that needs to be very precise, down to the mm. Two butting parts are made by 2 different manufacturers. While they may follow the same strict guidelines (in this case, JEDEC spec for memory design), they may not match up perfectly. If you REALLY want someone to blame for ICs not playing well together, blame the IC manufacturers for not working in perfect harmony with their fellow IC competitors, or better yet, the memory controller manufacturers for not designing their stuff to perfection. Ultimately, EVERYTHING in computers is an amalgamam of parts from a variety of manufacturers. I don't think you can really blame any one company for the issue you're having.

 

Now, no one ever said that Corsair would never test 4 sticks together, but looking at the forums and the low % of people asking about it, I'd guess that it's not going to happen any time in the immediate future.

 

 

What I was hoping could be done - and I think Corsair would win major points if you did this...exchange some RAM with me. Get me the "holy grail", though you didn't promise it I guess, you should be delivering it or PLAINLY STATING OTHERWISE!!!!! You can't be an industry leader in second, third - anywhere but first place, and it ain't a easy trip getting AND staying there. Besides, I paid a lot of money already - I almost expect it ya know. The RAM I have does not do what I want it to and I paid for RAM that does what I want it to. I guess I'd rather just through it in the garbage and get sets that will. Harsh?? Well, I'd rather that then either...

 

a) Drop down to single channel mode (which probably won't help anyway)

b) Use timing other than the ones I want (why, just - why?!!)

c) Lose 1GB of RAM (of which, what am I supposed to do with anyway)

d) See an itch of any error in any memory test, at least for a good long while

 

I also came to a conclusion in that, lol - if one stick goes bad on me, you replace both?? You have to, don't you. Or at least get them both back so you can FIND a stick that works with the original, that or just give me a pre-tested new pair and recycle. It'd be pretty grim to get a brand new stick back and find that, because of the replacment, I can't run the config I had set before hand....or does that happen too??

 

Well, as I said - this is what I want you to do to fix my 4x512 needs. And before you say, no way - not a chance, realize that I still believe Corsair can do this. I'd like to continue to praise Corsair and have others go, HOLY COW...I want another set of RAM now too! Maybe then you can make some money at this quad config, besides - other companies would go this mile, don't have me mention names now do I (that's just hideous to see in your forum isn't it)?

 

Correct, it's entirely possible that they may RMA your memory for you even if each TwinX pack tests fine by itself. I've seen RMAs approved for cust satisfaction before. Yes, TwinX packs are replaced by a NEW TwinX pack, and if one stick is bad, both are replaced. RMA'd memory is always replaced with new memory, and the version # is matched if possible, unless Customer Service overrides that I believe.

 

. The RAM I have does not do what I want it to and I paid for RAM that does what I want it to.

 

Did someone, somewhere, specifically state to you that the RAM would do what you want it to do, or is it something you assumed?

 

 

 

Now, let's assume you get an RMA, and that both TwinX packs are the same part #, version #, and hell, even LOT #. That's as matched as you can get. STILL doesn't mean they're guaranteed @ those timings together. In this situation, if you want to test intercompatibility between each pair, I'd take one stick from each TwinX pack and test them together. Now of course they haven't been pre-tested together, but it's a safe bet that they'll work together fine. I have no idea if Corsair's ever tried a test like that, but it'd be interesting to see what happens.

 

Now as for your testing, were you testing all 4 sticks at the same time, or each TwinX pack at the same time, or a stick at a time? Testing one TwinX pack at a time will help to quickly (quicker than one stick at a time anyway) identify if one pair or another has a problem at the guaranteed specifications. If they both pass like that, then the product is living up to what they advertise and guarantee. If not, well, then that's what a lifetime warranty is for.

 

Also, this next question isn't being asked to get around the situation or anything, but out of curiousity, why do you need 2 GB?

 

Also, the 250/66/33, I think that was supposed to be 200 like you tried it at. Probably just a typo.

Link to comment
Share on other sites

<EDIT>Awe, put it back dude - that was an awesome post!</EDIT>

 

And thank you for the very nice post...excellent read, excellent.

 

Anyway - I had a post just before yours...and for the record I'm sorry if I got a little steamed, I'm very tired - been at this for days...glad to finally have the answers I need and the help I asked for.

 

We'll see what happens now, and I'm sure Corsair will at least let me try - ONCE - an exchange to see if I can get them working. Agreed - it may not anyway...I got good expectations I guess, and am pretty optimistic, or I would have just junked it.

 

Why 2GB.

 

Lots of reasons. Photo editing, post video production (when I get to it), audio sampling, mixing and producting - games of course....gennnin'g primes...LMAO....catch ya later man, and again - thanks!

Link to comment
Share on other sites

What are the options now - both pairs are fine on their own, not together the way I want, need them to run. Please help me, I know this RAM will do it...just a matter of getting the right set....come on now, I've done my job to the letter. Lets see you do yours beyond the call, yeah Corsair!

 

Glad to hear the memory's running fine per specs. Ram Guy may approve an RMA for cust satisfaction, dunno. Keep in mind though that RMA'd memory is always replaced with new memory, but they try to keep the version the same, unless Customer Service overrides that I believe.

 

Now, let's assume you get an RMA, and that both TwinX packs are the same version #. The probability would increase with the versions being the same that they'd run @ 2-3-3-6, but the memory controller would still be a factor, as the more sticks you have, the more stress they are placed under.

 

Store credit would be up to the store. That's completely out of Corsair's hands.

 

And thank you for the very nice post...excellent read, excellent.

 

Anyway - I had a post just before yours...and for the record I'm sorry if I got a little steamed, I'm very tired - been at this for days...glad to finally have the answers I need and the help I asked for.

 

Thanks! It's no problem, we're all just trying to help each other out around here.

 

We'll see what happens now, and I'm sure Corsair will at least let me try - ONCE - an exchange to see if I can get them working. Agreed - it may not anyway...I got good expectations I guess, and am pretty optimistic, or I would have just junked it.

 

While of course I can't speak for Corsair, I can speak of what I've seen in the forum, and I'd say you have a better than even money's chance of getting a cust. satisfaction RMA. Kinda pity Ram Guy though for having to read all of this though LOL!

 

Why 2GB.

 

Lots of reasons. Photo editing, post video production (when I get to it), audio sampling, mixing and producting - games of course....gennnin'g primes...LMAO....catch ya later man, and again - thanks!

 

Definitely a good reason to have 2 GB. One of the guys around here, specmike, does lots of video production stuff. I believe he's dabbled in running 4 sticks in some of his systems. I'll PM him about this thread, see if he has any light to shine upon all of this.

Link to comment
Share on other sites

Hehe, editing posts with someone reading the original at the same time gets hinky at times lol!

 

PM'd specmike, also noticed his sig has this in it:

 

ASUS P4C800 D NON-E w/2.6c @ 250fsb, 3.2ghz DDR500w/4 X Corsair PC4000 Pro

 

Asked him for his input in running 4 sticks as well as video editing and memory. He's done a ton of research on this.

Link to comment
Share on other sites

  • Corsair Employees
I do apologize for the typo error that I have given you on the settings; yes the CPU external freq. setting should be 200/66/33 and not 250. If you have 4 modules installed on the system, then I will recommend you set it at 2.5/3/3/7, especially if you have different revision. I know that you have been frustrated for this past couple of days, but please try those settings; those settings should work on your system.
Link to comment
Share on other sites

I do apologize for the typo error that I have given you on the settings; yes the CPU external freq. setting should be 200/66/33 and not 250. If you have 4 modules installed on the system, then I will recommend you set it at 2.5/3/3/7, especially if you have different revision. I know that you have been frustrated for this past couple of days, but please try those settings; those settings should work on your system.

 

No apology necessary - I knew you didn't mean it (though I had a giggle when I thought maybe you were *trying* to get me to do that...jokingly of course...I have a strange sense of humor). The other configs were great, making up for the slight typo...a real newbbie though, they might have tried that...I'm sure the BIOS would just hang the machine...JUMPER RESET...,back in buisness, lol.

 

I will try the 2.5-3-3-7 (edit, had it wrong...holy cow) timings when I get a minute later as I'm *for now* running okay with 1GB at the lowest settings, all utilities report no errors thus far. I'm worried though, when I start hammering it with software (that's not so clean as these utils) that I'll be "jonesing" (wanting) that 1GB back in sooner than I can dbl click the icon to run the software. At that point I realized....will I really notice any difference between 2.5-3-3-7, and the 2-3-3-6....and I really think not (espec. with 2GB instead of 1GB)...

 

After coming through all this, reading lots of the great posts, I really think not. Truely though, I wouldn't mind if there was some sort of exchange offering since it's pretty darned obvious at this point that a matched revision stands a much better chance of working, even at 2.5-3-3-7 (since that's not much different than what I've already tried - with a smaterring of errors, I have little faith in that working well). Don't worry, I'll try it out - I've gotten pretty adept at switching sticks, though it still remains in the back of the mind - stress on those slots, not good...not at all....and it's very difficult to get ALL 4 sticks in there, VERY.

 

Also, as you've read - Wired knows someone who has accomplished this, blazed the trail ahead for guys like me...hopefully some more info will come into this thread, RAM_GUY - it may be a hidden market (cause it's so expensive) but there's a pretty big demand for machines running the 4 stick configs...though, understandably there's a lot of "big heads" involved from the factory, OS, software, hardware, etc, etc...Wired is bang on with his assessment of the situation....it's like a freakin' kindergarten...LMAO.

 

One serious question I hope you might be able to answer (I've taken up a lot of your time....sorry). A major issue I notice with 4 sticks is that such a config is seemingly disabling the PAT performance of the 875P controller at any timing, doesn't matter. Why? As pairs, PAT enables just fine. Please, any thoughts? (just thinking it's related, and if so - I need a set that doesn't disable it, if that is possible...hard to find info on PAT).

 

If you had a sec too, or anyone - why does test 7 fail in memtest86 so badly while all the others do okay...just curious (the random nature? the forward only limitation? what?)....it's so damn funny now with one pair, cause I can run memtest a good long time now, no errors...hilarious...(4 passes with the one pair - 5 passes with second - 3 hrs approx time).

Link to comment
Share on other sites

Is there any particular order these pairs have the best chance with...newest pair 1, 3 - older pair 2, 4? Or what?

 

For interest RAM_GUY

 

The RAM sticker identifies the RAM as follows:

 

first pair purchased Oct2004

 

0437093-3 Rev 4.1

 

second pair purcahsed Jan2005

 

0452050-3 Rev 5.1

 

(just the full monty, you know more about what those numbers mean than I ever would)

Link to comment
Share on other sites

  • Corsair Employees
You dont want to mix the modules, in other words if dual channel is slots 1-3 or 2-4 then one set that matches should be in slots 1-3 and the other in slots 2-4. And I would set the Dim Voltage to 2.8 Volts with 4 modules at Cass 3-3-3-8!
Link to comment
Share on other sites

You dont want to mix the modules, in other words if dual channel is slots 1-3 or 2-4 then one set that matches should be in slots 1-3 and the other in slots 2-4. And I would set the Dim Voltage to 2.8 Volts with 4 modules at Cass 3-3-3-8!

 

Understood - I meant what order the PAIRS go in though (of course I'll be keeping the pairs together, for sure). I was asking what order...newer pair in 1 and 3, older pair in 2 and 4 type thing, does it matter? Know what I mean?

 

If you have 4 modules installed on the system, then I will recommend you set it at 2.5/3/3/7, especially if you have different revision. I know that you have been frustrated for this past couple of days, but please try those settings; those settings should work on your system.

 

You suggest to run 2.5-3-3-7 eariler? Now you're suggesting 3-3-3-8, which does work, we've already got past that. I want 2-3-3-6 on all 4 sticks in dual channel mode with PAT enabled...even at 3-3-3-8, PAT does not enable and I have *some* system crashes and errors even with the higher voltage, all sticks in - though much fewer, errors nonetheless. Come on, after testing the 2.5-3-3-7 settings (which I have yet to do) I'd still be asking to switch up either the 5.1 for 4.1, or the 4.1 for 5.1 - that oughta to make it the most possible way...after that, I'm on my own - but for now, can't we agree on that?

 

Also, can I proceed with the 2.5-3-3-7 4x512 test? I'm a bit confused man... :confused:

Link to comment
Share on other sites

  • Corsair Employees

Sorry to confuse you, but since you are having a problem, and the modules to seem not to be the problem, and the fact there is more loading on the memory controller is the reason I suggested Cass 3-3-3-8. You are welcome to try any setting you like, but I think the goal would be to get a stable starting point and go from there. In addition, we do not suggest or officially support more than one set of Twinx in any system. Now that does not mean it cant be done, because it has, but there are to many other factors that may prevent the system from working with 4 modules that have little or nothing to do with the memory. For example with 4 modules you would need to consider about 48 Watts just for memory, so I would suggest a minimum of 450-500 Watt PSU depending on the system configuration. In addition, there is 20 loads per module on the chipset for each module and a MB will not run as well with 4 modules as it will with just 2 for the above reasons.

 

All that being said it's best to test the modules one and or one set at a time to isolate if you have a bad module. And I think you have done that but still have a problem. Is that correct? So I have to assume there may be some other issue and what I am suggesting is to try and isolate some other possibilities.

Also What is the make and model of PSU you have and please create a signature with your system spec as that would really speed things up in knowing what might be going on or what we might try next.

Link to comment
Share on other sites

Sorry to confuse you, but since you are having a problem, and the modules to seem not to be the problem, and the fact there is more loading on the memory controller is the reason I suggested Cass 3-3-3-8. You are welcome to try any setting you like, but I think the goal would be to get a stable starting point and go from there. In addition, we do not suggest or officially support more than one set of Twinx in any system. Now that does not mean it cant be done, because it has, but there are to many other factors that may prevent the system from working with 4 modules that have little or nothing to do with the memory. For example with 4 modules you would need to consider about 48 Watts just for memory, so I would suggest a minimum of 450-500 Watt PSU depending on the system configuration. In addition, there is 20 loads per module on the chipset for each module and a MB will not run as well with 4 modules as it will with just 2 for the above reasons.

 

Added Signature - hope there's enough info there, tell me if not, basically all of it besides disks. Is it safe to try 2.5-3-3-7 with all 4 sticks in? I have not tired that test yet.

 

All that being said it's best to test the modules one and or one set at a time to isolate if you have a bad module. And I think you have done that but still have a problem. Is that correct? So I have to assume there may be some other issue and what I am suggesting is to try and isolate some other possibilities.

 

Completed Tests:

 

All 4 Sticks 3-3-3-8 - PAT Disabled, 2.85 Voltage

 

Successfull memtest86 tests.

 

All 4 Sticks 2-3-3-6 - PAT Disabled, 2.85 Voltage

 

Unsuccessfull memtest86 tests. Fails test #7. > 100 Errors.

 

Single Pair (Rev 4.1) 2-3-3-6 - PAT Enabled, 2.85 Voltage

 

Successfull memtest86 tests. 4 Full passes.

 

Single Pair (Rev 5.1) 2-3-3-6 - PAT Enabled, 2.85 Voltage

 

Successfull memtest 86 tests. 5 Full passes.

 

Uncompleted Tests:

 

All 4 Sticks 2.5-3-3-7 - ?PAT Enabled?, 2.85 Voltage (Testing after this post)

 

Also What is the make and model of PSU you have and please create a signature with your system spec as that would really speed things up in knowing what might be going on or what we might try next.

 

Added.

Link to comment
Share on other sites

Anyway, board seems okay with 2.5-3-3-7 though PAT is disabled again...I believe this to be due to the revisions since the chipset knows this. I'm sure this will work with match revision...maybe even batch if available...

 

Can I exchange a pair now or what? I've done all the tests, I won't blame the RAM as bad entirely, just *different* - enough to be a problem for me. Please help me resolve this problem, at least give me one chance to exchange parts to see if I can get what I want...forget not, I'm ready to go get other stuff that will, and I'd be upset about losing THAT much money to this project - fault of my own or not...thank you!

Link to comment
Share on other sites

Also, as you've read - Wired knows someone who has accomplished this, blazed the trail ahead for guys like me...hopefully some more info will come into this thread, RAM_GUY - it may be a hidden market (cause it's so expensive) but there's a pretty big demand for machines running the 4 stick configs...though, understandably there's a lot of "big heads" involved from the factory, OS, software, hardware, etc, etc...Wired is bang on with his assessment of the situation....it's like a freakin' kindergarten...LMAO.

 

kindergarten? huh?

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...