XMS 3200 C2 512 Results

[mdzcpa]

Well, to sum up the results in one word it was....disappointing:(

I'm not sure if it was just a "bad" stick, or what, but it just didn't perform too well at all in my KX7 rig. After I popped out my 2 256 sticks of XMS 3000 C2, I slipped the new 512 stick right in (DIMM 1 to start)...with no other changes.

For reference, my 2 x 256 of 3000 C2 would go to 197 and be able to loop 3DMark endlessly at that speed, and be make banzi runs well above that (up to 203 or so would loop a few times before CTD). Also those pc3000 sticks could loop the Sandra Mem bench continuously up to about 203mhz. All of this was at a modest 3,6,3 2T but put the memory at about 3035/2820 at 202mhz with good stability. Using only one stick of the 256 XMS 3000 was good for another stable 4-6mhz accross the board on most tests.

The new 512 stick would go at CAS 2:
185 at 2,5,2 1T
190 or so at 2,5,2 2T
192 or so at 3,6,3, 1T
and 195 or so at 3,6,3 2T

Anything above that would bomb out on the Sandra Mem bench and/or not complete 3DMark without a freeze or CTD. I refused to run the memory at anything less than CAS 2 as I would never make that up in FSB with stability on a 24/7 all purpose rig (at least in my experience).

I tried every combination of voltages and timings to no avail. I tried different DIMM slots. Nothing worked. I was tempted to do the Vmem mod, but the fact that I couldn't get the mem to run even close to it's advertised CAS 2 3,6,3 200mhz at anything from 2.55v to 2.85v, I didn't bother.

After slapping my PC3000 back in (DIMMS 1 and 3), I immediately went right back to 197mhz and ran 3DMark in continuous loop, then memory benched it in a loop at 203mhz.

Next ,we put the new 512 stick into my buddies rig (KX7-333R & XP2000) and it did no better. wtf?

In the end, my buddy swapped me some parts in a trade for the new 512 stick. He doesn't have the need for very high FSB speeds with his Radeon 9700 right now and the new 512 went high enough for him to max his stuff out...so an RMA was avoided. But...even so...the stick did not even come close to advertised speeds and should have went back to New Egg by all rights.

I will say that seeing everyone elses success with this stuff makes me think I just had a bad stick or something. There is no other good explanation after not getting it to work well in two different boards.

Oh well....I guess I'll just order up whatever the best memory is out there when the next mobo order is placed in a few more weeks:)

[TheHunter]

That's some bad movie u had there mdzcpa...
Did u tried burn in the memory? i'm sure that would help.

[mdzcpa]

I'm thinking that some burn in would help. But, seeing I couldn't even come close to 200mhz at the advertised C2 3,6,3 1T (or even a less agressive 2T) I didn't think burn in would get me far enough to be satisfied.

In any event, the guy who has the 512 stick now is a close friend, and he will swap me for my 2 x 256 XMS3000 whenever i want to. So he is kinda doing the burn in for me now. The next time he pops over for some LAN gaming, I'll swap the memory back in and see if some burn in time has helped.

If it does...I'll be sure to report back:)

[twiggy]

wow that sux. i would have sent it back 4 sure..

[Marci]

I'm using the platinum edition version, which may be nicer to play with than what you had... I have no probs doin 200 at 2-5-2-1T @ 2.8v

[OPPAINTER]

Originally posted by Marci
I'm using the platinum edition version, which may be nicer to play with than what you had... I have no probs doin 200 at 2-5-2-1T @ 2.8v

Yea I think the difference is the platinum, I have the same version as mdzcpa only it's a 256 stick and it doesn't impress me, I never saw the word Platinum, the other day I saw a stick in my local store, Frys, and notice they had stuff called Platinum, it could be the hot Idem :D I think I'll try a stick.

OPP

[newhit]

Just curious but shouldn't these XMS sticks run 200 at default volts? Does the spec call for increased vdimm?

PS
mdzcpa, why don't u use your colleague farabombs good stick and test that against your problem stick? Also, from overclockers.com forum there is a number (4-5) of ppl also having trouble with the 512 sticks not reaching advertized speed. Not sure of the implications here but maybe there is food for thought.

[newhit]

Back again :)

Marci, the test spec on over-clock.co.uk shows the test done at 185 fsb. Is that a misprint?

[bmg]

Did you try 2-6-2-3? That works fine for me. I notice quite a few others use 2-6-3-2, which doesn't work above 200 for me. 2-6-3-3 should definitely work at 200, and I'd send the stick back if it doesn't.

[Bushboy]

That's strange because my XMS3200 C2 stick here is currently doing 217FSB 2-6-3-3 @2.7V. On this Bd7-II there's no option to select 1T or 2T so I don't know what it's running at:confused:

[TodB]

Originally posted by Bushboy
That's strange because my XMS3200 C2 stick here is currently doing 217FSB 2-6-3-3 @2.7V. On this Bd7-II there's no option to select 1T or 2T so I don't know what it's running at:confused:

The difference is that you're running an Intel based MB and he has AMD based - those AMDs are far more agressive as far as the memory goes.....

[mdzcpa]

[QUOTE]Originally posted by newhit
Just curious but shouldn't these XMS sticks run 200 at default volts? Does the spec call for increased vdimm?

Theoretically, the memory should do the advertised C2 3,6,3 1T at default voltage although I have no issue increasing the Vmem at bit for added stability.


PS
mdzcpa, why don't u use your colleague farabombs good stick and test that against your problem stick?

Would be an okay idea except that we live a good distance apart and the mail time/expense prolly wouldn't be worth it at this point. But...always still an option if I feel like investigating further.


Also, from overclockers.com forum there is a number (4-5) of ppl also having trouble with the 512 sticks not reaching advertized speed. Not sure of the implications here but maybe there is food for thought.

Thanks for the extra info. This may be an important trend. I wonder if they were ABit mobo users (Abit does tend to have odd memory timings having mobos with that 4th DIMM).


UPDATE-
Okay...my buddy came back again today after having had the 512 stick in his rig burning away for 2 days solid and it seems the burn in has helped (much to my surprise).

Now the stick will run to 198mhz at C2 2,5,2 2T...but it takes no less than 2.85v to do it.

1T still gives the system fits at anything above 185mhz regardless of voltage, but now I can at least run 2,5,2 close to 200mhz. It will now hit 200mhz at C2 3,6,3 2T. Still not what is guaranteed by Corsair though...but better.

If it get's any better...I'll let you guys know. Looks like I should have zero'd in on the "platinum" stuff.

[MrIcee]

Mike:)

A few thoughts my friend from my observations.

I haven't dabbled with the 512MB stick but have many hours with its counterpart, the 256MB stick.

The timings on not only the Corsair, but many other manufacturers Ram, are qualified on the AMD side at least, on Epox boards, and in particular the 8K3A+. What I found initially with my stick...is it strained to run the advertised timings at 200Mhz...and at that 2.85v. Couple of reasons there.

The Abit KX7 is much more aggressive in its register settings regarding memory timing calibration and the associated bandwidth it provides. I have found in all sticks I've played with and evaluated that an extended burn-in period was necessary...along with gradual increases in voltage during the process. These Winbond chipped sticks love high voltage....no doubt about that, on the Abit KX7...at least. At 3.25v this stick after extensive burn-in ran all the way to 226Mhz as you've probably seen in my bandwidth thread, max timings..2T command..and benched 3DMark at those settings at 216Mhz FSB.

1T command demands even higher voltage at lower FSB....but with a 1700+ XP aboard I ran as high as 210Mhz max timings 1T command in 3DMark benches. 3.38v was necessary to keep it all stable. I know alot of you guys will be alarmed at what I run...but I've not had a failure yet:) But I do push my equipment to the xtreme...these are not what the typical guy is going to run.

But the reality is...I believe ram manufacturers are finding it difficult with DDR 1 to run at that speed with max timings at DEFAULT voltage. It takes a very special stick to do that...there are probably a few out there that will. But the bulk need the patience and TLC of extended burn-in and high voltage to perform at the xtreme...and most sticks will.

The issue of voltage is, I believe, the reason you will finally see higher VDimm choices on upcoming Abit boards...much as already has been included on Epox boards and possibly others. The reality is these DDR 1 PC3200 sticks need more than the normally accepted "stock" voltages to perform on certain boards. This fact does not make any particular stick a POS...nor in my extensive testing have I experienced a failed stick from using high voltage, but again, I have always used heat spreaders and provided good airflow over the stick.

I did achieve Corsair's and OCZ's timing specs at 2.85v after burn-in of the Winbond chips...but more aggressive timings and/or xtreme FSB demanded higher voltage.

Randi:D

[Trevor]

Originally posted by MrIcee
The Abit KX7 is much more aggressive in its register settings regarding memory timing calibration and the associated bandwidth it provides.
I thinks this is very true, corsair test their ram on "EPoX EP-8K3A and/or Asus P4S533 and/or EPoX EP-8K5A", I found that with my stick of xms3200c2.5PT, at the same voltage, a huge difference between the 8k3a and the kx7. That being said, I would exchange it until you get a stick that out of the box runs at its published specs :rolleyes:

[mdzcpa]

Thanks for sharing your thoughts Randi:)

I'll give this memory more time to burn in. Perhaps I will do the Vmem mod to this KX7 based on what your sayin' (my old fully voltage modded KX7 died and I only have the Vcore done on this one so far).

I'll be sure to update everyone if I can wring more out of this Corsair XMS 512 3200 C2 :)

[OPPAINTER]

I went and picked up a stick of this Platinum stuff today, the old Corsair runs about the same as this ecxept the PT is running at Cas2 compared to 2.5, same timings. Of coarse this is on an Intel IT7 doing 230Mhz, I have now way to really test the stuff untill I get my AMD going.

Looks Cool though:D

[MrIcee]

OPP:)

That first stick you got that is pictured is not the same part # that I have. My Part # corresponds to the Platinum stick you have..XMS3200C2 Version 1.1. Of course I've got the black heat spreaders..but still rated cas 2.

Randi:D

[Bushboy]

Originally posted by MrIcee
OPP:)

That first stick you got that is pictured is not the same part # that I have. My Part # corresponds to the Platinum stick you have..XMS3200C2 Version 1.1. Of course I've got the black heat spreaders..but still rated cas 2.

Randi:D

I noticed that too, though mine is 023124. Perhaps that's the reason you haven't had much luck with yours, and also mine has C2 after the 3200.

[TheDude]

Originally posted by MrIcee
OPP:)

That first stick you got that is pictured is not the same part # that I have. My Part # corresponds to the Platinum stick you have..XMS3200C2 Version 1.1. Of course I've got the black heat spreaders..but still rated cas 2.

Randi:D

My 256 is the same as yours, Randi XMS3200C2 Version 1.1.C2...black spreaders

[OPPAINTER]

Well there you have it, I got ripped off online:D I hate buying online unless I'm doing a will call, of coarse when I got the stuff I wouldn't have known what to look for anyway:D

OPP

[QuadDamage]

for reference, my XMS3000 will do around 203fsb@most aggresive timings and 3.15V, around 218-221fsb at cas2/3/3/6/2T, but AGP tearing is bottlenecking me at these speeds. the same stick will do 243fsb on IT7Max2/Pee4 @ 3.0V. i haven't tried XMS3200 yet.

[newhit]

Originally posted by OPPAINTER
Well there you have it, I got ripped off online:D I hate buying online unless I'm doing a will call, of coarse when I got the stuff I wouldn't have known what to look for anyway:D
OPP
Just a bit of info i picked up at Cosairs forum. The "RAM GUY" states that the only reason for the "platinum" spreaders is that Soyo wanted them for use with their "Platinum" board. The thread is here.......
http://www.houseofhelp.com/v2/showthread.php?threadid=7743

I respect all you guys doing your volt mods and burn ins but are'nt u slightly annoyed that this is necessary. I want to buy some of this XMS stuff but am really getting put off with what's needed to get them to perform. Or is it just an "Abit thing"?

[newhit]

Of course i mean perform within spec, not overclocking.

[majormav]

Originally posted by MrIcee
OPP:)

That first stick you got that is pictured is not the same part # that I have. My Part # corresponds to the Platinum stick you have..XMS3200C2 Version 1.1. Of course I've got the black heat spreaders..but still rated cas 2.

Randi:D just a compare between the 2 sticks as i have both

[Torinalth]

I have used the 3200 3200c2 (both256 and 512)

sadly my amd rig is in pieces till I get out of debt, but my intel rig will push cas2-2-5-2-2t (cant figure how to make it go 1t) at 164 on a 3:4 ratio on all three sticks (default voltage) the cs and standard are the same as far as I can tell, though for some reason my 512c2 will let me go 171 (kinda disapointing, as that stick was for a clients PC)

also a note, ever since XMS ram came out the corsair mem loves voltage. my 2400 ran to 203 FSB on my kr7a at max timings and loved it with the 3.4 volts. they take everything you can feed and will never give it up. true everyone wants the ram to work at advertised speeds on stock settings, but not everything is perfect.

Torinalth

[Marci]

Mine perform worse above 3v... if I keep voltage to 3v then the ram works ace, but as soon as I go above 3 it starts giving me probs...

[toofast]

Just thought id toss what my lable details in from a stick i got yesterday..a bit different to you guys`s

sorry no pic its burning in atm

CMX256-A 3200 C2 "PT"
XMS "3202" V1.1
23514
Its got platinum heat spreaders on it

[CrawlingEye]

Odd, you have a 1.1 revision and it won't OC well?

[Marci]

OK... don't hold too much value in what Corsair put on the stickers. I labelled the XMS3200 on our site by it's part number, XMS3202, and got a call off Corsair telling me it was a misprint. Even the revision number is fairly irrelevant. The platinum stuff is handpicked standard XMS3200c2 that'll do a bit better than the black heatspreaded counterparts. The stickers are a rush job, with little relevancy other than they are handpicked v1.1 sticks.
I've had the spreaders off the 512 XMS3200c2PT and it has the exact same Winbond chips as my previous stick of XMS3200c2.5 & my XMS3200c2, on the exact same PCB.

[Marci]

newhit: cant find the mhz thing or test specs on my site that you're on about... where u looking?

[xgman]

I have a new 512 xms3500 corsair on the way. They have now moved to -5 winbond chips like the twinmos. It is rated at 433 at 2-3-3-6-3-1T, so I'll burn it in and we will see. Newegg has them.

[Marci]

still waitin for em to appear in UK... they should land tomorro or day after I think so my order'll get to me Monday... can't wait...

[newhit]

Originally posted by Marci
newhit: cant find the mhz thing or test specs on my site that you're on about... where u looking?

On the "Ram" page when selecting "Click here for full tech specs" a pdf comes up and when scrolled down there is a reference to the ASUS/Epox mbds and testing to 185MHz.

[Marci]

That's the official tech spec on CORSAIR's site, NOT mine... and it says 200 m8...
and I quote...


Each Module is tested in either Epox EP-8K3A and/or Asus P4S333 at 200Mhz... - 512Mb Specs (http://www.corsairmicro.com/main/products/specs/cmx512-3200c2.pdf)

ah, just spotted it on the 256 specs... still, not my misprint... speak to Corsair...
http://www.corsairmicro.com/main/products/specs/cmx256a-3200c2.pdf