My second kit of GT's after five months mostly of normal usage, started giving me troubles. BOSD and rebooting. Memtest v.4 reported lots of errors.
I just sent them back for a replacement.
What a disappointment for such an expensive kit.:(
Printable View
My second kit of GT's after five months mostly of normal usage, started giving me troubles. BOSD and rebooting. Memtest v.4 reported lots of errors.
I just sent them back for a replacement.
What a disappointment for such an expensive kit.:(
Another set of mushkin redline 2000mhz down the drain.
i wonder for how much longer we will see elpida kits go belly up... it seems some worked fine for over 6 months before they failed... and the problem was fixed almost 6 months ago... but some continued to use bad stock for a while longer, plus there was some inventory in shops that was never recalled... so i guess we will continue to see reports of dead elpida kits well within 2010 :eek:
really makes you wonder if we will see hardware degrade like this more and more often in future with smaller and smaller mfc processes...
OCZ Blade 2000 7-8-7 Kit buh bye. Third set to fail. At least this one lasted 6 months before failing. Never even took the things above rated specs. Was too afraid too. Guess I was prolonging the inevitable. Now there are none in stock for the forseeable future.
but that kit is from a "good old batch" isnt it?
or is it from one of the whimpy batches?
Yeah i suspect it's from very early/old batches indeed for GTs :) The kingstons no idea... been waiting till local disty has new replacements before i rma my kingstons.
well the khx2000cl8 kit that kingston sent me from rma now work great
it turns out that the msi eclipse sli and the foxconn bumrage gti dont support them very well
buy when in the clasy 760 they fly with low volts
have been running 7-8-7-20-69-1t @1.74v @ around 1100mhz +/- 20mhz
there still strong i only use them for benching so god knows how they would be on 24/7 lol
heres a hwbot cba to upload an image :)
http://hwbot.org/result.do?resultId=922871
Its a damn shame. I usually ran these things at 6-6-5-16-1T 1.60v @ 1600 for most of the time too. It only needed 1.24vtt set in the BIOS. 21x200 was no problem at those speeds. You just can't find memory like this any more.
I'm in the process of RMAing them but there is no replacement in stock, so I might give the 1600 Animals a shot if they have them...they are rated at 6-6-6 for 1600 and are supposedly 2000 binned.
I don't even care about 2000 memory any more. Too much of a hassle. I just want some good memory that can run 1600-1700 at tight timings so I can try to take this baby to 21 x 210 on air :)
Does anyone know if Microcenter was one of the places to have Corsair pull the hypers off the shelves? as i just ordered a kit of Dominator GT's from them. I don't want to get stuck with dead ram on my new build as i don't have any replacement DDR3.
Dragon
I don't think you will find any bad batches anymore. Besides, the Corsair GT's got a new type number after the bad batches so it's easy to confirm you got the right ones.
All the Dominator GT you can buy right now should be fixed, old stock has PNs starting with TR, the new ones start with CMG. :)
I can confirm that my Corsair 2000 cas 8 4gig GT kit is hypers :D, and they clock sweeet on AMD. I hope they dont fry on me. There rated for 1.65 what is the max safe voltage for 24/7 ?
I guess: 1.65V^^ :D
One module of my OCZ Blade 2000 CL7 died today :( Bluescreens and MemTest shows tonns of errors...
Too bad^^. :(
OCZ has no stock anymore and is replacing them with other modules or refunding the money.
Unfortunately, yes :(
Now i'm looking for a new kit with same performance. My OCZ kit ran at 970MHz CL6-7-6-18 1T @ 1,75V SuperPi32m stable.
What about the PI-Series from G.SKILL? e.g. http://www.gskill.com/products.php?index=251
Or something from corsair?...
Those G.skill are PCS, iirc, thus the high trcd and trp. I can only recommend Corsair's products, my GT 2000C7 are great. GTX2 should be even greater, if you can get a hold of them.
Okay. Yes i already found this one: https://shop.corsair.com/store/item_view.aspx?id=766411 but not in stock...
What about 3x https://shop.corsair.com/store/item_...spx?id=1454418 or 3x https://shop.corsair.com/store/item_...spx?id=1542371 ?
your second and third link are the same PSC of the g.skill PIS, they are all very funny but not the best choice if you want the best performance.
if you want good elpida hyper at decent price you should get the super talent 2000 cas7, 2000 cas8 or 2200 cas8.
they have the best value for bucks.
check in this section, here you can find many threads and results
Goodram Pro 2000 CL7 3 x 2GB kit with ofc Elpida Hyper still are strong. Mem is rated from 1.7 to 1.9 vdimm at label, running stron from some time 2k 7-8-7-21 at 1.74v
If You can get it, then thake it :up:
Had a kit of kingston hyper X 2000mhz 8-8-8 3gb die on me. No warning at all, PC refused to boot one day and it was the ram, eventually two sticks died before i could RMA.
I am sitting on another kit of the hyper Xs which i believe are MNH-Es and I also have a kit of Gskill Perfect storms which I know are MNH-Es and they are working perfectly
One of my sticks died recently :(
It was a MGH-E DIMM from my A-data DDR3-2000X cherrypicked kit.
I was using the kit on my everyday X58 setup at just DDR3-1066 8-8-8-31 @ 1.57V the last 5 weeks or so, and well... suddenly I started getting 4GBs of RAM in most of my power on/restarts.
I troubleshooted the issue, and found that one of the sticks is dead ( can't work at any other setting than... DDR3-1066 9-9-9-31 @ 1.62V )
wow after all this time heh...
I lost a stick of my mgh-e a week ago (t1 hyperX 3x1gb kit) they are run pretty hard though, usually around 1.8v. So now its a p55 benching kit ;)
so after all this time with the new type stuff
they still do this :)
http://i556.photobucket.com/albums/s...pretesting.jpg
wow 6/6/6/18 at 2000mhz, incredible!
man, 3:5 divider in realty is 2:3
your memory is running at 1800 MHZ
But what is, if he has run it in 200Mhz Strap? ;)
My RE can do this...;)
So you can't say by 100% that this is the bug...
But you are right, 400Mhz Strap and 2:3 is shown as 3:5 in CPU-Z...
what is the problem about hyper? I read on web that there are a problem about this ICs (or was a problem... )
lol my last stick of elpida hypers just died.
this one lasted 1 year 1 month 3 days running at 2100mhz 7-7-7-20 at 1.68..
and guess what no more elpida hypers on corsair website..
A sad day, mate^^ :(
Don't they replace them with hyper mgh-e's any more?
MGH-E is out of production unfortunately. Stupid Elpida. They better have something even better down the pipe :mad:
sorry to hear that
maybe i'm lucky but i still have a kit of A-Data 2133x MNH-E-Hyper and Kingston HyperX MGH-E-Hyper alive :D
orayt... i'll try some new rams next week...i wonder if it will be new "new" stuff from elpida...
Nah, you won't find any new high end IC from Elpida in any RAM out there.
Just pray that it won't be long until we have a new high end DDR3 SDRAM IC.
the lovo series i think will do ok...it will prolly die faster too once i push it hard lol
It's all on Elpida's EOL Products page mate! :(
the older die rev is eol...which is die rev "a"
they're using "d" "f" revs now. which means they're still in production i guess.
Why EOL something thats already good?
Unless they are switching densities I wish their engineers wouldn't be so r***rded.
how can they be retarded? the way i see it they're going for a die shrink from 50nm to 40nm and 35nm[later this yr]
Quote:
Reducing the size of the features in DRAM typically increases chip speed
and reduces power consumption, as well as lowering production costs as more chips
can be made from each silicon wafer.
Honda, MGH-E = EOL, trust me.
If you want MGH-E DIMMs rush it, and don't expect to see any GTX2's available again, so GT 2000C7 = instant buy button hit :p:
i'm not really looking for MGH-E.
looking @ sf3d's kingston hyper x ..they are EDJ1108BDBG DJ they're doing quite well. and they're the new die revision "D"
on air @ 1.84v ...if they were frozen maybe it can go faster. its still pretty good compared to PSC or Samsungs
I had a BDBG kit, Kingston HyperX too, but I can't say I was impressed by it for extreme benchmarking
The BDBG's are also EOL'd lol
they are eOLD lol
not sure if your rams were DJ / GN / GL Class....
sf3d's were DJ ...yours is probably class AE?
GN and GL ..will be better?
DJ, you can see it in the 2nd page pictures of the linked review :p:
nope no more
eventhough it was just one stick that was the issue
btw to that person who asked this was the second batch of elpida hypers after that temp worldwide hold april 09 batch
first batches were feb
u know its interesting
raja mentioned that the pcs strain the imc's more compare to elpida hypers
but i find better consistency with them on my e760 as to my hyper clocks
got some gskill pi 2k C6 atm.
and hmm was looking at corsair website
gtx1 looks like nanya
and gtx3 looks like pcs
so gtx4?? is the new revision of pcs.. the one that are doing 2k c6??
so any thoughts on gtx4 vs gskill pi 2k c6??
time to take out the p55 board again to check this
tweaktown i think the sample they had were rev 8.xx
thats nanya afaik
and the pic shown by a a guy in i4mem for gtx3's are 7.xx
which should be psc
tweaktown
http://www.tweaktown.com/popImg.php?...14_03_full.jpg
i4mem user snappy1
http://i4memory.com/attachment.php?a...p;d=1281257591
Then they changed chips...
Check matose's kit here. ver.7.1
A lil OT here ^^:D
Yes:D-but all my Hypers are still alive, so I concentrate on helping by answering questions:ROTF:
Which PSC did I say that about? The newer stuff is similar in all ways to Elpida apart from tRCD. I'm running 12GB of PSC here at 7-9-8 DDR3-1900 which is good enough. I think you are confusing this for something older or for the tRCD deficit (in that context the Hyper part is easier to drive at lower access latency).
Well the PSC and Elpida BBSE variants are ok for CAS, but worse than the original Hyper parts for tRCD. This was probably a pre-requisite for long-term reliability.
This is mine...Are they gonna to die?:rofl:
Sometimes fails on CL7 DDR3-1800+ on my PII X4 955, but works well at 2000 7-7-7 on my i7 920.:)
Running at DDR3-1685 6-6-5-18-24-1T 1.7v on PII X4 955 right now.
prolly amd imc... if with intel imc ok...
its the "MNH-e" that had problems
Iīm optimistic about my Hypers:D- no losses until now, and I cross fingers that it stays that way^^
Don't treat your MGH-E's carelessly.
They're not highlander.
They don't die as easily as the MNH-E's did, but they're not immortal as well ;)
don't go beyond 1.75v ...
1.65v-1.71v ....just to be on the safe side...
OK..Running 1.7v now. 1.74v just for benchmark
I have a lot in the closet:D- the ones who are in use at the moment work at 1600 8-8-8-24 1T at 1,3Volts- I think thats enough for 24/7- none of my Hyper have seen more than 1,66Volts for benching till now, time will come when I have to give a little bit more^^- but til now for 1150 C8 and over 2000 C7 this was enough
Yep... my guess is: hyper chips were not really made for continuous use of 1.65V+. So be careful with them, especially as they cannot be replaced like dead 980X. :D
Mine died ( 1 stick ) and the other 2 degraded while they were running at just 1.57V :D
The best advice ( as ridiculous as it might seem ) is Dino's.
If it's your only kit, the kit you're benching with actually, I would only use them when I bench.
Maybe... but I was assuming brainpower pcbs (many mf use) should not be the cause for it. If the pcb were the problem, wouldn't Elpida have told the manufacturers how to do it correctly?
Itīs a shame Goodram doesnīt sell in Germany- would have been fine to get a kit for testing them^^- especially as I have a real bad Corsair allergy:p: and it is hard to find good Hypers:D
a lot of memory is ref pcb these days... micron and elpida and samsung want to sell dimms, not chips... sometimes chips cost more than a complete ref dimm with chips on it... why? they have to create ref designs and dimms anyways for the oems, why pay for that and then enable other pcb makers to compete with you? instead they artificially increase the price of individual dram chips so it makes more sense to buy a ref dimm from them, which means their customers subsidize their ref design and pcb manufacturing :D
so i really dont think it was the pcb...
why would elpida even have gotten involved then... if customers use bad pcbs, thats 100% their problem, i dont think elpida would have cared...
Agree. Bad ICs are bad!^^ :D
Hope they learnt from it.
son of a biatch
all but one of my hypers which have been sitting in a draw for months have come back to life........i have a magic draw (like Marcus' CPU draw :D)
One stick of Hyper dead [Reaper 1866 C8], was runnign 1.63V at around 1800Mhz 8-7-8-20. Gotta RMA now but im wondering since OCZ dont stock hypers anymore what am I gonna get? Powerchips? :D
I know man. I sent them a ticket 2 months back when it died, they offered me OCZ Gold 1866 C10 (dunno which use they use but definitely a step down). When I told them I will not accept it and asked why they were offering me an inferior product, they said they didnt have anything left in stock and were developing new kits so he asked me to wait, which I did.
They should be offering powerchips at least. Thats the only replacement I can get for my hypers.
You mean they will offer me Micron? Sad. I had hypers, that I know :)
Well, they are offering me OCZ3RPR1600C6LV6GK, rated 1600Mhz 6-8-6-24. Should be an OK exchange I guess. I wonder what IC they use, most probably Micron but not sure which one
Sorry for the OT
My educated guess says that is PSC's bandwagon.