Printable View
:D Wish You nice holidays Mike - enjoy it - life is too short........:p:Quote:
Originally Posted by mikeguava
I've been in Venezuela in February - mid & south America is always worth to travel there......;)
SRY for :off:
Against all odds I managed to get these sticks run @ tRFC 60....:up::
http://img522.imageshack.us/img522/1603/trfc60gd4.jpg
:welcome:
I am not getting anywhere with my CellShock Blue :(
Before I installed the CellShock Blue, I had the following setup:
Asus P5E3 premium (bios 0503) + QX9650 + Corsair dominator PC14400
multi: x9
strap: 333
fsb: 433
ram: 7-7-7-20 1T
dram static read/write: enabled
clock twister: stronger
PL: 06
vcore: 1.41
fsb: 1.24
vdimm: 2.0
nb: 1.63
I have been rockstable with the above settings (16h+ hours prime95), see:
http://img403.imageshack.us/img403/6...pu39ao7.th.jpg
So I just replaced the Corsair dominators with the CellShock Blue, changed the timings to 8-8-8-16 and vdimm to 1.8V.
I immediately had tons of errors with memtest86+ in test #5....
And this is only for a frequency of 1730MHz :-(
To get rid of the memtest86+ errors I had to:
- go 2T instead of 1T,
- put dram static read/write to disabled,
- change clock twister to moderate
- drop the PL to 07
And I am still not completely stable because prime95 failed on 2 of the cores after 1 hour.
Also the bandwidth si so low compared to the dominators (went down from 11.5k to 10.3k!).
I have tried various vdimm from 1.7 to 2.0 with a step of 0.04V.
Memtest86+ test #5 gets better with more volts, but then errors start to pop up in other tests (#2 or #3à when vdimm > 1.9.
How can this be?
I just can't run the CellShocks Blue stable at 888 for only 1730 MHz :confused:
as i said via pm, the only thing i can think of is subtimings?
your results dont make any sense whatsoever to me... something must be very wrong...
id say start again from scratch, load setup defaults and then clock your way up.
make sure you use slots 2 and 4 and high vmch...
and try the sticks in single channel?
maybe you damaged one of them somehow?
Impressive DDR3, but i love CS D9GTR (PC14400) for drawn timings ;)
Thanks for your reply Saaya.Quote:
Originally Posted by saaya
I don't know what it is but as you said, something is really wrong...
The CellShocks can't do 1T even at 1700 cas 8...
I have not such problem with the dominators, which makes it difficult to accuse the MB or CPU.
I will tonite test each stick individually to see if the fault is really due to one of them.
I can't believe the sticks got damaged.
I received them back from RMA last friday and all was packaged very properly. I have been very careful with vdimm and I didn't go higher than 2.0V.
did you already try to set the bios to setup defaults after installing the cellshock mem?
Hmm.. Sounds almost like the issues I had with similar kit on P5E3 WS Pro.. :shrug:
Tested different mobos and eventually switched, that did it for me :yepp:
did you also use another memory kit in the same board that featured xmp?
i heard that when switching from an xmp kit to a non xmp kit there are weird bugs.. several people mentioned this already...
they reset the bios and suddenly it all works fine... hmmmm must be some memory timing id say...
I am still investigating.
The corsair dominators that I own are indeed the Intel XMP version, however I never enable xmp in bios and always set the timings manually.
I have tested each stick individually at fsb 466 8-8-8-16 1T and 1.86V, no problem.
When I put the 2 dimms in slot #0 and #2, I have to move PL to 7 and clock twister to moderate otherwise I can't post.
Then no memtest86+ errors but unstable system (prime95 fails or freezes my system within 10 minutes of stress).
It could very well be that my QX9650 has a fsb wall < 450, so I am not accusing the CellShock blue kit.
The CellShocks seem to be very sensitive to voltage, if I go higher than say 1.9V I will get memtest errors.
The kit cannot do 7-7-7 1866 no matter what vdimm and vmch is applied.
Saaya, you keep mentioning sub-timing, what would you advise me to change on the P5E3 premium?
your using the p5e3 premium and slots 1 and 3?
then that might very well be your problem :D
go for slots 2 and 4 man! :)
subtimings... i dont have a p5e3 premium... never had one...
id say check what mikeguava, sioux and metro posted, they all used the 1866 blue kits and the premium iirc :)
Thanks Saaya. Well I tried slots #2 and #4, sorry my post was not clear.
The best I could do is 1866MHz | strap 333 | 8-8-8-16 1T | PL7 to run with no memtest86+ errors.
This was with vdimm = 1.86V.
With vdimm < 1.8 or > 1.9 I get memtest errors so I would say the ram is on the verge of stability and very sensitive to volts.
My system is unstable at fsb=466 and I guess that's due to my QX9650 not handling high fsb so I can't blame the CellShock dimms on this one.
sure, i havent seen many qx9650s do better than that fsb wise... this is pretty much where they max out
hmmm.... ok, now try 777 with 1800 and work your way up, and see how high you can push vdimm :D
you should be able to reach about the same speed with cas7 :)
its normal that with 2,0v @ 1866 memtest give me a lot of errors ??
mobo is x48T-dq6 and settings are 467fsb divisor 1:2 strap 400 (4,00D) PL 8 and overvoltage to NB is +0,3v
have you tried with lower strap? set it to 333.
depends on the chips... some like 2v, some dont... and it depends on the timings you run, the difference between the max vdimm for relaxed timings and max vdimm for tght timings isnt as big as for gtr/gts anymore, but its still there.
and it also depends how much vmch and what slots your using.
on asus boards slot 1 and 3 dont scale with more vmch and act... weird... with vmch and vdimm bumps
slot 2 and 4 scale properly with more vmch and vdimm :)
on gigabyte boards 2 and 4 are better as well afaik, but they act weird with more vmch, like slots 1 and 3 on asus boards
on the blackops ive run some sc tests and it seems all slots clock really well
lol but on sticks there is write : 1866 8-8-8-16 2,0v id they dont run @ 2 volts at that speed i think i go for rma.
2,0 is garanteed voltage...
yeah there are definitely excellent sticks and not so good ones, that buyers like you and me will get...
I have already gone once through rma and received a replacement kit that is still miles away from the expected performance.
And I can tell it is not my CPU/MB/whatever because my corsair dominators totally outperform the CellShocks on the same system.
i thinking that black ones are better...
so, 2v is not guaranteed, and 2v isnt 2v anyways, on some boards 2v in bios is 2.1v real...Quote:
Originally Posted by aerie
thats the voltage range you can use without losing warranty, it doesnt mean the memory will scale up to that voltage.
more vdimm isnt always better, ive had some sticks that could run close to 2000 cas7 with 1.9v, but more than 1.93v and they only gave errors...
now think about it, what do you want?
sticks that can run with a certain voltage or sticks that can run a certain speed?
if all you want is sticks that run with 2v and you dont care about the clockspeed, go ahead and ask, and im sure they can send you some :D
and you know what i think? there are people who know how to push their memory and reach really nice results, and there are people who expect their memory to be plug and play and get the same results or even better than the best results they have seen with this memory so far, and if they dont, they will claim its bs and the other results were with handpicked hardware.Quote:
Originally Posted by MateoTTR
strangely your the only guy with this problem...Quote:
Originally Posted by MateoTTR
corsair has very well programmed plug and play spds, and yeah, they have some nice memory as well... if you have a quadcore cpu which limits you to 1866max clocks, and you can run 1800 cas7 with your older corsair kit already, then why did you buy a JNL based kit anyways?Quote:
Originally Posted by MateoTTR
buying a gtr/gts based kit would have made sense to try and get a good cas6 result, but JNL doesnt run that well with cas6...
some gtr clocks better than jnl, by far not all gtr/gts are as good or better than jnlQuote:
Originally Posted by aerie
i saw review that was screeming 1000 c7..... i cannot boot in anyway at that speed...
i saw that if i stay under 1600 memories can stay @ 2,0v but if i go near 1800 with 2,0v its totaly unstable and i must use 1,95v.
but with 1,95v mem dont boot at 1000 c7 but 1000 c8 yes...
i thought this memory for theyr cost 270 euro will be better... 270 euro.... i saw people with ocz 1600 tha out perform ,y 270 euro ram with 120 euro rams.... cant beleive that
This is just unfair. I have played around all the timings, sub-timings, vdimms, PL and all. This is not what I call plug n' play.Quote:
Originally Posted by saaya
The memory kit I have makes memtest errors consistently, can't run 1T, can't run cas 7 even at 1700, this is just fact.
If I replace the CellShock blue with the dominators, I can increase fsb, tighten the timings, go 1T and cas 7, and I have no memtest errors and I am rock stable.
Thanks for arguing that I am noob that can't handle the tweaking of memory. I am certainly not a pro, but neither I am noob expecting plug&play style.
Let me remind you that I got the jnl kit as a replacement by CellShock RMA, while I originally owned the gtr kit (8-7-6-21 1800). So I never wanted to buy the d9jnl kit.Quote:
Originally Posted by saaya
Some really great info in here fellas! Im learning alot about DDR3 since I will be doing my first benches with it this week (790i + E8500 + 8800gt SLI & 8800gts 512 SLI).
Sayaa I especially appreciate your explainations of how to find and edit registers! I have been trying to learn about this subject also, but most public info is about 965 strap tweaks.
aerie: There is no guarantee your RAM will run at tighter timings and higher clocks than rated. You stated a reviewer who had it at CAS7 @ 1000Mhz when the kit is rated CAS8 @ 933Mhz. Just because some people achieve these results is absolutely no guarantee every kit will. Just be happy you own a kit thats selected from the top 20% binned kits! Also getting your RAM stable at such settings requires alot of patience and tweaking, not of just your RAM timings/voltages, but also chipset/mobo voltages.
1866mhz 8-8-8-16 2,0v always blue screen nice....
and i am not a noob i have 150 points and i know how to manage a motherboard
my bad... if you cant get the kit to run fine no matter what, then send it back and get it replaced. i just find it very hard to believe that 99% of cellshock customers dont need an rma and you happen to have 2 bad kits in a row out of bad luck... but maybe you ARE that unlucky :PQuote:
Originally Posted by MateoTTR
in this case, the thid kit HAS to be good, cause nobody can be so unlucky to receive 3 bad kits in a row with the high testing standards cellshock uses.
your welcome :toast:Quote:
Originally Posted by TheKarmakazi
pm before and felix88, they might be able to give you some hints and tips on tweaking registers :)
exactly! it seems a lot of people look at the best results somebody managed to achieve with one piece of hardware and then think every piece of this product line can reach the same speed/performance...Quote:
Originally Posted by TheKarmakazi
and even the best kit out there might suck on one board and rock on another board... and as i said plenty of times, it doesnt even mean the board is bad, some boards work well with one kit and suck with another... theres a lot of weird variation from one board to another... between 2 good boards you might see a variation of up to 50mhz for one specific kit, another kit might work well on both etc...
why are you bumping 2v through the mem for just 1866 cas8? they should be able to run that speed with 1.8v or even less...Quote:
Originally Posted by aerie
Thanks for considering that it might actually be a problem with the kit, and not me/my mobo or my CPU...Quote:
Originally Posted by saaya
I agree and admit it seems impossible to receive two 'bad' kits in a row when the quality testing at CellShock has such high standard and good reputation.
This is why I keep on testing the kit on my system (3-4 hours each evening) to be 100% sure before I send it back.
I am exhaustively testing sub-timings and clock skews for a wide range of different vdimm in 1.7V-2.0V.