:clap: owiQuote:
Originally Posted by owikh84
Printable View
:clap: owiQuote:
Originally Posted by owikh84
3770K batch 3218B957 has a strong IMC
2666 11-13-13-25-1T 1.56v
http://i458.photobucket.com/albums/q...266611156v.jpg
Still struggling to post on 2800, code 23, 79, 30...
Both my 3770K's post at 2800 but i can't run 32M. I have tried both VCCSA and VCCIO up to 1.2V but still no joy. Highest i can go is 2720. Tried different bios' too to see if that was holding me back but that didn't help either.
Gotta go cold Zeus, it will open up imc for chip with (at least) 2800+ bootable air/h2o
Guys im sad having a problem i dont know what happened as i tried all
my OC is 4.8GHz HT@ 1.325 and Gskills 2800 stock = 11-13-13-35 1T @ 1.65
Some days ago PC wont start i have code 23 always and i turned off PC starts and i get "Press F1 OC failed..."
Its like BIOS get locked on timmings and no matter what i do they dont change unless i change mem speed, my IMC is strong i can boot direct from BIOS at 2933 so i dont know
So theres nothing i could do to make it start
- increased VCCSA no...
- increased VDIMM no...
- flashed new 1204 BIOS no...
- clear CMOS no...
- increased vcore no...
- changed relaxed did all with timmings no...
- run stock no...
So after all this i said OK mem are dead...so i tried 2600 with profile samsung and PC started perfect so i went and test HCI memtest DOS for 1 hour and passed...so mem are OK
then i just put 2800 adjusted timmings and PC perfect, restarted, turned OFF and start perfect BUT i leave all nite and the other day the SAME crap
Funny thing i can reboot, turn off, take current out, leave it 4 hours off and will start, but overnite more than 8 hours dont...
im just lost...can be BIOS ? mobo ? or mem ?
Thanks for your help in advance
Sergio
Corrupted OS possibly? Did you try a fresh install?
Thanks friend no aint OS its on POST...i now running them perfect at 2800 stock but seems i find a clue i touched the DRAM ref CYCLE (not trfc) and lower to 7936 and seems to affect a lot will keep testingQuote:
Originally Posted by deachus
i remember once i have a pair of crucial and one of them was dying and need more volts to start but PC didnt start at all, this ocassion gives ALWAYS OC failed press F1...
I picked up another chip. I'd really like to know for sure which ic are "under the hood"... Samsung or Hynix? I'm starting to think maybe Hynix. :shrug: Still no luck yet getting to 2800 (perhaps some more binning?) so I've been running/testing 2666.
This might help you mate its from the GSkill 2666 C11 thread & is thanks to "theix"Quote:
Originally Posted by PolRoger
0140 = Micron
0000 = Hynix
0240 = Hynix
0360 = Samsung
0640 = Elpida non-Hyper
0650 = Elpida (seen on blue PCB Tridents)
0660 = Elpida non-Hyper
0690 = Elpida Hyper
1040 = PSC
1240 = Nanya (?)
http://www.xtremesystems.org/forums/...=1#post5101204
PolRoger, Trp=11 is 100% sure Hynix. Samsung cannot run Trp that tight at 1333. Most 1400 C11 stuff is Hynix. 1200+ C9 or 10 is usually Samsung. This is why i went for the 2600 C10 Trident-X, it's samsung.
Here is newer G.Skill IC:Quote:
Originally Posted by C-N
0000 = Hynix
AB40 = Hynix/Samsung (?)
0140 = Micron
0240 = Hynix
0260 = Hynix
0340 = Samsung
0360 = Samsung
0640 = Elpida non-Hyper
0650 = Elpida (seen on blue PCB Tridents)
0660 = Elpida non-Hyper
0690 = Elpida Hyper
1040 = PSC
1240 = Nanya (?)
1400 = Hynix
1500 = Samsung
1600 = Hynix
2400 = Hynix
99.99% Trident 2666/2800 is Hynix.
2666 Samsung atm only Corsair and TeamX.
Yes after seeing Dumo's posts on the Corsair Platinum C10/C11 (Samsung/Hynix)... I then tried running 2666 C10 with Samsung like settings but it seemed to be a no go.Quote:
Originally Posted by Zeus
My kit has a "2400" series serial#...Quote:
Originally Posted by stasio
I'd love to score a set of the "high-bin" Corsair/TeamX Samsung based 2666 kits to compare with this one. :up:
I'm thinking that 2666 C11-13-12-28-1T might be a sweet spot for my combo. I'm looking to dial this in for a possible 46x daily stable o.c.
http://i695.photobucket.com/albums/v...R3-2666C11.png
Very good pol !!!Quote:
Originally Posted by PolRoger
Stasio that we have to look the IC right i mean take out HS ? or that is the serial nr ?Quote:
Originally Posted by stasio
Even,G.Skill has engraved the IC's with "GSK" to reduce recognition,some IC's can be read.
Btw,no need to take out HS,as per serial and list above you will know.
And what Zeus say above,if tRP can be lower then tRC (eg..2666 C11-13-12-28-1) must be Hynix.
samsung also do very low tRAS, ive been running my 2600 TridentX at 9-12-12-17.
Well seems i solved the issue i was pushing to hard Lat. Boundary and other timmings but still im not sure i was this, i just put VCCSA on 0.95 and VTT 1.1 and all went back to normal PC now resets perfect no errror code and was all nite off and turned off perfect at 2800 11-13-13-35 1T LB = 8
happy again :)
fixedQuote:
Originally Posted by stasio
Here is info from where I get AB40=Samsung (no pic).
Btw,
one of indication on the IC itself
if "GSKGSKGSK" are in 3 rows is Samsung
if "GSKGSKGSK" are in 4 rows is Hynix
http://mods.hardwareluxx.de/emissary...stunnedguy.jpg
http://www.ocaholic.ch/xoops/html/mo...D-16GTX_25.JPG
trans am posted this in one of the TridentX threads
Attachment 130069Attachment 130070
i also have single sided F3-2400C10D-8GZH AB40 (timings indicate Hynix ICs)
I don't know how i could have been so ignorant all this time. Today i finally tried the red slots and i could not believe how much better those are!
DDR2800 suddenly seemed so easy:
http://i.imgur.com/ss26x.jpg
With more volts Cas 10 should be within reach i guess.
Clearly, based on post #120 above, at least some versions of "AB40" are Hynix, as are some of the 3-row versions of the "GSKGSKGSK". I'm beginning to think that GSkill is just having fun with us and is mixing them all up in order to confuse us...Quote:
Originally Posted by stasio
Along the lines of GSkill trying to confuse us...I've also recently gotten some GSkill (Sniper DDR3-1866 9-10-9-28 @ 1.50v) with the "2100" numbering in the 5th-8th spot, so I guess that's one more to add to the list of unknowns. I can't match up the edges of the ICs to anything that has been posted so far. Once I test them perhaps I'll have some idea, based on their behavior.
@Zeus
Nice work getting your Samsung based kit up to 2800 :up: ... I wonder why the red slots on MVGene clock better than the black? :shrug:
Nice dimms Eugene !! Try 1.85 for 10-13-12 although I wouldn't waste them, Samsungs can suddenly die out of the blue... 2800C11 is already nice and overkill for 24/7 :D
Great Zeus CL10 but just for a benchQuote:
Originally Posted by Zeus
btw what Latency Boundary are u using or u are adjusting third timmings manual ?
this is my 24/7
http://i.imgur.com/6ZeK9.jpg
Thanks guys, i have to say i was really surprised.
I have absolutely no clue why the red slots are so much better. The black slots hardly got over 2600.
Albrecht, i know Samsung can be pretty vulnerable, i remember this from the old TCCD days. So i'm a bit reluctant to try high volts even though i know some have already pushed 2V into sammies.
Sergio, this was LB 6 iirc, haven't messed with third timings yet as the screen i posted was the very first try for 32M at 2800.
I'm afraid you're right... :rofl:Quote:
Originally Posted by Reefa_Madness
I haven't seen this number yet, but 1866 9-10-9 @ 1.5V could be Micron D9PFJ.Quote:
Along the lines of GSkill trying to confuse us...I've also recently gotten some GSkill (Sniper DDR3-1866 9-10-9-28 @ 1.50v) with the "2100" numbering in the 5th-8th spot, so I guess that's one more to add to the list of unknowns. I can't match up the edges of the ICs to anything that has been posted so far. Once I test them perhaps I'll have some idea, based on their behavior.
You're confused? Here's more confusion:Quote:
Originally Posted by Reefa_Madness
http://img.photobucket.com/albums/v6...ps1fac7758.jpg
It's probably some sort of Samsung C-die chip:
http://img.photobucket.com/albums/v6...ps5f080ea8.jpg
Stops scaling after 2200 C9.
It sure seemed like maybe you had pegged it, so I pulled out some OEM-type Crucial sticks that I have with D9PFJ and the edges of the chips don't match up at all with the GSkill "2100", therefore I think that we can rule those out.Quote:
Originally Posted by Don_Dan
More confusion...like I really need that. :)Quote:
Originally Posted by Zeus
So now we have 2150 to add to the list. Who is the Master List Keeper? Gonna be a full-time job soon at the rate GSkills keeps pumping out numbering schemes.
im glad someones keeping track, cuz yeh that is getting pretty confusing.
you can read between the lines on a lot of stuff, but yeh.
Question for those with experience clocking these 2800 sticks, did you have to do any advanced tuning in the Maximus V family BIOS settings at all (Swizzle, Latency Boundary, etc)?
With cpu capable to go 2850ish imc, it can do 2840+ xmp 2T @ 1.65V
http://img9.imageshack.us/img9/3185/44470134.png
http://www.xtremesystems.org/forums/...=1#post5097681
not at all Shiranu with good IMC you do what u wantQuote:
Originally Posted by Shiranui Gen-An
My CPU's IMC is certainly capable (no code 23) but it seems the DIMMs need some fine tuning, lots of code 55. Sometimes I'll even get to boot and bench but next boot with the same settings, Code 55. *shrugs*
Vaseline as insulation?Quote:
Originally Posted by Shiranui Gen-An
that happened to me Shiranu and i screwed timmings too tight once then i flashed back and fourth BIOS and same, so i started from scratch from 2600CL10 latency boundary 12 let it go...then i put 2800 again and all fine never happened againQuote:
Originally Posted by Shiranui Gen-An
my codes where 23 / 20 / 55 thanks god this i received only phew times this one
read here
http://www.xtremesystems.org/forums/...Q-16GTXD/page5
good luck hope it helps
Sergio
a 2800 capable chip doesn't need sub-timing tuning at all.Quote:
Originally Posted by Shiranui Gen-An
just set 2800 strap, main timing 11-13-13 & proper vdimm, vccsa, vccio that's all.
why not ? u can tight and gain better perfomance owi...Quote:
Originally Posted by owikh84
this is lat boundary 6
http://i.imgur.com/uBKMr.jpg
no need to tweak VCCSA or VCCIO
@ sergio, I was refering to the first boot up with 2800 without optimising
Because he asked whether the needs to tweak to make the chip posted at 2800.
ahhh sorryQuote:
Originally Posted by owikh84
Any pointers on how to stabilize this for 32M anyone?
http://img.photobucket.com/albums/v6...ps19e4d89a.png
i guess vdimm, and 4-4-4-4-22-22 on the trasmiter slewsQuote:
Originally Posted by Zeus
but maybe Dumo can help better i dont want to punish the memory with volts are very expensive
good luck
Sergio
^as sergio said, MOAR volts!
I guess you guys are right. This was already 1.79V, i don't want to kill those sticks even though i know they scale pretty well with volts. Maybe 2933 is asked a bit too much for Samsung? Most 2800+ overclocks we see are with Hynix, not with Samsung i think.
Lemme try to get 32M done at 1400 C10 first. 1.77V didn't seem to cut it.
My Samsungs 32m 2933 but need 1.84v for 11.13.13
Your old "2V kit" or your new 2666 kit? 1.84V is still a bit much for a :banana::banana::banana::banana::banana: like me. Gotta try C-N tactics and hit a few beers first. :D
Ha ha the 2666 kit & ye the beers work real good
^2v club ftw!
:rofl:Quote:
Originally Posted by Zeus
Some of my samsung won't jump 2900 unless jolted with 1.9V+, vccsa 2.5V and vttddr 0.85V
VCCSA 2.5V? I s sure hope that was a typo Dumo? If not, how many beers for that? :D But only 0.85V Vtt? Vtt=VCCIO right?
vttddr not vtt
Got the F3-2600C10 kit here. 32M @ 2800C10 on M5G with 1.75 vdimm:
http://img.photobucket.com/albums/v2...ps3c796fec.jpg
Very nice Bart but did you really need 1.25V VCCSA? Mine couldn't do that, not even at 1.83V but i didn't try that much VCCSA though, maybe i should have? I can finish 32M at 1400 11-13-13 at 1.71Vdimm though.
Left it @ auto buddy, sets it way too high. Rerunning now with vccsa & io set to 1.1v :)
edit:
http://img.photobucket.com/albums/v2...ps18754f8c.jpg
Nice, thanks for showing. Maybe i should try again with just 1.75V. ;)
edit: Bart, my friend, you're a genious! :up:
http://i.imgur.com/SFstn.png
Sometimes things are so simple that they just get overlooked. 1.75V in bios did the trick. This is in Windows 7 64 bit which is usually a bit harder to get stable than 32 bit OS.
Thanks for showing me the ropes Bart!
It is first run with 1.7V on M5G. i wish for some desent timings 32m.
Attachment 130439
very nice Viss !!!Quote:
Originally Posted by Viss
i agree with that ZeusQuote:
Originally Posted by Zeus
promising 32m result.
Attachment 130474
great result mate !!!Quote:
Originally Posted by kintaro
volts on mem ?
Thanks mate.Quote:
Originally Posted by andressergio
this is with 1.87v.
Promising indeed! Surprised to see you're using bios 1204. Very nice run!
wow i would be very scared to pump that on my gskillsQuote:
Originally Posted by kintaro
on air it is reasonable to be scared.Quote:
Originally Posted by andressergio
below 0 is the solution in terms of ''safe'' and not 24/7.
Hmm, so it IS possible to coax 9-11-10 out of CFR at higher clocks, you just have to gorge them with massive voltage lol
1.85V+ @ 2900+ for some cfr kits
Well 1.85v isnt enough to convince my RipjawsZ 2133C9 kit to do 9-11-10 at 1200 mhz lol
hahaaQuote:
Originally Posted by Shiranui Gen-An
8-11-9-16-1T @ 2728 on cold is so nice!!!
[IMG]http://imageshack.us/a/img850/904/capture1sl.jpg[/IMG]
Wait a minute, you're pulling this of with these Trident-X sticks?
i was testing mems on ln2 to find out max.Quote:
Originally Posted by Zeus
second and third timings are not tweaked. i was running out of ln2 so this is so far.
i know the cpuz mem tab is what you are after. next time i will post it.
That's great kintaro, G.Skill must be binning like mad for these kits.
By the way it's 8-11-9-16. You had me there for a second. :D
Still absolutely awesome for Hynix chips. Never knew Hynix could pull this off. Great stuff Kintaro! I bet you don't dare to tell Vdimm for that. :D
vdimm on DMM was quite low:D.Quote:
Originally Posted by Zeus
1.95v. an atempt of increase indicate no more scaling.
anyhow for the moment i'll squize 2nd/3rd timings.
7-11-9-20-1T @ 2722Mhz is realy nice out of (2x4GB).
Attachment 130984
Are you kidding me? CL7 at 1360MHz?! :up:
What was Vdimm and temperature?
Don i have that RAM its amazing but i think they have to be very high vcore to do that, mines do 3000 at 1,75 and i boot 2933CL11 at 1,725 but for that latency and speed must be high voltsQuote:
Originally Posted by Don_Dan
anyway congrats kintaro !
the only way to archive cl7 is more cold.for me below -100C.Quote:
Originally Posted by Don_Dan
Vdimm is slightly over 2V :).
Thanks andressergio but for the moment i can't use cl7 due to lower copy rate comparing cl8:confused:.Quote:
Originally Posted by andressergio
i think imc cant handle. i have to freeze them once more in order to retest.
Kintaro post pics of the mem on cold i luv pics ;)
you need to lower TCWL(cas write latency) in order to benefit the lower CAS value ;)Quote:
Originally Posted by kintaro
Thanks for the info, that's what I expected. 2V and plenty of LN2. :DQuote:
Originally Posted by kintaro
Nice work! :up:
Thanks Alex-Ro.Quote:
Originally Posted by Alex-Ro
Finaly they are efficient beyond of funcy Hz....
http://img838.imageshack.us/img838/1214/capture2uv.jpg
Uploaded with ImageShack.us
Wow these is really nice with Hynix man, congrats
Is that an old ATI 1950PRO?Quote:
Originally Posted by kintaro
I loved mine.
Very nice bench setup and impressive numbers. Hope to see more!:up:
haha thats what i like amazing pics
i dont know notha about cold but u have to open the RAM and isulate it right ?
thanksss
Good job,now run some 32M@5 ghz :),your bandwith numbers are huge,your score should be awesome
2837 11-13-13-25-1T 1.7v
http://i458.photobucket.com/albums/q...1-13-13-25.jpg
2856 11-13-13-25-1T 1.71v
http://i458.photobucket.com/albums/q...13-13-35-1.jpg
2933 11-14-13-25-1T validated, 32m failed
http://i458.photobucket.com/albums/q...cpu-zvalid.jpg
great owi this RAM kick ass ;) i have them and luv themQuote:
Originally Posted by owikh84
thanks sergio. yeah it is an awesome kit.Quote:
Originally Posted by andressergio
2866 11-13-13-25
http://i458.photobucket.com/albums/q...1-13-13-25.jpg
32M@5 ghz :)Quote:
Originally Posted by Alex-Ro
http://img35.imageshack.us/img35/5039/capture7dh.jpg
Uploaded with ImageShack.us
Daaarrnnn......nice run kintaro :up:
I have an 2x4GB Kit from these vali on air cpu and ram
http://valid.canardpc.com/show_oc.php?id=2612153
http://www.abload.de/thumb/2400er140...0biosuqrcn.jpg
My [ TridentX] F3-2800C11D-8GTXDG
Hynix CFR H9C
http://upic.me/i/17/20121217_110911.jpg
http://imageshack.us/a/img837/3413/dscn1275dd.jpg
I have no post errors with my gene but a lot of windows bsods at 2800MHz.
Here is with trident x profile 1.85 from bios.
http://imageshack.us/a/img846/818/ramzz.png
great !!! can you post some SPI32M ? i have the same RAM and they are AWESOMEQuote:
Originally Posted by zpaf
BSOD at 2800 ? on gene ? thats weird mate u sure u tuned VCCSA ? and Latency Boundary ?
http://imageshack.us/a/img100/6223/spi32m2600.pngQuote:
Originally Posted by andressergio
Yes bsods on gene but i have not tune any vccsa/vccio.
Gene gives 1.25 to vccsa and 1.15 to vccio.
Latency Boundary on auto.
And another one at 5GHz with trident X 1.85 profile.
http://imageshack.us/a/img90/2649/spi325000.png
Here is the box.
http://imageshack.us/a/img19/9586/2800j.jpg