thats nice Zeus yeah mines never heat up even at 1,65vQuote:
Originally Posted by Zeus
for a quad core u must use 4 instances on 1500 each as there ur not using the total mem of 8GB
Cheers
Sergio
thats nice Zeus yeah mines never heat up even at 1,65vQuote:
Originally Posted by Zeus
for a quad core u must use 4 instances on 1500 each as there ur not using the total mem of 8GB
Cheers
Sergio
Hey mate it's a 3930K so x79, I might go back to a gene-V for the mem clocking but I'm staying with x79 for now.Quote:
Originally Posted by andressergio
Actually, mine are the taller sticks, so I will post in Dumos HCH9 thread too.
:)
Tried some more, cas 9 didn't seem to happen with reasonable voltage. I tried up to 1.67V, the system booted but bluescreened right after the Windows splashscreen. Same thing with DDR 2600 10-11-11.
So for now i settled with DDR 2400 10-11-11-25 1T Latency boundary 10 at 1.55V.
http://img.photobucket.com/albums/v6...DR2400150p.png
Transmitter and receiver slew both at 3 for channel A and B seemed to do the trick for me.
Hey guys. Running into an issue. I have 4 sticks of this ram and it can't get it above 1866MHz. No matter what timing or volts. My old Corsair went above it fine. If I take two of the sticks out it will go past 1866MHz fine. The ram passes MemTest fine and I have also tried two different sets of this Samsung ram. Just got the second today in an RMA.
:(
My setup :
CPU : Ivy 3770k
Mobo : ASUS Sabertooth Z77
Try to raise Trfc. 200+ is not uncommon for 4 sticks of Samsung based ram.
Zeus - does that apply to x79 too? My HCH0 run 2400-10-12-12-28-1T 1.600v but dropped a prime thread after 5 hrs .... I know its 5 hrs but I like extreme stability, so should I raise my Trfc from 147 default?
:)
If you are running 4 sticks it's worth a shot. I've never owned a X79 set up but i think the rule applies to all platforms. (not 100% sure though)
Just tried 2000MHz @ 1.65v and 1.35 VCCSA @ 1.35v and the Trfc @ 200, then 400.Quote:
Originally Posted by Zeus
No post. :( Ideas round 2?
Try two sticks first. If that works fine try the other pair. Maybe you have a bad pair. You might try 2133 if possible with your board, maybe 2000 divider is bad. Did you try 2t? 2133 10-10-10-28 2t Trfc 147 should work with 1.5V easy. Try that as a starting point.
Good luck.
Ok. So I tried two sticks at a time. Both pairs in each slot. I could never go above 2200MHz. Both passed memtest86. So I don't think it's bad. Should I have any special ASUS Digi Power control settings? I have tried all settings but changing the memory power frequency. I am thinking my board is just a :banana::banana::banana::banana:... Thanks for your help.Quote:
Originally Posted by Zeus
What are you guys setting your transmitter and receiver skews as for 2133+? My Maximus V Formula with 3570k would not even post at 2133 until I set the transmitter skews as 3 and receiver skews as 4.
Please share!
Thanks!
I have a V Formula coming right now to replace my Sabertooth. The hell is a transmitter skew? Didn't have that option on my Sabertooth Z77... Maybe that is why this thing wont post above 1866MHz with 4 sticks.Quote:
Originally Posted by MaxT
I meant to type slews, not skews, heh.
Check out the rog formula overclocking guide posted by raja in the intel motherboards section.
Just went through the guide and never saw slews. Am I blind?Quote:
Originally Posted by MaxT
1.72v here.Quote:
Originally Posted by T0lsty
Attachment 128698Attachment 128699
Very nice Shiranui! 1.72V is quite a bit, did you measure voltage with a DMM? 1.70V set in bios on my board is ~1.72V measured by DMM. Why is your Aida bandwidth so low? Low CPU frequency perhaps?
Today my second kit came in from RamExperts, together with a €22,20 invoice from my friends at the customs. (about $25).
Oh well. These are week 19, lets hope they shine.
My 1215 kit can boot into Windows at 2600 10-12-12-28 1t but it errors out in the very first loop of Spi 32M. I have tried up to 1.69V with AI Suite, i get code 55 if i try to boot with anything over 1.68V in bios so i had to raise Vdimm in Windows.
2400 9-11-11 is absolutely no go, it won't even make it into Windows but maybe it just needs moar volts, dunno.
This 1219 kit is the first kit i´ve seen with the chips made in the very same week as the sticks itself. Most sticks use chips that are between 1 to 3 weeks older than the production week of the sticks. My 1215 kit has week 13 chips.
The production week of the chips is in the first line: SEC xxx HYK0. The xxx mark the week number.
I actually had the CPU set for 4500MHz but I left SpeedStep and the C-States enabled hoping to get higher Aida bandwidth, but it doesn't look like it worked. My board overvolts by about 0.01v so 1.72 was more like 1.73.Quote:
Originally Posted by Zeus
HereQuote:
Originally Posted by Bajawah
look under the Misc section where he talks about the transmitter and the receiver slews and how they need to be tuned before the voltage adjustments
Been testing the new week 19 kit today and it had exactly the same limit as my old kit. I'm pretty sure it's the CPU/IMC that's holding me back. I will find out soon as a new cpu is already on it's way to me.
I've been trying to get 16GB stable and it was not as easy as i thought it would be but in the end i got it to finish Hyper Pi32M:
http://img.photobucket.com/albums/v6...yperPi16GB.png
Had to raise Vdimm some and back of with Latency Boundary.
My Maximus V Formula comes in a few days to replace my Sabertooth. I will let you know how my 16 gigs fairs since they are newer than yours. If I can't get anywhere with them either I am going to just return them. :-/
Today i have been testing the sticks individually which was pretty interesting.
I tried for 2400 9-11-11-28.
The result:
2 sticks, no boot. Tried up to 1.68V but the system got into a rebooting loop without getting to post.
1 stick, booted but threw errors in Memstest 86, even with 1.675V
The last stick was golden, booted straight up, even with just 1.61V but threw the odd error in memtest 86 test 5 so i increased voltage to 1.635V and it was fully stable:
http://img.photobucket.com/albums/v6...perPismall.png
If only all 4 sticks were like that...
The good stick was from the week 19 kit, the other stick from this kit refused to boot. Overclocking results from these sticks are very random as the chips don't seem to be speed binnend at all.
Sure they are, they're binned for 1600 C11 :rofl:Quote:
Originally Posted by Zeus
Noob question:
Is HyperPi enough to test for memory stability ?
Well if you run HyperPi 32MB in a number of threads that equals the one your CPU can handle I would consider it a nice indicator of memory stability. Not that it means your memory is 100% stable but I would say that having a BSOD out of a memory problem gets pretty thin after that. You can also test y-crunsher, for me it showed up as another nice memory/IMC tester. Bother HyperPi/Y-crusher are fast tests as oposed to memtest, though the last will give you a more thorough insight on whether there is/isn't stability problems.
i agree with you mateQuote:
Originally Posted by st0ned
i passed hyperpi and memtest fails so as i open it...i always use memtest to test memory at least 300%
Very good Zeus !!! why you want to run 16GB may i ask you ? cauz if you put pagefile on your SSD wont be a problem as new SSD controllers tolerate a lot of write not as before and you wont be needing 8GB, and BTW OC on 16 not as easy on 8GBQuote:
Originally Posted by Zeus
cheers
Sergio
Thanks Sergio.
There's no particular reason to have 16GB other than just having it. :) Maybe there is a very small perfomance gain from Bank Interleave with 4 sticks in place.
lolQuote:
Originally Posted by Zeus
yes true when i have the 16GB of patriot on my MIVE it was a bit more than using 8GB on aida but now i want high OC on mem and just using 8GB@ 2600CL10 soon ill be on the gskills 2800 CL11 :)
Anyone have any quick tips for clocking this mem on an AMD FX/C5F rig?
(haven't been following this thread and it's huge!)
I played with it a little the other night and couldn't even get relative stability @ 2200 11-11-11 1.6v... :shakes:
It's a 1220 kit, maybe I just got a bad batch. :shrug:
Zeus, I assume those 2400 CL10 numbers were done on Intel...
Did you do any testing on the C5F?
Thanks,
Dave
The cheapest 2x4GB kit on Newegg that is rated 1200MHz C10 is $87.99. These Samsungs are nearly half that and even my weak 1205 sticks can do 1200MHz 10-12-12 with less than 1.65v. Crazy.
Attachment 128852Attachment 128853
Are the samsung F chips as good as these? or just thrash?
I haven't heard of "F" chips, at least not in terms of die revision.
http://webcache.googleusercontent.co...&ct=clnk&gl=noQuote:
Originally Posted by Shiranui Gen-An
At the very top and around the middle of the chart.
Hi Dave.Quote:
Originally Posted by Daveburt714
No i did not test the sticks in the CVF, sorry man. I sold my AMD stuff and went to the dark side (or is it the bright side? :D). Usually Bulldozer/Zambezi have very strong IMC and clock the ram very well. Try 1.5V and tRFC at 200+
I haven't really tried 2200 but i guess it'd be possible at around 1.4V on my rig, 1.5V for Cas 10.
I just found a data sheet for this HYK0 chips here: http://pdf1.alldatasheet.com/datashe...446D-HYK0.html
I don't know what Vdd and Vddq is but it's max rating is 1.975V. Does this mean we can put that much Vdimm through these chips?
edit: Upon scrolling down under Recommended DC Operating Conditions i saw max voltage 1.575V.
My samsung memory can do 2400mhz 9-11-11-31 @ 1.663V on a AMD FX-8120
Looks like older technology as it's 46nm. I doubt it would clock as well as the 30nm stuff.Quote:
Originally Posted by knopflerbruce
https://dl.dropbox.com/u/20116032/Re...msung/2600.jpg
Nowhere near stable. Just playing around with my sammys and new M5G. 1.675v
Tested one kit recently. Bought it from Amazon, waited two monts(in Ukraine this memory unavailable) and got epic fail. Worked perfectly at 2400MHz 10-12-12-30 1.5V but cannot 2600 with any timings and 1.75v Good for 24/7 use but bad for benchmarks ))
A payed with shipping, broker etc about 70USD. If its memory is sold in Ukraine and costs 50USD I'll buy two or three kits for 24/7 use
https://dl.dropbox.com/u/20116032/Re.../2600_16gb.jpg
Really relaxed just to achieve the Mhz, 16GB @ 2600 = nice !
good job amigo !!!Quote:
Originally Posted by st0ned
Done some more individual testing.
This time 2600 10-12-12-28. 1.65V
All 4 sticks booted!
First stick was the golden week 19 stick that did 2400 9-11-11 at 1.635V, this one seemed fine at 2600 but threw errors after a couple of memtest test 5 loops, even with more volts.
Second one was the week 15 one that was able to boot at 2400 9-11-11 but not stable, this one was fully stable at 2600 10-12-12 at 1.65V!!
The remaining two sticks both booted at 2600 but both threw errors in memtest.
Looks like 2600 10-12-12 is easier to obtain than 2400 C9 but you will still need good sticks to get it fully stable.
I'm working on as tight timmings as I can at 2400 cl10, 2400 cl9 with 4 sticks will ask for hardcore volts (or high binup chips ).
Atm I'm doing 2400 10-11-12-20, funny thing 10-11-11 will cause errors. Then I'll go for 2600 cl11 but idk if I'll be able to do it without too much volts.
Any one have any suggestions for Drive strength settings? I came up with this on my Asus Crosshair V w/1100T 1.3v cpu/nb, cpu 1.45v (needed for 4.0ghz)
also tried loosing the timings up some, but didnt help any.
DQS Drive Strength of X1.25
Data Drive Strength of x1.25
Mem Clock Drive Strength x1
ADDR/CMD x1.25
CS/ODT x1.5
CKE x1.5
This got me in the 970mhz range, stable at 943mhz
Look correct or did I missing somthing?
Thanks, Adam
Increasing the volts to 1.725V gives my 2x4Gb Kit this stable setting:
2470mhz 9-11-11-31 T1
Asus M5A97 PRO + FX-8120
Quote:
Originally Posted by Hotrodsun
Your settings gave me more 10mhz with same voltage.
My everyday use is now 2480mhz 9-11-11-31 T1 1.725V (board overvolts to 1,73v)
Tnx ;)
Dont know if your board has a VDDA voltage setting, I went from 2.5v to 2.6v gave me about 20mhz maybe it could help you too. Glad I could help someone, AdamQuote:
Originally Posted by andrenoites
What maximum VDIMM you guys think it's 24/7 safe for these 30nm Samsung low profile modules?
According to this datasheet it is 1.575V max.
TY.Quote:
Originally Posted by Zeus
Just for the record I'm running at 2133MHz CL10 1.5V on MSI Z68 motherboard. 2x 4GB...
Wow. Installed the Maximus, and I am at 2400 with more headroom. Sabertooth was a POS...
I'm running 2x4GB at 2133mhz 10-11-10-30-1T @ 1.4V on Asus P8Z77-M with Pentium G620. It probably can go lower, might need to do some further testing.Quote:
Originally Posted by andlcs
Haven't stress tested, but finishes Pi 32M and some h.264 video encoding without any problem
http://img843.imageshack.us/img843/3733/samsungyk0.jpg
Idk if this deserves publishing or not, but here it goes
Sammys 9-11-11-21 @ 2200, tight subs working on bandwith
https://dl.dropbox.com/u/20116032/Re.../2200_band.jpg
See what happens if you lower RAM REF cycle time ...one of my sets will do 96 happily at 2200 9,10,10
Another thing to try is tRRD to 4 and tFAW= 4*tRRD
Thank you ! :) I've been looking for ways to still improve my bandwith. I can go for 2400Mhz but eventhough I can pass HyperPi 32MB, I can't finish the y-crunsher run without errors :/ eventhough I was able to do it 2 or 3 days ago, oh well strange stuff.
I can indeed lower REF Cycle but it holds little bandwith improves, at least judging by AIDA's bench. As for tRRD and tFAW, will do.
EDIT:
Apart from latency it looks pretty much the same
https://dl.dropbox.com/u/20116032/Re...2200_band2.jpg
Yep. Some gains are bigger than others and some are not gains at all.
These were both tips I read here on the forums that gave me a little gain in superpi and aida 64. Just passing stuff on :)
Want to try tWR down to 9 or 8 next? Have a little fun and play with these settings... BUT.... I borked a windows install in the process so a recovery disk and that kind of stuff is useful ;)
Well I don't care much about this windows install, it's f***ed anyway, but I really appreciate the advice. I did some bios mambo jambo, aka 3rd timmings and boom :D
https://dl.dropbox.com/u/20116032/Re...2200_band3.jpg
I see ppl with 2600c10 with less bandwith then this :)
Your gain came from TRRSR=4. This gives a huge boost in bandwidth but renders the system unstable at higher frequencies.
I have 4 sticks from 1150 batch. Still trying to find a way to run them at more than 1866MHz. They boot until 2000MHz but they are not stable. Using bios F11 and cpu @ stock. Is there a magic timing for 4 sticks?
Tried CR=2T, tCL=12, tWCL=CL-2, etc...
I will test each stick alone to see the limits and using 2 sticks too when I have time.
What timings? Cas 9 requires quite some volts on these sticks. You might want to try Cas 10 for 2200+ speeds.
If anyone is interested in getting some (or more) of this ram, it is back on sale at the egg, $45 per 2x4GB kit, less $5 off code = net cost of $40 and free shipping.
$5 off w/ promo code EMCNBNA25, ends 8/15
http://www.newegg.com/Product/Produc...D=1fd966j3t763
a little birdie told me FULL compatibility for GB boards(so 2400+) will be here soon....
currently testing, but looking very good.
What are you testing on...the G1 Assassin 2?Quote:
Originally Posted by sin0822
You americans don't realize how lucky you are paying $40,- a kit. I had to pay more than double the amount for a kit here in Europe! :(
No currently just Z77 platform, Z77X-UD5H and Z77X-UP5 TH, I think it is a fix they might be able to apply to all boards, or it might be something that needs to be done board to board to tune for none XMp memory. I am not sure. All I know is that I just ran SPI32M at 2600mhz with 2 sticks on my UD5H at 10,12,12,25 T1 and 1.7v(i just set 1.7v to be sure) 2400mhz at 1.5v is easy too.Quote:
Originally Posted by Reefa_Madness
:bounces: Yay :up:Quote:
Originally Posted by sin0822
tell me the little birdie said the UD3H was getting the fix as well Sin. . .
Ill push for it to be put on all the Z77 boards, but I think they will do it anyways. UD3H would be an obvious choice for the fix, i think they just UD5H is a top seller and takes more attention.
That would be a nice change!Quote:
Originally Posted by sin0822
They maxed out at DDR3-2133 on GA-Z77X-UD5H with 9 of the 11 BIOS' I tested.
I had a brain fart...I meant to ask if it would be the G1.Sniper 3 (I figured that it would be Z77 chipset), since you did a recent review on it, but I had the Assassin on my mind due to the fact you have it posted for sale in one of the other forums (maybe here, too?) and I had just seen that thread. My bad.Quote:
Originally Posted by sin0822
Yea I have both boards, i am not selling either. Ill hang onto Assassin 2 for a while because it has the on-board CA20K2. I am currently debating which one to put in my main machine.
Perhaps Sniper 3.
They can do DDR2200 on current BIOSes and if you do it at a cold boot they can do 2400, but only on a cold boot.
Yup, and now let's show some stuff instead of talk!
First off the Z77X-UD5H running 2x4GB Samsung Green ULV@ 2600 Cas 10 T1
http://img545.imageshack.us/img545/7421/260032msam.jpg
Second we have the beautiful Z77X-UP5 TH-CF running 2x4GB Samsung Green ULV@ 2600 Cas 10 T1
http://img832.imageshack.us/img832/1050/up52600sam.jpg
Great news Sin! I hope they spend some time tuning 4 dimms compatibility as well.
Hey Sin you must have been pushing some serious volts (>1.7v) to those mems, never the less congrats it looks great !
yea it is there I am told. Already taken care of.Quote:
Originally Posted by thigobr
Beside a week 11 kit i see another week 5 kit putting up some nice numbers. Week 5 pops up a little too often when it comes to good clocking kits to call it coincidence.
What voltage for 2600 C10 if i may ask Sin?
1.7v
here are the BIOSes UP5 http://www.mediafire.com/download.php?2a68g1hg5iv08o7
UD5H: http://www.mediafire.com/download.php?c88l5bnj86gx37n
Very nice that this issue is finally fixed . :)
Has anyone ever tried the single sided 2GB sticks of this ram? I've found them for 22 bucks a piece. I know that's not cheap but at least they are availble with free shipping to my country.
Maybe it has bad performance due to no bank interleave?
single sided dimms tend to not have the same efficiency as their double sided counterparts.Quote:
Originally Posted by Zeus
I'm afraid you're right. I still ordered one stick, just to see how it overclocks. If it clocks like cr@p i'm willing to sacrifice it to see if these chips can take 1.8-1.85V :D
The IC's will take it but the pcb may not imo.
Due to the small nature of the PCB it has very short traces which in turn means less resistance so there is a chance the PCB will withstand the abuse. We'll see. Remember that these chips are rated for just 1.35V. ;)
True & you will soon find out.
Look up the Samsung data sheet, the ic's have a very high absolute max voltage 1.9 + if i remember correctly.
thanks for taking another $40 out of my wallet... I couldn't resist buying another set.Quote:
Originally Posted by Reefa_Madness
You are welcome!Quote:
Originally Posted by Rattle
Happy to have you join me as a member of the "I spend too much on tech" team. :D
Today i got one 4GB PC10600 1333mhz with nr M378B5273DH0-CH9 , can someone give me some tips on what i shall change, voltage and any other thing than the usuall timings? Running a MG5
That other thread that you posted in, the one started by Dumo about the generic Samsung, is the one that is going to have some of that info you need already posted. Check out the screenshots and comments in that thread for guidance. Your ICs are the HCH9, right?
http://www.xtremesystems.org/forums/...l.-sided-2X4GB
HCH9 i Think yea, it ends with H0CH9Quote:
Originally Posted by Reefa_Madness
Tried 2 dimms and no boot above 2133MHz using Giga UD3H. At least they seem stable at 2133MHz CL9-10-10-28-1T vdimm 1.50V.
That's Gigabyte for you. Keep an eye on bios updates, mem issue's will be fixed in upcoming bios'.
thanks for the info, gonna buy a few :P
anyone used these on x79 asrock board?
thanks
that's for sure with you.... you're a buying machine :)Quote:
Originally Posted by Reefa_Madness
XC
Working on a set of these on my Maximus V Formula.
So far I got 2 x 4GB running at 2200 with 9-10-10-27-T1 timings with 1.55V (not final yet, will be testing lower voltages). This is 17 hours prime95 27.7 v2 stable. VCCSA is at 0.925 and VCCIO is at 1.05. CPU PLL is at 1.7125. This is with the i7-3770k at 4.7GHz with 1.272v under heavy load.
Temperatures don't go over 69c. (3770k was delided).
Will be posting screenshots within a couple of days, once I finish lowering voltages and testing for stability.
Why oh why aren't my 4GB HYK0 sticks like this little 2GB stick? :(
http://img.photobucket.com/albums/v6...ts/DDR2800.png
There is no limit to this stick it seems, it even booted at 2600 cas 9-11-11 1.73V, not stable though but nothing volts cant fix.
I tried to run 32M at DDR 2800 but the system locks up, i think the memory controller is not up to it. If it was the memory i'd expect either an error or a BSOD.
I have another 3770K that might have a better memorycontroller but i don't feel like swapping cpu's right now.
Any experience with these stick on the asrock formula?
Mine do 2400CL10-11-11-21 1T 98 at 1.54V on my MVF!
Really dont want to dig back through this whole thread. Finally picked up an X79 / 3820, and curious what others were able to get with these on 2011.
It really depends on the batch you get. If your CPU's IMC is up to it, it seems like they'll all do DDR3-2400, it's just how tight you can set the timings that separates the really good sticks from the okay or bad ones. 11-11-11 at 2400 seems to be pretty standard.
i had this sticks of ram... tested on z77 /3770k and reached 2400 mhz @ 10-11-11 24 1T 1.50v
Good grief, I'd imagine with 1.65v they could do 9-11-11 at 2400 :oQuote:
Originally Posted by p3gaz_001
Finally got them running on Asrock!
http://www.abload.de/thumb/linxfrcva.jpghttp://www.abload.de/thumb/wow04pjc.jpg