Still getting errors tWR=12. Could this be motherboard-related issue?
Printable View
to run 2133 i had to start adjusting my drivestrengths .
Have you tried anything there ?
Try to set for example :
DQ = 1.0x
DQS = 1.0x
something is definately wrong there:
http://i40.tinypic.com/dy1x0g.png
my sammys at 2250 11-11-11 do about the same, so maybe timmings doesn't matter too much as before or maybe imc still not efficient
@Barrel
100 MHz CPUNB speed and faster RAM speed/timings (works better with BD)
He just has really bad efficiency
Oh, I figured thats what you guys were talking about. There wasnt much of a time difference, and there was a big difference in memory speed. Seems more speed is better for 32m, while tighter timings for 1m.
Also, what board was that with? Im using a Sabertooth, which is more of a "stable" line of boards, so Im guessing they relax the secondary timings some on it compared to the CHV or some of the Gig boards.
Trying to decide if its worth sending my RipjawsX 1600 6-8-6 kit back to newegg and trying out a 2x2 kit of these.
Ahem.
http://www.newegg.com/Product/Produc...82E16820148559
@Cecil
I bet they will do better than the RipjawsX. My RipjawsX kit was a little disappointing, I needed over 1.6v to run 1600 6-8-6 on an 1100T. 2000 7-10-7 needed 1.7v on that CPU. My Pi kit tore it apart, 2000 6-10-6 and 32M 6-9-6 on 1100T. I want to pick up a pair of them too. If the 2GB density modules/ICs can do 2000 C8-8-8-xx 1T (or better) I'd be happy. The price is dirt cheap. In 2010, Elpida Hyper were still $100-150 per 2GB D:
I've been doing an increasing amount of work in Photoshop and Hugin lately though, stitching and working with ~60MP images and may need to move to 8GB ram soon :(
Definitely go 2 x 4gb:
1) Its much easier on the memory controller to run one good pair than two good pairs.
2) If your board has one pair of slots that clock worse than the other pair you can find yourself losing a massive amount of speed.
3) 8 chip sticks are inherently slower than 16 chip sticks. Id make sure theyre double sided, but then if theyre double sided there's no guarantee theyre the same chips as the 2x4gb kits.
All things conisdered sticking to 4gb sticks is the safest option.
So, are these mem. 30 or 40nm?
30nm
got the mem today
got them stable at 5GHz HT @ 1,435 with 2133 9-9-9-27 1T @ 1,65 and have to tweak VTT +VCCSA
http://img19.imageshack.us/img19/286...33999271t1.jpg
http://img703.imageshack.us/img703/7786/ballistix.jpg
http://img714.imageshack.us/img714/6876/p1120539z.jpg
http://img515.imageshack.us/img515/6584/p1120488h.jpg
http://img35.imageshack.us/img35/9933/p1120485f.jpg
Uploaded with ImageShack.us
Ordered another 8gb kit of the Smart Tracer OBs for my HTPC, and a set of the non tracer 2x2GB for benching ram. I have a friend with Hypers if I ever need them, but mostly its just for video card benching with a 2500K, so these should do fine at 2k 888 or 2.2k 999. Maybe the 2x2GB kit will do better and give Hypers a run for their money?
The second kit won't boot I get 2E I tried all and my first which seems the better passes hyperpi which is 8 instances of spi32 and when I shoot mem test on windows I get all kinda errors :(
Btw bios 2302 didn't try on 2105
Screeny on Z68. Ill run a full test when I get the voltages as low as I can. Going to put the CPU under water tomorrow.
Attachment 124711
no luck i can pass hyperpi as shown but memtest windows fails in seconds...
tried all thanks FedericoUY for the help
bad luck this time :(