Well the sticks I got weren't good, they would crash with 1000Mhz 4-4-4-4 4:5 @ 2.4vdimm. I'll be selling the new ones cuz I scored some D9GKX (OCZ Platinum Extreme). So far I think the 10th Anny were better (yes I kept a pair).
Printable View
Well the sticks I got weren't good, they would crash with 1000Mhz 4-4-4-4 4:5 @ 2.4vdimm. I'll be selling the new ones cuz I scored some D9GKX (OCZ Platinum Extreme). So far I think the 10th Anny were better (yes I kept a pair).
So close, I set my sticks to 1000 4-4-4-12 at 2.0v and got it to boot into windows but then it crashed :( 2.1v seems to be the magic number. Now onto 1200!
Strap change occurs at 424>425MHz FSB (1696>1700)Quote:
Originally Posted by Dumo
i'll test my 680i tonight if a get a chance
Jeeze nicepun your bummed because they wouldn't do 1000Mhz at 4-4-4-4 on 2.4v, isn't that pretty impressive to run at that speed with those timings? Were you using 1T or 2T? Here I was bummed because my kit wouldn't do 900Mhz at 4-4-4-12 1T on 2.35v!
DDR2-1000 4-4-4 @ 2.25v or less should be possible with most good D9.Quote:
Originally Posted by Roger_D25
We're talking on 2T right? If not my D9's are just not up to snuff, lol!
I'm talking on 975/965...so yes, 2T. Although on 680i, with 2T...most good sets should be able to easily surpass DDR2-1200 4-3-3 w/ 2.3v, because the timings don't mean as much to the memory controller, according to results I've seen.Quote:
Originally Posted by Roger_D25
Thank you very much EnJoY, that info is greatly apprecaited! I continue to learn something different about this mobo/RAM combo everyday! The learning curve for me has been steep, being a Intel convert, lol!
Anyone else pair this up with a DS3? Results? I can't run 1200 4-4-4 even with 2.5v
Scratch that, can't even do 1200 5-5-5 w/ 2.5v. What's the highest ram would do on a DS3?
Well, since my XMS2 5400ULs do 1000Mhz @ 4-3-2-4 4:5 w/ 2.4vdimm I guess I set my standards a bit high....:slapass:Quote:
Originally Posted by Roger_D25
DS3's dislike micron d9's...you won't see 1200mhz on a DS3, no matter what the voltages.Quote:
Originally Posted by HeadOn
are you sure about that? my S3 loves my DDR667 D9's. or is there that much difference between the DS3 and S3? I got the DDR667's over 1000mhz. i'm sure i could get my DDR1000 D9's over 1200mhz on that board. i can try it tonight.Quote:
Originally Posted by EnJoY
I don't know what to say about your board, all I know is what I hear. There are endless topics about how DS3's hate D9.
I'm not saying you're wrong, because I personally don't know. I was just wondering if there were any tests done comparing RAM with that board? Or maybe comparisons between the S3 and DS3?Quote:
Originally Posted by EnJoY
I'll run my Ballistix 8000 in my S3 tonight and come back with some results.
I ordered this ballistix ram 12/8/06 from newegg. It arrived today 12/12 and it is D9GMH.
I ordered the Ballistix tracers on 12/8 from newegg and received them 12/12 also! :) Thanks for confirming its D9GMH, i really didn't want to take off the heatspreaders =)Quote:
Originally Posted by LiquidEric
Has anyone tried this on a 975x board, specifically the P5WDH? If so, how were the clocks?
1170mhz 2.2v 5-5-5-15
1070mhz 2.35(max p5wdh real unmodded) 4-4-4-4
Mine are running 1111mhz 2.1v 24/7 FAH 5-5-5-15 on XBX.
Quote:
Originally Posted by EnJoY
[QFT] what happened to your RMA? hope you got it back by now [/QFT]
:toast:
Ok mate thanks a lot, that's very re-assuring. I'm getting a pair of these in the near future, I'll post up how they score.
Hoping it get it back by friday.Quote:
Originally Posted by GrimReaper
My pair runs at 2.2v 1000MHz 4-4-4-12 and 2.2v 1067MHz 5-5-5-15 rock stable. :)
Edit: runs on Asus P5WDG2 WS Pro 0506 BIOS.
http://home.comcast.net/~riptide007/tracers.jpeg
all i need now is my e6600
AL1005's huh? do they have the same IC's as the AA1005's?
From what I understand they are exaclty the same modules except for heatspreader and use D9GMH IC's! Great looking sticks by the way, almost got them myself!
My came back from RMA a day early... :D
I'll give em a quick test today and see if this new set is better or worse than my last.
:woot: hehe finally :D waiting to see what they can do :toast:Quote:
Originally Posted by EnJoY
p.s
dont kill these too :p:
i just got my 667 kit back from RMA. the kit that i killed was old heat spreaders, the ones i just got from RMA are the new heat spreaders. on the S3 i was able to hit 1000mhz on 1st boot. :) S3 still playing nicely with the GMH
Quote:
Originally Posted by GrimReaper
I quickly tried them in memtest at 800mhz 3-3-3-8, and then just went to 1000mhz 5-5-5-10 to prime them in windows. They seem a little better than my last set actually, but I'm not going to be able to give them any tweaking time as I need to sell them, hopefully by tomorrow at the latest (they're already in the FS section).
Nice sticks, but I'm going fatbody. :)
Do you already have them EnJoY or do you have to find some? Before getting my Ballistix I tried to get ahold of some fatbody's but nobody who has them wants to sell them and trying to find retail sticks is next to impossible! Infact do you know if anyone is still making them, theres got to be a few 92-ball IC's left somewhere? One last thing, do you have any pictures of your current fatbody's?
2.2v 1067MHz 4-4-4-12 orthos blend test stop after 3h, up vdimm to 2.25v, still running, seems stable.Quote:
Originally Posted by bobov
2.2v 800MHz 3-3-3-8 dual superpi 32M
http://img273.imageshack.us/img273/5...i32mkz6.th.jpg
2.2v 1067MHz 4-4-4-12 dual superpi 32M
http://img450.imageshack.us/img450/1...spi3uk6.th.jpg
I made it to 450 3-3-3-4 @ 2.55 vdimm (set in bios)
http://www.importjdm.com/shipped/crucial_d9gmh.JPG
nice work irev. so i'm guessing the lower latencies will give you faster pi times?Quote:
Originally Posted by irev210
Quote:
Originally Posted by itznfb
nah, but i saw enjoy doing 3-3-3-4 at 450, i just wanted to be cool like him :cool:
I am going to rerun with 4-4-4-x and you will see faster pi times with the increase in Mhz i get out of the ram
that's what i figured. seems like ddr based AMD series is the last setup to really benefit from low latencies.
what's your max clock on these so far? mine has been about 615mhz. not sure if its the p5b dlx holding me back or the RAM itself.
just got my cpu in so heres my preliminary (and i mean VERY preliminary) oc results with Crucial Ballistix Tracer PC8000 2gb kit.
1000mhz 4-4-4 @ 2.3v
http://home.comcast.net/~riptide007/oc1.JPG
:toast:
will try 800 3-3-3 and 1200 5-5-5 later, dont have time now.
I have mine running at DDR2-800 at 4-4-4-12 right now, should I try 3-3-3-4 or 3-3-3-8?
What voltage would I need for that?
What is a safe voltage for ram?
http://img143.imageshack.us/img143/4452/ehtr8.th.png
Oh how I crave a bootable 400 FSB. :(
EDIT: Crap forgot to losen the timings.
I've bought 3 different sticks of Ballistix PC2-8000. The first stick's SPD says it's BW is 8000. The second said 5300 (and wouldn't run at 450MHz, I returned it to Newegg), and the third says it's 6400 (but it runs at 450MHz).
Can I trust the SPD? I'm using CPU-z to read it...
I'm running an E6600 on a bx2, FSB at 360. Ref freq is at 266, and Mem freq is 667. Ratio is 4:5 (which gives me better Photoshop performance than 1:1).
I'm running BIOS 2395.
Thanks, and I've really enjoyed all the great, informative threads on this site. This is my first home-built, and first OC.
My Ballistix claims to be PC2-6400 in CPU-Z, but in the timings table it says it can do 500mhz 5-5-5-15. Which are its rated timings.
But it says PC8000 on the tin, and mine at least are more than capable of doing 1000mhz (currently doing 1066mhz 4-4-4-16). See how far you can overclock them, that's the real test.
In case you guys did not know, the SPDs you see in CPUZ or whatever are there so the sticks can boot up on any mobo. Try to clear CMOS and see what you get when you first boot regadless of what RAM speed you have. As long as they can do 1000Mhz CL5 they have met their Advertised SPEED.
Sure, try 2.2v first, if not then try 2.3v. You should be able to do CL3 no problem.Quote:
Originally Posted by mawsoccer
Quote:
Originally Posted by nicepun
Um, problem is that these modules, the C5 pc8000, are supposed to be EPP profiled for 1000mhz CAS5. If your modules do not feature this SPD, and do not have a small white paper sticker on ieach stick denoting such, your retailer has ripped you off, or you did not buy the EPP modules.
I have one set of each of the ballistix with the new heatspreader. All feature the same SPD programming, except for the C3 modules. This includes the non-EPP profiled sticks....the difference being the EPP profile in those sticks. It has nothing to do with them booting, as the SPD should also be programmed for 800mhz 5-5-5-18. The PC6400 4-4-4-12 and pc8000 5-5-5-15 are EPP profiles, and will not show unless you have EPP profile board, or use cpu-z, hwinfo, or maybe everest.
the sticker i refer to is shown in the first post..i stole this pic from there(white sticker with "lead-free EPP"):
http://img361.imageshack.us/img361/7...shot199fn6.jpg
This is not 8000EPP but also Crucial 667 2 x 512MB kit @ 2.6V on P5B Dlx
Old good memory :D
http://images2.fotosik.pl/312/dc092bb3359c7643m.jpg
Ah, but then it is not prudent to compare your sticks with these...the sticks this thread is about are new, and 1gb in size, to boot. 512mb sticks will put less load on teh memory contorller, and therefore, can go much higher.
Here are mine at 616mhz 2.4v on a Biostar P965 Deluxe
http://img.photobucket.com/albums/v678/Doogs/616mhz.jpg
I am interisting with this Stick but what different ?:) :)
http://www.newegg.com/Product/Produc...82E16820148047
http://www.newegg.com/Product/Produc...82E16820146563
which better for OC perfomance ? any one can tell me Please !!!:toast: :toast:
They are almost the exact same moduels except the more expensive one (Tracer) comes with LED's on the heatsinks. Besides that they are identical (should be D9GMH)?
i just got these in today. They wont do DDR2 1000 4-4-4-12 but they will do DDR2 800 3-3-3-8 all day at 2.25v. 1000 would have been nice but i guess low latency will have to do ;)
Have you tried more vdimm?Quote:
Originally Posted by Budwise
Yea, i tried up to 2.35v at DDR2 1000 4-4-4-12 and it still would not get into Windows...Quote:
Originally Posted by nicepun
Hnmm. Did you try at different dividers? That's very odd because with just 2.45v you can get as high as 1200MHz CL5, not stable but you can run som 1M SuperPi and such.Quote:
Originally Posted by Budwise
anyone have a shot of the IC's on these? I saw a little bit of it but couldnt tell if it's still D9GMH. I thought I saw the micron logo but not really sure. However, they do run cas3 @ 2.25v so far.
Confirmed, they are still D9GMH. This is the CL11177.1L batch ;)
http://i55.photobucket.com/albums/g1...4/DSC00005.jpg
Hey Guys,
I've finally been able to start pushing these modules and I've been very impressed with them. At first I just left the voltage at 2.2v and 2.25v but the results weren't great. Well once I started to increase the voltage a bit this stuff started to scale amazinly well! I've been sticking with CAS4 at the moment but I'm currenlty stable at 4-4-4-5 @ 1000Mhz on 2.4v with active cooling.
After reading the above post I realized that each kit has different batch numbers, I just assumed all the moduels had the same batch number (yeah I know, stupid), lol! Anway here is my batch number and I can confirm their D9GMH although when I had pulled the spreaders off I didn't take a shot. Fortuenlty the "Ballistix" logo's weren't all in the same place so I was able to read the actual numbers. Mine were also -2F in the lower right hand corner as wel but the number above the "D9GMH" logo were different. I'll try to pull the spreaders again and take another shot.
PS, Does anyone know if these PCB's are 6-layer or 8-layer?
http://img353.imageshack.us/img353/342/img2up3.jpg
:rotf: nice sandals....^^^^ :lol:
Tracer's are $266 as of right now with the promo code EMC109BALLISTIX2GB. :D
Holycrap it worked! I am seriously considering picking up another kit! $260 is a steal for this RAM I think! Is it legal for me to use that promo code or not Gautam (Im' being serious)?
lol...of course. New years thing.
I put this info on the hotdeals thread. Excellent find Gman!!Quote:
Originally Posted by Gautam
Quote:
Originally Posted by Roger_D25
six layer if I am not mistaken.
Yes, ballistix rock. So far they are my favorite sticks.
They seem to do well with up to about 2.55V. 2.45V is the sweetspot for 24-7 for me.
Ordered:p:Quote:
Originally Posted by Gautam
Thanks a lot!:toast:
Still dunno if there's any reason for me to pull the trigger on these or not.
My Team 5300 does about 480 3-3-3 with 2.65v...anyone gotten round there with these? :p:
Glad to see you guys making use of the great deal though.
Thanks from me too, I ordered two sets.Quote:
Originally Posted by Gautam
What's the word on the Tracers versus "vanilla" Ballistix? Same IC's or what?
:banana: This shi.t cracks me up:
The bling-bling thing
How does Ballistix Tracer do that bling-bling thing?
Our custom-designed circuit relays bus activity to the LEDs. The chip drives the activity-indicator LEDs, while the ground effects LEDs are driven directly from the module power supply. However, Ballistix Tracer modules don't require a higher voltage than standard Ballistix memory. (The brightness of the LEDs will vary slightly with the memory module voltage, and higher voltages will produce brighter LEDs, but DO NOT exceed the recommended voltages.) The LEDs operate independently from the rest of the module and do not affect performance.
http://www.crucial.com/uk/ballistix/...asp?key=tracer
Well, i run them now at the 533 aka 1:1 divider on my AW9D-Max at 3-3-3-8 2.25v. Since i am at 400FSB when i bump it up to 667 divider it puts them right at 1000, but 4-4-4-12 was a no go with up to 2.35v. With my 24/7 OC of 3.6Ghz 400X9, i really only have those two options of memory dividers... I guess it could very well be a board issue. I cant imagine that they would d0 3-3-3-8 800Mhz so easily and not do 4-4-4-12 at 1000...
Quote:
Originally Posted by nicepun
well, it turns out in fact to be a mobo/bios issue. I set it to the 800 divider, and put it up to DDR2 1000 4-4-4-12 2.25v no problem whatsoever. I guess the new 1.3bios for the AW9D-Max board has a screwy 667 divider :(
Great......:banana: :clap:Quote:
Originally Posted by Budwise
here is my score on 8000 ;)
5-5-5-15 2.35vdimm
http://img154.imageshack.us/img154/8...235vqn4.th.jpg
bofors,
I could be mistaken but I'm pretty sure that both Tracers and "regular" Ballistix are the exact same modules. One has pretty heatspreader and one doesn't, lol!
On a different note I was able to pull the heatspreader off of my modules and snap a few shots. The quality isn't great but they are indeed D9GMH IC's!
http://img221.imageshack.us/img221/4498/img0086aq4.jpg
http://img142.imageshack.us/img142/7651/img0090qz3.jpg
Nicepun,
I didn't even realize my stupid feet were in that shot, shame on me, lol!
Thanks, that is my suspicion as well. It looks like they both black PCB's too.Quote:
Originally Posted by Roger_D25
Did you find a safe way to take take them off?Quote:
Originally Posted by Roger_D25
any body tested Ballistix Tracer max with 4-4-4-12 ? vdimm ? I ordered 2 pair from newegg.
Quote:
Originally Posted by Hazzan
max at 2.45 on my non-tracers was 1200mhz 4-4-4-5
see the ddr600 club post.
Nice kit there irev210, Congrats!
Pointer here.. 2.25V with Dominator AirFlow for 24X7, is it fine?? 3-3-3-8 @ wateva I can get... on my mind for now. :P
I don't think the tracers will perform the same, or maybe you got a good batch of regular ballistix. Anyway, below are a pic of the Tracers. I have not been able to do 1200Mhz w/ these as I've did with the non-Tracers, I tried with my P5W DH and P5B Deluxe. Maybe it has something to do with the lights robbing some of the performance because it has to power the lights...I don't know.Quote:
Originally Posted by irev210
http://img389.imageshack.us/img389/1...1550rg9.th.jpg
Interesting nicepun since their also D9GMH IC's? I guess it's safe to say that not all D9GMH modules are going to perform the same, like any other compoent some will do better than others. In my case I havn't yet pushed them for high clockspeeds (only tested at 1100), just been working on tight timings and 1T for the moment. Now that I've got my Xeon running stable at 3.6Ghz I'm looking forward to pushing the RAM now!
irev210,
I just noticed how many kits of DDR2 RAM you have, good times, lol! Please drop me a private message if your ever looking to get rid of any of your fatbodys!
well i ordered the normal balistics from newegg and i ended up with a set of tracers oh well i cant complain for $270 shipped!, right now ran memtest and pc-check on them for about 5 hours and its running at 1030mhz at 2.2 volts at 4-4-4-12, so i cant complain...
Quote:
Originally Posted by nicepun
yah that was just 1m stable.
my tracers do 570 dual 32m pi stable.
overall, i think it just comes down to the batch. I got two more sets to play with now :)
I really doubt that the Tracers are any different than the regular Ballistix in terms of design performance. Like Irev210 said, I think batch to batch variation is a much more likely explanation.Quote:
Originally Posted by nicepun
Anyways, I have two sets of Tracers scheduled to arrive tomorrow and also one set of regular Ballistix here already. I want to do some testing, but I usually only run 1:1 on my Xeon 3060 / BadAxe1 & 2 systems. That only gives DDR2-800 at 3.6 GHz, so I am wondering if there is any particular "divider" I should be using to test for maximum RAM speed here?
Quote:
Originally Posted by nicepun
yea i have a feeling these tracers we bought is an older batch than the orange one. The performance might not be as good as the latest orange ones. As matter of fact the orange ones I had was definitely fresh off production. It had that strong fresh pcb smell, you know what I'm talking about lol. ;)
Quote:
Originally Posted by syne_24
lol
even these tracers that I just got... one set is XXXXX.HB and the other is XXXXX.HC and one overclocks better than the other.
Quote:
Originally Posted by nicepun
nicepun, i have that same batch also.
I could not do 1200mhz either, dont feel bad. It is the batch. What did you max out at though? It is still a REALLY REALLY good set.
I have another batch, same 11178, but HC instead of HB tracers and I am dual-32m pi right now at 580 no problem.
I got a set of "HB" and a set of "SZ". Will be testing laterQuote:
Originally Posted by irev210
oooh, SZ lol
test!
Quote:
Originally Posted by Dumo
Whats this,i been reading to try to find out why are hes time slower than mine on Asus P5B E Plus,E6300@3795,Crucial 10th Anniversary @1084mhz D9GMH (B6-3)
Is it the divider?
Im not good at superpi,only disable what i dont use,17 processes rinning in win.
http://i94.photobucket.com/albums/l9...spi32m13-1.jpg
Wow... talking about "idiot lights"... :banana:Quote:
Originally Posted by bofors
Anyways... both sets have passed one round of Rember (Memtest) [I do not run Windows, but Apple's OS X] at DDR2-800 (3-3-3-4) @ 2.2V. These are the settings which I have been running my regular orange Ballistix at (batch#: CL11176.WD).
EDIT: These are 24/7 settings and I actively cool my RAM (2 x 1GB) with two 60mm Vantec Steath fans (12 CFM each).
The HB is on a BadAxe1 and the SZ is on a BadAxe2. The HB appeared to fail (my system crashed) when I initially installed them and booted at DDR2-800 (4-4-4-12) @ 2.2V. But that failure was not reproducible.
I will try to post some Xtreme Bandwidth results later.
Hey Guys,
You have talked a bit about your Ballistix batch numbers being either HC, HB, or SZ and how some might be better than others. Below is a picture of my modules, is the batch number for my modules (CL11174.8H) or is the batch number listed somewhere else? Basically I am just curious about these numbers and what they mean (is there any significance to this batch number)? Thanks for any help you can give guys!
EDIT: Just realized that the tiny type to the left of that number says "Batch" so I guess I now know what that number is, lol! In anycase the questions still stands, is there any significance to it?
http://img353.imageshack.us/img353/342/img2up3.jpg
Just in case this is not clear to everyone, several people here, including myself, have just received 2 x 1GB Tracer Ballistix sets from NewEgg at an apparently erroneously discounted price: http://www.xtremesystems.org/forums/...wpost&t=129867Quote:
Originally Posted by Roger_D25
These all seem to be batch# CL11178.??, where "??" is mostly "HB", but I also got an "SZ" set and other suffixes have been reported. These different suffixes may correspond to different batchs of ICs, which we should expect to have slightly different properties do to randomness in the fabrication process (like a "bell curve"). So, the idea here is to compare notes and see which of these batches are better or worse, and so far it seems that "CL11178.HB" is quite good, but not the best batch.
So far these can do DDR-800 3-3-3-4 @ 2.35vdimm all day long with in my P5W DH 400x9. I will try different configurations (get it above 1000Mhz but of course with lower FSB) but I doubt that you'll get the same results as the regular ballistix (Orange Heatspreader), it may has to do with the tracers (lights), I don't know but I can't think of anything else that maybe a factor for these tracers not to perform. Both sets have are "HB" batch so I'll messing with the on the P5B Deluxe and P5W DH. I believe 1100Mhz at CL4 with 2.45vdimm is doable. I will say that these are good for the price.
Crucial actually does address this concern here:Quote:
Originally Posted by nicepun
The LEDs operate independently from the rest of the module and do not affect performance.
http://www.crucial.com/uk/ballistix/...asp?key=tracer
I'm moving this stuff over the Ballistix Tracers thread.
I know this but where do the light draw their power from? Also, I've done some testing on @ 1120Mhz 4-4-4-4 @ 2.45vdimm (set in BIOS) with a P5B Deluxe. The P5W DH does not like this mem for anything when it comes to 1000Mhz, I tried different dividers with no luck, however it's doing quite well with 800Mhz 3-3-3-4 @ 2.35vdimm. I know it's not my P5W DH because I have some D9GKX and does well with it above 1000Mhz. I do not have my Regular Ballistix PC2-8000 here but I'll get a hold of it and test it in the P5W DH to see if there's a difference.Quote:
Originally Posted by bofors
I just got mine. I can't even pass memtest86+ 1.7 with stock settings. Anyone got similar setup as mine and did some memtesting on these and actually had it pass couple times? I can get mine to pass memtest with certain settings but it seems like these sticks are really sensitive and they will throw errors if you don't throw good voltage at them. I usually get couple errors in the same area on the same test. If you got them please do some memtest! I can orthos all day and it will be stable. My system feels stable, but memtest will give me couple errors.
Ok I got mine to 800 with 3-3-3-8 @ 2.2V and it passed 300%+ Win-Memtest & even went thru with Pi32M. Will post screenies once I get Home.
I don't think win-memtest will test your mem as well as memtest from dos. Because when you're in windows, windows uses your memory so you can't test the whole 2048mb you have.
Agreed, but I got bored of doing things. But after Hard-Work, testing things gets really BORING!
Ok, I was too quick to blame the memory for all my problems.
Just because you get errors in memtest doesn't mean it's always the memory. I found out that my overclocking settings were making my memory fail. There's a lot for me to say but some of you don't want to hear the rest. Just know that you should run your computer at stock speed and test the memory before you do any overclocking.
Right now it might be too early to tell but I'm passing memtest at 533mhz 4-4-4-12 2.25v. This memory is good stuff. I will let it run for the night then we'll see what sup in the morning.
By the way, anyone here wanna share their sub timings with me? What's a good sub timing for the 10-10-10-10 on my motherboard? I don't know what to set those? I think I need to loosen them at high clocks. Like 600.
Quote:
Originally Posted by bofors
I also think that is a great idea but how should we go about this? We should come up with some guidelines on how to test each kit so that we can actually do some objective comparsons (unless we can come up with some other way to compare the different batch numbers)? Anyway guys post your thoughts on this, I'd defienlty be game for posting results on these kits to get a better idea of what batches do better than others!
TO7N,
In my case I tested a bunch with voltage in the 2.0v-2.25v and results were less than optimal but as soon as I started putting some more voltage to them the fun began! These kits seem to scale very well to voltage and I feel quite safe running 2.4v-2.45v with active cooling, at least I havn't had any major problems yet (fingers crossed)!
I think I got mine very stable for 24/7 usage at 600mhz 5-5-5-15-5-35-11-11-11-11 2.35v in bios. Sub timing matters... I couldn't get my memory stable no matter what I did until I changed the tRFC and the other sub timings. I couldn't even get them stable even if I pumped more voltage into them.
I ran orthos and memtest for a couple hours and no errors. I'm gonna make sure even more by totally powering the system off and turning it on again and do some more test. But right now I'm very happy with my memory.
Here are some numbers:
http://img293.imageshack.us/img293/3317/mem6ye.gif
I don't know why the latency test have different results when they are done differently.