dddr2 prices are getting so low that one can just order a selection and pick the best ones for use:up:
1080MHz at 1.95 volts :)
http://www.edgeofstability.com/images/gskill/gskill.jpg
these memory modules use PowerChip IC's?
The PC6400 version has other IC's indeed: the F2-6400CL5D-4GBPQ uses Micron D9HNL.
Dont forget to post also memset settings just for reference!
1090Mhz with only 1.95V?...very impressive!. I had mine clocked at 1065Mhz with 2.1V...passed a 10 hour prime test. Must try lower voltage.
So we can buy the PC6400 ones and have the same scores
Does anyone run 4 of these sticks in same config ?
What are your results?
I´m thinking of getting 4x2GB of these nice sticks.
I saw some1 runned on a maximus with these sticks, 4x2GB with 4-5-5-12, with 2.1v I dont know it is stable or not, he didnt showed sshots, but just said that in the Maximus thread. But why 8GB? 4GB should be more then enough for now:X, but it so cheap, why not :up:
Got a pair of these for $115 from the 'Egg, what a deal. Currently running 1102MHz, 5-5-3-15 (nice tips in the first post EnJoY!), 2.15V think I'm going to stop there and try priming it over night. Unbelievably awesome RAM for the price though :D.
Okay, well, I'm going to go out on a limb here and say that this RAM doesn't like my Gigabyte DS3 (965P).
It does 5-5-5-15 @ 2.0v , 1000Mhz.
CPU is an E6420 that's doing 400x8, 3200 Mhz @ 1.375v. It's been the most consistent OC I've ever been to attain and it works quite well.
That being said, I absolutely cannot push higher than that. With my Crucial Ballistix 2GB DDR2-1000, I was able to hit 3.41 Ghz @ 1.45v on 2.2v (5-5-5-15, 10xx Mhz?) stable 4hours+ on Orthos. Now, I do the same - push voltage to 1.45v, 3.31 Ghz push with corresponding 10xx Mhz RAM timings on this G.Skill. It Memtests fine, but I get random BSODs within the OS.
Is it just that my CPU is finally beginning to suffer wear/tear? It's been on 1.375v and 3.2 Ghz since I purchased it in May and has rarely been off (only ~1 week in mid June). Other important specs include a Corsair HX520, Scythe Ninja on CPU (never breaks 60C, even under Orthos 100%), 8800GTS...
Its probably not the CPU but the memory controller on the MB being taxed beyond its limit. The move to 4GB from 2GB is smoothest usually sticking with 2 sticks as opposed to 4x1GB but even so, its not uncommon for somebody to lose their stable OC that they had with 2GB, as opposed to 4GB.Quote:
Is it just that my CPU is finally beginning to suffer wear/tear? It's been on 1.375v and 3.2 Ghz since I purchased it in May and has rarely been off (only ~1 week in mid June). Other important specs include a Corsair HX520, Scythe Ninja on CPU (never breaks 60C, even under Orthos 100%), 8800GTS
I had to lower my FSB when I moved to 4GB from 2GB and I have active cooling on NB.
I got 11 kits to test, this is good memory for the price.
One important thing for everyone to know,
neither Memtest86+ nor Prime95 aren't reliable memory tests, at least for modern chipsets + memory combinations.
Both can pass many hours without problems whereas Linpack fails in several iterations.
For those truly seeking rock solid stability, Linpack test is absolute must. Yes, it will cost some grief for many :).
Watch out for the temps while testing - 64bit version is the most demanding stress tester for Core2 ever known, temps will be substantially higher vs TAT(all cores), much much higher vs Prime95 Small FFTs. Even on water,
so air coolers beware!
I have four machines with 2 kits =8GB in each on P5K Pro boards.
This memory gets noticeably hot when running memory-intensive jobs. What is worrying me is very small clearance between heatsinks of adjusted memory modules. I wonder if it whould be better just remove the heatsinks, they are not particularly good quality and the contact they make with memory chips is not good either. Any thoughts on this?
Just an update. Seems I need 2.25V on this stuff to get it to 1100MHz 5-5-3-15 stable 24/7 in folding. Oh well, still good RAM and its still running very cool (granted I have a low speed 80mm blowing on them).
I can't get a pair 4x2GB stable 1000MHz 2.0V, it fails Linpack. Sometimes it takes several hours, but fails anyway. I hate to go higher 2.0V as it gets extremely hot even at 2V.
Strange at 1.98 it can pass several tens of iterations and at 2.04 fails in several first iterations. I am testing at 1.9V now in the crazy hope of desperate :)
Next one will be 2.1V, if it fails, i may as well start to panic :)
Update:
12hrs of 4x prime95 blend, all memory tested, passed no problem 1.9V
Memtest86+ passed numerous times.
Yet my infiniband network card hangs almost immediately together with Linpack.
So much for this "memory tests" :(
Now for 2.1V, the last resort.
Where can one download Linpack?
I don't think so. Increasing vMCH has no effect. There is no memory VTT on my board.
2.1V seems to be working, the test is in progress.
I'd rather think most ppl with high overclocks do not test thoroughly enough. This is also true for almost all high CPU overclocks.
Hate to think if those memory manufacturers also rely on tests like Memtest86+, this would be a total disaster and explain why there is so much complains.
Memtest86+ and Prime95 blend are not reliable enough and can only be used as coarse approximation.
I recommend always include Linpack. It is also extremely good CPU test.
Should become a standard test, like Prime95 is de-facto standard test for now.
Can be obtained here http://www.intel.com/cd/software/pro...eng/266857.htm
I have ordered a kit of this stuff for my rig and I hope to hit 1120Mhz. I'll definitely follow all the tips in this thread but how likely do you guys think hitting this speed will be? I've not seen any results with this RAM on AM2...
Just lost a Kit of this after 60 days of crunching on WCG @ 1.95V 850mhz 4-4-4-15 timings. Got something I never saw before, a Hardware Failure BSOD on Vista 64, replaced the kit and all is well.
Just a heads up for those tightening timings and running 24/7 rigs.
Its down to 114.99 at the egg.
Sadly, eventually fails 1000MHz 2.1V. :(
just got this mem :D
i can run 550@2,2v sub timming still auto
http://valid.x86-secret.com/show_oc.php?id=288157
http://img242.imageshack.us/img242/6258/imag0096zi9.jpg
http://img242.imageshack.us/img242/8307/55055515bk7.jpg
but i try to set the sub timming i cant even stable at 555?raising the vdimm doesnt make any difference only make memtest more worst.
any idea?
Let me assure you, you are not even stable at 550.
I ordered some yesterday.
Does anyone run these at 1:1?
You keep saying this, but you have absolutely know way of knowing this. Maybe it wasn't stable with your sticks, but it doesn't mean his sticks or anyone else's can't do those speeds completely stable. I personally can run 550MHz 5-5-3-15 with somewhat tight sub-timings with absolutely no problems whatsoever, 24/7.
I can run these 1:1 at 500MHz on my IP35 Pro no problem. :up:
Mine should be here today or tomorrow and testing will begin.:D
Is 2.2v to much for these sticks for 24/7 operation with active cooling eventho they are just rated up to 2.1v?
what is the max I should expect up to 2.1v? 1066 -1100 with 5-5-5-15 possible?
sure But I just got roughly a idea what I can expect from these sticks, thanks
Sorry, i am not trying to be rude or distrusting, but i think people should have as much information as possible.
I have no reason not to believe you (or anyone here to this extent ) when you say you have no problems, though there could be a really long way from this point to the real stability, i.e. continuous full-load error-free operation.
If you could elaborate on your testing method, this would be of very much help.
If you could run extensive Linpack test and confirm stability, you would have done a really great job. If you need any instructions, just ask .
On the DFI X38. The same 1.95 voltage as I posted with the P35 board. Haven't set any secondary timings yet. They are still on auto. The benches were run with the 8x multiplier. 6x is showing because of C1E being enabled.
http://www.edgeofstability.com/image...misc/1080a.jpg
Has anyone tried these rams with DFI P965-S?
Q6600 on the P965-S
Q6600 with 2x2GB GSkill DDR1000 memory
All three operating systems were run with the same hardware configuration. 3dMark 06 was run immediately after a fresh install of the OS. Because of C1E being enabled, screenshots were taken with Prime95 running so that CPU-Z would show the correct multiplier.
Vista 64 Bit
http://www.edgeofstability.com/image...imev64srun.jpg
http://www.edgeofstability.com/image...mev64sstop.jpg
http://www.edgeofstability.com/image...e/3d06v64s.jpg
Vista 32 Bit
http://www.edgeofstability.com/image.../primev32s.jpg
http://www.edgeofstability.com/image.../3d06v32ls.jpg
XP 32 Bit
http://www.edgeofstability.com/image...e/primexps.jpg
http://www.edgeofstability.com/image...me/3d06xps.jpg
Can you plz post your voltages and dram settings? (The first which affect drive strenghts etc..)
With the 4:5 divider it is hard for the chipset to maintain stability.
I don't have the settings/voltages any longer. But 4/5 isn't an issue until 420 FSB or so.
Hi!
I have these memories with the computer of my signature but I cant't let down the timmings. Why?
I have the bios config with 475*8 and the memories at 950 Mhz. It's stable with the memtest for windows with 1.9V. But if I try 5-4-3-12 or 5-4-412 with 2.0V or 2.1 V, it isn't stable even a minute with the memtest for windows. Is only stable with 5-5-5-15 timmings,not less...
Why?
Another pair of sets 4x2GB , still can't make them work reliably at 1000MHz 2.0V on ASUS P5K-PRo. Neither 5:4 nor 6:5 dividers are no go.
I wonder is it possible at all on P35.
I cant run lower than 5-5-5-15 but i can run 1140 Mhz 2.1 V 5-6-6-18 without error in 2 passes of memtest... and its stable without problems at 950 MHz 5-5-5-15 1.9V....but running with 5-4-4-12 fails with 2.1 V...fails at 2% of test... ?¿???¿?¿ Is it a bios configuration problem?I need change any parameter?
I can't even get the G.skill 2x2Gb sticks to see the full 4gb on my abit ab9 quadgt in the bios or on bootup. It will see 3gb, just not 4. The individual sticks will see 2gb in every slot, but not all 4gb in any slot combination. Anyone have this problem?
Man, even 200 passes in memtest86 means nothing, it can only be used as a first crude guideline.
I can pass numerous passes in memtest86+, yet Linpack fails very fast.
Of 11 kits i tested with memtest86+, at 2.0V 5-5-5 with 20MHz step, one passed 1060MHZ, 6 passed 1080MHz, 2 passed 1100MHz, one 1120MHz, one 1140MHz.
I have to make a notice guys.
The speed/timings you post, are they tested for stability? Or they are just for bench?
I' m saying this because i try to use EnJoY's timings but memtest fails. I run 3 instances o memtest for Windows simultaneously (so that 100% of ram is tested). My OS is Vista x64.
Orthos (blend) does not test the ram fully as it can't use 100% of it.
I'm having a hard time stabilizing the sticks at even 1000mhz. Memtest keeps failing.
Thought I would chime in on this, I picked up a set of these and I am running at 1.95v at 500.1 Mhz 5-5-5-15 at 5:4 on a P5K Premium.
I just got this ram in the mail about 45 minutes ago. ITS PRETTY!
So I downloaded linpack to test it along with using it as another CPU stability test. Well, where is the exe file?
This is where I got it: http://www.intel.com/cd/software/pro...eng/363184.htm
Is this the right file? It seems to only be for xeon cpu's. :shrug:
Synthetickiller
Here is the quick practical instruction on Linpack:
Open the file lininput_xeon(32,64) pertaining to you OS arch., 32 bit or 64 bit.
Determine the maximum problem size your memory can hold. The rule is simple : take the whole amount of memory you see in your OS, subtract 500MB for the OS needs (may need more for memory-hungry Vista), the rest can be assigned to Linpack job. In any case, do not let the test run from disk swap! The amount of memory Linpack uses is equal 8x (problems size)^2. So, for the problem size 10000 the amount of memory would be 800MB. For 4GB memory (needs 64bit OS to see), 3.5GB can be used for Linpack test, this amounts to sqrt(3 500 000 000/8) ~ 21000 problem size.
Set 1 for #number of tests field,
Set only the maximum problem size in the # problem sizes field,
Set the same number in the next field #leading dimensions ,
Set 100-200 for # times to run a test,
Leave the last alignment values as is.
My example file for 8GB RAM under Win X64 (Linux x64 actually, but the file is the same) looks like this:
Sample Intel(R) LINPACK data file (lininput_xeon64)
Intel(R) LINPACK data
1 # number of tests
30000 # problem sizes
30000 # leading dimensions
200 #times to run a test
4 # alignment values (in KBytes)
For 4GB, substitute 30000 with 21000.
To run the test, run runme_xeon(32,64). The best way is to run it from Far manager (google Far manager and download it, its free) or similar program.
The results will be in in .txt file with the name lin_xeon(32,64).txt.
There you will see the summary of input data, string
Size LDA Align. Time(s) GFlops Residual Residual(norm)
And for each test ran, there will be a result string like this:
21000 21000 4 129.620 41.1519 4.455000e-10 3.943651e-02
Error: info returned = 1
The fifth value (41.1519) is gigaflop rating obtained in this run, the seventh (3.943651e-02)
is Residual (norm) value. The last is most important part. It should be small, of the order of 10-02 (0.01),
and should not change from run to run with the same problems size. Any abnormal value here indicates
Linpack fails the test, though it will continue to work. This is very sensitive test, actually.
Ignore the Error:info returned field.
P.S.
If you have 64bit OS, run 64bit Linpack but beware! Monitor your temps, you will see something you have never seen before :)
32bit Linpack taxes CPU considerably less, produces much much lower temps as well as benchmark results (Gigaflops rating).
P.P.S.
I was assuming you have Intel CPU :). 32bit Intel Linpack will probably work on AMD, 64 bit -probably not.
There is a way to run non-Intel Linpack on any CPU, of course , but this will require re-compiling the program from the source code.
Thanks for the explanation I was entirely lost.
I opened linpack in CMD and i select data and input the proper values (I have 4 gigs and vista 64, so its all good) but once everything has been input, the CMD just closes. My temps are fine and not moving nor is my CPU usage above 2%.
I am opening the linpack xeon64 app.
Anything stupid I might be doing or overlooking?
81 reviews and counting!
http://www.newegg.com/Product/Produc...82E16820231145
Thanks, and you were right to assume I have an intel (in my tag). I will get to this when I have a chance. Hopefully I won't be a bonehead and I'll figure it out.
I really appreciate the help.
@ either R*s or EnJoY....
What mobo are you using? I'm using a p5k Deluxe. In the bios, I can't get 5-5-5-20-55-10-11-11-11. I can only go to 42, not 55. I have that option set to auto instead (which might be causing a stability problem). Any hints?
Well, 42 is waaay too tight for these sticks. I can do 48 on my P35, but no lower with any stability. Try auto and look in memset to see what it sets the value as.
48 is as low as I can go with the DFI X38. Anything lower produces errors in Memtest.
Can i get them anyware else besides newegg, with credit card maybe?
Thanks EnJoY, its set at 52 right now. Any hints? I've never used memset before.
http://i13.tinypic.com/6xg7vyo.jpg
Oddly, I'm not stable with my current set up for longer than 25 minutes. I noticed moving the tRAS from 15 to 20 made a difference of about 10 minutes. So strange. This is at 2.0v btw. I figure thats more than enough for 840mhz.
I know the cpu can 420x9 so its not a cpu limitation. Blue screens at 425 right now.
The patch is a windows update thing or its a manual install? Know the name of the patch?
I don't have Vista and don't know the details. Google for this.
Hahaha!F U C K!
I set at 1140 Mhz...570 x 2 with timmings 5-7-7-20 2.1V...In 4 hours since memtest for windows an error appeared!!!!($%·%·"!!!!)
http://img144.imageshack.us/img144/6...t800pd2.th.jpg
¬¬
Sorry!xD I think 20 hours of test and 1 error is stable... i think...
Here another tests:
950Mhz(475fsbx2) 5-5-3-12 1.9 v
http://img246.imageshack.us/img246/2...5312wr2.th.jpg
950Mhz(475fsbx2) 5-5-3-12 1.8 v:
http://img175.imageshack.us/img175/7...5312pq4.th.jpg
If i try set 5-4-3-12 with 2.1 V(or lower) memtest fails in 1 minute...Stranger
Try memory orthos there to test more.
Loosen to 5-5-5-x to see what they can do...and then start to tighten
I'm 4 hours prime stable. This is the best I've had all day. Sadly, the timing are really loose. I'm hoping that I can get the timings down to something like 4-4-4-12 so I can game w/o any real downside. If I get 8 hours, I'll up the FSB. I'm hoping I can get 4.0ghz 24/7 stable (445x9 or 500x8) although I think my voltage will stop me. I'm at 1.55v now, though I know 3.825 will be stable at that voltage.
http://img114.imageshack.us/my.php?image=primesa1.jpg
Here is a CPUz screen.
http://img.villagephotos.com/p/2005-...cpuzscreen.jpg
I have to agree, but also note this Maximus would be the only board to ever get along 100% w/ my X-Fi, have a perfect layout, and of course good lookin' through a Window. :D
We'll see what happens when I plug it up w/ a pair of ATI's finest and a nice Penryn tho. Judging by the sheer number of posts/interest in the board, I think I'll be alright. :D
Interesting, i am not yet 100% sure but it seems tightening tRAS from 15 to 12 noticeably helped with stability in 4x2GB configuration at 1000MHz.
Needs further testing.
Cronos, I might try that. I'm going to do some stability testing tomorrow w/ 450x7 1:1. I can't boot 475x7 sadly or even 500x6, but i think i need more Vcore for that.
This is stable now? :D
http://img175.imageshack.us/img175/7...5312pq4.th.jpg
http://img99.imageshack.us/img99/8533/24hsu3.th.jpg
This is with 1.8 Vmem 5-5-3-12 950 Mhz.
so what speeds is everone else seeing...has anyone reached 1100+ besides enjoy? also would the 800 ones oc to the same point as these?
Has anyone booted 500mhz 1:1? I can't even boot 500x6 w/ 1.5v vcore and 2.05v for the ram.
A screenie would kick some major @$$!
Don't opt for 1:1, this is pointless with Intel chipsets like P35.
First, i highly doubt anything close to 500MHz can be stable for serious 24/7 use with Quad-cores. ~450MHz is practically reasonable FSB speed with P35, and likely with X38 too. Lets hope X48 opens the road for really stable 500+ FSB.
Second, at least P35 works best with 5:6 memory divider. At the same memory speed, there is virtually no improvement 1:1 provides over 5:6 for bandwidth-sensitive apps.
Ah ok. I'm actually doing fairly well w/ OCing now. Dropping the timings to 5-5-5-12 has made a HUGE difference. I'm almost 7 hours prime95 stable. Right now, I'm testing 400x9. I'm hoping I can push it eventually to 445x9. You know, we all would love to break 4.0ghz. If I can even only get to 3780 like I was before, I'll be happy.
I think once I get down to brass tax and find the highest OC w/ a 9x multiplier, I'll drop back down to 6 and see if there is a way to do 500x6. 450x6 was no big deal, although the yield for the bandwidth SUCKED.
If anyone out there can do 500x6 or higher 1:1, please post it along w/ mobo and CPU (conroe c2d or c2q please). I'd love to see if anyone has done this yet.
EnJoY, thanks again for posting info on this ram. It runs MUCh cooler than my ballistix. Awesome ram. I've never tried G. Skill before this, but I am glad I did.:up:
If I want to test this kit with either HCI or Prime95 Blend, How much should I put in Memory to use: ? Standard Prime blend is 1750MB.
And should I change to MIn FFT size (in K) ? ( default = 8 ) And Max FFT size ( in K) 4096. Any help is appreciated.
Or any suggestion for a sollid test? I had in mind 12 hours blend and 1 hour HCI.
my freind got his today and said after 3 hours or so he simply can not get anything good out of them. he said he couldnt even run then at 1000 with the stock timings.. he eventually got them to 1040 but didnt check them to see how stable they were anything over this even with 5 5 5 12 or looser he could not get more out of them???
Zfactor, it took me a few days to really nail anything stable. I'm sitting on 400x9 right now and I'm going to test out 420x9 tomorrow.
What speed are you trying to run your processor at? What processor is it? Is it 1:1 ?
I'm on my second pair of kits now (that is, second 8GB set) and it also looks stable 1000MHz 5-5-5-12 2.0V. tRAS 12 seems to make a real difference at least with 4 DIMMs.
this will for now be used with the e6750.. but ill be buying a e8500 when they are for sale.. for now the e6750 though.. not sure of speed since i have yet to try a oc on it cause i have standard hp ram i pulled out of a system i was given so i can oc at all right now with that stuff if i even change anything at all it just reboots or bsod's.. i want to know that i can use the ram at close to anything i want though once the e8500 comes in..
I am building my this-year project (2007 actually) - computing cluster with infiniband network. It will include 4 computing nodes based on Q6600@overclocked, P5K-Pro (for max. energy saving), 8GB RAM each (32GB total). All will be watercooled and placed into custom-made compact chassis. Plus infiniband switch of course, if this is saying anything to you :)
This G.Skill ram cost me about a week in delay and lots of frustration. I hope this tRAS 12 finding will help me out of trouble.