Magic Mem: post your samsung results

[Soulkeeper]

I just picked up 4x4GB samsung 1600 30nm
running 1600 8-8-8-22 1.35v right now
7-7-7 and 7-8-7 weren't stable even with 1.5v
tried 1866 speed, but latencies under 10 weren't stable
Many factors to consider as I try other speeds/timings.

what results are other AT 30nm owners getting ?

 

[sequoia464]

1866 1.35 volts 9-9-9-24-45

This is on an Asus AMD 990FX board

Edit ; Just using two 4 GB sticks.

 

[Kodongo]

8GB (2 x 4GB)
DDR3-2133 @ 9-10-10-27-1T (1.6V)
w/ 2600k and Asus P8P67-M Pro

This thread is your cheat sheet to overclocking this memory.

 

[Concillian]

2x4GB @ 1.5v
Mine were pretty iffy at 1866 9-9-9. Usually stable for short tests of 10-12 hours, but would ALWAYS fail for 24 hour tests.
Had to run 10 at 1866 or 8-8-8 at 1600 for total stability.

 

[Soulkeeper]

thanks guys

I'll have to read that link
I just tried 1866 with various timings up to 10-10-10-27 1.35/1.45/1.50v
it don't seem to like that speed.
I guess if I went higher volts it'd stabilize

I am impressed with 8-8-8-22 1.35v 1600, that alone should beat any of the 1866 speeds with 10+ timings

for $30 a kit, this mem is a good deal imo

 

[readymix]

P9X79
4 x 8 dimms
1866, 1.5v, 9-9-9

 

[sequoia464]

P9X79
4 x 8 dimms
1866, 1.5v, 9-9-9

Where did you find the 8gb sticks, Been waiting on those to come out.

 

[readymix]

Where did you find the 8gb sticks, Been waiting on those to come out.

correction- 4gb x 8 dimms

 

[Soulkeeper]

I can't get 4x4GB stable at 1866
11-11-11-28 with 1.52v didn't even do it (always fails at test #5 in memtest86)

passed 12.5hrs of memtest at 8-8-8-22 1600 1.4v just now

my mushkin (hynix) 4x4GB kit did much better at high speeds in the same system
but couldn't touch these low latencies at 1600, even hold TrFc at 90ns which i've never seen before.

 

[Soulkeeper]

I figured out why I couldn't hit 1866

one of the two kits is the kulprit

1866 9-9-9-24 1.4v
booted with 2x4 failed memtest at #5 like every other attempt.
swapped it out same settings/slots and passed fine

Both kits have the same numbers and month/year markings
Maybe i'll order another set and chalk this up to luck of the draw

 

[Z15CAM]

You really should not run Dimms over 1.5v's. Might be able to push it to 1.55v without damaging the SB & IB Processors IMC or up the VCCIO Voltage a touch & the CPU Voltage a notch to stabilize the Ram at Higher Clocks.

I just purchased 2 x 8 GB Kits of Samsung Green DDR3L MV-3V4G3D/US for $91 including Tax & Shipping from NewEgg.CA to go with my ASUS P8Z68-V Pro Gen3 / i7 2700 / Corsair H100i build. Sure looking forward to getting that Rig up and running ;o)

I've heard about incredible stable timings like 1600MHz ( 7- 7- 7-20 1T) @ 1.50v and 1866MHz ( 8- 9- 9-24 1T) @ 1.50v and 2133mhZ ( 9- 9- 9-24 1T) @ 1.50v and has been pushed to 2760MHz (11-13-13-35-2T) @ 1.65V with that Ram.

 

[Soulkeeper]

I did further testing on the kit that won't do 1866 speed
turns out it's DOA, passes memtest at 1600 but if I really stress it IE: compiling code it'll segfault or other random errors
Not even worth returning given 2 weeks round trip and being out shipping costs.
I'll just try to pickup another kit locally

Z15CAM -- I agree, above 1.5v is not something I wanna do
I see people pumping 1.65v or higher into these on forums ... crazy

 

[Soulkeeper]

Ok, I got another set and the results are similar.

I suspect that the bios in this motherboard (gigabyte A75M-UD2H) is incompatible with samsung's new memory. Either that or this mem isn't so magic afterall.
I can't even get 1600 fully stable with 4 sticks
I suppose I could pester gigabyte to see if they can recreate the errors that i'm consistently able to recreate and hope they release a bios update ...

I guess i'll just go back to the hynix based mushkin memory I have


Any thoughts ?

 

[Z15CAM]

You got me concerned now - LOL

What CPU you running?

 

[Soulkeeper]

You got me concerned now - LOL

What CPU you running?

A8-3870
I really think the memory can do good speeds, but there is an issue with the bios/compatibility. I'm gonna try the mem in a different motherboard and see what happens.

 

[spinejam]

Samsung Green -- 4x4gb @1.5v



Uploaded with ImageShack.us




Uploaded with ImageShack.us

 

[Z15CAM]

passed 12.5hrs of memtest at 8-8-8-22 1600 1.4v just now

That's not to shabby

AMD A8-3870 Llano 3.0GHz Socket FM1 100W - Not familiar with it but I heard if you add a little voltage to the VCCIO & add a little more voltage to the vcore of an Intel CPU it tends to stabilize the IMC for better Dram OC's - Don't know how this equates with an AMD CPU or if you have such BIOS Tweaks.

 

[chimaxi83]

I had been running 8GB at 2133, 9-10-9-27-1T at 1.55v with my 2600K. I now have a 3770K and 16GB, and am running the same speed at an even 1.5V.

 

[Soulkeeper]

Ok, after testing the memory in a different motherboard (asrock) same cpu/chipset I'm able to get 1866 9-9-9-24 1.4v fully stable
8-8-8-22 wasn't stabilizing even up to 1.5v but was worth a shot

So this confirms that the gigabyte bios just didn't like the samsung stuff

 

[WiseUp216]

4x4GB @ 1866 9-9-9-24 2T

That was straight out of the box with no tweaking. I just set the values and they've run stable for a couple of months now.

I've thought about dropping to 2x4GB and pushing those as far as they'll go and keep the other 2x4GB as spares.

 

[GPz1100]

Ok, after testing the memory in a different motherboard (asrock) same cpu/chipset I'm able to get 1866 9-9-9-24 1.4v fully stable
8-8-8-22 wasn't stabilizing even up to 1.5v but was worth a shot

So this confirms that the gigabyte bios just didn't like the samsung stuff

There's more to ram settings than just those 4 parameters. In fact, there's at least a dozen or more other parameters that need to be set properly for the ram to function properly at higher speeds. 1.5V is standard for ddr3. If your other board shows what those extended ram parameters are, you could record them and enter them manually in the other board.

Bios does play a big role. In the board in my sig, I wasn't able to get the ram @ 1866 unless the most recent bios was flashed (recent as of last month). Setting those extended parameters manually had no effect. It was as though the bios was ignoring/not applying them.

Ymmv.

 

[Soulkeeper]

There's more to ram settings than just those 4 parameters. In fact, there's at least a dozen or more other parameters that need to be set properly for the ram to function properly at higher speeds. 1.5V is standard for ddr3. If your other board shows what those extended ram parameters are, you could record them and enter them manually in the other board.

Bios does play a big role. In the board in my sig, I wasn't able to get the ram @ 1866 unless the most recent bios was flashed (recent as of last month). Setting those extended parameters manually had no effect. It was as though the bios was ignoring/not applying them.

Ymmv.

Yes, I realize this. And trust me I spent 2 days trying every possible timing, subtiming, voltage, cpu setting, etc. Where 4x4GB hynix was stable(fully) samsung was not with the gigabyte A75 motherboard. I tried 2 different bios versions. It is worth noting that there are various settings which aren't exposed in most any bios, and thus require a bios to be updated/validated with new memory from time to time by the manufacturer.
Gigabyte needs to release a bios update to the GA-A75M-UD2H Untill then samsung 30nm will not do 1866 and will not do 1600 (fully stable) but will post/boot/benchmark at 1600 with the fastest timings(or any looser variety), but always fail when compiling code or gaming.
As i've said above: Ohh well, it's out of my control
I have other plans and have moved on.

I'll post a picture showing the tight subtimings I was able to get with 9-9-9-24 1866 1.4v in the asrock board in a bit
EDIT:
One
Two