Sign in with
Sign up | Sign in
Your question

1.35v DDR3 and possibility of tighter timings

Last response: in Memory
Share
March 7, 2012 12:48:28 PM

Hi all,

I am in the proccess of upgrading my system. As of now I have:

i5 2500k
Asrock Extreme 4 Gen3

So now I am looking to add the ram to the system. I am getting 8gb (2x4). Trolling Newegg I see pretty much pickup most cas 9 1600 for $50. My question is if I can pickup 1.35v Cas9 9-9-9-24 for $50 particulary:
http://www.newegg.com/Product/Product.aspx?Item=N82E168...
or
http://www.newegg.com/Product/Product.aspx?Item=N82E168...

What are the chances that I could tighten the timings down a tad if running this ram at say 1.5-1.65v instead of getting ram that is already rated to run at 1.5v-1.65v @ 9-9-9-24? Any help is appreciated.
a c 347 } Memory
March 7, 2012 1:15:00 PM

Both kits are DDR3-1600 9-9-9-24-2N @ 1.35v, and if by tightening down you mean i.e. DDR3-1600 8-8-8-24-2N with a high voltage?

Then I'd say a slim chance. My experiences with 1.35v RAM is they 'can' run @ 1.50v and if you run them say at 1.65v then that's dangerous to both your Sandy Bridge CPU and RAM. About the highest voltage you want is 1.55v. By default until you set the XMP or otherwise set up the RAM it runs @ 1.50v. Further, don't exceed a VTT or VCCSA > 1.20v and in most cases the testing was done at 1.0v~1.1v.

In general, most 1.35v kits aren't 'good' for OC'ing the RAM much if any higher than what they're rated. However, the litho of the IC is very efficient and typically even at Rated out perform their 1.50v counterparts.

Further, on Sandy Bridge that is unless your plans are to OC the i5-2500K >4.2GHz~4.5GHz+ then anything >DDR3-1600 CAS 9 does nearly zip to enhance performance other than useless synthetic testing; see - http://www.bit-tech.net/hardware/memory/2011/01/11/the-...

If your plan is to OC the i5-2500K >4.2GHz then I'd look at Corsair Dominator or similar ultra-high quality IC's and DDR3-1866 or higher with low CAS, but they're expensive and off only a tiny increase in overall performance. Most RAM testing is done at the CPU @ stock and not OC'ed.
m
0
l
March 7, 2012 1:31:14 PM

I Appreciate the feedback!

I do plan to overclock the i5 to about 4.5ghz maybe 4.7ghz. I am not to worried about ram overclocking, I'd be actually happy to run them at stock as I know ram overclocking provides minimal benefit.

I really don't want to drop $ on cas 8 or 7 ddr3 1600 ram. I think the cas 9 should work just fine for me. I was just curious if the lower voltage stuf would yeild the possibility of tighter timings.

So with all being equal and not wanting to overclock the ram but I am overclocking the CPU would it be better to go witht he 1.35v or the 1.5v rated ram?
m
0
l
Related resources
March 7, 2012 1:58:31 PM

@jaquith
in your experiences, could you raise 1600 to 1866 between 1.35V & 1.5V? of course, the timings would be higher
and have you used low-voltage ddr3 on amd platform? maybe you could help me at this thread low-V ddr3

@stylez777
since you're on SandyBridge & not planning on OC'ing the ram, then it would be better with the lower voltage
m
0
l
a c 347 } Memory
March 7, 2012 3:17:42 PM

@drabun, in my case I have 32GB (2 Kits) of Corsair CMZ16GX3M4X1600C9G which is their X79 kits, and mine failed to OC trying DDR3-1866 9-9-9-24-2N using 1.40v with a VTT and VCCSA of both at 1.10v and then both at 1.20v. However, in my instance I have 8xDIMM and it only takes (1) stick to botch the OC. Most importantly, I need STABILITY using both a SSD + RAM Cache and RAM Drive; the last thing I want or need is data corruption!

In general, the Corsair Vengence work @ Rated fine, and maybe with (1) kit going from DDR3-1600 to DDR3-1866 without some goofy DRAM voltage 'fix'; in other words they're not good to run much higher than @ Rated.

In contrast, the Corsair Dominator's are indeed fine to OC above Rated. The improvements though, other than synthetics or Super PI, aren't worth IMO doing. The X79 does benefit 'more' from higher frequencies (>DDR3-1600) IF the CAS Timings can remain low, but that's no so much the same for Z68/P67. The LGA 1155 default is DDR3-1333 and the LGA 2011 DDR3-1600.

Bottom-line, get expensive RAM >DDR3-1600 IF everything else is the 'fastest' it can be, and otherwise RAM, to a point, dollar-per-dollar yields the lowest performance increase. I.E. use the extra money for a better CPU, GPU, SSD, etc...for 99% of the the builds.

For example, say I could get the CMZ16GX3M4X1600C9G (DDR3-1600 CAS 9-9-9-24-2N) to run stable at DDR3-1866 CAS 10-10-10-30-2N then that would defeat the purpose. Here's a good chart and I know I need to update through DDR3-2600 but you'll get the idea. The idea BOTH Frequency + CAS Timings determine overall 'speed' as shown:
m
0
l
March 8, 2012 6:11:22 AM

ah yes comparing the timings in the actual seconds would be the ideal way in OC'ing ram, but currently i'm planning to build amd APU sys, & there's a confirmed increase in graphics performance that's too good to be ignored, by going from 1333 to 1600 even while sacrificing 1 point of CL (benchmarked here llano review)

but thanks man, you make me give up the idea of OC'ing cheap low-voltage ddr3, i was attracted by the range of safe voltage (eg. 1.35V to 1.5V), thinking there could be great OC at the range. now i can simply grab cheap low-/normal-voltage ddr3-1600 &, as you said, spend money where performance increase is more noticeable & guaranteed


btw, you might want to change the label 'Column Address Strobe (cycles) / CAS Latency (ns)' in the chart to something more general like 'tCK in ns' coz most other timings (such as tRCD, tWR) are also multiplications of tCK which is the mem IO period

thanks
m
0
l
a c 347 } Memory
March 8, 2012 11:46:01 AM

Ah yes, this is a Sandy Bridge thread.

The AMD is not a good place for 1.35v RAM, and DDR3-1866 is the default frequency for the FX. Further, regardless of the RAM it's not going to touch the performance of the Intel Sandy Bridge.
m
0
l
!