• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Looking for 16gb (4x4gb) ddr4 kit that will do lowest timings at 2400mhz @ 1.35

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Brando

Member
Joined
Jan 9, 2006
I decided to stick with 100bclk for my 5820k oc to avoid random odd behavior and rule out problems with straps. My cpu doesn't want to go over 2400mhz ram without going to a 125 strap so I'm going to aim for 2400mhz with the lowest possible timings. so far my ripjaws V kit seems ok at 2400mhz @ 13-14-14-30 @1.35v but i'm wondering if i can do better without spending a ton. is there anything new out there now that can do better? i remember back in the day the bare chip green samsung ram was the go to for manual tweeking. what's the best now?
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
I wouldnt bother, honestly.. for such lititle gains. I'd rather work off 125 strap (no issues here when using it) and push what I had or leave it alone.
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
I wouldnt bother, honestly.. for such lititle gains. I'd rather work off 125 strap (no issues here when using it) and push what I had or leave it alone.

the problem is 4375mhz isn't stable like before and will take more voltage. it also seems to make more heat for some reason. with 100bclk @4.3ghz 1.23v i'm idling under 30c now. at 125 strap with 4375mhz and 1.22v i was idling 35-40c and not 100% stable. i was thrown off because aida and realbench seemed pretty stable and if i had a random problem i didn't think about the oc because of the illusion of stability. when i ran prime95 it was almost instant blue screen. so basically i have to settle for another 100mhz drop to 4200mhz with a 125 strap or use even more voltage to stabilize 4375mhz and end up with high temps. i'd rather just keep it simple at this point and stick with low latency 2400mhz ram speed. 4300mhz seems to be the sweet spot with my cpu. if you think it won't make any difference to lower latency by one more notch i'll just leave it how it is at 13-14-14-32 @ 2400. if i can get it down to 12-13-13-28 or something close for cheap then why not but it's not worth a lot to me.

EDIT: I may be wrong about cpu strap effecting temps. i need to double check. either way i was already at 79c at 4375mhz but maybe i'll give a a shot and see if i can still make it. i'm just not super comfortable with going over 80c.
 
Last edited:

Lochekey

Senior Pink Member
Joined
Sep 13, 2015
Have you just tried changing the voltages on that ram kit and lowering the timings further?
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
Have you just tried changing the voltages on that ram kit and lowering the timings further?

above 1.35v? not yet. i'm pretty sure i already tried cas12 and got a no boot but i'll give it one more shot.

edit: cas12 is good to go at 2400mhz. i think i tried it at 2600 and failed to boot but remembered wrong and thought it wouldn't work at 2400. so where can i find a kit that does cas 10 at 2400mhz? :D
 
Last edited:

Lochekey

Senior Pink Member
Joined
Sep 13, 2015
yes above 1.35. you can push 1.4v pretty safely as long as you have good airflow over the sticks. Are you trying to run at a command rate of 1 or 2?
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
yes above 1.35. you can push 1.4v pretty safely as long as you have good airflow over the sticks. Are you trying to run at a command rate of 1 or 2?

1t is nice but i don't think haswell-e likes it. if i'm wrong please say so. i've been using 2t so far.

EDIT: i was under the impression that the danger was to the cpu when going over 1.35v on ram correct?
 
Last edited:

Lochekey

Senior Pink Member
Joined
Sep 13, 2015
I was going to suggest running at 2t if you were not already. As for the dram voltage effecting the cpu itself I have not heard this before but I am not intamently familiar with the x99 platform.

According to a post on legit reviews here, intel listed that ddr4 voltage is allowed up to 1.5v for xmp settings so you should be good to push past 1.35v.

“1.5v is the absolute max we allow for XMP certifications. However, good DDR4 memory will run at 1.35v up to 3200. Technically, no “safe” (guaranteed) OC over-voltage but 1.35v or lower is best.” – Intel
Read more at http://www.legitreviews.com/what-is...emory-overclocking_150115#kwErvybIZLzHwORM.99
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
i think ill stick to 1.35 and under. thanks for the link.

EDIT: after some googling there are lots of answers about ddr4 safe voltage depending on who answers. some people say 1.5v 24/7 is fine. some people say that's the maximum they can take including momentary spikes under load and you should leave some wiggle room. in my bios anything up to 1.39v shows in yellow numbers. 1.4-1.49(i think) shows up in purple. 1.5v through 1.9v show voltage in red numbers. this more or less falls in line with people that say 1.4 and under is most likely okay with most chips and after that it's chancy. i think i can live with the "safe" yellow voltage numbers up to 1.39 or so since asus seems to feel that way if i understand their ram voltage setting color scheme correctly.

anyways it may not be necessary since the old girls seem to have some life in them even at 1.35v. not bad for cheap ram that came as part of a newegg package deal. currently doing fine at [email protected] i tried to do 11-11-11-30 but couldn't post. this makes me wonder what some of the nicer new kits can do but this is probably enough.

Capture.JPG
 
Last edited:

Lochekey

Senior Pink Member
Joined
Sep 13, 2015
I don't know if it is directly comparable since I'm on z170, but I can try and clock my 4x8gb kit down to see what it can do. It is a 3600 c16 kit.
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
I don't know if it is directly comparable since I'm on z170, but I can try and clock my 4x8gb kit down to see what it can do. It is a 3600 c16 kit.

i imagine a higher quality kit will have a better chance but i wonder if being binned for higher speed will help it's chances or hurt it? i haven't come across any kits personally that were binned for low latency since my first ocz ddr kit that ran at 2-2-2-5 back in the day with my athlon 3500+. if you feel like trying i'll watch! i'm curious. as for me i think i'll try a few cpu benchmarks and see if it makes a difference. so far i've only done superpi 1m and it seems to drop by about .3 sec finish time per unit of cas latency lowered. i'll think of something else to bench with in a minute.
 

Lochekey

Senior Pink Member
Joined
Sep 13, 2015
Welp it looks like my kit ends up at about those same timings(my tRFC is higher) if I lock it to 1.35v. Granted this is a 4x8gb kit though so it tends to be a little hard to stabilize anyways.


Are you running any programs that will really benefit from higher bandwidth anyways?
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
Welp it looks like my kit ends up at about those same timings(my tRFC is higher) if I lock it to 1.35v. Granted this is a 4x8gb kit though so it tends to be a little hard to stabilize anyways.


Are you running any programs that will really benefit from higher bandwidth anyways?

i'm not 100% sure but if i can get some free performance i'll take it even if it's only a little which seems to be the case. i know some games like different ram speeds more than others (fallout4?). i ran sandra lite cpu arithmetic and memory bandwidth tests after a fresh restart each time to compare the effect of latency. not a perfect way to test but it's just to see if anything moves one way or the other. i did 2400mhz at 15-15-15-35 vs 11-12-12-28. the lower timings got between a .5 and 2% boost in the various scores in the cpu arithmetic (horsepower) test. in the memory bandwidth test the lower timings gained about 5-8% gb/s. looks like earthdog is correct. it's not worth spending money on lol.
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
is it normal for super pi scores to get higher if you run them one after the other? it seems to fluctuate and i'm wondering if going a bit over 1.35v is causing them to slow down from heat or something.

sp1m.JPG
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
everythings been running like a dream since i dropped the 125 strap and went to 2400mhz 11 cas with 4300mhz [email protected]. i'm not 100% sure if it was the strap or just not being truly stable and 4375mhz but i can't feel a difference. still feels fast to me and i don't have random weird behavior. i think i was barely stable at 4375 even though realbench and aida said everything was fine. 5 minutes of prime95 at large ffts seems to be enough to tell since it would blue screen almost instantly without 1.23v at 4.3ghz. i guess the moral of the story is i'm getting old and don't feel like struggling for that last drop of speed that i'll never notice at the expense of extra heat and mysterious issues. i just want it to work. i'll probably eat my words on the next build but stability is great.
 
OP
Brando

Brando

Member
Joined
Jan 9, 2006
hate to sound like a newb here (well hell, I am) but what are you guys referring to when you say "strap"?

it's the number that gets multiplied when overclocking. if you have a 100 strap with a 44x multiplier you end up with 4400mhz. if you want to run a higher memory speed on x99 with an xmp profile it goes to 125mhz times multiplier which means you have to have a lower multiplier to get the same speed and the final speed must be divisible by 125. i don't know exactly why this is necessary to be honest. it just is. it changes a bunch of offsets and voltages behind the scenes and effects bandwidth.