Overclock.net banner

1 - 20 of 36 Posts

·
Registered
Joined
·
518 Posts
Discussion Starter #1
Me and my brother are playing with some RAM OC, and we've kind of both ran into the same issue: At very tight timings and subtimings, RAM seems to become very sensitive to temperature.
I have a meshify C and he has a define S.
Our RAM stays at around 50°C+ in stress test and between 45 and 50 under normal loads, with 25 ambient. When ambient temps are around 20-21°C we don't have any errors at all in stress test, but when is 25-26°C, they start popping up.

If we add a fan right over the RAM, are we gonna get a 5-6-7 or maybe 10°C drop, or the temps are too low for something to happen???

Speeding up the front intake (we both have fully populated front intake), there's only seems to be a couple of degrees of difference.....
 

·
The last VRM burner
Joined
·
2,590 Posts
If it's B-Die then you must control temperatures. Temperatures can only be too high.
 

·
Just hangin' out
Joined
·
1,071 Posts
20°-25°C over ambient with normal use?

What kind of voltage are you giving those DIMMs?
 

·
Overclock the World
Joined
·
1,904 Posts
@Shenhua Memory starts to get unstable after 42c
It will help for sure :)
 

·
Registered
Joined
·
518 Posts
Discussion Starter #6
20°-25°C over ambient with normal use?



What kind of voltage are you giving those DIMMs?
Not really that high to be honest. 1.36-1.37.
25°C over ambient in stress test. But trfc it's at 264. It seems like the lower is the trfc the more sensitive they become to heat.
.....also while normal use seems like low usage it's not. We have 16gb of RAM and our usage stays between 12 and 14.5gb.
Sent from my Redmi K20 Pro using Tapatalk
 

·
Just hangin' out
Joined
·
1,071 Posts
Not really that high to be honest. 1.36-1.37.
25°C over ambient in stress test. But trfc it's at 264. It seems like the lower is the trfc the more sensitive they become to heat.
.....also while normal use seems like low usage it's not. We have 16gb of RAM and our usage stays between 12 and 14.5gb.
Sent from my Redmi K20 Pro using Tapatalk
That is definitely not too much voltage.

Does the TRFC need to be at 264? I mean, are there substantial and significant gains in performance?
 

·
Registered
Joined
·
518 Posts
Discussion Starter #8
That is definitely not too much voltage.



Does the TRFC need to be at 264? I mean, are there substantial and significant gains in performance?
Yes. We are running ryzens.

Sent from my Redmi K20 Pro using Tapatalk
 

·
Just hangin' out
Joined
·
1,071 Posts
To answer your initial question; I would say if you can pull cooler air into the chassis and direct it over the DIMMs, it could not hurt.
 

·
Just hangin' out
Joined
·
1,071 Posts

·
Overclock the World
Joined
·
1,904 Posts
45C idle is really high , are these RGB rams ?
They should move near the 26-30c range, or around 2-3c over ambient

tRFC is not everything and not only voltage dependent
tRC is voltage dependent , and tRP is the modifier for that
tRAS is tRAS, but tRP needs to be able to (p)recharge it in time - this one is voltage dependent
tRFC does play a role, but it's heat that defines how fast cells lose charge, not really voltage

42+ they start to get unstable
up to 1.45v memory can cool itself passively
If the case is not a closed box - i'd start to question and check if these units even have thermal pads and if they are sticking
1.4v you can cool memory that's barebones fully passive
1.45v will also cool itself passive with just a heatsink
soo check if you aren't having an issue somewhere else
30c is common idle temp not 45+
 

·
Registered
Joined
·
518 Posts
Discussion Starter #12
45C idle is really high , are these RGB rams ?

They should move near the 26-30c range, or around 2-3c over ambient



tRFC is not everything and not only voltage dependent

tRC is voltage dependent , and tRP is the modifier for that

tRAS is tRAS, but tRP needs to be able to (p)recharge it in time - this one is voltage dependent

tRFC does play a role, but it's heat that defines how fast cells lose charge, not really voltage



42+ they start to get unstable

up to 1.45v memory can cool itself passively

If the case is not a closed box - i'd start to question and check if these units even have thermal pads and if they are sticking

1.4v you can cool memory that's barebones fully passive

1.45v will also cool itself passive with just a heatsink

soo check if you aren't having an issue somewhere else

30c is common idle temp not 45+
45+ not idle At 80-90%~ load (normal load for me) and with 25-26°C ambient. At idle, they indeed stay 2-3 over ambient.
I'm running 3200 with 14-14-14-28-42-264 at 1.36v. im running the fast profile from ryzen dram calc, for 3200.

My RAM is not hot by any means, but it's hotter than i need it to be.
 

·
Premium Member
Joined
·
2,301 Posts
As for temp drop with fan over ram....on mine, hci stresstest temps decreased from 47C to 43C max with small 80mm fan on quiet speed (1900 is quiet for 80mm and doesnt move much air). Already had some moving air via top 140mm rad fan at 950rpms. With neither fan, ram temps 7C higher stress testing and 4C higher gaming...this with 1.49 vdimm 4133c16.

next time I redo my watercooling loop, going to leave space for a bracket over my ram for twin 80mm+ size fans.
 

·
Registered
Joined
·
518 Posts
Discussion Starter #14
As for temp drop with fan over ram....on mine, hci stresstest temps decreased from 47C to 43C max with small 80mm fan on quiet speed (1900 is quiet for 80mm and doesnt move much air). Already had some moving air via top 140mm rad fan at 950rpms. With neither fan, ram temps 7C higher stress testing and 4C higher gaming...this with 1.49 vdimm 4133c16.



next time I redo my watercooling loop, going to leave space for a bracket over my ram for twin 80mm+ size fans.
Tnx for feedback


Sent from my Redmi K20 Pro using Tapatalk
 

·
ٴٴٴ╲⎝⧹˙͜>˙⧸⎠╱
Joined
·
6,078 Posts
@Shenhua Memory starts to get unstable after 42c
It will help for sure :)
45C idle is really high , are these RGB rams ?
They should move near the 26-30c range, or around 2-3c over ambient

tRFC is not everything and not only voltage dependent
tRC is voltage dependent , and tRP is the modifier for that
tRAS is tRAS, but tRP needs to be able to (p)recharge it in time - this one is voltage dependent
tRFC does play a role, but it's heat that defines how fast cells lose charge, not really voltage

42+ they start to get unstable
up to 1.45v memory can cool itself passively
If the case is not a closed box - i'd start to question and check if these units even have thermal pads and if they are sticking
1.4v you can cool memory that's barebones fully passive
1.45v will also cool itself passive with just a heatsink
soo check if you aren't having an issue somewhere else
30c is common idle temp not 45+



43-44c on 2x16 b-die, no problem here


though if my AC is turned off the temps will surely rise up to 49-52c


DRAM power consumption when stress test around 9.5w avg, 10w max. testmem5 anta777 profile



rear intake has filter









 

·
Overclock the World
Joined
·
1,904 Posts
@TK421 your settings don't look that tight :)
But it's only a chance to happen, after 42c
Every degree higher increases this chance
Just to keep accuracy up , 42c is the entry level when they start to spill out errors as Cell discharge happens faster by increased temps
Which in plain words mean, the tighter tRAS,tRP, tRFC - the higher the chance of choking under increased thermals
Higher tRP would mitigate low vDIMM, where tRP logically increases tRFC.
Longer time to (p)recharge, less affected by faster discharging as wait time is plain longer~
 

·
ٴٴٴ╲⎝⧹˙͜>˙⧸⎠╱
Joined
·
6,078 Posts
@TK421 your settings don't look that tight :)
But it's only a chance to happen, after 42c
Every degree higher increases this chance
Just to keep accuracy up , 42c is the entry level when they start to spill out errors as Cell discharge happens faster by increased temps
Which in plain words mean, the tighter tRAS,tRP, tRFC - the higher the chance of choking under increased thermals
Higher tRP would mitigate low vDIMM, where tRP logically increases tRFC.
Longer time to (p)recharge, less affected by faster discharging as wait time is plain longer~

It's already at 1.522v and I can't increase freq or reduce tCL without error.


Not sure what to do in order to tweak this any further.
 

·
Registered
Joined
·
1,071 Posts
It's already at 1.522v and I can't increase freq or reduce tCL without error.


Not sure what to do in order to tweak this any further.
for day to day I'd be turning the voltage down for benchmarking I'd be looking at fans and different heat spreaders.
 

·
Overclock the World
Joined
·
1,904 Posts
It's already at 1.522v and I can't increase freq or reduce tCL without error.


Not sure what to do in order to tweak this any further.
You might be one of the few to try fixing accurate tREFI instead of Max Amount

Try to use tRFC (if you can even tRFC 2 and 4) + tREFI max
SuperPi 1.5 SX 32million should take 7-10min and error fast out if there is a tREFI issue
 
1 - 20 of 36 Posts
Top