CptT95
how do I post images here?

I almost always use imgbb.com then insert the img tag on either side, or use the interface to post the link.  Uploading images here is so hit & miss I stopped trying. lol
 
 

 Tomorrow... or should I say, in about 10 hours...
 

 
https://www.newegg.com/g-skill-32gb-288-pin-ddr4-sdram/p/N82E16820232978?Item=N82E16820232978
 

 
3800MHz CL14 with 4 sticks here I come, hopefully. If not they gettin' returned :-D lol

New personal best by a smidge: https://www.3dmark.com/3dm/61872130
 

 
That bumps me up to 4th best leaderboard score for 5950X/single 3090. Yeeeeeet.

Nice!

Dabadger84

alpharius194
Did you end up ordering the 4000cl15 kit?

Yeah, I spent about a day fighting with it because it was having to train just to run what should be pretty normal settings, but it's given me no issues in the last 2 days or so.  Set at CL14 3800MHz with 1.5V, basically the same timings as the other kit, but with all 4 slots populated, tested for an hour then I've played Metro Exodus EE & CSGO a fair bit with zero WHEA problems, so I'm fairly confident it's stable.
I haven't taken the plastic protection off them yet since initially I thought I might be returning them when they started the whole F9 loop thing at first, probably take those off today.  I'm sure it won't make a huge difference in temps, but it might help a little.  At present they idle (with fans on low speed for idle) at 28-31C and under load it depends on how high the top radiator fans ramp up, they run between 36-38C at high fan speeds, and 38-41C if they fans are in a more quiet setting.  I've thought about putting a fan more directly on them again, but given they're not running especially warm, and part of the reason I bought the kit was for the looks, I'd prefer not to cover them partially with a fan.

Did you try them at DOCP before going down? I think RAM stability is fine to 50C so if you’re at max 41 with low speed you probably don’t need the fan for just those with how much air flow you have.

alpharius194

Dabadger84

alpharius194
Did you end up ordering the 4000cl15 kit?

Yeah, I spent about a day fighting with it because it was having to train just to run what should be pretty normal settings, but it's given me no issues in the last 2 days or so.  Set at CL14 3800MHz with 1.5V, basically the same timings as the other kit, but with all 4 slots populated, tested for an hour then I've played Metro Exodus EE & CSGO a fair bit with zero WHEA problems, so I'm fairly confident it's stable.
I haven't taken the plastic protection off them yet since initially I thought I might be returning them when they started the whole F9 loop thing at first, probably take those off today.  I'm sure it won't make a huge difference in temps, but it might help a little.  At present they idle (with fans on low speed for idle) at 28-31C and under load it depends on how high the top radiator fans ramp up, they run between 36-38C at high fan speeds, and 38-41C if they fans are in a more quiet setting.  I've thought about putting a fan more directly on them again, but given they're not running especially warm, and part of the reason I bought the kit was for the looks, I'd prefer not to cover them partially with a fan.

Did you try them at DOCP before going down? I think RAM stability is fine to 50C so if you’re at max 41 with low speed you probably don’t need the fan for just those with how much air flow you have.

Problem with DOCP is it would set to CL15 & then that requires Gear Down disabled.  CL14 at 3800MHz 1.5V seems to be perfectly happy, zero issues thus far.
 
B-Die doesn't like temperature, especially if you're pushing it.

CptT95
I have hit almost 50c with my setup, no fan blowing on ram and it was VERY warm in my room so that contributed to the high temps
Is liquid cooling mem worth it? how exactly does that work, curious since I am pushing my mem

Not really no, when it comes to liquid cooling.  B-Die is temperature sensitive enough that hitting over 40C may will influence your stability at more aggressive settings.  Having a fan blowing air over the sticks from the top or where-ever will make a BIG difference, any air flow at all will significantly reduce temperatures because typically in between the sticks, where the heat will actually build up, is what's needed.
Disabling the RGB while you're doing stability testing specifically will lower temperatures a degree or two in my experience, and that's with cooling/fan airflow already being a thing over them.
 
The issue with "liquid cooling" memory is that the cooling devices for them rarely actually cool what needs to be cooled properly, usually it's just liquid being channeled through metal, which then connects to metal that touches the chips.  It's not flowing liquid directly over the memory modules, which is what would really be needed.  Basically DDR4 runs about as warm as DDR5 on video cards, so doing liquid cooling on it would be overkill. 
 
Just a plain ol' fan blowing some air in to the spaces between the sticks to blow that hot air out & fresh air in will probably drop your temps 10C or so.  When I had a dedicated fan blowing over my sticks last they peaked in the low to mid 30s celsius.  Now, with the top fan of the top radiator exhaust blowing over them semi-directly, I'm seeing as high as 41C on 1 of the sticks.  So more direct airflow that's closer to them is better, but any airflow is better than no direct airflow at them.

CptT95
well hopefully putting a gpu under water in will eliminate a source of hot air blowing into the case and lower overall air temps down to help
Edit-not everyone has a supersize pc like yours lol hard to fit a fan in there with all the tubes in the way

Oh, yeah, if you have an air-cooled GPU that is increasing your RAM temps by at least 5-10C, guaranteed.  I had the same issue when I had the 3080 FTW3 Ultra - my RAM temps went from being low to mid 30s with the RAM fan I had to being in the low to mid 40s.  Unsurprising since said 3080 was venting 50-60C air in to the case directly on the M.2 & RAM areas.  Going to the liquid cooled HC card will lower your RAM temps & M.2 temps a good bit would be my guess.

Okie I stand corrected:
 


That's with fans capped at 80% as their highest possible speed during gaming, for almost 2 hours of CSGO (which is CPU intensive).  32-35C max vs 38-41C without the dedicated fan... of course, SOME of that reduction could be from removing the plastic protective covers.  Who knows.  I will probably take the fan back out & test that because I am curious.

Also 1500 on mem works for me too

Would the psu be to blame for everything going dark then whole pc restarting? Didn't think I would have this issue with an 850 t2 psu:/

I think I had some dust in psu causing higher heat and possibly unstable clocks, I'm still learning this thing 😅
Haven't seen over 480w off gpu so idk

5800x
Asus crosshair viii formula
4 x 8 gb gskill trident z rgb 3200cl14 at 3800 cl15
D5 pump
9 x unifans 120mm from Lian li through control hub
2 HDD that don't get used much
1tb 980 pro
Oh and one exhaust 120 ml120 pro corsair fan lol

CptT95 ()<-from me, not CptT95
5800x (~100Wmax (# from you))
Asus crosshair viii formula
4 x 8 gb gskill trident z rgb 3200cl14 at 3800 cl15 (~12W)
D5 pump (~20W at 12V)
9 x unifans 120mm from Lian li through control hub (total of ~50W with hub)
2 HDD that don't get used much (~20W)
1tb 980 pro (~8.5Wmax)
Oh and one exhaust 120 ml120 pro corsair fan lol (~2.7W)

So that's ~693W max (also throwing in 480W on the GPU). Normal conditions, yeah, I'd think the 850W was more than plenty. The thing we don't know is how much of a current spike the GPU could have added to that, or how much of a current spike the PSU can actually handle.
 
Looking at this review of that supply, it looks like the one negative they gave it was for the 3.3V drop during the transient testing. Is that your actual problem? I don't know. Just to be clear, I'm not saying "this is it, replace that thing". I'm just saying you have a little bit of evidence to say that it could be an issue. Some reviews show peak power for the 5800X hitting 140W. That still wouldn't take you to the limit, but if for reasons I can't explain COD spikes everything beyond normal values simultaneously for no reason...maybe? I haven't played COD since they got rid of private servers. I think BO (original) was the last COD game I bought so I can't really run any testing with it to give you some info from my system.

CptT95
Do you think using boost lock on X1 would help mitigate those spikes? This keeps the gpu at full power state and hold boost clock all time 😒🤔

Probably not. There are two different kinds of spikes that seem to be prevalent. One is due to the weird voltage jump when a game jumps from a low power state to a high power state. That is what seems to be happening to people who's 3080's and 3090's crap out with the red light of death during Halo or League of Legends or something...what your describing might help those folks, but I couldn't say for certain. Then there are the high power usage spikes that happen when actually using the card in high power loads like games where the GPU is at 100% utilization. Those are just there...I don't know what scenes or gaming events cause them, but it seems like the whole 30-series has higher spikes than any cards before them, which is why people have had issues with 750W and even some cheaper 850W supplies even though the spec says 750W should be enough. I haven't seen a lot of documentation about what they look like or what you can expect from them, but I'd personally be hesitant to run any 3090 with less than a 1000W PSU (let alone a KP lol), especially while overclocking everything in the system. Now, you do not have a cheap 850W PSU, and I bet if everything was stock you wouldn't have a problem, but the KP is already not "stock" and if you threw some extra +'s in the speeds, it would definitely hit higher than stock. You do have an efficient CPU (especially compared to mine that peaks ~250W during an r23 run lol), so that should definitely help. Another question I would have for you is how old is your PSU? Something like transient spike handling capability could degrade over time. Usually the T2's are built like tanks so I would think it should take quite a while before it had trouble, but there are very limited data-points out there for exactly what the spikes look like, what power supplies can handle (exactly), and if/when you might be in trouble. The 30-series totally changed the game in my mind, as for years people were buying way bigger PSUs than they needed...now, not so much.