[DAM20]

I'm having an hard time keeping my g.skill kit under 42c while gaming, I already have 2 noctua 40mm fan on them, but the open back gpu still heats things up when reaching 240W+.

Do you guys have any advice or something I could buy to get better temps?

Oh the case is a Dark Base Pro 900, front is open, 3 140 in + 1 120 in near the ram and 2 140 out.

Yes, I could reduce the oc or limit my gpu, but that's for plebs.

 

[Ichirou]

There's a reason why I've mostly avoided the latest gen of nVidia GPUs; I dislike the open-back design. Really anti-RAM overclocking, lol. Just build some sort of cover to block off the open-back. Won't look pretty, but it'll work.

Also, undervolting/underclocking the GPU isn't necessarily bad; better consistency in performance is better than fast initial clocks most of the time due to the inefficiency of GPU Boost above 70C or so. Try it out in 3DMark yourself.

 

[SmOgER]

There's a reason why I've mostly avoided the latest gen of nVidia GPUs; I dislike the open-back design. Really anti-RAM overclocking, lol. Just build some sort of cover to block off the open-back. Won't look pretty, but it'll work.

Personally I see DDR4 B-die needing to be under 45C for stability as a design flaw.
There were no issues of this kind in previous generations of ram. Blaming VGAs now is just ridiculous IMHO.

 

[SmOgER]

OP, you could try some kind of DIY case separator to stop the heat from rising as much.

 

[DAM20]

There's a reason why I've mostly avoided the latest gen of nVidia GPUs; I dislike the open-back design. Really anti-RAM overclocking, lol. Just build some sort of cover to block off the open-back. Won't look pretty, but it'll work.

Also, undervolting/underclocking the GPU isn't necessarily bad; better consistency in performance is better than fast initial clocks most of the time due to the inefficiency of GPU Boost above 70C or so. Try it out in 3DMark yourself.

It's a Nitro+ 5700 XT, I could try to undervolt to mb 1.15v, but the bios target 91c hot spot temp, so no metter what the card will stay there, unless I touch the fan curve but it gets noisy.

 

[DAM20]

OP, you could try some kind of DIY case separator to stop the heat from rising as much.

Hmm, idk where to put it, but maybe another fan on top pushing out could help?!

 

[Ichirou]

It's a Nitro+ 5700 XT, I could try to undervolt to mb 1.15v, but the bios target 91c hot spot temp, so no metter what the card will stay there, unless I touch the fan curve but it gets noisy.

You have to both undervolt and underclock to cap off the temps. The GPU will only clock as high as you allow it to.
For my card, I made the GPU cap off at around 70-71C at most. I start getting issues if I accidentally leave it uncapped and the GPU heats up to 75C+, causing RAM corruption =\

 

[EastCoast]

It's a Nitro+ 5700 XT, I could try to undervolt to mb 1.15v, but the bios target 91c hot spot temp, so no metter what the card will stay there, unless I touch the fan curve but it gets noisy.

A Nitro? What are your temps in game? Because if edge temps are above 65c at stock clocks you might need to reapply thermal compound imo.

But that shouldn't effect ram temps though.

You could by a ram cooler though. But I haven't seen those in sometime.

 

[SmOgER]

Hmm, idk where to put it, but maybe another fan on top pushing out could help?!

Take a photo of how it looks.

It's a Nitro+ 5700 XT, I could try to undervolt to mb 1.15v, but the bios target 91c hot spot temp, so no metter what the card will stay there, unless I touch the fan curve but it gets noisy.

That seems overly high. BIOS target is to limit the temp as well as prevent damage in case standard cooling fails to keep the temp under it.

 

[DAM20]

You have to both undervolt and underclock to cap off the temps. The GPU will only clock as high as you allow it to.
For my card, I made the GPU cap off at around 70-71C at most. I start getting issues if I accidentally leave it uncapped and the GPU heats up to 75C+, causing RAM corruption =\

I'll stress test 1.15v tomorrow, but I doubt is stable, mb 1.175.

A Nitro? What are your temps in game? Because if edge temps are above 65c at stock clocks you might need to reapply thermal compound imo.

But that shouldn't effect ram temps though.

You could by a ram cooler though. But I haven't seen those in sometime.

Nitro+ SE, have a 240W bios, but in games like Divinity: Original Sin 2 can hit 255W with 15% limit increase.
I've repasted a few time, even tried LM but temps are kinda always the same, under full 240W load 70-73c and hotspot 91-94c

Take a photo of how it looks.


That seems overly high. BIOS target is to limit the temp as well as prevent damage in case standard cooling fails to keep the temp under it.

Well considering is the hot spot and not the actual temp, should be fine till 110C according to AMD.

My idea is to put another 140 on top of the ram as exahust, the 120 intake is kinda useless... Using the 2 40mm reversed gives me worst temps.

 

[Ichirou]

Just cover the back opening of the GPU lol

 

[EastCoast]

I'll stress test 1.15v tomorrow, but I doubt is stable, mb 1.175.


Nitro+ SE, have a 240W bios, but in games like Divinity: Original Sin 2 can hit 255W with 15% limit increase.
I've repasted a few time, even tried LM but temps are kinda always the same, under full 240W load 70-73c and hotspot 91-94c

Hmm, what thermal compound are you using?

Does it looks like this when you take it apart?

 

[chris89]

I recommend running the memory at the timings & voltage printed on the module itself. Then test temperatures, but 42C is ice cold bro. Memory & all components are safe up to 84C, anything hotter than 84C is too hot even if an engineer says its okay, it isn't.

Try these EverCool MSW/B Strong Wind Memory Cooler (Blue) 4711158194003 | eBay

 

[DAM20]

Just cover the back opening of the GPU lol

I thought about it, but will increase gpu temp, I think

Hmm, what thermal compound are you using?

Does it looks like this when you take it apart?

Used kryonaut, conductonout and now NTH2.
Yep but I've noticed that whit too much paste the hotspot increase, I always need to spread a lil of paste evenly to cover the whole die.

I recommend running the memory at the timings & voltage printed on the module itself. Then test temperatures, but 42C is ice cold bro. Memory & all components are safe up to 84C, anything hotter than 84C is too hot even if an engineer says its okay, it isn't.

Try these EverCool MSW/B Strong Wind Memory Cooler (Blue) 4711158194003 | eBay

Unfortunately B-Die needs to stay cool at some point, my wall is 42c.
The gpu hotspot should be the hottest point in the die, while the avg is always like 20c~ lower.



Anyway, bumped up a bit the fans, was able to keep the ram at 38c while gaming, I'll test some more.

 

[chris89]

I run my Samsung memory module's @ 55C all day no problems for years. 42C is only 107.6 Fahrenheit. Which is ice cold even to the touch its barely warm at all. So I know you have nothing to worry about, your golden.

I also found these fans which might be able to be used to cool all 4 memory modules directly on top.

ASUS Motherboard Heatsink Chipset Heat Pipe VRM Cooling Fans-YD124515MB | eBay

 

[DNMock]

Water cool your GPU or better yet, your GPU and your RAM.

Only other alternative I can think of would be to remove your heat spreaders, replace your thermal interface with some better thermal tape and put some copper heat spreaders with fins on your ram. That's a good way to accidentally rip of a ram module though and dollar per dollar, you would just be better off slapping an AIO on your GPU to have it vent outside the case instead of inside.

Outside of that, turn your thermostat down in the house.

 

[o1dschoo1]

I recommend running the memory at the timings & voltage printed on the module itself. Then test temperatures, but 42C is ice cold bro. Memory & all components are safe up to 84C, anything hotter than 84C is too hot even if an engineer says its okay, it isn't.

Try these EverCool MSW/B Strong Wind Memory Cooler (Blue) 4711158194003 | eBay

Wrong b die throws error above 45c ish

Try running cl15 3800 with a 290 trfc and 20000 trefi with 55c.

 

[chris89]

I ran SAMSUNG module's on my AMD Opteron 1389 system & the memory had to have been getting warmer than 42C @ load @ DDR3-1600Mhz. Then again its not a fair comparison Ryzen vs old school Opteron memory temperatures.

 

[o1dschoo1]

I ran SAMSUNG module's on my AMD Opteron 1389 system & the memory had to have been getting warmer than 42C @ load @ DDR3-1600Mhz. Then again its not a fair comparison Ryzen vs old school Opteron memory temperatures.

That's ddr3. Samsung ddr4 bdie is temperature sensitive. My cl14 3200 runs 32c idle at 1.35 1.45 and 3800 runs 35c idle and 38 load.... Ddr4 is a totally different ballgame

 

[Blameless]

Do you guys have any advice or something I could buy to get better temps?

Watercool your GPU.

Personally I see DDR4 B-die needing to be under 45C for stability as a design flaw.

When being run anywhere near design spec, they can get a lot hotter without becoming unstable.

The fastest bin of B-die from Samsung is 2666MT/s CL19, with a 260ns tRFC (347 cycles), and the standard 7.8us refresh interval. Hard to call it scaling well with reduced temperature a flaw.

Memory & all components are safe up to 84C, anything hotter than 84C is too hot even if an engineer says its okay, it isn't.

Safe is not the same as stable, or vice versa, especially when running way beyond rated spec.

I ran SAMSUNG module's on my AMD Opteron 1389 system & the memory had to have been getting warmer than 42C @ load @ DDR3-1600Mhz.

Those are a totally different product and have no bearing on the heavily OCed DDR4 B-die people are talking about here.