Presenting my new invention - RAMGUARD

[Marcus L]

This whole thread is a breathe of fresh air

Temperatures may not be enough

If the RAM is receiving cool but stagnant air, this may not be as good as warm moving air.

Seriously hope you are being tongue in cheek/sarcastic like the OP

 

[Shrek]

Good point, I am taking this too seriously.

 

[Kwadratowicz]

Thermistor and thermocouple sensors are two different things!

I've learned something

yay
and back to google again
hmm thermo couple (cause of couple inside the probe)
thermi stor (what the hell is stor? )

 

[nomdeplume]

I appreciated stipulating the exact brand of product cardboard must be sourced from.

 

[lilhasselhoffer]

I've learned something

yay
and back to google again
hmm thermo couple (cause of couple inside the probe)
thermi stor (what the hell is stor? )

I...want to pretend this is real.

Thermistor - an element which when the temperature changes will change its resistance. Therm (al res) istor - Thermistor.
Thermo couples are usually bimetallic strips. You bond two metals with different coefficients of thermal expansion and you can figure out the temperature of an area by measuring the stress produced by the difference in expansion rates.

None of this to be confused with Memresistors. The fabled but only recently found electronic component which can change measurable state by heating, producing a non-volatile bit of memory that doesn't require constant refreshes like modern RAM cells.


Of course, this is a thread about the fun of alkaloid compounds.

 

[Kwadratowicz]

Thermo couples

cause of couple of 2 metals, that are at each others warmth

 

[95Viper]

Stick the topic.
Stop the thread trashing.

 

[iSpeakVeryWell]

That's funny, I made something similar this week!

I used the lid of an Amazon box tho, certainly not as good as your coco box!

The cardboard sit on top of the 2 180mm that come with the fractal torrent. I made sure to not block the exhaust's with it.

I also added the black fan over the CPU pump 5 hours ago to see if it would change something. Seems to help a tiny bit, I think I'll keep it. I removed the backplate of the GPU (I hate backplate!) so I had free anchor on the frame of the GPU cooler.

 

[eidairaman1]

Ok so an Air Shroud basically. DIY modifications work when they do, otherwise back to the drawing board.

 

[lexluthermiester]

I am happy to present my new invention, made from a paper carton from a Kakao box. It needs to be from Kakao, Cacao or any other product would not work!

You may also want to install a sheet of some transparent plastic to be less distracting and ugly, but that would not work either. Only paper from kakao boxes works.

The function is simple - with the air in the case moving towards the back of the case, this simple barrier prevents the hot air coming from the GPU to hit RAM sticks and significantly lowers their temperature, probably also lowers the temperature of the VRM heatsinks on the motherboard.

The fan installed on the card has nothing to do with the RAMGUARD, it is just a coincidence.

View attachment 400734

This is excellent! It would fit right in with the Ghetto Mods thread(and that is a complimentary suggestion)!

That's funny, I made something similar this week!

I used the lid of an Amazon box tho, certainly not as good as your coco box!

The cardboard sit on top of the 2 180mm that come with the fractal torrent. I made sure to not block the exhaust's with it.

I also added the black fan over the CPU pump 5 hours ago to see if it would change something. Seems to help a tiny bit, I think I'll keep it. I removed the backplate of the GPU (I hate backplate!) so I had free anchor on the frame of the GPU cooler.

View attachment 400850

Unsightly, but I'm guessing very effective.

 

[iSpeakVeryWell]

Unsightly, but I'm guessing very effective.

It's ugly indeed, but I only care for temps and overclocking capacity. RGB and immaculate cable management are 2 things I don't care about!

 

[BoggledBeagle]

I think that the topic of this thread is:

HOW TO PREVENT THE GPU FROM HEATING UP THE RAM STICKS

The above post by iSpeakVeryWell does exactly the opposite, it seems it has been made to put as much of the air coming from the GPU on the RAM sticks.

BTW that 43°C air temperature of the air leaving the GPU I measured is very low compared to what is possible today. For example the temperature of the fins of the FE 5090 card cooler can reach 65°C (Igors lab measurement), the air leaving this cooler will be just single digits degrees cooler.
This card would be scorching the RAM sticks with 60°C air. This is not good at all.

Honestly I am disappointed that card manufacturers abandoned the approach of discarting the hot air from the GPUs in the back of the case, but are dumping all of it in the case (and blowing it right on a critical part of the PC).

But at least you can pretty easilly guide the hot air away from the MB and RAM sticks with a sheet of carton or plastic.

 

[iSpeakVeryWell]

I think that the topic of this thread is:

HOW TO PREVENT THE GPU FROM HEATING UP THE RAM STICKS

Oops, sorry, I thought the topic was "how to use a cardboard to control airflow". My bad!

One of the best, simplest and cheapest way to prevent the GPU to heat up the RAM sticks is to put a fan over them. I put 2 in my build, but one is more than enough.

The above post by iSpeakVeryWell does exactly the opposite, it seems it has been made to put as much of the air coming from the GPU on the RAM sticks.

It has been made to cool the RAM so I can OC it to it's limit without having to worry about the GPU cooking it while gaming.

Air coming from AIO or GPU doesn't matter as long as the air is moving.

IMO, if you don't OC your RAM, don't worry about it's temperature. If you OC it, put a fan over it and call it a day.

I'd be curious to see who have the lower RAM temp after 1 hour of Prime95 Large FFT + Furmark.

 

[BoggledBeagle]

One of the best, simplest and cheapest way to prevent the GPU to heat up the RAM sticks is to put a fan over them. I put 2 in my build, but one is more than enough.

No, that does not work if you blow hot air on the RAM sticks.

Air coming from AIO or GPU doesn't matter as long as the air is moving.

IMO, if you don't OC your RAM, don't worry about it's temperature.

If my CPU draws 150W while gaming, it will warm the water in the AIO with 360 rad by 5°C at most. So the air incoming in the case through the AIO will be just 5 degrees warmer than ambient at worst.

I just showed that 5070 Ti (with performance bios) produced 43°C warm air. There are a lot of cards in use, that produce hotter air in larger volume and will just flood the RAM sticks with it.

There is a huge difference between the temperature of air coming from these two sources and you need to use the cooler air to cool your RAM sticks. There is no need to warm them by 20-30°C with no good reason.

 

[iSpeakVeryWell]

I was curious so I made some testing. I ran Prime95 Large FFT for 100 minutes to see what temps of my RAM would look like. Later, I made a second test running Prime95 Large FFT and Furmark together. Ambient temp was 20C in both test.

Idle temps: 24C
After 100 minutes of Prime95 Large FFT: 41C
After 100 minutes of Prime95 Large FFT + Furmark: 48C

Prime95 by itself added 17C, Furmark added 7C.

No, that does not work if you blow hot air on the RAM sticks.

I'd say that it work pretty well. There is no way my OC would be stable without at least one of the 2 fans, temps would be way higher.

There is no need to warm them by 20-30°C with no good reason.

This is not how it works. It's not because that air exhausted by the GPU is 43C that it will make everything it touch catch fire. There is no way my 9070 XT exhausted air is 7C.

But all this doesn't mean that your idea is bad. It doesn't mean that it doesn't work. You just need to test it and provide data for it. Would you be willing to do 2 runs of Prime95 Large FFT + Furmark? One without the cardboard and the other with it. I'm very curious to see how it would turn out.

Idle

Prime95 Large FFT alone

With Furmark

Also, keep in mind that my RAM is OC'ed. Here a stock vs OC'ed aida64 "benchmark" to give you an idea by how much it is OC'ed.

Stock

OC'ed

 

[harm9963]

I use brutal forced

 

[BoggledBeagle]

I made a proper test now - with the thermocouples placed in the same places and conditions, the only difference is removing the RAMGUARD.

I measured air temp in the RAM region, temperature of the RAM heatsink surface and also monitored RAM temp in HW info.
Loaded by real gaming - Cyberpunk, case closed, fans fixed speed (3 intake fans on the AIO at 1700 RPM). Ambient temperature 20,3°C.



Results without the RAMGUARD:

AIR temp: 41,3°C
RAM heatsing temp: 43,5°C
HW-info RAM temps: 48,2°C, 48,5°C

With the RAMGUARD:

AIR temp: 27,8°C
RAM heatsing temp: 37,1°C
HW-info RAM temps: 46,5°C, 44,2°C

So the result is pretty conslusive. You still need to keep in mind that 5070Ti is not the worst hot air producer you can put in your case. I wonder how this test would go with a 5090. 5090 FE must a be a horrid RAM roaster.

 

[rusty caterpillar]

I made a proper test now - with the thermocouples placed in the same places and conditions, the only difference is removing the RAMGUARD.

I measured air temp in the RAM region, temperature of the RAM heatsink surface and also monitored RAM temp in HW info.
Loaded by real gaming - Cyberpunk, case closed, fans fixed speed (3 intake fans on the AIO at 1700 RPM). Ambient temperature 20,3°C.

View attachment 401025View attachment 401026

Results without the RAMGUARD:

AIR temp: 41,3°C
RAM heatsing temp: 43,5°C
HW-info RAM temps: 48,2°C, 48,5°C

With the RAMGUARD:

AIR temp: 27,8°C
RAM heatsing temp: 37,1°C
HW-info RAM temps: 46,5°C, 44,2°C

So the result is pretty conslusive. You still need to keep in mind that 5070Ti is not the worst hot air producer you can put in your case. I wonder how this test would go with a 5090. 5090 FE must a be a horrid RAM roaster.

View attachment 401028View attachment 401029View attachment 401030View attachment 401031

Just drawing your attention to chipset temps, IMO is important as much as the RAM.
I would test the chipset temp, the one behind the GPU exhaust. Don't know how much your guard will stop exhaust of the GPU shroud towards the chipset.




Also I tested my self Chipset temps and RAM temps CrystalMark 1GB for M2 (gen3)drive and Furmark 13min. In my situation modified case Frontal 2x 180mm intake fans which does much better for chipset and RAM than the side intakes config. 2x14cm +1x12.




IMO we have to take in consideration that my M2 is a Gen 3 so is pushing the chipset less than your M2 which is Gen4, also my RAM goes max to 4800 with a meh latency of 79-80ns

Hope it helps.

 

[BoggledBeagle]

Just drawing your attention to chipset temps, IMO is important as much as the RAM.
I would test the chipset temp, the one behind the GPU exhaust. Don't know how much your guard will stop exhaust of the GPU shroud towards the chipset.

That is a good point, not only RAM sticks suffer because of the GPU air heaters, but chipset and SSDs as well.

The paper covers almost half of the back side of the GPU, but the GPU cooler can still eject hot air direcly on the motherboard in the other half.

I think I should place some additional fan under the GPU blowing on the lower part of the motherboard.

 

[rusty caterpillar]

That's funny, I made something similar this week!

I used the lid of an Amazon box tho, certainly not as good as your coco box!

The cardboard sit on top of the 2 180mm that come with the fractal torrent. I made sure to not block the exhaust's with it.

I also added the black fan over the CPU pump 5 hours ago to see if it would change something. Seems to help a tiny bit, I think I'll keep it. I removed the backplate of the GPU (I hate backplate!) so I had free anchor on the frame of the GPU cooler.

View attachment 400850

Is this the Fractal Torrent Compact or the big one?

 

[freeagent]

High airflow cases that no one wants to use, so paper it is?

 

[BoggledBeagle]

High airflow cases that no one wants to use, so paper it is?

I set the intake P12 MAX fans at 40% (around 1700 RPM) and the case had a good airflow, if was not like the air was stagnant there.

 

[iSpeakVeryWell]

Results without the RAMGUARD:

AIR temp: 41,3°C
RAM heatsing temp: 43,5°C
HW-info RAM temps: 48,2°C, 48,5°C

With the RAMGUARD:

AIR temp: 27,8°C
RAM heatsing temp: 37,1°C
HW-info RAM temps: 46,5°C, 44,2°C

This is very valuable data you got us here.

Ambient air arround RAM dropped by roughly 14C
Heatsink dropped by roughly 7C
Hotspot of RAM dropped by roughly 3C

While ambient air dropped by a lot, RAM hotspot dropped by only 3C.

If you could do the same test but with a fan instead of the RAMGUARD it would be very awesome!

Thank you for the data you bridged us! It's very interesting!

Is this the Fractal Torrent Compact or the big one?

It's the big one

 

[ir_cow]

I'm glad someone else noticed the ram overheats from the GPU exhaust. Been saying this since the RTx. 30 series (even wrote it my reviews). Depending on the case I either put cardboard in-between or a fan on top of the GPU pointing at the ram.

I asked Nvidia about it and they said I'm the only one that has this problem lol. That was with 350w GPUs. Now we are at 600

 

[sram]

That is a good point, not only RAM sticks suffer because of the GPU air heaters, but chipset and SSDs as well.

The paper covers almost half of the back side of the GPU, but the GPU cooler can still eject hot air direcly on the motherboard in the other half.

I think I should place some additional fan under the GPU blowing on the lower part of the motherboard.

View attachment 401075

Time for ChipsetGUARD and SSDGUARD. I can invest in the project.