How hot is your Ryzen 3700X ? Air Cooler comparision.

[Deleted member 185158]

Wraith.

I wouldn't do glass tubing. I am not even sure glass tubing is available. Bending it would be a nightmare.



I want to but I am already going to need another pump/res combo or figure out how to mod a bracket. I hate ghetto looking mounts so that is going to be the toughest. I have been looking at the Maelstrom D5 V2 from Swiftech or possibly the Tt Pacific pr22 (I think was the name). I really like my Photon with D5 though but no vertical mounts exist for it.

The ups and downs of liquid cooling.
Stopped running a case some years back now. Gave a good handful of gear to a local buddy and team mate at warp9. 2x 120.3 and 2x 120.2 rads pump and mini res. The 5th res in my case was sized to the back panel of 80mm x2. Water cooled 2x 580s and aquires a 3 pack of 8800 Ultras with blocks which the lool couldnt handle all 3 with a cpu lol.
Tons of work with tubes and cable management. Many hours.
All housed in a full Aluminum TJ07 case with dual PSUs. I think it was close to 100 pounds fully loaded. Ran raptors in raid and 2-4 storage drives. Had it all man. Even my Ageia Physx card but wasnt on the loop.

Now. Water from the tap on a bench table. I run my Ryzen in a lower Pstate of 3.2ghz 1.05v nice and cool. 0 fan rpm.

 

[Taraquin]

Maybe mention before but a tips to lower temps on the 3000-series is to use the best thermal paste possible (not the stock stuff which). Safe approach is Thermal Grizzly Kryonaut which often reduces temp by 3-5C during load compared with popular stock paste like MX-M4 etc.

Second tips is to undervolt with an offset. Don`t set a fixed voltage, it ruins performance. Most will be able to set a -50 to -100mv offset which can lower temps by 5-10C. Undervolt With offset has generally no impact on performance. In some circumstances it can improve performance since Ryzen 3-series has certain temp-threshold where the lower clocks. In general there is several such threshold between 55C and 80C of temp according to gamersnexus. If a offsett og -80mv lowers temp by 8C it can possibly make the CPU run 100MHz faster if you are close to one of the temp\frequency-points.

 

[Midiamp]

Out of curiosity are these 70c+ temps constant or just a spike? My 9700K @ 4.9ghz all cores pretty much stays at 55-65c while gaming but if I look at max temps there could be spikes as high as 74c in gaming sessions.

my room temp is 28 degrees C, and I use Fractal Design Define C with a Micro ATX board. So my case temp is governed by the hottest component which is my GPU. 72 degrees C is still okay in my book, the trade off is fair, I look for a compact case with good sound insulation, and I got my wish... Probably going to be much worse temp with Be Quiet offering.

 

[moproblems99]

While I know this is for air cooling, I just got my temporary loop setup for 3900X. On the wraith, I was getting spikes as high as 90C and gaming was averaging about 75C - 80C. With my loop, my highest spike has been 68C and the average temp looked like 58C to 64C . This has been in a room that sits at about 28C. Idles were about 45C to 55C on the Wraith and on water it is more like 37C to 47C. Idles are tough because about every 10 seconds it takes a 10C jump. I can just about count it down and watch the temps drop till about 37C and then it will jump back up to about 47C and repeat.

The most interesting thing out of all this is that my boost speeds were actually about 25mhz to 50mhz better on the Wraith....

 

[Deleted member 185158]

The most interesting thing out of all this is that my boost speeds were actually about 25mhz to 50mhz better on the Wraith....

Lol. This goes way way back.... some chips just like to be warmer.
or
you've got different settings in bios going on between the testing?

 

[moproblems99]

Lol. This goes way way back.... some chips just like to be warmer.
or
you've got different settings in bios going on between the testing?

Same settings. Weird but 6 of my 12 cores still boost to 4599 or over. The other 6 sit at about 4349 to 4399.

 

[Deleted member 185158]

Well the only suggestion I have to give you is to apply a fan to the VRM area. You took away that air flow from removing the stock cooler. See if that helps.
edit: can you update your system specs please?

 

[Divide Overflow]

My 3900X runs games at around 53 - 60 on air cooling at 22 ambient.

 

[moproblems99]

My 3900X runs games at around 53 - 60 on air cooling at 22 ambient.

Unlucky I guess? I mean mine doesn't even do that on water.

 

[Rahnak]

I recently got a 3700X and decided to give the stock cooler a chance, to see if I could avoid spending extra on a cooler, but it doesn't seem like that's gonna happen.
While it does a pretty acceptable job cooling the cpu - I'm getting low 60s playing AC Unity, which is slightly below what I was getting on my 2500k with a Hyper 212+; the fact that temps oscillates so much at supposedly idle, means the fan keeps ramping up and down in the 900 to 1100 rpm range and it's pretty audible. It's been driving me nuts.

So I think I'm getting a Scythe Fuma 2, seems like it has a pretty stellar price/performance ratio.

 

[Bill_Bright]

Some times all it takes is another case fan.

 

[moproblems99]

I recently got a 3700X and decided to give the stock cooler a chance, to see if I could avoid spending extra on a cooler, but it doesn't seem like that's gonna happen.
While it does a pretty acceptable job cooling the cpu - I'm getting low 60s playing AC Unity, which is slightly below what I was getting on my 2500k with a Hyper 212+; the fact that temps oscillates so much at supposedly idle, means the fan keeps ramping up and down in the 900 to 1100 rpm range and it's pretty audible. It's been driving me nuts.

So I think I'm getting a Scythe Fuma 2, seems like it has a pretty stellar price/performance ratio.

The oscillations were annoying as crap. It happens when one or more of cores boost and drive the core voltage to or near 1.5v. the cores seem to boost for very trivial background tests.

 

[Rahnak]

Some times all it takes is another case fan.

Opening the side panel doesn't make a difference, so I don't think adding another fan would help much either. It's just a small fan that's too audible for my personal taste above 1000 rpm. Maybe if the case was on the ground, under the desk it wouldn't bother me.

The oscillations were annoying as crap. It happens when one or more of cores boost and drive the core voltage to or near 1.5v. the cores seem to boost for very trivial background tests.

It's certainly a different experience than what I was used to with Intel. Just opening file explorer makes the cooler ramp up. And having Firefox open makes it "idle" between 40~50ºC, which is quite a considerable delta.
Yesterday night I just closed everything except for hwinfo and I just sat here looking at the cpu temp and fan speed and wondering "What the heck are you even doing?".

 

[Bill_Bright]

Opening the side panel doesn't make a difference, so I don't think adding another fan would help much either.

In terms of case cooling, these do two very different and unrelated things. You should not use the effects of one to judge the other.

Adding another case fan can significant increase the desired front-to-back "flow" of cool air through the case.

Opening the side panel can totally disrupt, or even eliminate that desired flow. I never recommend opening the side panel unless you are also going to blast a desk fan in there.

It's just a small fan that's too audible for my personal taste

Well I hate, I mean I really hate fan noise so I can relate to that. Small fans tend to make more noise because they generally have to spin faster to move the same amount of air as larger fans. So I recommend you inspect your case to see if it will support additional and larger (preferably 120mm or larger) case fans. Generally you want front case fans drawing cool air in, and rear fans exhausting heated air out. This is the most common direction used because power supply fans already push air out the back, and that direction cannot be changed. So "to keep with the flow" case fans are typically oriented to maintain that same flow direction - though you will sometimes see bottom to top.

If your case does not support larger fans, not all fans are created equal. You may just need to replace existing fans with better quality, quieter fans.

It should be noted another responsibility of the case is to suppress noise. So upgrading the case may help here too. I really like Fractal Design cases. Many models have sound deadening lining built right in. Plus FD fans are very quiet. The R4, R5 and R6 come with excellent 140mm fans.

 

[Rahnak]

Well I hate, I mean I really hate fan noise so I can relate to that. Small fans tend to make more noise because they generally have to spin faster to move the same amount of air as larger fans. So I recommend you inspect your case to see if it will support additional and larger (preferably 120mm or larger) case fans. Generally you want front case fans drawing cool air in, and rear fans exhausting heated air out. This is the most common direction used because power supply fans already push air out the back, and that direction cannot be changed. So "to keep with the flow" case fans are typically oriented to maintain that same flow direction - though you will sometimes see bottom to top.

If your case does not support larger fans, not all fans are created equal. You may just need to replace existing fans with better quality, quieter fans.

It should be noted another responsibility of the case is to suppress noise. So upgrading the case may help here too. I really like Fractal Design cases. Many models have sound deadening lining built right in. Plus FD fans are very quiet. The R4, R5 and R6 come with excellent 140mm fans.

I do have a Fractal case, the Define S, which is kind of a Define R lite, if you will. Got some sound deadening on the side panels. And I did just upgrade the stock FD fans it came with, because of some vibration noise I had on the front. I now have three 140mm bequiet! Silent Wings 3, two in the front, one in the back, running at ~1000rpm. There is space to another front fan on the bottom, but it would be mostly pointing at the psu.

 

[killster1]

You should still care about temps - as noted by the 4th line in my signature, "Heat is the bane of all electronics!". You sure don't want anything to "over" heat. It just is not necessary to care about achieving the lowest temps possible.

Since a processor can go from cool to over-heated in just a few clock cycles, and since many processors today easily run at 2 - 3+ billion clock cycles per second, I keep an eye on my temps with a system tray app running full time. I use and recommend Core Temp for that.

Depends on the phone. I don't think most have local thermometers. Just checked my Sony Android and it doesn't.
Yeah, hands don't work. They are pretty good at determining which of two objects are warmer, and they can tell if something is hot or cold. Beyond that, not so much.

I agree with checking the home thermostat if nothing else is available. However, my computer room (office/spare bedroom) has a couple large windows facing the morning sun, my computer and 2 monitors cranking out heat. This room is currently 7° warmer than the thermostat reading out in the hall and my computer and monitors have only been awake for less than 30 minutes. It will soon be warmer yet. And if I wake the other computer and its two monitors, it can soon be a bit toasty in here. So you really need a thermometer in the computer room, and ideally near the main intake vent of the computer. You typically can get a simple wall thermometer for the computer room which is hardly a budget buster. These are typically accurate to within ±2C. That's close enough.

Actually, I recommend users get a decent laser guided infrared thermometer gun. These are great for checking and verifying temps inside and out of your computer case and other electronics. And they are very handy for checking temps in the kitchen (for hot and cold), grilling, the wall next to your HVAC thermostat, kid's foreheads (Not a toy! Avoid eyes!) and more. The better models are more accurate and perhaps more importantly, have much more accurate laser targeting (so your temp reading is for precisely where you are pointing).

Or, if looking for a decent multimeter too, many also have temperature probes. My meter is the Mastech MS8209 which includes a very accurate K-type thermocouple probe on a 40 inch cable - plenty long to dangle inside a closed computer case (or oven). This one is accurate to within ±1%, which is very accurate for this type of device. The thermocouple itself is small enough you can easily put it between the fins of most heatsinks.

you think your cpu is going to die from heat? i used to think that too 20 years ago.. since then i never have had a cpu die, even the ones i built for friends that had no fan and throttled the whole time at 100c where fine when replaced the fan/ pump / clogged with smoke dust dirt grime heatsink. Its nearing winter so im rather happy to leave the pc on and use it for a space heater, during the summer i underclock/undervolt.

you can always buy a smarter thermostat for your house that measures each room anways this is such old reply sorry i was on vaca in italy/france/ireland living it up.

 

[Bill_Bright]

you think your cpu is going to die from heat?

Huh? Where did I EVER say that? I didn't. Please don't try to put words in our mouths. That's not cool. Go by what we actually say.

I NEVER said the CPU will die from heat. But it can over heat which can indeed shorten the lifespan of any electronics device. In fact, if allowed to stay excessively "warm" for extended periods of time, that heat can actually damage the socket or even the motherboard underneath. This is rare but can happen. You can sometimes see the early effects of this by the darkening of the socket materials.

And a CPU (or any electronics device) could die from heat if allowed to get "too hot". This is exactly why thermal protection is hard coded into the processor itself - so it will never get "too hot".

 

[moproblems99]

you think your cpu is going to die from heat?

There aren't a lot of things that kill electronics but heat is one of them.

 

[killster1]

Huh? Where did I EVER say that? I didn't. Please don't try to put words in our mouths. That's not cool. Go by what we actually say.

I NEVER said the CPU will die from heat. But it can over heat which can indeed shorten the lifespan of any electronics device. In fact, if allowed to stay excessively "warm" for extended periods of time, that heat can actually damage the socket or even the motherboard underneath. This is rare but can happen. You can sometimes see the early effects of this by the darkening of the socket materials.

And a CPU (or any electronics device) could die from heat if allowed to get "too hot". This is exactly why thermal protection is hard coded into the processor itself - so it will never get "too hot".

because i asked if you think heat will kill the cpu doesnt mean you said that heat is going to kill the cpu. Like i said i used to care about temps and all that then i realized everything is designed to run at extreme temps and i usually buy decent hardware with decent stock cooling anyway with airflow of course. of course heat will make solderpoints break and brittle but as far as the cpu from heat nah not worried at all (of course ive never spent more than 400 on a cpu and by the time it would die i would be happy to upgrade

 

[Deleted member 185158]

because i asked if you think heat will kill the cpu doesnt mean you said that heat is going to kill the cpu. Like i said i used to care about temps and all that then i realized everything is designed to run at extreme temps and i usually buy decent hardware with decent stock cooling anyway with airflow of course. of course heat will make solderpoints break and brittle but as far as the cpu from heat nah not worried at all (of course ive never spent more than 400 on a cpu and by the time it would die i would be happy to upgrade

The electronics are built to support supposed "heat" in the terms of high temperatures. Your high temp will differ from mine.

So generally VRMs at 90c is within operating range. Fear not, they probably good to 110c.
Ryzen Cpu for example, 3000 series good for 95c before throttle, ThermTrip (board shuts down) around 110c.
Minimum run time on many PC electronics at 100,000 hours at any temp below threshold. That's 4166 days, or 11 1/2 years...... sustained.

You can safely run a PC pretty "hot" below it's threshold and have quiet fans for quite a long time. Just important to stay below threshold.

Really it's just a matter of preference.

 

[tabascosauz]

Opening the side panel doesn't make a difference, so I don't think adding another fan would help much either. It's just a small fan that's too audible for my personal taste above 1000 rpm. Maybe if the case was on the ground, under the desk it wouldn't bother me.


It's certainly a different experience than what I was used to with Intel. Just opening file explorer makes the cooler ramp up. And having Firefox open makes it "idle" between 40~50ºC, which is quite a considerable delta.
Yesterday night I just closed everything except for hwinfo and I just sat here looking at the cpu temp and fan speed and wondering "What the heck are you even doing?".

Boost is aggressive; no way around it unless you want fixed freqs. You can try offsetting voltage. Mine is lowered 0.075v.

Make sure you're on 1.0.0.3ABBA and have the latest chipset drivers. Between Ryzen Balanced and Windows Balanced, just pick whichever works best.

Your best bet is just to have more cooling at your disposal, and also fans that ramp up the speed without increasing much noise or the pitch of the noise. I'm much happier on the U9S with 2 x A9s than I was with the D9L with 2 x A9s and 1 x A9x14.

Mine idles at about 28-35, but as soon as I start actively doing stuff, it's up in the 40s and low 50s. But the A9s are quiet enough that I don't really hear the load.

The voltage is not the problem, because at 1.5V it's not putting much current through. When the load actually needs the bulk of the current, you'll see Vcore come down to the 1.3s if your board does not overvolt.

Another tip is to tie your case fans to something that doesn't spike if they are PWM controlled. My A12x25s initially were linked to either case temp (the speeds wouldn't ever come down again after the air inside got warm) or CPU temp (diabolical noise), so I put them on VRM instead. The CPU still gets the airflow it needs when it needs it, but the case fans don't spike on transient loads.

 

[robot zombie]

Maybe coincidental, but I did have a CPU 'die' from heat once. IIRC it was an Athlon II X3... 450? I think... not that important. I can tell you the mobo was also cheap! The CPU fan died, but the machine kept running. For all I know, it was like that all day. I went to use it, not noticing. Start seeing signs of instability... at stock settings. After a little while, it crashes. After booting back up, I notice the fan isn't spinning. Not having a spare, I popped off the side panel and put a desk fan to it. Temperatures are amazing... stability is not. Not long after I finished the order for a replacement, it went down and never booted again. Fortunately everything went well once I got the replacement CPU and cooler in.

Much older tech at this point and those CPU's had lower throttle points... I think most mobos now will actually not even allow things to run without a CPU fan, though that may only be at boot. I know this to be true with the Strix B350-F - it will not boot without a CPU fan connected and spinning. In the case of my poor Athlon II, it seems the heat strain from just idling like that for however long, though clearly not enough to make it shut itself off, was enough to rapidly and severely degrade it.

Do I think that's relevant here? Probably not. Like I said, I think the protection is probably much better now. I have witnessed a couple of hard shutdowns from excess heat on my own Ryzen 2 system. Just saying that it apparently CAN happen! Ideally, everything works like it's supposed to, but that doesn't mean you count on it to do everything for you. Think of it like a circuit breaker in your house. Just because you have one on a circuit doesn't mean you can overload the circuit and keep tripping the breaker. Eventually, your wiring and/or receptacles are going to overheat. It's still stressing everything every time it gets hot enough to trip. Or maybe for some inexplicable reason the breaker fails. Either way, your house still burns down, even though it was built not to and you're technically well-protected.

 

[Bill_Bright]

because i asked if you think heat will kill the cpu doesnt mean you said that heat is going to kill the cpu.

Sorry. My bad. I thought you were suggesting I did.

then i realized everything is designed to run at extreme temps

Ummm, sorry but that is not true at all. Electronics are designed to operate within a specified operating temperature range. The top end of that range may be extremely "warm" but should never cross the threshold into "hot" unless it is a heating element or something similar.

of course heat will make solderpoints break and brittle

Technically, this is true. But if this happens, it means (1) there was "excessive" heat that never should have occurred. (2) The wrong solder formula was used. (3) The solder was exposed to some unexpected "mechanical" stress. Or (4), it was a bad solder joint in the first place.

Maybe coincidental, but I did have a CPU 'die' from heat once. IIRC it was an Athlon II X3... 450? I think... not that important.

But it is important as that gives us a clue as to the generation and release date of the CPU. And that series came out a decade ago. Today's processors have more sophisticated thermal protection features that likely react much more quickly too. But also, that CPU could have been faulty already. Hard to tell at this point and really, it would have to be examined under a microscope to see what damage actually occurred.

A CPU can go from cool to overheated in just a few clock cycles. And when there's 3+ billion clock cycles per second, it does not take long after a fan seizes to overheat.

 

[Rahnak]

Boost is aggressive; no way around it unless you want fixed freqs. You can try offsetting voltage. Mine is lowered 0.075v.

Make sure you're on 1.0.0.3ABBA and have the latest chipset drivers. Between Ryzen Balanced and Windows Balanced, just pick whichever works best.

Your best bet is just to have more cooling at your disposal, and also fans that ramp up the speed without increasing much noise or the pitch of the noise. I'm much happier on the U9S with 2 x A9s than I was with the D9L with 2 x A9s and 1 x A9x14.

Mine idles at about 28-35, but as soon as I start actively doing stuff, it's up in the 40s and low 50s. But the A9s are quiet enough that I don't really hear the load.

The voltage is not the problem, because at 1.5V it's not putting much current through. When the load actually needs the bulk of the current, you'll see Vcore come down to the 1.3s if your board does not overvolt.

Another tip is to tie your case fans to something that doesn't spike if they are PWM controlled. My A12x25s initially were linked to either case temp (the speeds wouldn't ever come down again after the air inside got warm) or CPU temp (diabolical noise), so I put them on VRM instead. The CPU still gets the airflow it needs when it needs it, but the case fans don't spike on transient loads.

I am running 1.0.0.3ABBA and Ryzen High Performance power plan with latest drivers. Didn't notice any diference in temperatures/fan ramping up between high and balanced.

The main issue was the stock cooler just being too loud for my personal comfort. I just installed the Scythe Fuma 2 and now I can barely hear it. CPU fans at around ~450rpm and case fans ~850rpm. Idle temps are lower too. I may play around with voltages and stuff when I have more free time. I'd also like to tinker with the ram a little. But very pleased

Good tip for the case fans, I did just that.