AMD Wraith Question

For how long did you run each of those Prime 95 tests? It takes a while before Prime 95 starts actually taxing the cpu. It sounds to me like you ran it 5 minutes at a time. Let it go 15 minutes and see what happens.
Yeah I think Im good with Prime 95, I already have a idea what the result will be. Im pretty happy with the performance of the 3900X in games, and also just as happy with how the stock cooler has performed.
 
Yeah I think Im good with Prime 95, I already have a idea what the result will be. Im pretty happy with the performance of the 3900X in games, and also just as happy with how the stock cooler has performed.
C'mon man, you've run every other test we asked of you and you're going to stop here? ;)
 
Prime 95 Small FFT’s, 30 min run, all CPU cores stayed at 3800GHZ and never broke 67C for temps, ambient temps are 64F.
 
Last edited:
Prime 95 Small FFT’s, 30 min run, all CPU cores stayed at 3800GHZ and never broke 67C for temps, ambient temps are 64F.
Seems pretty good to me. Almost unbelievable numbers on stock cooling. Your ambients are very low though which helps.
 
Seems pretty good to me. Almost unbelievable numbers on stock cooling. Your ambients are very low though which helps.
We run a cool house year round, even summer we keep it at 68F, which really benefits computer components. That and the AMD Wraith is a pretty solid cooling solution combined with a well ventilated case and low ambient temps.
 
Prime 95 Small FFT’s, 30 min run, all CPU cores stayed at 3800GHZ and never broke 67C for temps, ambient temps are 64F.

It sounds like the processor is being throttled. With temps like that it should be boosting higher until the temps go up quite a bit more. I'm willing to bet there's a setting in BIOS for throttling once you reach a specific temperature and it's set to around 67C instead of the normal 95C or so.
 
It sounds like the processor is being throttled. With temps like that it should be boosting higher until the temps go up quite a bit more. I'm willing to bet there's a setting in BIOS for throttling once you reach a specific temperature and it's set to around 67C instead of the normal 95C or so.
There is no setting like that I’m aware of, maybe cool and quiet? Usually when someone is hitting 95c they have high ambient temps, a poorly ventilated case or bad mounting on the CPU. When I ran the 30 min test the frequency was 3800ghz, which is the base operating frequency of the 3900X. Normally when running Prime 95 With all the cores at 100%, a CPU will default to its base clock, at least that is what I have experienced in the past.
 
There is no setting like that I’m aware of, maybe cool and quiet? Usually when someone is hitting 95c they have high ambient temps, a poorly ventilated case or bad mounting on the CPU. When I ran the 30 min test the frequency was 3800ghz, which is the base operating frequency of the 3900X. Normally when running Prime 95 With all the cores at 100%, a CPU will default to its base clock, at least that is what I have experienced in the past.

This is my point. My 2600x with the Wraith Spire cooler pretty much never goes below 3.8Ghz all core even when it hits 85C in a high ambient temperature room. That's 200mhz above the base clock and it's always boosting. You're not boosting at all even with low temps and that's not right. It should be boosting quite a bit unless there is something throttling the CPU to keep temps down. In the BIOS of my Asus B450 Gaming-F board there is a setting under the PBO sub menu to set a different throttling temperature. The default is 95C but it can be set lower. More than likely you have something similar in the BIOS and it's probably set quite low. The other option is there is a power limit setting in the BIOS and it's set quite low. Something like a 65 watt or 95 watt limit.

No matter what, you're not getting anywhere near the performance you're supposed to since the CPU isn't boosting except for rather light CPU loads. At the temps you're reporting my 2600x is normally boosting to at least 4Ghz on all cores and under a quite high CPU load.
 
This is my point. My 2600x with the Wraith Spire cooler pretty much never goes below 3.8Ghz all core even when it hits 85C in a high ambient temperature room. That's 200mhz above the base clock and it's always boosting. You're not boosting at all even with low temps and that's not right. It should be boosting quite a bit unless there is something throttling the CPU to keep temps down. In the BIOS of my Asus B450 Gaming-F board there is a setting under the PBO sub menu to set a different throttling temperature. The default is 95C but it can be set lower. More than likely you have something similar in the BIOS and it's probably set quite low. The other option is there is a power limit setting in the BIOS and it's set quite low. Something like a 65 watt or 95 watt limit.

No matter what, you're not getting anywhere near the performance you're supposed to since the CPU isn't boosting except for rather light CPU loads. At the temps you're reporting my 2600x is normally boosting to at least 4Ghz on all cores and under a quite high CPU load.
I took a look and there is settings in there regarding throttle limits, but the option looks to be Auto or Manual which I guess you enter your own. Anyway Regardless I dont feel like blowing up my system so I'll just let it be. I really dont care about Prime 95 or Cinebench, the CPU seems to be boosting just fine with games which is exactly what I want.
 
If your TDP limited by Cooler .. then the pre CCX would allow one CCX or more to drop the clocks while keeping others boosted without dropping all at once I think is how it works .. re run Ryzen Master set up
 
It sounds like the processor is being throttled. With temps like that it should be boosting higher until the temps go up quite a bit more. I'm willing to bet there's a setting in BIOS for throttling once you reach a specific temperature and it's set to around 67C instead of the normal 95C or so.

There is throttling in steps. Around 80C it starts to throttle of how far it will boost. Then at 95C it throttles to whatever keeps the temps down.

After running P95 Smalls on my 3900X I get all core of 3.95Ghz and temps at 75-76C in a room that is 67F, 19C. According to Ryzen Master, with CPU power between 155-160 Watts. I am even typing this while it is running. I have a 280 AIO on mine though.

But I would think that something is strange since kilroy67 67 is at 67C and there seems to be something limiting it since it is not the Temp. Maybe change the power from the standard AMD settings to the MB setting which will raise TDC and EDC and allow higher boosts. My board supports EDC of 280A and TDC of 210A.
 
After running P95 Smalls on my 3900X I get all core of 3.95Ghz and temps at 75-76C in a room that is 67F, 19C. According to Ryzen Master, with CPU power between 155-160 Watts.
Running the same test all cores running at 3800GHZ using 113 watts, 65-66c for temps according to AMD Ryzen Master. At 3.95ghz and 155-160 watts that seems very high, we are talking a difference of 150mhz between our CPU speeds and a extra 42-47 watts to achieve it.

I also ran Cinebench for 15min and all cores remained at 4000GHZ and the total watts used was still 113w, while heat stayed at 65-66c. Again this according to Ryzen Master, with HWinfo running alongside.
 
Running the same test all cores running at 3800GHZ using 113 watts, 65-66c for temps according to AMD Ryzen Master. At 3.95ghz and 155-160 watts that seems very high, we are talking a difference of 150mhz between our CPU speeds and a extra 42-47 watts to achieve it.

I also ran Cinebench for 15min and all cores remained at 4000GHZ and the total watts used was still 113w, while heat stayed at 65-66c. Again this according to Ryzen Master, with HWinfo running alongside.

And that is fine if you are ok with not getting the most out of the chip. That was running solid to get that wattage and temps. While during normal day to day you will not be running 100%. Running Cinebench I get 4250 all core. I get single core bumps to 4550 as well. I wanted to get the most from my processor so I allow the motherboard to supply what is needed to get it. The chip does the rest of course. I just removed the 115W limit by allowing the motherboard to decide instead of the chip. Some people may want that extra Mhz, I am one I will admit. I am not concerned with the extra power consumed since it does not run 100% all core all day. My temps rarely ever bump above 55C during normal operation and if they do it is a split second. They hover in the mid 30's to mid 40's. I don't even have my fans ramp up until 55C, they radiator fans sit at 40% under that temp. Anyway I was just stating that the power settings were most likely why you were getting lower all core clocks and lower temps.
 
And that is fine if you are ok with not getting the most out of the chip. That was running solid to get that wattage and temps. While during normal day to day you will not be running 100%. Running Cinebench I get 4250 all core. I get single core bumps to 4550 as well. I wanted to get the most from my processor so I allow the motherboard to supply what is needed to get it. The chip does the rest of course. I just removed the 115W limit by allowing the motherboard to decide instead of the chip. Some people may want that extra Mhz, I am one I will admit. I am not concerned with the extra power consumed since it does not run 100% all core all day. My temps rarely ever bump above 55C during normal operation and if they do it is a split second. They hover in the mid 30's to mid 40's. I don't even have my fans ramp up until 55C, they radiator fans sit at 40% under that temp. Anyway I was just stating that the power settings were most likely why you were getting lower all core clocks and lower temps.
Yeah defiantly I understand, I am more cautious just because the VRM's on my board are not that great, and I am using air cooling. So that said I just let it run at default settings with PBO on, which suits me fine for gaming. My next upgrade is the Nvidia 3000 series if that's what they are called, at some point I may get a higher end X570 like the Gigabyte Master, with a more robust VRM solution .
 
Yeah defiantly I understand, I am more cautious just because the VRM's on my board are not that great, and I am using air cooling. So that said I just let it run at default settings with PBO on, which suits me fine for gaming. My next upgrade is the Nvidia 3000 series if that's what they are called, at some point I may get a higher end X570 like the Gigabyte Master, with a more robust VRM solution .

Understandable. I am on my second X570 board. I wanted a hair better VRM in case the 4000 series will work on the X570 boards when then come out, not the APU ones. It really did not increase anything mush but the board has more features, it should it was twice as much as the first. Lol. As long as your happy that is all that matters.
 
Here's a bit of anecdotal evidence of what I was talking about with regards to leaving performance on the table.

I just picked up a Cooler Master Hyper 212 on sale on Amazon to replace the Wraith Spire which came with my 2600x. With the Wraith Spire my typical workload with all 12 threads maxed out with either distributed computing projects or video encoding or gaming or a mixture gave me a CPU temp of around 80-85C and boost clocks of 3.8-3.9Ghz. The base clock of the 2600x is 3.6Ghz.

The same approximate workload with the CM Hyper 212 with a push/pull fan configuration has the temps at 58-65C and boost clocks of 4.05-4.15Ghz. Remember, this is all core boost and full processor load.

Just to see what it would do I had a light load of only a couple threads and the boost on the working cores was the max at 4.25Ghz and didn't leave there. The temps were maybe in the low 40s.

While you would need a better cooler than a CM Hyper 212 to properly cool a 3900x I think the comparison is still apt. The Wraith Prism is good for a stock cooler but it is going to leave a lot of performance unrealized compared to a good air cooler or water cooler. In my case I'm looking at an approximate temperature drop of 20C and a boost clock increase of around 200mhz and that's quite a lot of performance (especially for highly threaded work) that I was leaving behind while using the stock cooler.
 
Here's a bit of anecdotal evidence of what I was talking about with regards to leaving performance on the table.

I just picked up a Cooler Master Hyper 212 on sale on Amazon to replace the Wraith Spire which came with my 2600x. With the Wraith Spire my typical workload with all 12 threads maxed out with either distributed computing projects or video encoding or gaming or a mixture gave me a CPU temp of around 80-85C and boost clocks of 3.8-3.9Ghz. The base clock of the 2600x is 3.6Ghz.

The same approximate workload with the CM Hyper 212 with a push/pull fan configuration has the temps at 58-65C and boost clocks of 4.05-4.15Ghz. Remember, this is all core boost and full processor load.

Just to see what it would do I had a light load of only a couple threads and the boost on the working cores was the max at 4.25Ghz and didn't leave there. The temps were maybe in the low 40s.

While you would need a better cooler than a CM Hyper 212 to properly cool a 3900x I think the comparison is still apt. The Wraith Prism is good for a stock cooler but it is going to leave a lot of performance unrealized compared to a good air cooler or water cooler. In my case I'm looking at an approximate temperature drop of 20C and a boost clock increase of around 200mhz and that's quite a lot of performance (especially for highly threaded work) that I was leaving behind while using the stock cooler.
Im not understanding how your comparing a Wraith Spire to a Cooler Master Hyper 212, its like night and day. Ive seen head to head comparisons of the AMD Wraith Prism vs the Hyper 212 and they are pretty much even, in fact they both perform as well as a lot of the AIO solutions out there, which you can also find head to head comparisons of. The problem with these big massive hunks of metal for a air cooler is eventually they will warp your MB over time due to their weight, its gravity no way around it. That and they will really perform not much better than the Wraith Prism. So really the best solution is a custom loop, which takes some skill and knowledge to master, plus a good chunk of money. One also must remember ambient temps play a huge role in performance, high ambient temps=high computer component temps.
All in all for what I do with the 3900X, which is gaming, the AMD Wraith serves me just fine, I never see temps hit above 65c and Im maintaining a average clock of 4325ghz all cores with boost clocks hitting 4475-4550ghz. In desktop situations I see even as high as 4650ghz on a handful of cores.
 
Let me just add that my buddy has a top end custom loop, when its comes to gaming both out 3900X's behave pretty much the same. Where he really shines is in full workload situations with all cores maxed, he can maintain a higher boost clock than I can for a longer period of time. But gaming we are very similar in performance, so it really boils down to what you use your system for.
 
Im not understanding how your comparing a Wraith Spire to a Cooler Master Hyper 212, its like night and day. Ive seen head to head comparisons of the AMD Wraith Prism vs the Hyper 212 and they are pretty much even, in fact they both perform as well as a lot of the AIO solutions out there, which you can also find head to head comparisons of. The problem with these big massive hunks of metal for a air cooler is eventually they will warp your MB over time due to their weight, its gravity no way around it. That and they will really perform not much better than the Wraith Prism. So really the best solution is a custom loop, which takes some skill and knowledge to master, plus a good chunk of money. One also must remember ambient temps play a huge role in performance, high ambient temps=high computer component temps.
All in all for what I do with the 3900X, which is gaming, the AMD Wraith serves me just fine, I never see temps hit above 65c and Im maintaining a average clock of 4325ghz all cores with boost clocks hitting 4475-4550ghz. In desktop situations I see even as high as 4650ghz on a handful of cores.

I have anecdotal evidence right here that a hyper 212 evo doesnt warp anything, even after 6 years.
 
I have anecdotal evidence right here that a hyper 212 evo doesnt warp anything, even after 6 years.
I’m not talking about the Cooler Master 212, I’m referring to the much larger ones. Maybe I didn’t make it clear, my apologies.
 
I’m not talking about the Cooler Master 212, I’m referring to the much larger ones. Maybe I didn’t make it clear, my apologies.

I have a Thermalright Ultima90 and TRUE120 which were hanging off motherboards for more than ten years, the TRUE120 still is as a matter of fact. Both of these heatsinks weigh a lot more than the 212. The TRUE probably weighs twice as much as the 212. Even after more than a decade of hanging from motherboards neither one of the heatsinks warped the motherboards they were attached to.

Based on your posts you seem to be operating under some really odd and incorrect information. My question to you is why did you post this thread in the first place if you were going to tell everyone they are either wrong or that you don't care and you're not going to do it anyway? You asked if there would be any differences and we answered yes. We told you you were leaving performance on the table, especially when you literally couldn't run higher than the base speed when you had any sort of real load on the processor.
 
I have a Thermalright Ultima90 and TRUE120 which were hanging off motherboards for more than ten years, the TRUE120 still is as a matter of fact. Both of these heatsinks weigh a lot more than the 212. The TRUE probably weighs twice as much as the 212. Even after more than a decade of hanging from motherboards neither one of the heatsinks warped the motherboards they were attached to.

Based on your posts you seem to be operating under some really odd and incorrect information. My question to you is why did you post this thread in the first place if you were going to tell everyone they are either wrong or that you don't care and you're not going to do it anyway? You asked if there would be any differences and we answered yes. We told you you were leaving performance on the table, especially when you literally couldn't run higher than the base speed when you had any sort of real load on the processor.
No one has been able to show me any real benefit to getting a bigger air cooler or AIO for what I use the system for which is gaming. A handful of MHZ to gain a bigger score on Cinebench or run Prime isn’t important to me. That and the thread pretty much just went off the rails.
 
No one has been able to show me any real benefit to getting a bigger air cooler or AIO for what I use the system for which is gaming. A handful of MHZ to gain a bigger score on Cinebench or run Prime isn’t important to me. That and the thread pretty much just went off the rails.

If you are happy with it then great. I did have a Noctua NH-D15S on mine and it held its own. Sure during something like Prime95 it would peg the processor at 95C but who is running Prime95 all day. Lol. I did add another Noctua Fan to it but in the end it was so huge I could not even get to it to remove it easily when I wanted to replace it. I actually had to bend the one side by the RAM out of the way to get the middle fan out so I could get to the screw to remove it. Needless to say it is useless now, my old case was kind of on the small side so the top fans made it hard to get to the stupid fan clip on the top. I do like the look of that Corsair Air Cooler. Not sure how good it is. I put my Wraith Prism from my 3900X on my 2600 that I have sitting in a back room waiting to see who I can pass it on to.

Anyway the Wraith Prism works for normal day to day, but if I were you I would try a little better Air Cooler just for better cooling while gaming. The cooler it runs the longer it shall last.
 
This is my point. My 2600x with the Wraith Spire cooler pretty much never goes below 3.8Ghz all core even when it hits 85C in a high ambient temperature room. That's 200mhz above the base clock and it's always boosting. You're not boosting at all even with low temps and that's not right. It should be boosting quite a bit unless there is something throttling the CPU to keep temps down. In the BIOS of my Asus B450 Gaming-F board there is a setting under the PBO sub menu to set a different throttling temperature. The default is 95C but it can be set lower. More than likely you have something similar in the BIOS and it's probably set quite low. The other option is there is a power limit setting in the BIOS and it's set quite low. Something like a 65 watt or 95 watt limit.

No matter what, you're not getting anywhere near the performance you're supposed to since the CPU isn't boosting except for rather light CPU loads. At the temps you're reporting my 2600x is normally boosting to at least 4Ghz on all cores and under a quite high CPU load.

Thorttling doesn't only happen with temperature, it also happens when you hit your TDP limit. Your comparing your 6 core to his 12. I'm willing to bed if he only loads half his cores, he'll have higher boost levels.
 
Thorttling doesn't only happen with temperature, it also happens when you hit your TDP limit. Your comparing your 6 core to his 12. I'm willing to bed if he only loads half his cores, he'll have higher boost levels.

If this was the case, every single review of the CPU would have shown throttling to the point of no boost. Instead I haven't seen a single review that showed this. The six vs twelve core also doesn't matter much since the 3900x has a higher TDP in the first place. Hell, when I have all twelve threads fully loaded on my 2600x I'm pushing 4.0Ghz+ and still within the 95w TDP. As a matter of fact, there isn't much more power or heat when going from 6 to 12 threads in most cases.

The fact of the matter is he's being throttled rather badly when the CPU has more than a couple threads running and much of that throttling is going to be due to thermals.
 
If this was the case, every single review of the CPU would have shown throttling to the point of no boost. Instead I haven't seen a single review that showed this. The six vs twelve core also doesn't matter much since the 3900x has a higher TDP in the first place. Hell, when I have all twelve threads fully loaded on my 2600x I'm pushing 4.0Ghz+ and still within the 95w TDP. As a matter of fact, there isn't much more power or heat when going from 6 to 12 threads in most cases.

The fact of the matter is he's being throttled rather badly when the CPU has more than a couple threads running and much of that throttling is going to be due to thermals.

Not sure what review's your reading, but the ones I read are typically benchmarking games, encoding and productivity tasks, not Prime95 which is what your basing your "something is seriously wrong with your setup" claim. There's also a signficant difference between a 6 core and 12 core CPU's power consumption. It's no accident that one comes with a much better cooler than the other.
 
Not sure what review's your reading, but the ones I read are typically benchmarking games, encoding and productivity tasks, not Prime95 which is what your basing your "something is seriously wrong with your setup" claim. There's also a signficant difference between a 6 core and 12 core CPU's power consumption. It's no accident that one comes with a much better cooler than the other.
In games I get the same performance typically as my friend who has a 3900X and a custom loop, we both maintain a similar all core clock and boost to pretty much the same range of 4400-4500ghz. Running Cinebench R20 I can maintain a clock of about 4000-4100ghz for a score of around 7200, which is right around many reviews I have read including Guru3D, which they score 7155 with the 3900X. So I guess if that bad performance, I'll take it...Im not going to base my system performance on Prime 95 which is great for loading your CPU to 100% and producing a lot of heat, but not much else.
 
Something to think about, when I was binning my 3900x chips, which basically turned into running all kinds of tests with different voltages and what not on one of them, I found that the lower the temperature at a given speed, the lower the voltage I could also get away with running. This contributed to the chip not bumping into the wall at the top of the power envelope as quickly. 3900x Binning Conclusion . Spoiler: With one perfect seating of my AIO, I got the 3900x to complete 5 minutes of CB R20 at 4.25 GHz with voltage set at 1.19375v with no clock stretching. Temperature delta above ambient was 36.4C. I didn't mess with PBO settings, this was at stock settings for power.
 
In games I get the same performance typically as my friend who has a 3900X and a custom loop, we both maintain a similar all core clock and boost to pretty much the same range of 4400-4500ghz. Running Cinebench R20 I can maintain a clock of about 4000-4100ghz for a score of around 7200, which is right around many reviews I have read including Guru3D, which they score 7155 with the 3900X. So I guess if that bad performance, I'll take it...Im not going to base my system performance on Prime 95 which is great for loading your CPU to 100% and producing a lot of heat, but not much else.

My 3900X would never hold that sort of boost in Cinebench R20 with stock power restrictions, temperature didn't matter at that point as I was bumping TDP below 4 GHz.
 
My 3900X would never hold that sort of boost in Cinebench R20 with stock power restrictions, temperature didn't matter at that point as I was bumping TDP below 4 GHz.
Ive run it numerous times and averages 4025-4075ghz all cores with latest score of 7249, in the beginning of the test boosting to 4100ghz and by the end at 4025-4050ghz, temps are typically 65-66c. My ambient temps are around 63F.

Capture.PNG
 
Ive run it numerous times and averages 4025-4075ghz all cores with latest score of 7249, in the beginning of the test boosting to 4100ghz and by the end at 4025-4050ghz, temps are typically 65-66c. My ambient temps are around 63F.

View attachment 224781
I believe you. You have a more efficient 3900X than mine. My 3950X is also not a great clocker as it scores lower and boosts lower in CB R20 than most others I've seen. The silicon lottery is alive and well. You got a nice one.
 
This run goes up to 71C. As you can see it hits 4.275Ghz all core. I get 517 single core at 4.6Ghz. This is obviously not with a Wraith Prism but show what the processor can do. I use an AIO but I am pretty sure one of the bigger Air Coolers could keep the temp in the low 70C range as well.
 

Attachments

  • R20-02202020.jpg
    R20-02202020.jpg
    170.9 KB · Views: 0
  • R20-02202020RM.jpg
    R20-02202020RM.jpg
    129.4 KB · Views: 0
This run goes up to 71C. As you can see it hits 4.275Ghz all core. I get 517 single core at 4.6Ghz. This is obviously not with a Wraith Prism but show what the processor can do. I use an AIO but I am pretty sure one of the bigger Air Coolers could keep the temp in the low 70C range as well.
Thats a nice score, Im thinking at some point of moving up to a good AIO or maybe even a custom loop if I can afford it. Water seems to be the way to go, I really don't dig those massive air coolers.
 
Thats a nice score, Im thinking at some point of moving up to a good AIO or maybe even a custom loop if I can afford it. Water seems to be the way to go, I really don't dig those massive air coolers.

Yeah I had the Noctua on it and really it was huge and covered the RAM somewhat and just filled the case above the video card. There was like an inch between the back fan on the cooler and the exhaust fan on my case. I had to mount the fan on the back side because it would not clear my RAM and let me put the outside cover on the case since the second fan hit the case to clear the RAM. With the AIO or a custom loop the case is much more open and air flow nicer. If I go with say a 4000 series when they come out I may do custom loop. I have all Corsair lighting and fans so I would do a custom Corsair loop and i priced that up and it is another $500, which I am currently between jobs so that is not going to happen. Lol.
 
So what would be a good air cooler that is easy to mount using the existing back plate. Ive looked at the Noctua and its just massive, also considered one of the Be Quiet models.
I'd read all of the cooler reviews, then make a decision on that. 65fully loaded seems pretty damn cool. Like
I found some stuff out tonight, by putting my CPU VDDP voltage to Normal, and enabling PBO under AMD CBS the 3900X hovers around 4325ghz while boosting to 4475-4600ghz while playing BL3. Temps are staying around 55-60c with Ambient temps of 64F.
I read somewhere that lowering the voltage by .1 you can get those lower temps.
 
Back
Top