저자: OlavDeng2 · 외국인
요약 보기
결론: 스톡 설정의 AMD R9 Nano는 과열로 인한 서멀 스로틀링이 아니라 전력 제한(파워 스로틀링)으로 클럭이 떨어지며, 전력 제한을 올리면 성능이 개선된다는 주장이다. [테스트 배경] 댓글과 일부 리뷰(JZ2Cents 등)에서 R9 Nano가 과열되어 클럭이 떨어진다는 주장이 있었고, 이를 확인하기 위해 직접 테스트했다. [테스트 환경] 노르웨이에서 실내 온도 약 18.1도 상태. 시스템은 FX 8320(4.3~4.5GHz 오버클럭), 커스텀 수냉 루프와 360mm 라디에이터 사용. Nano는 공랭. 여러 앱을 열어둔 상태였다. [테스트 1: 스톡 설정] 클럭 1000MHz, 전력 제한 0%, 메모리 500MHz 기준으로 Fire Strike Extreme을 약 10분간 돌렸다. 코어 클럭은 약 900~920MHz 수준이었고 GPU 사용률은 대부분 100%, 팬 속도는 약 41%, 최고 온도는 74도였다. [테스트 2: 전력 제한 +50%] 같은 조건에서 전력 제한만 50%로 올리고 그래픽 테스트 2를 반복했다. 클럭은 최저 940MHz 수준까지 떨어졌고, 평균 클럭은 약 800MHz 높아졌다고 설명했다. 온도는 약 4도 상승했다. [테스트 3: 팬 100% 고정, 전력 제한 0%] 팬을 100%로 고정해 온도 문제를 제거하고 전력 제한 0% 상태에서 다시 측정했다. 클럭은 약 900MHz 근처로 다시 스로틀링되었고, 온도는 약 70도 수준, 평균 클럭은 약 920MHz 근처였다. 전력 제한을 +50%로 올리면 스로틀링 없이 1000MHz 유지가 확인되었다. [테스트 4: 오버클럭 설정 1040MHz, 전력 제한 +50%, 사용자 팬 커브] 약 10분간 실행했고, 클럭은 대체로 1040MHz 근처를 유지했다. 온도는 약 75~78도, GPU 사용률은 대부분 100%였다. 이 결과를 근거로 온도가 높아도 클럭이 유지되므로 스톡 조건의 스로틀링은 열보다 전력 제한 때문이라고 주장했다. [결론] 발표자는 R9 Nano가 스톡 설정에서 열 스로틀링은 없고 파워 스로틀링만 있다고 결론지었다. 4096개 스트림 프로세서를 175W TDP 안에서 운용하기 위해 전력을 제한하며, 이것이 에너지 효율이 좋은 이유이기도 하다. 다만 전력 제한을 올리면 추가 성능을 얻을 수 있다. 발표자는 주변 온도가 높은 환경(예: 40도)에서는 과열이 발생할 수 있다고 덧붙였다. 발표자는 틀렸다면 댓글로 지적해 달라고 요청했다. [한계] 발표자 스스로 과학적으로 엄밀한 테스트는 아니라고 인정했고, 자막 인식 오류와 촬영 문제로 일부 수치는 부정확할 수 있다. 실내 온도가 낮은 환경에서의 테스트라는 점도 한계다. [종목·자산 관련] 증권 관련 발언은 없다. 언급된 대상 자산(삼성전자, 엔비디아, AMD 등)에 대한 가격 방향 예측도 없다. AMD는 제품 하드웨어(R9 Nano)로만 언급되었다.
다듬은 글
Hello everyone and welcome to this short video. So , maybe I'm a little late. Nano came out quite a while ago. But I'll make a little video to prove that I have a Nano. Here she is. It's in my system now. Here we have Afterburner settings. I don't know if you can see clearly. In any case, the settings are stock.
Core frequency 1000 MHz, power limit 0. Memory frequency, oh, here. 500. I also apologize. I'm filming on my phone. And one more thing, this is not my usual content, but I want to highlight it. And here we have Fire Strike, which I'm going to configure . Also characteristics. Sorry again for confusing me .
This is my first time doing this. So the reason I'm doing this is because in some of my comments on YouTube and on some subreddits, AMD hardware is maybe also PC master race. There were comments that the R9 Nano overheats and drops frequencies. And there were some reviews where this was also stated, for example, JZ2 Cents reviews said that the R9 Nano overheats.
So of course I have uh I live in Norway so it's a little colder here than other countries. But I have a stock Nano. Also keep in mind that I have a custom water circuit and a 360mm radiator in the front as an intake that pulls air from outside into the case, cooling my FX 8320. The Nano is not part of that, as you saw, it is air- cooled.
And, I must admit, I have quite a few apps open. But I don't think it matters. If people ask me to redo with closed applications, I can do it, but I'm a little lazy right now . So, the plan is this: do two activities. One event where we run Fire Strike on stock settings. Actually, three measures on stock settings.
Without increasing the power limit, without overclocking or anything else . The second measure is only with an increased power limit. The third event is with overclocking, and we will see what the core frequencies will be during all three events. And I will use MSI Afterburner, it is my favorite overclocking software.
I usually overclock my R9 Nano to 1040 MHz. And the power limit is 50% and the fans' own curve. Of course, the fan curve, as you can see, is not custom right now . So, this is a test. I will be running Fire Strike Extreme. Also, I'm really sorry for my terrible, terrible, uh, what's it called? The recording is going on, but as it is.
So , I'm going to run this test for, uh, 10 minutes. And then I'll, uh, come back. So , one thing to remember: I don't know why this happens, but the second screen goes black during Fire Strike, and only during it. I don't know why this is happening. This doesn't happen in other games. So I can't follow it live, of course, but I can always check out the schedule after it's over.
I'll be back once Fire Strike is finished and show you the results for stock settings . So, as you can see, the test is underway. I will be back soon. This has been going on for about 10 minutes. So, this was a completely stock R9 Nano. Well, let's go out. And let's see . If we go to the right here, let the focus adjust.
I hope you see the numbers. Come on, focus. Just don't let me down now. That's it. 900, approximately 910-920. Somewhere in between, say 900. GPU usage was usually 100%, with small dips . Fan speed 41% at drain. The temperature is a maximum of 74. So, if the theory that the R9 Nano is dropping frequencies due to overheating is correct, then at this point the R9 Nano would have, uh...well, we would see this result if it were dropping frequencies due to temperature or due to a power consumption limit.
That's why we're here. We will check if it is thermal throttling or power throttling. So, go to the Power Limit slider, increase it to 50. Apply. And let's run the test again. Fire Strike Extreme. I also switched to graphics test #2 because it's longer. I'll loop it for another 10 minutes. And then we'll go check the results.
How are you, okay, it's late now, but as you could see, it's still 1000 MHz core frequency and 500 MHz for HBM memory. So, I'll be back in 10 minutes on my countdown. After that, we'll see the results, and then I'll test the overclock. And if my theory is correct, it should now hold 1000 MHz, although the temperature should be higher, because higher core frequencies mean higher voltage and, as a result, higher temperatures.
But the core frequency will be 1000 MHz if I'm right. If I'm wrong, the frequency should drop. So, we'll see now. And yes, I'll be back soon. Hello everyone. Okay, I'm back. Let's see. Let's get out of here. Theoretically, the frequency should be one thousand. And if I focus, well , focus. Stop these prompts.
I need to see if I can turn that off. Well, focus. Is it well focused? Here , now it's good. You may see some drawdowns. The lowest value drops to 940. Yes . However, the temperature is about 4° higher. And the frequencies are on average about 800 MHz higher. So, I see two reasons for this. There used to be no thermal throttling, but now there is, or it's still a power limitation.
But in any case, the power limitation is less than it was before. But the temperature also increased by 4°, so it's quite possible that it's thermal throttling. I don't know at what level the R9 Nano starts thermal throttling. So I'll do two more tests. I'll lower the power level back to zero. I will turn off the automatic fan mode.
Oh, not user-friendly. I will put it at 100%. Here's my theory, and I'll do another test with 50. Um, plus 50 to the power limit. My theory is this: if at 100% fan speed and zero power limit there is still throttling, but when the limit is increased there is no throttling, then in the previous test with the increased limit and standard fan speed there was thermal throttling.
However, at standard fan speed and zero power limit, it was throttling due to power limitation . Although I could be wrong. And if that's the case , excuse my finger. Don't be shy about writing your counterarguments. But anyway, I'll run the test again. Let's see the results. I'll be back in 10 minutes.
And if my current theory is correct, the frequency should drop back to 900 or so MHz. And then I have to do another test. And with this test, if there is no thermal throttling, but there is power throttling, then the next test after that should produce a stable 1000 MHz. But this is just a theory. And after that, I can do a test with the current overclocking settings, just for fun.
So , yes, I'll be back soon. Okay, we're back. So, this went on for about 10 more minutes. Let's get out of here and try again . Now, if I was right, it should throttle just like it would at standard fan speed. And if we look, no more of these damned clues. If only it would focus. Damn camera. That's it.
I'm really sorry. Oh, look. It resets the frequency. These are the settings, so let's try the same thing with plus 50. So, as we saw, it reset the frequency. So let 's get back to the test. And let's try again. Unless it drops the frequency here, I'd say I have strong evidence that at stock settings the R9 Nano doesn't have power issues or overheat, but is still power-limited.
But this will of course also depend on the ambient temperature. Now, I had to start with this. Sorry. So, my indoor temperature, sorry about the outdoor temperature, the sensor is broken. But the temperature in the room is 18.1 degrees Celsius. So, uh, if you live in a place with an ambient temperature of 40 degrees Celsius, chances are it will overheat.
But anyway, I'll leave it for the conclusions. So, yes. I'll be back in about 10 minutes of my time. We'll see the result, and then I'll do another run with overclocked settings just for the sake of curiosity. To see the core frequency results. But yes. I will be back soon. Okay, the test has been going on for about 10 minutes now.
So this test should be convincing, I think. But if you don't believe the results of this, the following will definitely be convincing . So, that's where I use overclocking. OK. Apart from that slight dip, it's out of focus, right? Well, focus. You can do it. A camera that failed, apparently. That's it.
Thousandth, thousandth. This is a very large subsidence. But this is more the exception than the rule. 70 ° C. 1000 MHz. I forgot to check the temperature. GPU utilization is not always 100%, but most of the time it is 100%. So, yes . That's pretty convincing, but here are my overclocking settings. I'm going to use user-defined fan settings.
Uh, is the car on? This is a car, as far as I'm concerned. So if those results weren't convincing enough, here are my overclocking settings . I will run the same test with these settings. As you can see, here is the test again. So if this result, I think, I think the results that we just got are quite convincing.
But if you still have doubts, this test should be the final nail in the coffin to prove that at standard frequencies, nano cards have no overheating problems, but I'll talk about that in the conclusions. So I'll be back soon. And we are ready to see the final results . So, this test lasted a little over 10 minutes, I think.
My overclocking results. And let's take a look. Let's take a look. The focus is there and working at about 1040. So what can we make of this ? It runs at 1040 with these settings. GPU usage is mostly like this, I think my CPU might be limiting the system a bit. This is an FX 8320 overclocked to 4.3 GHz, it seems, at the moment.
Or 4.4. 4.5, I mean. Whether it's 4.3 or 4.5. I don't remember exactly. Uh, let's see. 4.5 GHz processor. So, just to confirm, R9 290 at stock frequencies, custom water cooling. The characteristics are also in the description. So, what can I summarize? So the conclusion is, uh , let me turn the camera around, use the front camera to get to the conclusions.
So, I'll be back soon. I have a thick beard. But in any case, the conclusion. As you can see, even with overclocking, i.e. over 1000 MHz, there was no throttling . At stock settings it was throttling, but it was a moot point, looking at just the stock settings, whether it was overheating. But when we increased the power limit, it still throbbed, but not as much.
So, the temperature rose by 4 degrees. Now, if it were throttling due to overheating at 74°C, it's likely that power throttling or overheating would cause it to drop frequencies even more, even if the power limit is increased . So the frequencies would be even lower. At least that's how I imagine it.
I could be wrong. Please correct me in the comments if I'm wrong. But that didn't happen. The frequency increased by about, was it 80 MHz? Which is a significant increase. Now that we've done the same thing with 100% fan speed, we've eliminated, sorry, completely eliminated the overheating. And we still saw the same throttling at stock settings.
Except for the fan speed, of course. It was still throttling at almost 100 MHz, let's say. A little less. So, even at +50% of the power limit, it didn't throttle. And with my overclock settings, I could see it reach 1040 MHz. So, what was the temperature? I agree, the fan curve is manually adjusted, I admit it.
But the temperature was still 75°C. So, given the results, I have no idea how there could be thermal throttling at 74°C if it can reach even more than 900MHz at a higher temperature, reaching 78°C and around 980MHz. But with standard power limits it was producing around 920. So, let's say 60 MHz more.
I apologize for the inaccuracies. This isn't the most scientific test, I admit, but in any case, there is a difference of 60 MHz. And the only difference was the power limits. I'm not saying...Well, actually, that's exactly what I'm saying. On the default Nano settings, depending on, um...I'm lost, sorry.
Depending on idle temperature or ambient temperature, it's around 20°C here right now. The Nano doesn't throttle. It has no thermal throttling at default settings. There is power throttling, and it occurs because it is a large chip, 4096 stream processors . They keep it at 175W TDP I believe. And to meet this power limit, they are forced to limit the power at the drain.
So that's why the Nano is so energy efficient, if you will. But this also means that increasing the power limit without additional changes gives a significant performance boost. So, this is actually proof that there is. At stock settings, at least according to my conclusion, and write in the comments if I'm wrong.
The R9 Nano has no thermal throttling at default settings. Thank you all for watching. See you next time. See you later.
Comments 0