How do I find out how hot my car engine is?
How do I find out how hot my car engine is?
Hi everyone, I am using MSI Afterburner to try and get more FPS for CS GO on my GeForce 635M. (I know I shouldn't overclock laptop GPUs, but someone has done it before.) The issue is that when I raise the Core Clock between 70-85 degrees C (85 gets hot enough during games), the core clock drops down to 662 Mhz (which is the base speed of the GPU). But the memory clock stays the same. So, I think this means I am getting thermal throttling that slows things down. Question: How do I find out exactly when it starts getting too hot? The in-game monitoring from MSI Afterburner isn't super precise, so it's hard to tell because there is a small delay between when the temperature goes up and when the clock speed drops. Thanks so much in advance.
There is always a delay. It cannot be avoided. Intel checks how hot their CPU gets every 256 milliseconds, which equals four times per second. If you tried to read those exact temperatures right away, it would look like just the number '88', nothing but a blur. So there is a natural pause of a few seconds between when the temp is checked and when it shows up on your screen. Just so you can see that moment clearly. Only if you graph it will you get everything shown, but this gets confusing with multiple cores because the reported temperature is usually just for the hottest one, which changes depending on what you are doing. Because of how the silicon is made and all the tiny impurities inside, you will inevitably end up with some cores that are stronger and others that are weaker, along with a few that have different traits. So while one core might slow down at one temperature...
There is always a delay. It cannot be avoided. Intel checks the temperature every 256 milliseconds, which equals four times per second. If you try to read those exact numbers right away, you would only see an '88' on the screen, nothing else but a blur. This means there is naturally a small pause of several seconds between a polled temperature and the next one that gets reported. Just to make it easy for you to watch it happen. Only if you graph the data will you see everything, and even then it can get confusing because with multiple cores, the reported temp is usually just the hottest core, which often changes depending on what you are doing. Because of how the silicon is made and all those impurities inside it, your cores will naturally be some stronger than others or have slightly different features. So while one core might slow down at a certain temperature, another core might throttle at a slightly higher or lower temperature. I do not think there is an exact measurement for exactly what total temp the whole cpu will start to throttle on. Intel says they guarantee it throttles around Tjunction, but that's just my guess and could be wrong; however, you can be sure it will throttle near that mark. How nvidia calculates its own throttle temp is a complete mystery to me, but I think they probably do something similar to intel since the chips come from the same maker. But keeping your gpu temps below 80 degrees is much safer and better for you. It makes getting high fps in csgo very easy, even when it exceeds refresh rates. There is really no real advantage to trying that on a gpu because you would only see maybe 5 to 10 fps gains at most depending on which version of the 635M chip you have, and there are four or five different versions just like that. That's not much of a difference and it won't affect your gaming at all.
Grab Gpu-Z from this link: https://www.techpowerup.com/gpuz/ Then hit the "Advanced" button. Pick the drop-down menu and select "Nvidia Bios". This will tell you your right now, lowest, and highest temperature settings.
Thank you for your reply. It's really tough to keep the laptop cool while playing CS GO. My cooling pad might help a little bit or maybe two degrees of heat. A vacuum cleaner could also help, but usually, nothing works better than replacing the thermal paste on the back of the laptop so it cools down faster. That extra 5-10 frames per second will make a huge difference when I play in 1080p because my game lags way too much when the frame rate drops below thirty. Right now, I'm playing at 720p and it's fine for me, but the picture quality isn't great anyway.
Is this okay? The current setting is 90 degrees Celsius, but the minimum can drop down to 0 C. My default setting is set to 90 C, and I thought there was a maximum limit of 105 C somewhere on the specs sheet. If my actual max temp is higher than that, then something is definitely wrong with my cooling setup.
Another question: GPU-Z says both the GPU clock and default clock are running at 475 MHz. The memory speed is set to 1015 MHz because I did an overclock, which makes sense. But the boost clock is hitting 855 MHz while the default is still 800 MHz. That also looks good because I was tweaking my OC settings. Why isn't the core clock showing up at the GPU clock or default level? The specs list the default as 660 MHz, but it's currently showing 475 MHz. I thought both clocks should be around 715 MHz given those numbers. Thanks again!
Different things have different temperature limits. Some numbers change depending on which one you look at. Tjunction is the point where transistors start losing their steady working, so that's when you see throttling happen (around 5°C above or below TjMax). TjMax is where you risk permanent damage and throttling stops working, causing a forced shutdown after a set time. If you push a CPU past TjMax, it goes into thermal runaway, meaning the shutdown isn't fast enough to stop the actual burning inside the chip. So if you see a number like 105°C, that's almost certainly TjMax. That puts Tjunction at about 100°C plus or minus Tcase. But that rule only applies to test samples because production models don't have sensors there, so you can't measure it accurately. The heat from the core has to travel out through any timers used and end up hitting the IHS (Intel Heat Spreader). That temp is basically useless for normal use. My advice is to stay as far away from Tjunction or TjMax as possible. For a much better guide on how To check your CPU temperature, see this Intel update from February 17th, 2026. This guide helps explain why things can be confusing with processor temperatures due to different opinions about terms and specs, leaving users unsure how to check their cooling properly. This guide provides more info... forums.
The real speed you get at 475 MHz is the slowest possible rate in a bad case. The OC (Overclock) setting only matters if you are trying to push things faster than normal. Using it at low speeds just uses more power and drains your battery faster. If you have a huge fan, the temperature can actually reach 105 degrees Celsius when you get direct from Nvidia's software. In that case, the laptop maker went into the chip's settings and set the max safe temp to only 90 degrees instead of letting it hit the higher limit.
I took a look at that thread and got some good info there. I can't really find temperature specs for the 635M. It looks like I must have mixed it up with my CPU, which is actually running at around 105 degrees Celsius. So if gpu-z is correct, which I think it probably is, then I am pushing that laptop GPU to its limit pretty much. It doesn't seem to go higher than 85 or 86 degrees. It isn't hot all the time though; usually during play, the temperature stays between 80 and 85 degrees. Thanks for the help!
Got you on the speed stuff, thanks! I guess I messed up on what the temps were for. 105C is actually my CPU, not my GPU. I'm pushing it pretty hard right now near that 90C limit. I'm waiting on the Arctic MX-4 to arrive so I can reapply some thermal paste there too. Ever since I bought this laptop way back in 2012, I haven't changed that old stuff yet, so hopefully that'll help! Thanks again! P.S. I'm also building a new GPU for my laptop using a GTX 950. Just waiting on the exp GDC beast to show up there too. That might even boost my frame rates anyway. For now though, let's just try to overheat my old dGPU so I can get more performance out of it. It seems like around 80C is when it starts slowing down and going back to normal speeds. So unless I figure something else out, overclocking won't really help much.