RaazP

I need to try the new "Target Curve". Until then: I wanted to do exactly the same and after some time of tinkering with it, I've found a good solution for my special needs. Maybe something similar might work for you? Mix multiple graphs: I have one very very slow curve and one that simply ramps up to 100% when the CPU hits a high temperature. Ofc you need to set the fan controller's step %/s to the maximum to make it work (50). SlowCurve: Hysteresis 8°, Response time 15s, 30% up to 65°c, then linearly up to 100% at 90°c HighBoost: Hysteresis 4°, Response time 1s, temp to trigger: 82°c Yes, my cpu is running quite hot.. Planning to change to a better case. Although the 10600k simply spikes a lot. During rendering with Magix Vegas with nvenc, the CPU is between 65-80°c, during gaming mostly 60-75°c. During writing a forum post, the cpu will be between 35-50°c, clicking on a youtube video will give a 70°c spike for one second or less. Fans stay very very steady due to the "SlowCurve", while it's protected from overheating spikes like when loading a savegame or when starting a rendering process by the HighBoost. Yes, my fans are spinning too fast or a bit too slow most of the time but I don't really care. I want steady speeds and just my CPU within certain limits. Looks like this: You will need different values, but it should by easy to achieve the same after some experimenting.

Should be okay, airflow and heat wise. Of course a mesh front is a lot better but you got intakes from top to bottom and 3 fans so enough air should get sucked in. At 30 degrees ambient, I think all case fans should be at 100%. Then normally a gpu fan will match the noise level of the case fans at 60-70%. Depending on the gpu and case fan models of course... Against your legs though? The 3 at the front should suck air into the case, while the rear and the top Fan push the air out of the case. Not sure how your case is positioned that you can feel the hot air on your legs? But anyway, gpu temperature is okay so feel free to just keep things as they are! Have to say I worry a bit about your cpu now though... What Temps are you getting on it?

What case and where/which/how many case fans do you have? If the heat can't get out of the pc case, the gpu will only push hot air onto the cooler... Also what level of fan speed were you on at the end? Is the fan curve maxing out to 100% fan speed?

Ryzen 4xxx will be a big boost at the end of the year and Nvidia 3xxx will probably be released in August/September and also be a big boost in performance. So you might want to get a B550/x570 board with the ryzen 3600 to be able to easily swap it out for a ryzen 4600/4700 when they hit the market. And maybe hang in there with your current gpu and lowered settings. Apart from that: looks good!

What do you mean by "maxing out"? The bios doesn't allow the slider to be raised higher in msi afterburner but that's all. It's called Temperatur target. It's more like a limit, I agree. However there's no issue with high temperatures apart from "degrading" over long periods and if it won't shut down at 99°c, the silicons might leave the working point and the resistances might change and cause a shortage. But nothing will meld.. And I trust the manufacturers that they will know when the change of the silicon's resistance will become critical. However the latest cards are so efficient and easily cooled, that the cooling required to stay below 70°c won't cause noise issues. So I absolutely agree that the OP should create a fan curve and drop the temperatures. But it's not like he is risking his card or pc when he plays a few hours of rust at below 90°c. What I find a bit weird about this card is that it will apparently: - run above 80°c on default settings - might clock down instead of raising fan speeds - seems not to increase fan speeds... My msi card will go up to 100% fan speed before throttling when reaching 80° and won't reach more than 81°c but instead throttle down if 100% fan speed isn't enough (tested this by locking fan speed at minimum).

The modern Nvidia cards will simply clock down when reaching a temperature "target". You can manually raise it. If you don't touch it, it will be at around 80 degree. So as long as you aren't bothered by the reducing fps after some time (and then they should stay the same), it's totally fine. My gtx 1070 is running every few days at 84 degrees Celsius for about 2-3 hours (I have a pretty hot silent case....) since 2016. No issues, still at the same 2050 mhz overclock. Jay's 2 cents did a video on hit.. The bios has some heat protection built in. His 2080 shut off at 99° after disabling the fans and pointing a heat gun at it.. Lol. So don't worry! As long as it's below 90° it should last longer than you ever gonna use it! About fan curve: install msi afterburner from guru3D and copy this: In most games the card stays below 75°c, in some games it reaches 84°c. Easy to adjust. With the slightly plateau and the hysteresis it should stay kinda consistent and not ramp up and down the fans all the time.

The fps don't sound too bad then. But yeah, looks like your 1050 ti wasn't such a massive bottleneck then. What's your resolution btw? Resolution basically only causes load on the graphics card so if you're on 1920x1080, a 2070 is a beast and the 2600 will almost always be the bottleneck. I'm playing at 3440x1440 and my 1070 is crying for help, while my new 10600k is cruising around. All games set to run between 60-90 fps. About overclocking your ryzen cpu: Sadly I can't help.. Just watch some YouTube videos from the big and good channels. However what I would recommend first: Go into your windows powerplan and set the maximum cpu level to 80%. I guess you have min and max at 100% and are in the high performance power plan? Reduce min and max to 80% and compare your fps. Your cpu frequency should be a bit lower and if you are really cpu limited (and not something else just screwing up or being incorrectly set up), your fps should be about 20% lower. Maybe play around with that power plan setting. Try 70%, then back to 100%. If your fps match what you set, it's definitely a cpu limit. If your fps stay the same, there's some issue with your system not using your hardware correctly. This is a good way to test something in general by the way. Instead of overclocking and searching for an improvement, simply downclock one part and see if the performance goes down. No stability issues or heat problems this way

1. Gpu load looks good in kombustor so everything should be working fine. 2. Gpu frequency looks good with 1725. Although these cards can run 2000 when overclocking. But since you don't get full gpu load in your games, no oc is needed here! 3. The boosting has nothing to do with this single thread performance limit. On Intel, all cores will boost after another. But there's a small latency for boosting up, which is why overclocking all cores to the same frequency mostly gives better and smoother fps. But yes in theory, as I explained in my first post, you could boost 2 cores for the 2 game threads and clock down the "not so highly loaded" cores. Actually amd cpu's tend to do this. So on intel you'll see the frequencies for all cores going up and down, while Taskmanager will show pretty evenly spreaded load across all cores. On amd cpu's you might often see a few cores constantly being loaded and boosting, while other cores are constantly down clocked and not really used. Sadly there's no tool that will show you when you're hitting the single thread limit of your cpu. You only see averages. But as I mentioned, with 8 cores (cpu threads), you might be already at the limit at 12.5% overall cpu load. Depending on how much this is averaged, you might see only 5% load on all cores. Still cpu limited! Crazy, isn't it... 4. Fortnite: I don't play it, so no clue about the cmd stuff working. Also the affinity doesn't do anything. It just tells windows which cores not to use for that game. But from using only 4 cores instead of 8, your game won't magically use 4 cores instead of only 2. I never saw an improvement by telling windows how to do stuff. Only if there's a bug in the game regarding smt or hyperthreading, setting the affinity to core 1/3/5/7 only helped a lot. Or you go into the bios and disable smt or ht. 5. Higher Single thread performance helps though. The latest amd 3xxx beat the non overclocked Intels in 1 thread scenarios but sadly the AMDs clock down if you use more cores. While on Intel it's pretty easy to overclock all cores to constantly boost to the maximum. That's why you see Intel cpu's being better for most games. Only in games where you can really use all cores, amd will be better sometimes. But AMDs are using less power and are cheaper so you get more for your money. For maximum fps though, intel is the way to go, sadly. 6. What fps are you getting in your games? You only said the performance would be "not good". What exactly does this mean? 40 fps in fortnite? (that would be bad). 100 fps in battlefield 5? (that would be awesome). We need more info! 7. In general your ryzen 2600 should be enough for all games. Depending on what's enough for you of course! My main games are racing simulations. You want 90 fps or more in these due to the high speeds. Due to being "simulations", the cpu has to do a lot of physics. Due to being a niche hobby, the simulations aren't optimized so mostly running on 2-3 threads. Meaning a 4 core cpu is mostly fine for them and you just need to overclock your cpu to the maximum. That's why I know so much about this damn single thread performance...

Guys the Psu is clearly not his issue. He got your point. If he wants to know more about his Psu quality issue, he can search for information himself. So many replies and not one of them was helpful towards his issue... Rant over and actually replying: Psu can't really bottleneck a system. 650w is plenty enough too. What he means is that the corsair Psu you have is just low quality in general and if might fry your system at some point (although I'd say I'd it's not 10 years old that's very unlikely to happen). Also the maximum overclock might not be as stable as with a high end Psu. The voltages, ripple etc are not as good. Anyway, that got nothing to do with your performance issues! I wonder if your cpu is limiting. Yes, it's not used to 100%, not even a single core. But windows is shuffling the load in the cores around and you simply can't see the single thread performance limit. If you could measure cpu load at the cpu frequency. So 4 GHz for example, you would see each core being either at 100% or at 0%. CPUs can't run "only at half load". It's always an average. They either calculate something or they don't. So if you have 8 cores and the game you're running runs on 2 threads, only 2 cores can be used at the same time. So maximum cpu load would be 100% divided by 8 = 12.5% per core. X2 = 25% maximum overall cpu load. Windows is so good at scheduling though, that you'll probably see around 40% overall cpu load. That's likely due to the next scheduled core caching something in preparation for its turn. Can you download the latest msi kombustor version (4.1.6 from geeks3d) and just run the benchmark - preset 1080 without adjusting anything? I have a gtx 1070 and my result is 2339 points / 38 fps. Also this will show you the graphics cars load at the bottom. It should show 99%! This is the most important thing to get here actually. Doing a basically only gpu stressing benchmark and seeing if the gpu will be used to full extend (99% is full load!). About cpu limit: In general, if you don't have an fps limit active (vsync or a fps limiter), if the graphics card isn't at 90% or more load, it's your cpu limiting. Sadly all the cores of today's cpu's aren't used in most games. So apart from having headroom, 4 cores or 10 cores barely make a difference. Games like assassin's creed origins and odyssey however can use 8 cores pretty well! Hope that helps!

Thanks guys, I see now where I was wrong. Always good to learn!

To break this down, your thought is: Hysteresis 20, response time 1s, stepsize 1 Linear fan control 50° at 0 seconds 71° at 1s Fan speed goes up 1 step 2 seconds: point on curve at 55°, temperature still 71° Fan speeds are stuck since the difference is now only 16°, hysteresis not reached 75° at 3s Fans still won't respond 77° at 4s Fans will go up to the 57° point on there curve? So in theory you need to set the fans to the maximum for "target temperature minus hysteresis"? With a big stepsize (50%), the fans would jump to their "correct" point on the curve though and won't get "stuck"? Anyway to give you a quick solution: just do what I did, create a second fan curve that hard limits the temperature via hysteresis and response time at minimum so if that get reached the fan speed goes definitely up. While I'm at it: Stepsize per fan curve instead of per fan would be awesome! Would make my slow curve even smoother, while my upper limit can spin the fans up swiftly. But I'm happy as it is now anyway.

Just to report back: Yes, that's what I did and thanks for calling it smart hehe. After I read that you have a completely different perspective I thought I might just need to change mine. So I went back to what fan behavior I want exactly and what fancontrol gives me to work with. Thought if I can't achieve it via temperature curves, I might be able to mix another parameter. Works very well since then! Very stable and steady fan speeds with some bursts when loading save games from idle or rendering a video without editing right before. Still adjusting here and there but it works very well to mix response times!

Just to report back: Thanks for telling me that my approach was "weird". I thought about it for a while.. Thought I'd try the different approach then and it runs nicely as far as I can tell from only 2 days. One graph (gonna transfer this into just a linear one): flat at 35% until 50°c, going up in a straight line to 100% at 83°c. 10°c Hysteresis, 10s response time Second graph: flat at 0% until 85°c, then 100% at 86°c. 4°c Hyst, 1s response time Mixed into a Max mix. I know, still pretty weird for you probably but I found out that my cooler is big enough to passively cool my CPU long enough to sub 90°c that the second graph is quick enough if I get 100% load on all cores from complete idle. Which is more a fail safe.. Apart from doing a benchmark out of idle, there's no program I'm using that wouldn't cause some load before maxing out. Once the slow curve responses, the temperature will always be okay. Overall the fan speed history (openhardwaremonitor for 24h logging) looks very nice and I didn't get annoyed get by ramping up/down fans.

Hi, oh wow, thanks for the detailed reply! Guess I'll try to make a linear fan curve and see if it annoys me or works well. It could all be solved by putting the response time higher to ignore peaks. But not ignoring peaks above 80°c would also be needed. Maybe mix a linear graph with a 5s response time and high hysteresis with a graph that triggers above 80° (0 hysteresis and 1s response time)? Anyway, long post again, sorry.. Hope you find it interesting to hear my "weird thoughts" so they may become a bit less weird for your perspective. Btw I'm halfway through my engineering bachelor so while ofc still beeing a noob, I have a basic understanding of the things you mentioned. Or if you think I have not, then feel free to tell me straight up so I'll learn. To 1) After your post I understand where you're coming from. Must look weird from your perspective. What I want to achieve is basically a constant fan noise for my 3 different use cases without short load peaks (if below 80°c) increasing the fans at all: 1. Desktop/Office stuff: CPU is most of the time at about 45°c, barely any load. But when I load an excel sheet, the temperature will peak up to 70° for 1-5 seconds, not really higher so I'd like my fans to stick at the 35%. Only if something odd is happening and increasing the CPU load a lot, I'd like the fans to go up, of course. 2. Generic gaming: most of my games, mostly racing simulations won't max out the cpu. At about 70% fan level, the cpu will vary between 60-80°c. So I'd like my fans to stick to the constant level of 70%. Not dropping down when I'm in the pitbox and the cpu cools down to 62°c but also not ramp up for a short moment during fights with multiple cars (probably around 75°c). 3. Really stressing the cpu: when I'm rendering a video or playing a bit of Assassin's Creed, running around in a big city, the CPU might hit more than 80°c. Still varying between 65-84°c but anyway, constantly hitting high temperatures. So I'd like my fans to stick at 100% all the time I'm in that big city or rendering the video. And only lowering the fan speeds when the cpu will stay at a lower temperature. Probably still kinda weird how I'm approaching this but in SpeedFan with the old 2600k it worked exactly like this. Might've been weird too but it worked and I had a very constant noise level depending on what I did. To 2) Probably enough said about this in 1): I don't want burst cooling, quite the opposite. I want my fans so stick at a certain level, if an application will frequently heat up the CPU above a certain temperature. Like "okay, if that game heats up the CPU to this level: Stick to a higher level until the situation really changes (cpu temperature going down a LOT).". Just a note: the 94° won't happen with fan speeds being in "level 2" (85%). Thks high peak only happened because the fan level dropped before (for whatever reason..). I know you can't burst cool like that. The cpu will heat up way too quickly. To 3) I really wondered about this happening too.. Probably down to afterburner/rtss being at 500ms refresh cycle, showing the peak. I tried to replicate and test my fan curves with the kombustor cpu burner and cinebench at different amounts of rendering threads. I see this little dot on the curves (very nice btw!) and depending on how I set the plateaus, the dot will stick at different points. Example hysteresis of 20°, below 70°c = 35%, then 100%. The dot might stay be at the 50°c point. Then the temperature goes up to 70°c and the fan level goes up. Then the temperature goes up to 89°c and drops down to 69°c. The fan level will go down to 35% and the "dot" will be at 69°c. Now the temperatures need to go up to 89°c instead of 70°c for the fan level to change. Anyway, I apologise for so much text! It seems that with the current functionality of FanControl, I should simply use a linear fan level graph and keep it from fluctuating too much via response time and hysteresis. My approach would be kinda simple though I guess. I'm only able to code very basic stuff but for my needs the fan code would probably only need some if/else combinations for two cases: upwards or downwards cpu temperature change: (rough coding. I know there's a lot more needed... I'm a noob at coding but it might give you an idea?) About my cooling: - Cooler Master silencio 550 - Front: be quiet silent wings 3, 140mm - Rear: silent wings 120mm - CPU: Thermalright Le Grand Macho RT The case is really bad... And the 10600k is a lot hotter than the 2600k, although that ran at 4.4 GHz all-core.

@Rem0o Thanks a lot for your program! Going from Win 7 on an old CPU to Win 10 with a z490 board and losing SpeedFan compatibility let me search for an alternative. Got my PC silent and cooled, thanks to you. However I would like to request a feature that would make my life a lot easier but I have no idea how much effort it would take to code. All I'd need would be 3 steps that trigger/go up instantly when the CPU hits a certain temperature but not go lower until reaching another certain temperature, not being time related. For example: 35% up until 75°c / 85% up until 83°c / 100% when above 100% until below 75°c / 85% until below 60°c / 35% when below I tried to achieve this with response times and hysteresis, even mixing graphs with different hysteresis but I either end up with not consistent fan speeds or my CPU having temperature spikes if a bad combinations of temperature changes happens. Or maybe I'm just stupid... So I'll now make a longer write up but it's not about the feature, but about my exact case so feel free to ignore it. So I'm gonna post my 2 current graph curves: Graphs are mixed = Max 1. Graph: Hysteresis = 15°c, Response time = 1 2 points: 75°c = 35%, 76°c = 85% 2. Graph: Hysteresis = 30°c, Response time = 1 2 points: 83°c = 0%, 84°c = 100% Last night I had a temperature spike from 72° to 94°, while the fans jumped from 35% to 100% but it wasn't quick enough. The problem was that the temperature went like this: 72° (35%) -> 83° (85%) -> 71° (35%) -> 94° (100%) What I don't get is with hysteresis at 15°, why got the 1. Graph triggered from 72° to 83° and why dropped it down to 35% when going down to 71°? Anyway, I appreciate any help I can get. If it's just how it is I guess I'll have to find a way to change the config for every cpu intensive game or program at launch and back when I quit it. Thanks in advance, Rasmus PS: 10600k with a pretty hot & dampened case. Old 2600k was fine but this new i5 goes up and down in temperature a lot faster... Not willing to buy a better case just now though