MSI Afterburner Causing The Infamous "AMDip"?

[@Mostly_Positive_Reviews]

***Important*** This is NOT exclusive to MSI Afterburner. It happens with CapFrameX and HWInfo too, as per the CFX dev. CapFrameX informed me that it's an issue with the NVAPI. The GPU in the video is a 4090. Also, my 12700K and 14600KF do not exhibit the same issue as my 7800X3D with MSI Afterburner, and I believe it is due to the e-cores. Due to time constraints I havent been able to test my Intel CPUs with e-cores disabled, but I will do so and update via a community post and via Twitter.

[@omegajvn1]

Nice catch! I heard that’s all of the X3D chips were experiencing this kind of problem. I have a 5800 X3D myself in one of my systems. But it has an AMD 6800 XT GPU. So I use the built-in driver monitoring software. I’ve never experienced anything like this before. I do have another workstation system that has an RTX A6000 and a 3700X along with MSI after burner. Still haven’t experiencing anything like that either on that system. But of course it does not have 3D Vcache

[@Thunderstyle7]

This also affects AMD GPUs. I stopped using MSI Afterburner years ago exactly because I figured out it was the problem. Rivatuner Statistics Server on the other hand works perfectly without any issues. It is just afterburner.

[@aboutthat413]

​@@Thunderstyle7 It's possible it has in the past, but I have a 7700XT as well and it doesnt happen there, only on my 4070 Ti.

[@Mostly_Positive_Reviews]

@@aboutthat413 Yeah, I had a few people test Radeon 7000 GPUs and none could replicate the issue, but they could on 40-series GPUs.

[@robertmyers6488]

​@Mostly_Positive_Reviews That makes sense as I think the NVIDIA cards offloads much more scheduling to the CPU. When testing the 4080 against the 6900XT, 7900XT, and XTX on older hardware (i.e. prior to 5000 series Ryzen,PCIE 4) the 4080 just gets smoked. The AMD cards are more self sufficient.

[@unorthodox5171]

It's quantum mechanics, it acts differently when being observed.

[@Mostly_Positive_Reviews]

Hahaha! Schrodinger has entered the chat

[@lordcommander3224]

Being an IT guy I feel this comment 😂

[@silentx9709]

Yes, i found this out yesterday. I went from a 5600x with no stuttering to a 5700x3d with stuttering and come to find out it was the msi afterburner cause the micro stutters. Now it's running like butter in every game. Thanks!

[@THEMISOGYNISER]

i feel like i had stutters too with my 5600 , i dont think its an x3d problem only

[@mondodimotori]

I was experiencing sudden stuttering during ANYTHING I DID ON MY PC. It still is my main rig, with a Ryzen 3600 and an RX5700, and I lived with this issue for months! Until I wrote a post on the Linus Tech Tip Forum and, suddenly, here's the solution: Disable the mobo firmware TPM. I had enabled it for more safety, but didn't connected that the stutter was caused by that. It was finally fixed by a BIOS update of the mobo. These kind of issues are a real pain in the butt to troubleshoot, and it's incredible that they are still happening with other softwares suites.

[@Mostly_Positive_Reviews]

I was just telling a friend of mine that between the Nvidia App issues, Windows issues, BIOS issues, and now this, it's become impossible to know what is the cause of any stuttering in games. It can be so frustrating at times :( Glad you got your issue sorted though! I also wouldnt first think of TPM.

[@marsovac]

Isn't this fixed? On my 5600 a BIOS update fixed the TPM USB stutter.

[@cant_Comment]

i have an intel i5 6500 on a dell office computer. how do i disable this TPM thing? this dell bios is weird

[@DragonOfTheMortalKombat]

​@@cant_Comment its laptop bios, those are generally locked. You should try to once to go to the UEFi settings. Who knows you might find an option.

[@ondrejsobek6565]

Great tip about the profiler and thanks to Unwinder for explaining it. I knew about the power usage problem since the video you linked, but now I checked using the profiler and discovered that dedicated GPU memory counter was eating around 3-4 ms on my system (9800X3D + 4090) and sometimes spiking as high as 10 ms. That's totally not worth it, so I disabled that and now all the remaining sensors are fairly balanced with no single big performance hog.

[@unwinder]

No need to optimize it just to optimize. That counter's timing is pretty normal. Also, don't forget that CPU polling time depends on CPU clock. So it can be easily let's say 10ms for CPU idling at 1GHz and 2ms for the same CPU running on full 5Hz. So fluctuations in some sensor's timings due to dynamic CPU clock changes (and multitasking) are pretty normal and expected. Personally I always pay attention to total polling time and don't bother to disable anything if total polling time is below 50ms in idle.

[@ddd45125]

Thanks brother, I have a very similar setup, Ryzen 9800x3d and Gigabyte gaming 4090. This certainly fixed the occasional micro stutter I had in certain games as well. Dankie!!!

[@Mostly_Positive_Reviews]

Glad it helped! Dis n plesier 🥳

[@GlennsHardWired]

This has been well known for years. GFE used to do this until nvidia patched it so it wouldnt monitor power states of gpu. Now the nvidia app has the same issue when some of the ai settings are turned on

[@TheRealDlo]

Gonna go test this right now! Thanks for the video

[@Mostly_Positive_Reviews]

Let me know how it goes!

[@KiltedCritic]

So if you're not actively running afterburner, this shouldn't have any impact? I'm asking as I usually just apply settings to up the power limit and OC a bit, then I close it.

[@Mostly_Positive_Reviews]

If it is closed entirely it wont cause it, correct.

[@stealthhunter6998]

I always have stuttering issues in every UE5 game (which Ik is a UE5 issue) and some other games like space marine 2 and dead space remake. So over stuttering troubleshooting. I use RTSS to cap all my games for a flat frametime to combat the issues. Wonder if this will help (I only use RTSS without MSI open thou so maybe it won’t??) Should I seperate game and is ssds is that an issue? Should u use ssd connected directly to cpu PCIe lanes only for gaming? Any tips to avoid stutter in general pls hand them over haha. (Assume I’ve updated everything and every driver).

[@TerraWare]

Great video. Saw Dannys video but didn't get a chance to play around with it much. This makes it easier to look into. Interesting to see how this effects AMD GPU's.

[@Mostly_Positive_Reviews]

Thanks man. Yeah, would be interesting to see, and I actually think Radeon GPUs wont have the same issue as CapFrameX just replied to my tweet and said it's an issue with the NVAPI. Would be awesome to test though.

[@TerraWare]

@@Mostly_Positive_Reviews No this explains why my XTX gets smoother frametimes. I just tested INdiana Jones with power percent disabled which seems to make the most difference. And perfectly smooth. 125 FPS AVG 116FPS 1%low

[@Mostly_Positive_Reviews]

@@TerraWare You know, between the Nvidia App, Windows 24H2 updates, and now MSI Afterburner it's becoming quite a chore to identify stuttering issues and who is responsible for it 😥

[@TerraWare]

@@Mostly_Positive_Reviews Very true lol. Im hoping Steam OS can save by debloating all the nonsense with their built in performance metrics. I love my Steam Deck. It's such a nice well packaged game machine.

[@Mostly_Positive_Reviews]

@@TerraWare Yeah, very much looking forward to an official SteamOS desktop version. Would be fun to make comparisons against Windows, and I am pretty sure it's going to be a bloodbath when it comes to performance consistency.

[@pullmeifinger]

Thank you! I had this issue with my 9800x3d. After days of messing i narrowed it to msi afterburner and just stopped using it. I never knew that little trick to see cpu usage time.

[@Mostly_Positive_Reviews]

You're welcome. I also didnt know about the profiler until very recently. There are more options that you can enable to view even more specific info regarding possible performance hits. Will do a follow up video showing that soon.

[@mikehawk6918]

No one should be using msi afterburner anyway. They literally stole RivaTuner software, which is the basis of their program, from the original creator and refuse to pay him. So the actual dev isn't developing the software any more. No idea who's doing it now.

[@unwinder]

@@mikehawk6918 where do you read all that?!? This statement is also absolutely disconnected from reality. Nobody stole anything. Nobody refused to pay licensing fees, MSI couldn't do it due to sanctions, there are still some problems with that but the project is alive. Nobody else besides me can develop that project because source code doesn't belong to MSI.

[@adi6293]

That's mental, I have a 7800X3D and never had any stutter issues but I also have a 7900XTX and I never monitored that power, recently I stopped using MSI all together because the AMD overlay is good enough

[@Mostly_Positive_Reviews]

I was informed after the video was published that Radeon GPUs dont exhibit the same behaviour. I left a pinned comment but seeing that you have an AMD GPU it makes sense why you dont see this issue.

[@Profetorum]

Would be interesting to check if this kind of performance hit is noticed with any sort of background app (with or without polling)

[@Mostly_Positive_Reviews]

I'll work on a video regarding this. Any specific background apps you want to be shown / tested?

[@rasporilac]

You got a new subscriber 😊Any idea why hwinfo64 and hwmonitor can't display cpu package power correctly on my ryzen 5 5500, it's like 4-5w under full load while if I look at all 6 cores where are like 10-11w per core but package is not megring them. I had ryzen 3 1200 before ryzen 5 5500 and no issues with that.

[@Mostly_Positive_Reviews]

Thank you very much, really appreciate it! Ryzen CPU power reporting is a strange one, some apps report correctly and others dont. No idea why 🤷‍♂️

[@olivr_wtf]

Happens with 5950X as well - CRAZY!

[@Mostly_Positive_Reviews]

That sucks 😔

[@lordcommander3224]

Yeah, I think this is an AMD issue across the board.

[@desmond3245]

Thanks for the info! For me it's GPU1 usage (iGPU of the 7600) constantly hitting over 10ms.

[@Mostly_Positive_Reviews]

Welcome! I also noticed GPU usage would cause it to spike every so often.

[@bobloblawlove]

Does this eat performance even if you aren't monitoring stuff? I mean it shouldn't, but just in case I turned that stuff off. I assume HWInfo and stuff share the same issues if they're accessing the same sensors?

[@Mostly_Positive_Reviews]

Yeah, HWInfo has the same issue as it works the same way in reading sensors. If you just untick it from monitoring as shown in the video it resolves it so you can still have the software monitor other sensors without causing a big impact. That said, there will still be a performance impact monitoring other sensors, but it might be too small to really notice.

[@XDDX_NZ]

I think I had this same issue a while back on an Nvidia and Intel system. Stumbled upon the same fix and since then i've been careful with the performance overlays/metrics I use. Thankfully I use an AMD card now so I just try to use the AMD overlay instead of afterburner since it is so fully featured, and the driver overlay is likely to have less issues. Unfortunately the Nvidia overlay still doesn't show some metrics and doesn't have a lot of options so you kind of need to use afterburner with Nvidia cards.

[@Mostly_Positive_Reviews]

Yeah, this is actually a lesson for me too, to be careful about what metrics you enable. Radeon GPUs arent affected by this issue as it's caused by the way the sensors poll power percent through the NVAPI. Unfortunately I only found that out after I already uploaded the video, but I pinned a comment stating the same.

[@overseasinterp]

i dont have msi afterburner on my pc, but i do have msi center. could it cause any stutters? i cant see any options that i can change on there.

[@Mostly_Positive_Reviews]

You use it for RGB control etc? Basically every actively running app has some performance cost, some big, some small. If it doesnt negatively impact your gaming experience I wouldnt worry too much about it.

[@overseasinterp]

@@Mostly_Positive_Reviews mostly just for monitoring yeah. if there isnt anything substantial i guess ill keep it on. thanks for ur help

[@ultraplus_]

Thank you for this video! You have no idea how much time you saved me from redoing benchmarks. However, another issue has come up. Are there currently any alternative ways to show GPU power usage during benchmarks if I turn this off?

[@Mostly_Positive_Reviews]

Glad I could help! I'm trying to find an alternative as well. At the moment I think physical sensors are the best way, combined with PCIE power draw, but will test some software to see what works. For me the biggest culprit was power percent. Using normal power in watts didnt have this issue, but it can vary per system.

[@ultraplus_]

@@Mostly_Positive_Reviews Noted on this. Thank you once again!

[@ismailornekol]

hey thanks for the video. Is it only effect x3d series or all amd cpu? like 5700x 5800x? and what about the nvidia app overlay? Its also showing us gpu power when we press alt+R

[@Mostly_Positive_Reviews]

I had a few people say they only started experiencing it when they moved to X3D CPUs, but others have said their Zen3 non-3D CPUs are also impacted. I dont have a non-3D CPU to test with though but it seems to be affected in some cases.

[@ismailornekol]

​@@Mostly_Positive_Reviews i testi it with 5700x watch dogs legion in game benchmark. didnt see a huge different. %1 lows are better without afterburner power but not much. Maybe effecting more cpu bound scenario i dont know.

[@Izanagi-Arsene]

another thing to note, if you're using a wireless keyboard or mouse and game with a controller, some games will experience a micro stutter as they go to standby mode or become active again. it is incredibly annoying that these issues crop up and there's several and i mean several things that could cause them. the most annoying part is when you can't identify the fault and cannot successfully replicate the problem when troubleshooting.

[@Mostly_Positive_Reviews]

Good tip!

[@GlennsHardWired]

You can fox this by changing your usb power state in power option profiles.

[@qtipp7306]

@@GlennsHardWired change it to what?????

[@rentojad]

Damn, i had a lot of things using around 15ms. GPU Temp, Hotspot temp, Core Clock, Memory Clock, Fan rpm. Very useful video, thanks.

[@unwinder]

Which GPU is that?

[@rentojad]

@@unwinder AMD RX 6600

[@unwinder]

@@rentojad then you're mistaking, it is not a case of a lot of things using around 15ms (and that's why you should display both profiler and status bar to see total polling time), it is cumulative value for all sensors read in single AMD ADL call, I've commented in under DannyzReviews video: "- Also, if you try to peek inside performance profiler on AMD GPUs and compare it to NV, don't get confused after seeing non-zero polling time just for one GPU sensor and close to zero times for the rest GPU related sensors. That's a result of batching access for all sensors at once into single AMD driver API call. So you see cumulative value for the first GPU sensor and the rest ones are just read from cache (so polling time is effectively zero). NVIDIA GPUs use different approach and poll sensors independently, but some of them are also grouped/batched (e.g. NVIDIA GPU core clock and memory clock sensors are read into single call, so it is also normal to see non-zero time for the first clock but close to zero (cached) time for the second one). Intel Arc GPUs use implementation similar to AMD's one, so all sensors are also batched and read into single call. So it is also normal to see a pattern when just one GPU sensor is non-zero and the rest are close to zero on such platform."

[@rentojad]

@@unwinder Ok, i kinda get it. Now, when i switch on any of GPU sensors (outside of GPU usage and fan speed (not rpm)) my polling time goes from 1,5 ms to 15-20 ms

[@rain8478]

@@rentojad Exactly the same behavior. Any GPU sensor puts it to 15ms.

[@Machistmo]

This has been a 'thing' for like a year? It took people a year to figure this out? Just asking

[@joniqst]

i just uninstalled all these monitoring softwares including the new nvidia app and now everything is smooth again

[@Phys1csMan]

Does this affect older versions of MSI Afterburner as well? I am running and old build from late 2022 and I have not noticed this issue with power monitoring turned on. System is a R7 7800x3D + RTX 4090

[@Mostly_Positive_Reviews]

I'm getting conflicting reports on that front, sorry. Some people are saying older versions dont exhibit this behaviour, but others are saying it's to do with how these sensors read the data within the Nvidia driver, so the version of the monitoring tool in that case shouldnt make a difference. It also doesnt happen in every game for me, but a few I have tested that do include Stalker 2, Cyberpunk 2077 and Indiana Jones, although to a much smaller degree.

[@ActionBrother]

I only see people say that they have the same issue with X3D CPUs, but is it only an issue for X3D CPUs? I mean, is it only possible to experience this issue with X3D CPUs or can this also happen on CPUs without X3D from AMD or even maybe Intel CPUs?

[@Mostly_Positive_Reviews]

I wasnt able to replicate it on my 2 Intel systems, and neither could DannyzReviews. There are people that said it impacted even their non-3D Zen3 CPUs, but others have said it was only on their X3D CPUs. Mixed bag really.

[@ActionBrother]

@Mostly_Positive_Reviews Oh wow, thank you very much for the fast response! As soon as I can I'll have a look on my system to check if I see something similar. Thanks again for the well delivered video!

[@michaelfalabella6296]

yes i always have turned off the ones i don't use otherwise it causes a dip. i've been saying this to framechasers and media alike. it's always a program or some sort of setting. however i still can't explain why this doesn't happen on intel. wattage could be a factor as well since amd is trying to be efficient. if it's fixable with turning off MSI Afterburner settings for example, then that's good enough for me as i find it a very rare thing to occur anyways.

[@Mostly_Positive_Reviews]

Yeah, being selective in your probes / sensors is the best way to go about it. They all have some impact, but some a lot less than others, so it's importsnt to find that sweet spot. With regards to Intel vs AMD, I have no clue really. I've seen some people say they have this issue on previous gen Intel CPUs, specifically 10th gen laptop CPUs, but I havent been able to reproduce it on either my 12th and 14th gen systems. I still need to test with e-cores off though.

[@TheOne214]

Does this affect Intel cpus as well i have an i7 14700k ?

[@Mostly_Positive_Reviews]

I couldnt replicate it on my Intel systems, and I've had many people say the same. I talk about it in the the video as well.

[@charleyweinhardt]

Makes me question all power monitoring softwares...

[@Mostly_Positive_Reviews]

I pinned a comment regarding that. This is not unique to MSI Afterburner and I'll do a follow-up video regarding everything that came to light since I posted this video.

[@mytech6779]

It would be nice if some tools tracked outlier frame times rather than the pseudo-inverse 0.1%fps. FPS in the context of games is a rolling average and a single frames can't rationally be said to have an FPS. (Assuming a 1 second rolling average for easy math.) A 0.1% low of 40fps could be either 40 frames of 1/40th second each, or 36 frames of 1/80th second plus 4 frames of 1/8 second, and those two dips are very different events. A graph of frame times would clearly show them as different too.

[@Mostly_Positive_Reviews]

Yeah, it becomes very technical very quickly and a frametime plot would be the best way to show how long individual frames take to be rendered and displayed. I guess for simplicity the rolling averages are okay, but a more accurate plot of each individual frame would be very useful, especially when troubleshooting issues.

[@simonwest6002]

Well I did this for the highest usage one as you suggest and then the next option just jumps up to the top for usage and also the same amount of time it takes. So this is not a solution. You have to disable all of them, for me at least

[@Mostly_Positive_Reviews]

Interesting. In a lot of the feedback I received deselecting power percent fixed it for everyone, but polling can affect different systems differently.

[@Xamiakass]

i tried using msi afterburner throughout the years on diff hardware... and decided that it is worse than just using the bios settings and native (to hardware) software.

[@Ayannnnnnn]

İ have 5800x3d and 4070. This effect me?

[@Mostly_Positive_Reviews]

Highly likely that it would yeah. It seems to be Nvidia specific, but for me the issue is only on my AMD system. Worth it to check it out if you are also impacted.

[@DualPerformance]

I just disabled power usage of gpu and cpu, I have the 5700X3D

[@west5385]

Christ almighty, a 4090. Whats that like?

[@Mostly_Positive_Reviews]

Eh, it's okay I guess... Okay okay, it's fantastic! And the best part is I basically got it for free from my boss, so having the fastest GPU without spending a dime is something I am extremely grateful for!

[@elinbarrage9638]

hello man ,i hope u are doing fine. ive' had this exact same issue eversince months ago after updating rivatuner to the latest version and reverting the version fixed it!but for me it didnt take long to find out that msi afterburner was the culprit

[@Mostly_Positive_Reviews]

Hey man, doing well and you? Merry Christmas (a few days late, but it's the thought that counts, right?) It's often so frustrating troubleshooting these issues as it can be anything that cause it lately. Driver issues, Windows Issues, BIOS issues... It can be quite tedious to resolve these problems!

[@elinbarrage9638]

@@Mostly_Positive_Reviews yea that counts:D merry christmas bro

[@KoseChris]

and that is what i love about pc gaming, the community, they always show me something i did not know 😊

[@Mostly_Positive_Reviews]

Always learning something new every day too!

[@aboutthat413]

Hi, good video! Just tested my 13700K in this same game and didnt see the same behavior. RTX 4070 Ti Super here. Interesting that it seems to only affect AMD CPUs.

[@Mostly_Positive_Reviews]

Yeah, very weird actually. Thanks for providing your feedback. Which version of the game do you have? Steam or GamePass?

[@aboutthat413]

@@Mostly_Positive_Reviews I just tested with the GP version.

[@BorealSine]

i7 13700k with E-cores disabled (5.6 ghz, 4.9 ghz cache, 7000 ram, lowered timings) + RTX 4090 - I've turned on Profiler to see my latencies and saw ~9-16ms latency caused by GPU power compared to just ~0.9ms being Core Clock, then goes GPU load - ~0.5ms and temp metrics being ~0.1-0.3ms. In cs 2 that gpu power draw metrics caused lower fps when I got killed, when disabled fps remained about the same as AVG and i didn't have that sudden fps drop, fps got more stable and frame-time graph got cleaner, its now almost a perfect dead line like it should be. I have ~51ns RAM latency in aida64 and ~112 / 109 gbps read / write / copy speeds and maybe that affects how its less relevant on intel system than on amd where ram latency is a bit higher ? But anyway it does affect BOTH systems if you look at Profiler + Frametime graph + 0.1% lows. Glad I found that, i've msi ufterburner for years but didn't know about these little things, imagine how many are suffering and thinking its game developers fault. In my case I was stuggling in cs 2 over a year playing with frame time spikes and fps drops when killed, it wasn't perfectly smooth experience at all espcially in death match and then i found out it was caused my E-cores, the only game which had big problems running with them, I've then did huge analysis in many other games and yes i do get much higher fps and much smoother frametimes without E - cores, CS 2's source engine would run on E-cores even if only 2 were left running, it would throw 100% of workload on them instead of P-cores. And trust me these ''little'' things are only a tip of an iceberg of what can cause stuttering and frame time drops in our systems, there are many things in windows 10/11 itself, that's why i was always running debloated windows (60 processes) and debloated nvidia drivers to have clean frametimes and lows being close to AVG, that's when you achieve that buttery smoot experience.

[@Mostly_Positive_Reviews]

I still need to test my 12th gen and 14th gen with e-cores disabled, but with e-cores enabled I see no difference with power % on or off. Appreciate you sharing your findings!

[@Kuliaao]

my 5800x3d definetly had this issue, thanks

[@Mostly_Positive_Reviews]

Glad I could help!

[@elinbarrage9638]

idk how much to thank you man!i thought there is something wrong with specifig version of afterburner/rivatuner on my pc!

[@Mostly_Positive_Reviews]

No need to thank me, just happy to spread the word and hopefully help some people in getting their systems running as it should!

[@BJovanovic1987]

7800X3D here, I don't even use the Afterburner myself, so I should be good to go? EDIT: I did fire up the Afterburner just now (in desktop) just to see which monitor/sensor could cause any issues and immediately found "GPU1 Temperature" hovering between 15 and 16ms while everything else was way lower, second highest value (Framerate) only showing 0.25, and then the rest of the stuff getting smaller and smaller. Kinda inconvenient, GPU temperature is one thing that I pretty much always keep an eye on! Luckily, I use Radeon Adrenalin software's own overlay to monitor these things instead so it's not that much of a biggie.

[@Mostly_Positive_Reviews]

The values for AMD GPUs are shown differently and those values seem about right, you neednt worry about those. You can check unwinder's comments on here, he explains it well (the MSI AB dev).

[@NVMDSTEvil]

Yeah, any monitoring software can cause this. A way to monitor output FPS using a dongle on the HDMI/DP video output might be interesting.

[@Mostly_Positive_Reviews]

Yeah, what DF does is use a capture card but they have their own software that marks each frame. Them when they play it back in their software it can tell you what your FPS and frametimes are. I know Nvidia has something similar to identify duplicate frames, and I think their software is kind of based on similar technologies. Could be completely wrong though. But in that case you are limited by your capture card's refresh rate.

[@ElTortla]

LOL. Where's Mr. Frame Chasers ?

[@Mostly_Positive_Reviews]

Rumor has it if you say his name 3 times he'll appear. This is now the second mention of him 🤣

[@2ndLastJedi]

@@Mostly_Positive_Reviews Jufus

[@Mostly_Positive_Reviews]

Oh boy, now you've done it...

[@FatetalityXI]

Frame chaser's content getting outdated now. RIPBOZO

[@Mostly_Positive_Reviews]

Sorry, I'm not sure I fully understand your comment. Are you saying RIPBOZO to me, or to FC?

[@gaav87]

@@Mostly_Positive_Reviews hair chaser

[@FatetalityXI]

@@Mostly_Positive_Reviews to FC cuz his whole content is based on AMDip.

[@ZomgZomg007]

@@FatetalityXI who is this FC guy? never heard of him before. EDIT: I regret looking him up..

[@fefsefefesfsefsef2733]

@@FatetalityXI You know It's still dipping af 0.1% low are still low af it just don't stutter XD

[@Raptor_i81]

I just checked my msi afterburner and it was reporting 12-13 also on power percent, my CPU is 13700K, thanks for the heads-up.

[@Mostly_Positive_Reviews]

You're welcome? Did disabling it improve your lows? I saw the same behaviour on my Intel systems but disabling it had no effect on the lows.

[@Raptor_i81]

@Mostly_Positive_Reviews I was on desktop watching your video, no game was running, will check Stalker2 later today on and off. Thanks again.

[@unwinder]

It is expected timing for this sensor, you don't need to disable it if it gives you no problem with stuttering feel. Every single sensor eats some CPU time to be polled. Some more, some less, but absolutely none is free. There is always some performance hit for any background hardware monitoring application. That's axiom, that's how it always worked in every single app and that's how it will work in future. The same apply to any kind of RGB control software, which is also periodically accessing hardware and eat performance on hardware access events. The same apply to any kind of software fan control applications, like FanControl, software fan curve in MSI AB, EVGA Precision etc etc. It also periodically accesses hardware to send new fan speed to fan controller IC and can cause pretty similar performance hit. Majority of users simply don't understand how PC and such things works. And the only question for such kinds of performance hit is "how much is too much?".

[@aboutthat413]

​@@unwinder agreed. If it has no impact there is no reason to disable. What also happens a lot in my experience is people have monitoring tools running, RGB software, mouse software, etc, and they all add up, even if each one by itself has little impact. But no actively running tool is completely free, they all have an impact. Some larger than others.

[@Nihilius87]

Cool, thank you!

[@Mostly_Positive_Reviews]

No problem!

[@yunodiewtf]

Why in the goddamn fk isn't all the monitoring fully asynchronous???

[@unwinder]

Why do you need to comment something you don't understand _at_all_? Any kind of monitoring lives in its own process, it is not done in context of game you play. So of course it is fully asynchronous.

[@rain8478]

I'm on a full AMD system and any kind of GPU measurement causes this issue for me. Power, temps, anything. By the time its resolved all I have GPU related is gpu total percentage usage. Its disappointing but I'll learn to live without it I suppose. The system specs are 7800X3D and an RX 6800XT.

[@unwinder]

No, it is not. I'm afraid that those who never peeked inside context help before and never read about performance profiler will start genrating new myths about it due to _misunderstanding_ what they see: "- Also, if you try to peek inside performance profiler on AMD GPUs and compare it to NV, don't get confused after seeing non-zero polling time just for one GPU sensor and close to zero times for the rest GPU related sensors. That's a result of batching access for all sensors at once into single AMD driver API call. So you see cumulative value for the first GPU sensor and the rest ones are just read from cache (so polling time is effectively zero). NVIDIA GPUs use different approach and poll sensors independently, but some of them are also grouped/batched (e.g. NVIDIA GPU core clock and memory clock sensors are read into single call, so it is also normal to see non-zero time for the first clock but close to zero (cached) time for the second one). Intel Arc GPUs use implementation similar to AMD's one, so all sensors are also batched and read into single call. So it is also normal to see a pattern when just one GPU sensor is non-zero and the rest are close to zero on such platform."

[@rain8478]

@@unwinder Yeah , I must be hallucinating the stutters, ofc.

[@unwinder]

@@rain8478 the profiler is just not for you. You don't understand the basics and I give up to explain it.

[@rain8478]

@@unwinder I never said I'm interested in learning about what profiler does. I'm not a dev, I do not do the drivers, I'm not some engineer working to do X3D CPUs. I'm trying to kill off stutters.

[@unwinder]

@@rain8478 what's a point of trying to use the profiler if you don't need to know what kind of info it provides to you? It is not a magic stick to fix any kind of stuttering, you're using wrong tool, analyze wrong data with it and interpret it wrong way.

[@quintrapnell3605]

The lore behind the AMDip is mostly oriented around power states and how windows power plan interacted with Ryzen in the past. It’s just big tech trying to offset its carbon footprint for tax rebates and it’s almost a myth presently with no shot of the latest CPUs having such problems compared to Intel which is how this all was started years ago. Comparing to Intel.

[@Mostly_Positive_Reviews]

Yeah, I personally believe the whole AMDip narrative is blown out of proportion. While true that some games have better lows on Intel CPUs, the opposite is also true in other games.

[@gaav87]

Its not only power. 6800xt+ 9800x3d here CPU/GPU temp and cpu/gpu power gives me the "dip" every 0.6sec (had 600 pooling rate) Disabling all 4 above gives massive 0.1% low gains.

[@Mostly_Positive_Reviews]

Yeah, if you use the profiling option it's easy to see what causes it. The Nvidia API is known for causing issues specifically with power %, but not always, and not always only that. The AB dev mentioned that these polling options can affect different systems differently, that's why I showed how to use the profiler to identify potential issues. Very useful option that.

[@gaav87]

@@Mostly_Positive_Reviews ye i used the profiler 17ms for cpu temp crazy...

[@pigzera]

This should invalidate any AMD CPU/nvidia review lol, unless they did unchecked those bunch of monitoring. The effect in 1% low is astronomical.

[@Mostly_Positive_Reviews]

I think most bigger reviewers use external tools for monitoring power usage, and strangely enough it doesnt affect every game the same. But yeah, it would be interesting to know which reviewers had this issue unknowingly. I certainly did, at least my last few videos... Guess I'll have to rerun those benchmarks.

[@spenmac]

Interesting, I've been using Afterburner for years

[@Mostly_Positive_Reviews]

Same. It's not just Afterburner though. Most apps accessing the NVAPI for power percent will have the same CPU cost, I pinned a comment in this regard and will do a follow up video early next week.

[@gamingfromjohnwayne]

Download gpu tweaked i did this for my 4070ti super this had no problems. 😊

[@BlackZero1891]

It also has a serious effect in Hogwarts Legacy on my i9-10940X. This is with an Nvidia GPU, and can especially affect 1% lows in other games like The Great Circle.

[@Mostly_Positive_Reviews]

Yeah, I noticed the same in Indiana Jones. Will do a follow up video with more games and some more in depth sensor settings.

[@notwhatitwasbefore]

Short answer: YES! Longer answer: Just swithed from a 3070 overclocked with afterburner and RTSS running to a 7900xt had a list of instabilities and performance weirdness going on so I disabled the Adreneline software but not the driver and got the 7900xt to work as expected with afterburner and RTSS, as if by magic the issues went away. So I tried it the other way round and also worked like a dream so I enabled afterburner and BOOM dips in 1% lows driver crashes OC values not accepting after a reboot and probably more weirdness had I left the system in that setup. Can introduce issues when using any hardware polling like HWinfo but afterburner and RTSS is by far the worst in my limited testing, you can work around this by only using the AMD driver and not the full adrenaline software but seeing as that software works well and has a bunch of features built it to do these things I would just use it and not use other programs unless you need a specific function and have no other work around. Also and this might be way out of date so take with a big caveat the guy (I think) who was developing Afternurner for MSI was in Russia and things got weird with payments due to world events and I'm not sure if thats a factor or what state the dev of afterburner even is now but yeah might be a less/lack of development making this problem worse.

[@ForzaE2]

Thank you.

[@Mostly_Positive_Reviews]

No problem, glad I could help.

[@gamingfromjohnwayne]

Yes it's doing this to my 7 7700 cpu and my 4070ti super it will not go to 100 percent poerbut I used gpu tweaked and everything is working better and at full blast.

[@cybergaming3030]

Not using MSI Afterburner after knowing about CapframeX tool

[@Mostly_Positive_Reviews]

Unfortunately it also happens with HWInfo and CapFrameX too. I pinned a comment regarding that. According to the developer it's the way Power Percent is read via the NVAPI, and it happens with all monitoring tools.

[@superior96]

So it might be fine with Radeon GPUs then ​@@Mostly_Positive_Reviews

[@gaav87]

this also with capframex and hwinfo man xD

[@Mostly_Positive_Reviews]

Yeah, from what I'm told by the CapframeX dev it is related to the NVAPI and how the power percent sensor is accessed. I had a few people with Radeon 7000 series GPUs test it for me and they couldnt replicate the issue on their end, but could with Nvidia GPUs.

[@gaav87]

@@Mostly_Positive_Reviews Idk about 7000 series but 6800xt has this for sure +hwinfo

[@Accuaro]

Huh, I just tried it and it works..

[@Mostly_Positive_Reviews]

You sound surprised? 🤣

[@Accuaro]

@@Mostly_Positive_Reviews Yeah, why TF is Afterburner doing that (rivatuner)

[@Mostly_Positive_Reviews]

It's got to do with how the NVAPI is called for this sensor. It's not limited to Afterburner though, most monitoring software have the same issue reading this value on Nvidia GPUs.

[@PCZONE1]

Frame chasers that clown lol

[@Superdazzu2]

god fkcing dammit.. here i was wondering why these periodical stutters were killing my gaming experiences..(7800x3d..)

[@Mostly_Positive_Reviews]

😔

[@DaveTheeMan-wj2nk]

MSI after burner is terrible for metrics. MArvel rivals on the B580 would crash on start up, couldn't figure out why. Closed MSI, funny enough works perfectly. Loaded MSI back up and yet again, wouldn't load. That thing is so glitchy. There are better things to use.

[@unwinder]

I share my “so glitchy” tool with those who have some basic PC tuning skills as a hobby during 30 years. You can try do the same and offer your own solution. But in this case you’ll have no time for games and such toxic comments.

[@DaveTheeMan-wj2nk]

@unwinder You have a very fragile ego. The tool is incredibly out of date. The font looks like crap, to this day it still causes issues and weird FPS issues and crashing. Everyone's supposed to just not mention that and ignore it? Get over yourself

[@unwinder]

​@@DaveTheeMan-wj2nk Sure. Incredibly out of date tool, which got last update just 3 months ago. Which is getting 50x0 series support before they even announced and before you can see them. And LOL at fonts statements. The fonts in overlay and overlay itself are not customizable too. Probably the channels like @h1ttz, who use RTSS overlay editor to create such layouts, are just a _bit_ more educated? Sync your ambitions with your knowledge (or a lack of it).

[@DaveTheeMan-wj2nk]

@unwinder My you're quite a petulant one. You're incapable of writing a response without insulting somebody at the end of it. You're very insecure. I don't care how long it's been updated for, it sucks. The font is archaic, it's full and bloated with features no one can even use anymore and no one cares about. It lacks any kind of support for custom skins you would think something that's been out for 30 years would have that right about now, since it's been updated so very much. Seems to me it's had very minimal updates with very minimal effort and you've just kind of been riding the wagon of attention for far too long. It's likely you have a forum in many other people who defend you no matter the criticism because it's "free" Trash is free too, doesn't mean I want it. The incompetence required to have so many issues after 30 years of development is quite astonishing and a testament to the person designing it. It's just not very good.

[@niezzayt3809]

Quick heads up! MSI Afterburner is developed only by ONE GUY. Despite the name "MSI Afterburner", they couldn't constantly update the tool as needed, because that one Guy is a Russian who escaped from Conscription. No, I'm not kidding. If you really need on-screen Overlay, use something else

[@Mostly_Positive_Reviews]

Not sure I understand correctly. Are you saying that because he didnt want to be forced to go to war we shouldnt be using his software?

[@niezzayt3809]

@@Mostly_Positive_Reviews No, He couldn't Update MSI Afterburner frequently because of that reason. You might want to blame MSI Company instead. Why don't they hire a more qualified programmer instead of just relying on 1 Guy?

[@niezzayt3809]

@@Mostly_Positive_Reviews do you even remember how "Frequent" MSI Afterburner gets Updates before 2022? Very often! And then after 2022, the updates becoming Rarer and Rarer?

[@unwinder]

What are you smoking?

[@niezzayt3809]

@@unwinder facts? If you still demand a reliable On-screen overlay, just use Digital Foundry overlay. It's significantly better than MSI Afterburner. And you can use Ryzen Master to adjust clock speed, Overclock, undervolt etc. And then you can also use Nvidia App to adjust GPU clock. Or Radeon Panel to adjust Radeon GPU. Basically not MSI Afterburner since the dude not frequently updating the app.

[@steven7297]

What did you expect playing a cpu intensive game and using applications that use the CPU. duhhh your gonna get dips. It’s not rocket science here

[@Mostly_Positive_Reviews]

We have a new contender for dumb comment of the year 🥳

[@teamvoidhacks]

wish people would benchmark games like PUBG, Fortnite with 0.01% and 0.1% lows.

[@Mostly_Positive_Reviews]

I have videos for both those games up on the channel actually. Will do a more comprehensive video about this subject that'll include more games.

[@KillaGorilla-l7z]

There is no amdip f lie

[@Mostly_Positive_Reviews]

That's why I put "AMDip" in inverted commas. I find the use of that term very funny and overblown.

[@UKKN516]

why would you use MSI afterburner if theirs a whole wattman tool in the AMD drivers? i dont understand people you think a third party APP is going to run better than AMD's own implementation? lmfao USE YOUR DRIVERS.

[@Mostly_Positive_Reviews]

Now why would I use AMD drivers when I use an Nvidia GPU in this video?

[@johnhughes9766]

Everyone knows about the amdip and framechasers but no one gives him the credit

[@Mostly_Positive_Reviews]

My issue with him is I think he's blowing it out of proportion. Many of the AMDip instances he showed back in the day when it all started with the 5800X3D my 12700K would exhibit the exact same issue in the same place. When I told him that he hid me from his YT channel so all my comments would be hidden. Havent watched one of his videos since, but the AMDip term keeps popping up all over social media. I havent seen a video other than the one Danny did to actually resolve the issue, instead everyone just throws that term around for anything.

[@johnhughes9766]

@ his 9800x3d video is worth a watch, it’s all real ingame footage and testing compared to a 14900k

[@gaav87]

@@johnhughes9766 Oh you mean the one where he preloaded the textures on 14900k to make it look "good" ? xD

[@johnhughes9766]

@ the cope is real lol

[@chrissimao14]

There is a reason why no one gives him credit.. he has no credibility. Lol