Coil whine related to micro-stutter?

[Andynor91]

My finished mini-ITX build works great except for this weird micro stutter going on when I play games. When I play games, I can hear coil whine from my GPU in sync with when it stutters. That is - at the exact point when the game micro stutters, the coil wine stops, then starts up when it's all smooth again. It's basically like the frame freezes for some ms, then continues on.

Anyone have suggestions for solutions in regards to this? Give the GPU more voltage/undervolt it? Raise/decrease the power limit? I have isolated the coil whine to the GPU, as the noise is 100% coming from there. However, when I pair the GPU in a different setup with a TX750V2 PSU, the coil whine is not being emitted. Might it be that the SF450 PSU I have in my build is not providing the GPU with enough voltage/a steady stream of voltage?

The specs are:
PCPartPicker part list / Price breakdown by merchant

CPU: AMD - Ryzen 5 1600 3.2GHz 6-Core Processor ($197.19 @ SuperBiiz)
CPU Cooler: CRYORIG - C7 40.5 CFM CPU Cooler ($26.99 @ Newegg Marketplace)
Motherboard: ASRock - AB350 Gaming-ITX/ac Mini ITX AM4 Motherboard ($108.99 @ SuperBiiz)
Memory: Corsair - Vengeance LPX 16GB (2 x 8GB) DDR4-3200 Memory ($162.88 @ OutletPC)
Storage: Samsung - 840 EVO 250GB 2.5" Solid State Drive
Storage: Seagate - BarraCuda 2TB 2.5" 5400RPM Internal Hard Drive ($84.99 @ Amazon)
Video Card: Sapphire - Radeon RX 470 8GB NITRO+ Video Card
Case: Silverstone - Sugo SG13B Mini ITX Tower Case ($39.99 @ Amazon)
Power Supply: Corsair - SF 450W 80+ Gold Certified Fully-Modular SFX Power Supply ($89.99 @ Amazon)
Case Fan: be quiet! - Pure Wings 2 51.4 CFM 120mm Fan ($10.89 @ SuperBiiz)
Other: SilverStone Technology Universal ATX to SFX Power Supply Bracket RL-PP08B ($10.99 @ Newegg)
Total: $732.90
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2017-09-07 21:11 EDT-0400

 

[jboydgolfer]

iirc, coil whine is at its loudest when theres a loading screen, or title screen. maybe when the stutter happens its a similar reaction to that "frozen screen" situation and becomes its loudest during the stutter?

 

[Andynor91]

iirc, coil whine is at its loudest when theres a loading screen, or title screen. maybe when the stutter happens its a similar reaction to that "frozen screen" situation and becomes its loudest during the stutter?

The coil whine completely disappears during the micro-stutter. So the sound disappears for some ms at the exact time when the screen stutters, then keeps going when the micro stutter is over with the coil whine going at it again. In addition to this, it becomes less intense when I set the FPS limit to e.g. 30 in Overwatch. The change of sound is very evident when I switch between e.g. 60fps and 30fps.

 

[jboydgolfer]

The coil whine completely disappears during the micro-stutter. So the sound disappears for some ms at the exact time when the screen stutters, then keeps going when the micro stutter is over with the coil whine going at it again. In addition to this, it becomes less intense when I set the FPS limit to e.g. 30 in Overwatch. It's very evident when I switch between e.g. 60fps and 30fps.

I may have gotten it reversed , but my guess is during the micro stutter rendering is stopped because the frames have dropped to zero , & for that millisecond the load on the gpu is zero for a split second,therefore the microstutter stops

I may have it backwards but I'm almost 100% certain that the freezing of the FPS for a millisecond is what's causing the stopping of the wine

 

[Andynor91]

I may have gotten it reversed , but my guess is during the micro stutter rendering is stopped because the frames have dropped to zero , & for that millisecond the load on the gpu is zero for a split second,therefore the microstutter stops

I may have it backwards but I'm almost 100% certain that the freezing of the FPS for a millisecond is what's causing the stopping of the wine

Makes sense. But why does it happen, and how can I fix it though?

 

[MrGenius]

Power limit. Raise it.

 

[Andynor91]

Power limit. Raise it.

By how much would be a good number?

 

[jboydgolfer]

By how much would be a good number?

If by "it" and you mean the coil wine, your guess is probably as good as mine I've heard people say hot glue on the side of a capacitor as well as stress tests for long durations.

Regarding the micro starter, i'm not sure. My AMD experience and did when I got rid of my 6950. I still have a few AMD cards but not in my personal PC. I always thought micro stutter stemmed from crossfire and SLI , that shows how little i know.

Is it all games? Have you tried windowed mode changing settings doing what Mr. genius said increasing power limit etc.

 

[eidairaman1]

Your games on that 5400 drive?

 

[Andynor91]

Your games on that 5400 drive?

Games is on the EVO 840 SSD

 

[MrGenius]

By how much would be a good number?

Until it stops causing micro stutter/power throttling. Or just dial back clocks until it stops power throttling. One or the other.

 

[Andynor91]

Until it stops causing micro stutter/power throttling. Or just dial back clocks until it stops power throttling. One or the other.

Only the GPU clock or the Memory clock too? The memory clock is at 2000MHz. What would be a safe number to start with to test if that can have an effect?

Until it stops causing micro stutter/power throttling. Or just dial back clocks until it stops power throttling. One or the other.

So I tried turning the GPU clock down. I can hear quite the difference in coil whine when I turn it down 600MHz or so, but that's basically cutting the performance in half I guess.

 

[cadaveca]

So I tried turning the GPU clock down. I can hear quite the difference in coil whine when I turn it down 600MHz or so, but that's basically cutting the performance in half I guess.

So, "coil whine" is induced by the frequency of the power delivered. It's usually an inductor causing it. It's not really a problem, just something that happens because of the switching frequency of the power delivery, and as such, it's largely unavoidable 100%. The coil whine changing when the FPS drops is natural... because the power being used by the GPU when this happens has changed. It is commonly heard obnoxiously when the work being done by a GPU is slight, and FPS is high, but the reverse can also be true too, I suppose. The two are related because the power delivery changes, and neither is a "fault"...

Looking at PSU is a good idea, but what's funny about that PSU is that it's 450W rating is only good at 40c, but the marketing material on Corsair's site suggests that it's good to 105c, because of "Premium internal components ensure solid power delivery and long term reliability." and "Zero RPM mode for virtually silent operation, and 105°C All Japanese capacitors for better reliability.", but I can't understand how a PSU is quality if it says "Engineered to meet maximum power output at a 40°C temperature rating." I'd be more looking for 65C or 70C myself to call something "quality" (55c is a normal de-rating point), especially in SFF use. I'd says PSU is crap and is causing GPU to throttle as you surmised. If your GPU is internally exhausted, it could be that the case inside is too hot, and this causes the PSU to flake out, which then causes the GPU to stumble a bit.

 

[Andynor91]

So, "coil whine" is induced by the frequency of the power delivered. It's usually an inductor causing it. It's not really a problem, just something that happens because of the switching frequency of the power delivery, and as such, it's largely unavoidable 100%. The coil whine changing when the FPS drops is natural... because the power being used by the GPU when this happens has changed. It is commonly heard obnoxiously when the work being done by a GPU is slight, and FPS is high, but the reverse can also be true too, I suppose. The two are related because the power delivery changes, and neither is a "fault"...

Looking at PSU is a good idea, but what's funny about that PSU is that it's 450W rating is only good at 40c, but the marketing material on Corsair's site suggests that it's good to 105c, because of "Premium internal components ensure solid power delivery and long term reliability." and "Zero RPM mode for virtually silent operation, and 105°C All Japanese capacitors for better reliability.", but I can't understand how a PSU is quality if it says "Engineered to meet maximum power output at a 40°C temperature rating." I'd be more looking for 65C or 70C myself to call something "quality" (55c is a normal de-rating point), especially in SFF use. I'd says PSU is crap and is causing GPU to throttle as you surmised. If your GPU is internally exhausted, it could be that the case inside is too hot, and this causes the PSU to flake out, which then causes the GPU to stumble a bit.

Do you think the SF600 would make any difference? I opted for the SF450 because I thought it was more than enough power for this build.

Also, if I turn the power limit down, the coil whine becomes less. However, the temperature go higher. Could it be an option to lower the power limit, put on a more aggressive fan curve, and keep it under 75-80C?

 

[cadaveca]

Do you think the SF600 would make any difference? I opted for the SF450 because I thought it was more than enough power for this build.

Also, if I turn the power limit down, the coil whine becomes less. However, the temperature go higher. Could it be an option to lower the power limit, put on a more aggressive fan curve, and keep it under 75-80C?

I would test this by taking the entire PC apart, and assembling it outside of the case. This would allow the PSU fresh air all the time. If it works fine in those conditions, then likely PSU is culprit.

I can't say for sure it's the PSU though, but SFF cases usually don't have great airflow, and that PSU's relatively low temperature rating is odd to me, and your system is likely very close to 400W if you have the CPU OC'ed. The GPU might use 175W, say 50-75 for the board, memory, cooler and fan, 85W for CPU (maybe a touch less, or a touch more), drives use only a little bit, so you're looking at like 300W - 350W so the PSU would be fine... if it is remaining at 40c or less. So testing it with lots of airflow should allow it to push that 450W.

It is also possible that your CPU is throttling, so while the first thought might be the PSU, it might not be. Could also be some weird BIOS bug for the board... hard to tell. Update board's BIOS, perhaps. If it has the launch BIOS, the CPU could be pulling more power than it needs, common thing on early BIOSes for this platform, and that might explain the PSU being overwhelmed. I'd be investigating this should running the system out of the case does not fix the issue.

 

[MrGenius]

First of all, coil whine is not a problem, or even related to any kind of problem. Get that thought out of your head entirely. It sounds weird to me that using a higher output PSU seems to make it go away. Makes no sense at all frankly. But I'm not going to argue with you about that. I'm also not going to say it(the "micro stutter" not the coil whine) couldn't be caused by the lower wattage PSU. You didn't mention anything but the coil whine being "fixed" with the higher wattage PSU. So I went with the assumption of the power limit being too low causing power throttling(which I'm still assuming might be true). Stock clocks shouldn't power throttle with a stock power limit(+0%). But I don't know what clocks you're running...so that's why I mentioned lowering the clocks. Lowering the power limit causing higher temps makes no sense either. But again...whatever. I'll take your word for it. As far as this or that being safe...you're worrying to much IMO. The graphics card BIOS has limits set that will protect the card from being damaged by any adjustments you want to make(unless you modded the BIOS). Power limit maxed, voltage maxed, overclock it until it crashes and/or starts throwing artifacts...you're fine. You're not going to hurt anything.

BTW Sapphire recommends a 500W PSU for your card. So that's something to consider. And he's got a point about it possibly being the CPU throttling too.

 

[eidairaman1]

I would test this by taking the entire PC apart, and assembling it outside of the case. This would allow the PSU fresh air all the time. If it works fine in those conditions, then likely PSU is culprit.

I can't say for sure it's the PSU though, but SFF cases usually don't have great airflow, and that PSU's relatively low temperature rating is odd to me, and your system is likely very close to 400W if you have the CPU OC'ed. The GPU might use 175W, say 50-75 for the board, memory, cooler and fan, 85W for CPU (maybe a touch less, or a touch more), drives use only a little bit, so you're looking at like 300W - 350W so the PSU would be fine... if it is remaining at 40c or less. So testing it with lots of airflow should allow it to push that 450W.

It is also possible that your CPU is throttling, so while the first thought might be the PSU, it might not be. Could also be some weird BIOS bug for the board... hard to tell. Update board's BIOS, perhaps. If it has the launch BIOS, the CPU could be pulling more power than it needs, common thing on early BIOSes for this platform, and that might explain the PSU being overwhelmed. I'd be investigating this should running the system out of the case does not fix the issue.

I wonder if the other psu he tested can fit in the case of his, because it seems to have the right stuff for power delivery. And dave I take corsairs descriptions as a grain of salt.

By the way, at op has the drive been upgraded to the latest firmware?

When you installed the OS did you have the HDD disconnected?

 

[Andynor91]

I would test this by taking the entire PC apart, and assembling it outside of the case. This would allow the PSU fresh air all the time. If it works fine in those conditions, then likely PSU is culprit.

I can't say for sure it's the PSU though, but SFF cases usually don't have great airflow, and that PSU's relatively low temperature rating is odd to me, and your system is likely very close to 400W if you have the CPU OC'ed. The GPU might use 175W, say 50-75 for the board, memory, cooler and fan, 85W for CPU (maybe a touch less, or a touch more), drives use only a little bit, so you're looking at like 300W - 350W so the PSU would be fine... if it is remaining at 40c or less. So testing it with lots of airflow should allow it to push that 450W.

It is also possible that your CPU is throttling, so while the first thought might be the PSU, it might not be. Could also be some weird BIOS bug for the board... hard to tell. Update board's BIOS, perhaps. If it has the launch BIOS, the CPU could be pulling more power than it needs, common thing on early BIOSes for this platform, and that might explain the PSU being overwhelmed. I'd be investigating this should running the system out of the case does not fix the issue.

Any way I can confirm it's the GPU or CPU power throttling? Any apps specifically which can address this? I'm not running any kind of OC by the way. Everything at stock settings.

I wonder if the other psu he tested can fit in the case of his, because it seems to have the right stuff for power delivery. And dave I take corsairs descriptions as a grain of salt.

By the way, at op has the drive been upgraded to the latest firmware?

When you installed the OS did you have the HDD disconnected?

SSD has latest firmware. Installed OS with the other HDD connected - any reason why this could have an effect?

 

[eidairaman1]

Any way I can confirm it's the GPU or CPU power throttling? Any apps specifically which can address this? I'm not running any kind of OC by the way. Everything at stock settings.


SSD has latest firmware. Installed OS with the other HDD connected - any reason why this could have an effect?

Because the install medium puts certain bootfiles on that drive instead of keeping them contained on 1 drive, its been like that since XP

 

[Andynor91]

So I did some research for my GPU, and it runs at 1.185mV at stock settings 1260MHz/2000Memory.
Figured out that this is a very high voltage to be running at for stock settings, so lowered it down to about 1.100mV, and coil whine is almost gone. Will try to get it lower, but afraid it will throttle the GPU/impact performance and stability.

 

[INSTG8R]

In general terms coil whine happens when either A: which I think is your issue is insufficient/dirty power or the one that you can induce whine on the best of cards is letting it run "free" with no vysnc and pushing really high FPS as someone mentioned things like Game Intros where the FPS is pushing 500fps or greater. That is the only situation where mine wines is when it's pushing frames well above and beyond reason and its pretty much "straining" the coils trying to "keep up"

 

[Flow]

My coilwhine goes away when I run a game, or in simple terms, put the gpu to work. It does not cause microstutter though.
Would coilwhine go away with sufficient power? I doubt it, it would be posted many times already if so.
microstutter however is something else, which could be solved with more power, or optimized drivers/game patches, or better graphicscard.
If this would also eliminate the coilwhine, then that's a bonus, but purely coincidental.

 

[infrared]

Coil whine is normal tbh, you can be lucky and get totally silent cards, or unlucky and get a horrifically bad one. It's just the card's VRM doing it's thing. The micro-stuttering is completely unrelated, the coil whine stopping during the stutter is normal because the card briefly isn't doing anything so there's no load and no whine.

Did you check the RAM is running at a decent speed? Normally if you just put some ram in and boot it'll likely be running at 2133mhz, and with Ryzen, the infinity fabric runs at the same speed as the RAM so if it's running slow like I suspect it could definitely be holding your computer back and attributing to the stuttering. Overclocking the CPU would most likely help a little too, but that's your choice. Also some games just suck and are more likely to have stuttering.

 

[Andynor91]

Coil whine is normal tbh, you can be lucky and get totally silent cards, or unlucky and get a horrifically bad one. It's just the card's VRM doing it's thing. The micro-stuttering is completely unrelated, the coil whine stopping during the stutter is normal because the card briefly isn't doing anything so there's no load and no whine.

Did you check the RAM is running at a decent speed? Normally if you just put some ram in and boot it'll likely be running at 2133mhz, and with Ryzen, the infinity fabric runs at the same speed as the RAM so if it's running slow like I suspect it could definitely be holding your computer back and attributing to the stuttering. Overclocking the CPU would most likely help a little too, but that's your choice. Also some games just suck and are more likely to have stuttering.

As I previously remarked, my GPU runs at 1.185mV at stock settings 1260MHz/2000Memory. After looking it up, it seems like this is a very high stock voltage for stock settings for an RX 470 Sapphire 8GB. Lowered it down to 1.075Mv and coil whine is almost gone.
Figured out that this is a very high voltage to be running at for stock settings so lowered it down to about 1.065mV, and coil whine is almost gone. Will try to get it lower, but afraid it will throttle the GPU/impact performance and stability. What do you think? I checked the ASIQ score of the card, and it's 79%, which as far as I know is pretty decent. GPU temps under load is 75C, while CPU temps while playing e.g. Witcher 3 or BF1 for some hours is 62-65C. CPU temps while running the blend test of P95 for some hours is around 77-80C. Are these temps alright in a SFF build?

To add to the CPU temps, the voltage seems to also be set higher at stock than what is considered normal. While it's set at 1.2375V in F-Stream as well as in BIOS manually, HWiNFO64 shows otherwise if I'm not mistaking. From what I can gather from the readings, it sometimes spikes to 1.3V. Haven't done any OC on any components except for memory, which is set at 3066MHz with the XMP profile.

https://imgur.com/a/zXFnG