Are streams killing my NVME?

A

alf717

Gawd
Joined
Dec 8, 2006
Messages
906
Hey fellas, I have a PC I use mostly for watching streamed content from YouTube, Twitch and Amazon. The NVME drive I use is a 512GB Inland from microcenter I bought back in 2019. As of 2021 I leave this machine on 24/7 so I can watch shows or streams as they come up. I've noticed since about 2024 the drive's health has been degrading slowly since then. I mostly use both Firefox and Vivaldi and under the details tab in task manager they can show very large numbers under I/O write bytes sometimes up to 50,000,000,000 or more after some time of watching streamed content. I usually go days without closing the browser. My question is are streams causing the drive to slowly loose more health? For the time being I redirected cache to my D: by editing the target section of the exe with --disk-cache-dir="D:\Vivaldi-Cache". Not sure if it helps but this machine is running Windows 10 and has been the initial install from back in 2019.
 

Attachments

  • nvme.png
    nvme.png
    162.6 KB · Views: 0
kram182 said:
streaming mostly gets stored in ram/vram
Click to expand...

Interesting I guess the drive is just going out on its own? Maybe power on hours or something.
uOpt said:
Are you swapping from RAM shortage?
Click to expand...

I have 16GB of RAM in this PC but I never seem to go above that limit useless I'm playing a game.

Probably not related but I did have an HDD drive go out on me last year. I found a very large indexing cache file that was over 70GB so I turned that off completely. I try to stay around 200GB of free storage on this drive but for some reason I tend to lose a few GBs every time I go to that limit. It's currently sitting at 196GB free of 476GB.
 
The average life of a drive is around 4-5 years, depending on how much data you've written to the drive. And based on your CrystalDiskInfo stats ... you should really replace that drive ... it's on the way out.
 
Typical 512GB NVMe TBW is between 300 and 650, depending on the specs. You've got close to 125 at 71% health. It took 6 years, 9 months, and 15 days to get there. So far there's been no errors, so at this rate, it'll last another 9 years, 8 months, and 6 days. You've got time.
 
El_Capitan said:
Typical 512GB NVMe TBW is between 300 and 650, depending on the specs. You've got close to 125 at 71% health. It took 6 years, 9 months, and 15 days to get there. So far there's been no errors, so at this rate, it'll last another 9 years, 8 months, and 6 days. You've got time.
Click to expand...
This is nonsense and isn't the only metric for determining a drive's health.
 
I am not convinced that streaming only buffers to RAM. Isn't youtube using local storage?
 
Shoganai said:
This is nonsense and isn't the only metric for determining a drive's health.
Click to expand...
Then pray tell, what is?

I haven't had any NVMe SSD go bad on me, yet. A few SATA SSD's, but never from write wear. I still have one that's at 63%, maybe even lower. Still works fine.
 
alf717 said:
Interesting I guess the drive is just going out on its own? Maybe power on hours or something.


I have 16GB of RAM in this PC but I never seem to go above that limit useless I'm playing a game.

Probably not related but I did have an HDD drive go out on me last year. I found a very large indexing cache file that was over 70GB so I turned that off completely. I try to stay around 200GB of free storage on this drive but for some reason I tend to lose a few GBs every time I go to that limit. It's currently sitting at 196GB free of 476GB.
Click to expand...
If you don't have enough memory (RAM), your operating system will use swap space to expand it; by default, this is located on your main drive (NVMe), so it uses it in addition to RAM > yes, that's your problem.
 
Both firefox and chrome have massive databases that they write to constantly, with cached pages, images, etc. I wouldn't expect it to kill a drive normally, but if it is failing it could accelerate the death.
 
El_Capitan said:
I haven't had any NVMe SSD go bad on me, yet.
Click to expand...
That has zero relevance whatsoever. I've replaced many of these for clients with failed NVME drives. NVMEs aren't forged from magic. They're still hardware with an expiration date like anything else. I've seen brand new drives fail after a few months. I've seen drives last a decade. You never know. Your empirical bubble isn't reality. His drive could very well outlast every component in his PC or it could fail tomorrow. That's why it's important to keep a backup regardless of what kind of internal drive you have. Because failure is an inevitability. I'd also recommend a backup that prevents bit rot.
 
Shoganai said:
That has zero relevance whatsoever. I've replaced many of these for clients with failed NVME drives. NVMEs aren't forged from magic. They're still hardware with an expiration date like anything else. I've seen brand new drives fail after a few months. I've seen drives last a decade. You never know. Your empirical bubble isn't reality. His drive could very well outlast every component in his PC or it could fail tomorrow. That's why it's important to keep a backup regardless of what kind of internal drive you have. Because failure is an inevitability. I'd also recommend a backup that prevents bit rot.
Click to expand...
Well, no shit, Sherlock.
Shoganai said:
This is nonsense and isn't the only metric for determining a drive's health.
Click to expand...
You still haven't answered the question.

Backups, redundancy, yeah, no duh. Our car that has 150K miles could fail any day. Maintenance and repairs will help it last longer. Same with any computer component. A component that has lasted almost 7 years with a CDI of 71% isn't indicative of failure, it's indicative of success and wear due to TBW usage. There's no errors found on the CDI screenshot. It's a healthy drive. Talk about nonsense, just tell us the other metric determining a drive's health.
 
El_Capitan said:
Well, no shit, Sherlock.

You still haven't answered the question.

Backups, redundancy, yeah, no duh. Our car that has 150K miles could fail any day. Maintenance and repairs will help it last longer. Same with any computer component. A component that has lasted almost 7 years with a CDI of 71% isn't indicative of failure, it's indicative of success and wear due to TBW usage. There's no errors found on the CDI screenshot. It's a healthy drive. Talk about nonsense, just tell us the other metric determining a drive's health.
Click to expand...
"My NVME drive never failed so NVME drives don't fail" is your argument. I've been doing this for over 20 years and drives that have no errors in the S.M.A.R.T. data whatsoever can still fail. And they have. Many times in my experience.
 
Shoganai said:
"My NVME drive never failed so NVME drives don't fail" is your argument. I've been doing this for over 20 years and drives that have no errors in the S.M.A.R.T. data whatsoever can still fail. And they have. Many times in my experience.
Click to expand...
Lol, Jesus Christ, you're dense. It's not an argument, it's an anecdote. Just because an NVMe drive never failed for me doesn't mean it doesn't ever fail.

So, your "metric" is randomness. Gotcha. Go back to drinking your Kool Aid.
 
El_Capitan said:
Lol, Jesus Christ, you're dense. It's not an argument, it's an anecdote. Just because an NVMe drive never failed for me doesn't mean it doesn't ever fail.

So, your "metric" is randomness. Gotcha. Go back to drinking your Kool Aid.
Click to expand...
It's not an anecdote, it's just false. Why are you swearing? The metric isn't randomness. The metric is things can fail at any time regardless of their health status in a S.M.A.R.T. diagnostic and if the the behavior of the drive changes and performance starts being random or slow or terrible or spotty, that is a better indicator than what the S.M.A.R.T. readout is. That's from doing this for over two decades. Please spare me whatever nonsense this rant is. Your anecdote is completely unhelpful and irrelevant, no matter how useful you falsely seem to think it is. Drives can also perform fine one day then completely fail the next. "Randomness" sometimes is the metric, because components can fail randomly without explanation ... things that you would need to send the component to lab just to find out what actually happened.
 
Shoganai said:
It's not an anecdote, it's just false. Why are you swearing? The metric isn't randomness. The metric is things can fail at any time regardless of their health status in a S.M.A.R.T. diagnostic and if the the behavior of the drive changes and performance starts being random or slow or terrible or spotty, that is a better indicator than what the S.M.A.R.T. readout is. That's from doing this for over two decades. Please spare me whatever nonsense this rant is. Your anecdote is completely unhelpful and irrelevant, no matter how useful you falsely seem to think it is. Drives can also perform fine one day then completely fail the next. "Randomness" sometimes is the metric, because components can fail randomly without explanation ... things that you would need to send the component to lab just to find out what actually happened.
Click to expand...
You don't know what an anecdote means. When is Jesus Christ a swear word? Spare me whatever nonsense you just wrote, might as well write with an auto pen.

Randomness can never be a metric. It can never be measured. Random is random, and if it isn't random, then it's not random, and even what's not random is random, it's part of the random process.
 
Last edited:
El_Capitan said:
No, don't. It might randomly fail, same as any drive you replace it with, which may also randomly fail. It's best to just back it up with a drive, that might, also, randomly fail.
Click to expand...
That's not what I was saying, you dingus. Or your advice "do nothing ... cuz I've never had a drive fail."

The drive health doesn't look great at the moment. I'm saying make sure he has a back up just in case as drives can fail regardless of the health status. Does that offend your sensibilities or something?
 
uOpt said:
I am not convinced that streaming only buffers to RAM. Isn't youtube using local storage?
Click to expand...

The web page part of the page? Sure. The video however is being handled by the video decoders/encoders on the GPU/iGPU, using its VRAM/RAM. Especially with DRM streams. And then sent to the screen/framebuffer is handled by VRAM.
 
  • Like
Reactions: uOpt
like this
You must log in or register to reply here.
Back
Top