SSD in low ram computer

[tortillasoup]

I actually have 8GB of ram which should be sufficient but due to using 32bit Windows 10 for some very legacy software (16bit), I only have 3.2GB of ram available. The issue I have is, while this computer is very speedy with its 512GB SSD despite only have 3.2GB of ram available, my commit charge is exceeding 8GB regularly, so approximately 4+GB of the SSD is being used as the swapfile. I'm concerned with the longevity of this configuration and while I have wiped the SSD and created the initial partition to be only 100GB, I'm still concerned with how long this drive will survive in this current configuration. On a mechanical drive, 8GB of ram would be bare minimum but because of this SSD, 3GB works pretty well which disturbs me a bit because it makes me wonder how many writes I am making to this drive per day if I were to continue like this.

 

[pcgeek11]

I just keep good backups and don't worry about it.

 

[VirtualLarry]

I don't really think that it matters that the partition is only 100GB, or 500GB. But you're right, you're putting wear and tear on your SSD. Just watch your SMART values change, if you're worried.

 

[sdifox]

You are probably ok

https://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

 

[VirtualLarry]

Then again, we had a member claim that he burned out two Samsung 850 Pro 128GB drives, in less than 2 years.

 

[IntelUser2000]

The issue I have is, while this computer is very speedy with its 512GB SSD despite only have 3.2GB of ram available, my commit charge is exceeding 8GB regularly, so approximately 4+GB of the SSD is being used as the swapfile. I'm concerned with the longevity of this configuration and while I have wiped the SSD and created the initial partition to be only 100GB, I'm still concerned with how long this drive will survive in this current configuration.

I'll tell you what I think. You are being a bit paranoid. If you have got an SSD from top tier manufacturers(Intel/Samsung/Micron for example) then then drive is more likely to die of being old than due to having too many writes.

There's a system my family uses that I replaced the HDD with a Intel 520 SSD back from 2012. It's running on Vista and has only 2GB of RAM. Despite the SSD, its swapping quite a bit and using virtual memory so the system can be quite slow in everyday operations like web browsing and watching streaming videos. But, the system works fine.

What did I mean by drive dying being old? Well all components succumb to the process of natural wear and tear. It might take 10 years, or last 20 or 30 years.

 

[WhoBeDaPlaya]

I wouldn't worry about it with any contemporary MLC drive.
Even TLC should be fine, but I don't particularly like those (vanilla 840s left a sour taste in my mouth).

 

[UsandThem]

I'd personally just keep an eye on every six months or so with a utility like CrystalDiskInfo. As long as you keep backups, even if you wear it out faster than typical home users, you can just get a new one. It should still last you a long time, unless it is defective. I'd just set a nice over provision on it (10% or about 52GB), which should help with a large number of writes.

 

[XavierMace]

Then again, we had a member claim that he burned out two Samsung 850 Pro 128GB drives, in less than 2 years.

And I claim I'm the Queen of England.

I've been using SSD's for almost 8 years now.

OCZ Apex
OCZ Onyx
OCZ Agility
OCZ Agility 3
OCZ Vertex
OCZ Vertex 2
OCZ Vertex 3
OCZ Vertex 4
Intel X25m
Kingston SSD V300
Samsung 840 EVO
Samsung 850 EVO
Samsung 960 PRO
Crucial MX100
Crucial MX200
Crucial MX300
Plextor M8Pe

I've had at least two of all of those. I'm probably forgetting some too. I've had a grand total of 3 fail. 1 Apex (Running in RAID0 so OCZ RMA'd both), 1 OCZ Onyx after 6 years of daily usage, and 1 MX100 (being used as a SLOG). The MX100 is the only recent model drive to have a failure. The rest have just been replaced by larger, faster drives. Multiple of these drives are being used as SLOG's in my ZFS SAN's which have 6-12 VM's running on them at all times, 24x7. I couldn't even guess how much data those have written to them. I stopped caring about the page file or any of the other "concerns" after the Onyx's.

 

[VirtualLarry]

Hey, I'm with you, man. I personally thought that the claim was a bit strange and unusual.

I can't say that I've had much SSD failures either, except in a particular high-temp environment, that made even a new drive go bad in months. (SSDs don't much like 90C ambients.)

Edit: No, wait, I did have a Mushkin SSD go bad in about 2 months in my laptop at the time. It was a time of great upheaval (and progress) in SandForce-firmware land.

 

[Ken g6]

I used a swap partition on a system almost exactly opposite of yours for years without a problem. It's a 64-bit Linux system with only 4GB RAM. And although I moved on to a newer system a couple of years ago, the old system and drive still work.

But it seems like you should be able to set up a RAM drive on the extra memory or something.

 

[WhoBeDaPlaya]

As an anecdote, I abused the crap out of a 240GB Crucial M500 for a year or so
(it served as a DL drive for HIGH (90MB/s+) speed Usenet NZB downloads)

Using it as an OS drive on one of my seedboxes. Absolutely no issues or SMART errors

 

[coercitiv]

Let's go full paranoid and assume the worst: from the moment the OP starts using his computer until the moment he stops, the SSD is continually writing data with 512MB/s, destroying NAND cells at the fastest rate possible for a 32bit Win machine running 16bit code. (it's continually and viciously writing "I'm the Queen of England" to the pagefile)

It's a 512GB drive and in the absence of clear specs we can assume NAND inside is good for 2000 P/E cycles. We'll also assume a write amplification factor of 2X. That puts the SSD at 500TBW before we can expect to go out of JEDEC spec. At a rate of 512MB/s for let's say 4 hours a day, the drive will reach it's max rated P/E cycle count in... 4.67 years.

So this settles it OP: your drive can only last 5 years under the extreme load you're putting it through. Look back to the exact date your started using the SSD and mark the 5 year deadline in your calendar.

That day your SSD usage counter will probably drop from 98% to 97%.

 

[Glaring_Mistake]

I actually have 8GB of ram which should be sufficient but due to using 32bit Windows 10 for some very legacy software (16bit), I only have 3.2GB of ram available. The issue I have is, while this computer is very speedy with its 512GB SSD despite only have 3.2GB of ram available, my commit charge is exceeding 8GB regularly, so approximately 4+GB of the SSD is being used as the swapfile. I'm concerned with the longevity of this configuration and while I have wiped the SSD and created the initial partition to be only 100GB, I'm still concerned with how long this drive will survive in this current configuration. On a mechanical drive, 8GB of ram would be bare minimum but because of this SSD, 3GB works pretty well which disturbs me a bit because it makes me wonder how many writes I am making to this drive per day if I were to continue like this.

Don't think you would have to worry about your drive wearing out just because you have your swap file on it - especially not with a drive of that size.
But information on the wear it has seen and what drive it is should allow us to calculate when (at the current pace) it is going to reach the amount of P/E cycles it is specified for.
And remember that a drive may fail long before then as well as long after that - though it will have lower retention and be more sensitive in general after having seen much wear.

Let's go full paranoid and assume the worst: from the moment the OP starts using his computer until the moment he stops, the SSD is continually writing data with 512MB/s, destroying NAND cells at the fastest rate possible for a 32bit Win machine running 16bit code. (it's continually and viciously writing "I'm the Queen of England" to the pagefile)

It's a 512GB drive and in the absence of clear specs we can assume NAND inside is good for 2000 P/E cycles. We'll also assume a write amplification factor of 2X. That puts the SSD at 500TBW before we can expect to go out of JEDEC spec. At a rate of 512MB/s for let's say 4 hours a day, the drive will reach it's max rated P/E cycle count in... 4.67 years.

So this settles it OP: your drive can only last 5 years under the extreme load you're putting it through. Look back to the exact date your started using the SSD and mark the 5 year deadline in your calendar.

That day your SSD usage counter will probably drop from 98% to 97%.

Mind checking your numbers again?
Because to me it looks like it would write about 7TB to the drive daily which means it would take a little over 70 days to reach 500TB.

 

[coercitiv]

Mind checking your numbers again?
Because to me it looks like it would write about 7TB to the drive daily which means it would take a little over 70 days to reach 500TB.

Oh dear, I don't know what I was thinking: that SSD is surely dead by now.

PS: the 4-hour day got me, I ended up multiplying instead of dividing. My sincerest mathematical apologies, especially considering the entire little joke is ruined.

 

[whm1974]

I highly doubt that the OP has anything to worry about as he is probably not even writing a TB of writes to his SDD everyday.