Which 4TB NVME Gen 4 for gaming

Apart from the QLC drives (NV2 & P3 / P3 Plus), I think you're good, just check if you need single/double sided, since this capacity is very large.

Personally, I'd take WD's SN850X or a Samsung 990 Pro over the Phison controllers which Sabrent and Corsair use.
 
Apart from the QLC drives (NV2 & P3 / P3 Plus), I think you're good, just check if you need single/double sided, since this capacity is very large.

Personally, I'd take WD's SN850X or a Samsung 990 Pro over the Phison controllers which Sabrent and Corsair use.
I've just bought an SN850X 2TB for a game drive as they are pretty cheap now. You're not saving that much with the cheaper drives now prices have dropped.
 
Last edited:

Seagate FireCuda 530 is a reasonable price at the moment for 4TB.​


I wouldn't rule out a P3 Plus to be honest. It's endurance is low but you're unlikely to exhaust it in its lifetime and they're pretty cheap for 4TB.
 
Last edited:
Yeah i saw this morning for about £215.. Great deal. That said if you already have a reasonably quick OS drive then something like the 4Tb P3 Plus can currently be had for less than £150.

Be quick tho as prices on nvme are on the rise, particularly Samsung.
 
If you can tolerate 2gb I'd highly advise the wd black sn750. Why? Because not only are they very reliable and fast but they have insane endurance. They have twice the endurance of a Samsung drive. The sn750 has a tbw rating of 2400 tb. In other words its good for 2400 terabytes of writes. The closest Samsung equivalent is only 1200 tbw.

Buy hang on. Isn't that a pcie gen 3 drive? Yep. Those drives read at 3400 mb/s and write at 3000 mb/s. That is MORE than fast enough for gaming. A higher gen for gaming is a waste of money and effort. Above gen 3 for gaming is at the point of diminishing returns. You will see no difference in load times between gen 3, gen 4 and gen 5. That's because there is more to loading a game than just reading from a disk. Truth is most games will barely hit top gen 3 speeds never mind gen 4. The ONLY way you can get full use of gen 4 is if you were transferring large files from one gen 4 drive to another.

ALL gen 4 drives are HALF the endurance of gen 3. For example the sn750 has a tbw rating of 2400 tb where as the gen 4 variant of the same drive, the sn770, only has a tbw rating of 1200 tb. Its the same with Samsung gen 4 nvme. Its half the endurance as well. That's because of the temps they hit and the way the speed hammers the nand. So if you want a drive that's going to last many years for gaming then wd black gen 3 is def the way to go.

My own experience of these drives. I have a 1 tb sn750 as a boot drive and a 2 tb sn750 for games. I've had the 1 tb one for about 2 years and the 2 tb sn750 for about a year. I constantly keep swapping out game installs (because I literally have hundreds of pc games) and yet both drives are still showing 99% health in both crystaldiskinfo and the wd dashboard. How many years are these things going to last if I'm still at 99% on the 1 tb after 2 years of being a boot drive and hundreds of game installs?

Thats how good these drives are. That's why I went out my way to get them. The 2 tb sn750 is currently selling at 110 from wd themselves. If you MUST have 4 tb then why not get 2 2 tb ones? One as a combination boot drive and gaming drive for less hdd demanding games and a seperate 2 tb for the more hdd demanding games like starfield etc. That's exactly the way mine is setup. Gaming from a boot drive, you have to compete with Windows for access times so a separate nvme for more hdd intensive games is a wise move.
 
Last edited:
Does anyone ever get remotely near the TBW of an NVME drive? I'm a pretty heavy user and have never even remotely close (like not even 10%) before they get upgraded. Write endurance really wouldn't influence my decision much.
 
Last edited:
You probably won't see a difference for gaming between any of them perceptually - I don't know if they are going to bring out a 4TB variant but the SK Hynix Platinum has extremely strong game loading performance, though is more average in other areas.

I've been testing a 2TB SK Hynix side by side with a 990 Pro and perceptually there is little in it and TBH I don't really see much difference to the 970 Pro and 980s in my other setups when loading games.
 
Does anyone ever get remotely near the TBW of an NVME drive? I'm a pretty heavy user and have never even remotely close before they get upgraded. Write endurance really wouldn't influence my decision much.

It depends on how often you uninstall/install games and how long you plan to keep them for. Most games nowadays are 100 gb + and if you're constantly swapping games out... can you see my point? Also an nvme used as a boot drive gets hammered by windows as well, not just games. And then there's all the file ops. If you use your drives like I use mine then endurance becomes an issue. Also how long will these drives last? Why would I ever need to upgrade them? Maybe one day when gen 3 nvme is no longer supported but they'll still last beyond that. I can't ever see the speed being an issue and i shouldnt need to drop cash on a drive for many years now.

In comparison. I used to have a 2 tb sabrant rocket. That was a big mistake. Not only was it a buggy mess but the health dropped to 74% after just one year of use. Where as my 1 tb sn750 has been used in just the same way for 2 years as a boot and gaming drive. It has almost a full year of up time, constantly reading and writing and it still has 99% life left. What would you prefer? A drive that holds out for 3 years max or a drive that can easily last a decade and still be healthy?

I don't think this has any real world effect since we moved away from mechanical drives.

Actually it still has a big effect. For a start while gaming, Windows is constantly doing many reads and writes in the background, trying to keep the os alive, trying to keep up with the game and possibly even paging. (Yeah people called me a mug for going 32 gb of ram a few years back even though I pointed out many uses for it. "16 gb is more than enough". Em no. Its not. You're just starting to wake up to that.)

The point is, the boot drive is constantly swapping between tasks and even though the drive is an nvme it's still adding latency and even hitching to certain games and trust me if you try to run forza horizon 5 or starfield at max settings, 1440p and you only have 16 gb of ram and worse for forza, only 8 gb of vram, then your os WILL page like crazy, hitch all over the place and watch your load times sink like the titanic.

I actually have a post on here that highlights exactly that. Due to a bit of os corruption and a bad nvidia driver my nvme boot drive was being hammered with writes during gaming. This caused chaos tbh. Very long loading times, hitching, lockups etc. All because the drive was being hammered by a bad nvidia driver. That's why I put the more hdd demanding games on a seperate drive. That way the windows drive can do its thing while the game has unrestricted free access to a seperate drive. It's a valid point I'm making here.

I've got adhd. I've no patience. I want it to load fast and run silky smooth at the same time. Tbh who doesn't want that?
 
Actually it still has a big effect.
Evidence please. Don't believe it for a moment.

I actually have a post on here that highlights exactly that. Due to a bit of os corruption and a bad nvidia driver my nvme boot drive was being hammered with writes during gaming. This caused chaos tbh. Very long loading times, hitching, lockups etc. All because the drive was being hammered by a bad nvidia driver. That's why I put the more hdd demanding games on a seperate drive. That way the windows drive can do its thing while the game has unrestricted free access to a seperate drive. It's a valid point I'm making here.
This isn't a valid point. A software issue crippling drive performance is not the same as having an OS on the same drive and you wouldn't fix the software problem by putting the broken software on a separate drive, you'd fix the software issue.

There's also a reasonable likelihood depending on your motherboard chipset that the 2 drives will be running over a shared bus anyway (like on X570 if you have more than 2 NVME slots).
 
Last edited:
Strange, Western Digital themselves have the 2TB SN750 down at 1200 TBW:
As does TPU's SSD database:
The 4TB version does have a 2,400 TBW rating.

There are drives which have higher endurance ratings like there Seagate Firecuda 530 and so on, but the Western Digital SN750 isn't such s drive.
 
Nonsense. Windows doesn't create substantial write loads (that will get anywhere near TBW levels in my lifetime) on any NVME I've ever used.

I take it you don't pay attention to what your os is doing with your drive? Loads of constant tiny little writes happening all the time. Then there's all the downloads, documents folders, windows updates, apps and their updates, the page file, the hybernation file etc. I could go on forever here. It all combines and adds to the total writes of your drive. It eats away at it more than you realise.

Evidence please. Don't believe it for a moment.

Cyberpunk, starfiled and forza horizon 5 to name a few. If I have them on my boot drive then I get occasional hitching and loads of microstuters. If you look at the frame timings, they're a bit all over the place. However if I run them from my seperate drive then they are silky smooth.

Take for example Cyberpunk. I can show you seperate screenshots of the game installed on either drive. One rts shows not only much smoother frame timings but you will also notice its a good 3 fps faster based on my save point. The ONLY difference is the drive its installed on. I've actually had a similar debate about this recently and posted exactly that. 2 screenshot in the same place showing different frame timings and fps. The guy I was debating this with couldn't answer that.

I can show you if you want? Let me reinstall cyberpunk on one drive then the other and I'll grab screenshot to show you. I challenge you to offer another logical realistic explanation for that.
 
Sorry I don't believe it without evidence, it sounds completely make believe.

And that's what the last guy said. OK. Fair enough. Guess I'm installing cyberpunk again and I'll show you the difference. Check back in the early hours or the morning and I'll show you what I mean and again, I challenge you to offer an alternative logical explanation for it. Again the only difference is the drive its installed on. Those frame timing graphs and the fps readout don't lie.

And fyi what I'm about to show you, I can name at least 4 other games with the same problem. However most of my games it's not an issue with. Just a select handful of games.

And btw, what makes you think that extra drive access wouldn't mess with it that way? Do you honestly believe that a nvme has unlimited full duplex bandwidth?
 
And that's what the last guy said. OK. Fair enough. Guess I'm installing cyberpunk again and I'll show you the difference. Check back in the early hours or the morning and I'll show you what I mean and again, I challenge you to offer an alternative logical explanation for it. Again the only difference is the drive its installed on. Those frame timing graphs and the fps readout don't lie.

And fyi what I'm about to show you, I can name at least 4 other games with the same problem. However most of my games it's not an issue with. Just a select handful of games.

And btw, what makes you think that extra drive access wouldn't mess with it that way? Do you honestly believe that a nvme has unlimited full duplex bandwidth?
By the sound of your previous issues I'd worry more that you have a system issue rather than a bandwidth problem. You're welcome to show your figures but you'd have to control for other variables as much as possible especially if analysing at the level of frame times:

- identical drives
- drives connected the same (not one via the CPU and the other via the chipset for instance)
- have to be careful for random background activity during each run (not just effecting drives but CPU cycles/RAM bandwidth)

I'm sure there's plenty of other factors but it's not impossible but it's easy to get it wrong.

The reason I doubt it has a major affect is looking at baseline on my main windows system, reads are less than 1MB/s and writes are typically in the regional of 2-50MB/s a second once booted up and I gave it another few minutes it's showing sub 1MB/s for both, mostly sub 100KB/s. I can't imagine any of that activity is going to cause significant issues in a game (especially one that is noticeable in the real world) and have never seen anyone make similar claims. Yes there will be spikes here and there with antivirus and other software but enough to impact on gaming I'm highly dubious.

I've just tried running a game installed on a second SSD to see what activity is like on my main drive - running MW2 I see usage on my OS drive I see 100KB/s read and write typically - which I was genuinely surprised at. Maximum peaks upto 10MB/s. Those aren't very impressive numbers.

Regarding endurance, my main drive was purchased 2 years ago and despite heavy use gaming/photo editing/WFH and the windows "hammering" it has a grand total of 13TB written, it'll take another 90 years of use to get that to the 600TBW limit. My oldest NVME drive is a 2TB 660P from 2019 ish has a grand total 12.9TB and is used as both game and photo storage - so that has well over 100 years until it hits it's really quite low TBW of 400.
 
Last edited:
By the sound of your previous issues I'd worry more that you have a system issue rather than a bandwidth problem. You're welcome to show your figures but you'd have to control for other variables as much as possible especially if analysing at the level of frame times:

- identical drives
- drives connected the same (not one via the CPU and the other via the chipset for instance)
- have to be careful for random background activity during each run (not just effecting drives but CPU cycles/RAM bandwidth)

I'm sure there's plenty of other factors but it's not impossible but it's easy to get it wrong.

The reason I doubt it has a major affect is looking at baseline on my main windows system, reads are less than 1MB/s and writes are typically in the regional of 2-50MB/s a second once booted up and I gave it another few minutes it's showing sub 1MB/s for both, mostly sub 100KB/s. I can't imagine any of that activity is going to cause significant issues in a game (especially one that is noticeable in the real world) and have never seen anyone make similar claims. Yes there will be spikes here and there with antivirus and other software but enough to impact on gaming I'm highly dubious.

I've just tried running a game installed on a second SSD to see what activity is like on my main drive - running MW2 I see usage on my OS drive I see 100KB/s read and write typically - which I was genuinely surprised at. Maximum peaks upto 10MB/s. Those aren't very impressive numbers.

Regarding endurance, my main drive was purchased 2 years ago and despite heavy use gaming/photo editing/WFH and the windows "hammering" it has a grand total of 13TB written, it'll take another 90 years of use to get that to the 600TBW limit. My oldest NVME drive is a 2TB 660P from 2019 ish has a grand total 12.9TB and is used as both game and photo storage - so that has well over 100 years until it hits it's really quite low TBW of 400.


I've no other issues in my system. I just had to run sfc /scannow and update my graphics drivers to fix the issue i had. It was mainly caused by a known bad nvidia driver that hammered the boot drive during gameplay logging thousands of false errors per second. It's all fixed and the drive is running as normal again.

Right i used starfield for this. I already had it installed and was able to easily swap drives with my installation (right click in steam then move). Not only is starfield a better game to demonstraight this (because it streams assets from the drive at run time and doesn't use windows read caching. Why do you think it wont work properly on a standard hdd?) but wouldn't you know it, windows update kicked in in the background, downloading a few updates. So this is a typical normal usage scenario with windows downloading updates in the background. Same game install files. Same drivers. Same conditions. Same save point. No mods. Nothing else but windows update running in the background. I tried to keep it as fair a comparison as possible given the variables.

Running from a seperate drive:

mFS4I6.jpg



Running from the boot drive:

xQZtmn.jpg



Fyi i caught the boot drive screenshot mid hitch. You can see gpu usage and fps dip a little in the boot drive pic. Couldn't have timed it any better if i tried. Can you imagine what that feels like during playback? All that hitching and stamering. So far i've identified starfield, cyberpunk, forza horizon 5, hogwarts and forza motorsport as being games that are susceptible to this in my computer.
 
You'd need the same Windows update to be running in the first run to be more valid otherwise there are multiple factors that could be in play (storage/RAM bandwidth/CPU cache/thread priority issue etc).

Forgive my cynicism but you could just also wait for a stutter to come along before capturing the second image.

I haven't got Starfield but will try with Hogwart's to see if this is replicatable.
 
Last edited:
You'd need the same Windows update to be running in the first run to be more valid otherwise there are multiple factors that could be in play (storage/RAM bandwidth/CPU cache/thread priority issue etc).

Forgive my cynicism but you could just also wait for a stutter to come along before capturing the second image.

I haven't got Starfield but will try with Hogwart's to see if this is replicatable.

It was the same update. There were 3. 2 silly small ones and a cumulative update. I'd been ignoring it for a while but that's it just been forced on me. Update finished the small ones first then moved onto the cumulative update. Both pics were taken during downloading the same cumulative update in the background. For various reasons (bt's stupid fiber to the box) i only get around 48 meg of internet so it took a while to download that update. I did this during that time to show how even the simplest of thigs in the background can mess with certain games frame timings. As i said, i did this as fair as i possibly could.

Eh that's more than 1 stutter it's showing. 2 on the graph, 1 mid stutter (will show on the graph a split second later) and you can see loads of micro stuttering and fluctuation as well. It's no where near as smooth or as stable all the way along. It's like that constant. If you compair pics (look at the shadows) you can see both were taken around the same time. I loaded the game, waited 10 seconds for it to settle then hit the steam screeshot button. I did that both times.

Please try hogwarts. You should litteraly see a periodic noticable hitch. Hogwarts isn't as bad but it still happens. Starfield and cyberpunk are the worst offenders here.

Look you doubted me. You asked for proof. I gave you it and yet you still doubt me. You're a stubborn one but that's not a bad thing. Question everything. Form your own opinions. Don't just trust some randomer on the internet. Although i will say im a 45 yo father of 3, not some silly little kid talking out his back side. With those 5 games it's def like that for me if i run them from the boot drive. Every other game works fine though. Just i need to play those 5 games on a seperate drive.

My theory is this. Based on data on how nvme duplexing works. So an nvme has 4 lanes but it's full duplex meaning it can read and write at the same time. However that doesn't mean reading AND writing get 4 lanes. No. Windows has to split the load over those 4 lanes meaning aprox 2 lanes will be used for reading, the other 2 for writing. It's not exactly that but is roughly about that. So many variables. But what happens is those games that need an ssd, if your drive suddenly starts writing, windows tries to split the load meaning your game just got less reading speed when it needs it so that's where the hitching and stuttering come from. That's just a theory though.
 
Last edited:
Back
Top Bottom