Any reason why my SSD would be going so slow?

no_1_dave

no_1_dave

no_1_dave

Soldato
Joined
7 Jul 2004
Posts
8,284
Location
Gloucester
This is my Crucial P5 2TB NVME drive which I use as a games drive. I've had the drive about 18 months or so but it's had little use because it's only some games from my Steam library sat on it.

Q5ssx26.jpg.png


I can't understand why it's so slow, sequential writes should be 10x those figures and all the read tests apart from the top one seem very off as well when compared to results people have online.

For reference here is the results from a similarly spec'd Sabrent Rocket that I'm using as an OS drive:

2uKq2zN.jpg.png


I've checked trim is enabled etc which it is by default on win10. My motherboard is an 'msi z390m gaming edge ac' which as far as I can tell should run both NVME sockets at full bandwidth so I'm not sure why the P5 is running so slowly... Do any of you have any ideas?
 
How full is it? 80%?

Have you checked the driver 'n stuff is working properly?

Looks kind of like what I'd expect if it was downloading an update, or something like that.
 
Snapshot said:
Does Crucial have a suite of tools like Samsung's Magician. Have you run CrystalDiskInfo?
Click to expand...

Info is fine in CrystalDiskInfo, health is all good. Not showing anything that stands out

Tetras said:
How full is it? 80%?

Have you checked the driver 'n stuff is working properly?

Looks kind of like what I'd expect if it was downloading an update, or something like that.
Click to expand...

80% is 371GB free out of 1.81TB
Driver is fine
It only has Steam games on it and Steam was closed when running the benchmark

Edit - Weird... I moved the folder with CrystalDiskMark onto the P5 instead of running it from C drive and the results have gone to what I'd expect and trade blows with the Sabrent.
It appears that as long as the program is being run from the drive it is testing it gives the expected results. I guess this means the bandwidth of the 2 NVME slots is shared
 
Another thing to check out is the temperature during the benchmark. From what I can see that motherboard doesn't offer any m.2 cooling so it could be thermal throttling. My Samsung m2 (also PCIe 3) previously throttled before on my older X370 board since it was right next to the GPU, but adding on a m.2 heatsink significantly helped.

Based on your latest comment, I'm wondering if the bottleneck could be how the m.2 slots are wired up. According to the manual both slots are chipset controlled, I'm not sure how the benchmark works but maybe there's not enough bandwidth on the chipset to share across the two as well as the CPU/RAM?
 
Quartz said:
Do I recall correctly that NVME drives can have problems if they’re too cold?
Click to expand...
There at least used to be an issue with them being too cold and it affecting performance.

As to the OP issue, benchmarks on ssd's can be 'variable' in my experience, one test will be fine, the next test could be completely different.... in all honesty there are a LOT of variables when it comes to ssd benchmarks and I'd only worry about it if you're actually seeing slow speeds in real world use.... but being 80% full isn't likely helping.
 
We had a Corssir SATA drive that slowly lost write speed after just a year of use, read speed fine. The drive was around 60% full. I don't think some last as long as they like you to believe !
 
digitaldreams said:
We had a Corssir SATA drive that slowly lost write speed after just a year of use, read speed fine. The drive was around 60% full. I don't think some last as long as they like you to believe !
Click to expand...
Some drives slow right down when they are nearing being full, Ive read of other people with problems when they run their SSDs too full, that means when windows shows the drive on red
 
You must log in or register to reply here.
Back
Top Bottom