Currently £136 for the 480gig version on Scan.co.uk, very tempted as an upgrade for my gf`s rig, one thing i would like to know though, page 16 shows read/write speeds for various drives....were these reading taken when the drives were empty, reason for asking is i have in my own rig 2 Samsung 960`s, a 960 pro 512gb and a 960 evo 512gb, my pro which i use for os and a couple of games has 131gb free and gets 1800 seq read with As ssd, my evo which has 8 of my games on has 151gb free and gets 2100+ seq read, in both cases read/write speeds are no where near the empty drive speeds when i first in stalled them. So i`d like to see populated test reading for the mp 300 480gb please.

The Goose:

Currently £136 for the 480gig version on Scan.co.uk, very tempted as an upgrade for my gf`s rig, one thing i would like to know though, page 16 shows read/write speeds for various drives....were these reading taken when the drives were empty, reason for asking is i have in my own rig 2 Samsung 960`s, a 960 pro 512gb and a 960 evo 512gb, my pro which i use for os and a couple of games has 131gb free and gets 1800 seq read with As ssd, my evo which has 8 of my games on has 151gb free and gets 2100+ seq read, in both cases read/write speeds are no where near the empty drive speeds when i first in stalled them. So i`d like to see populated test reading for the mp 300 480gb please.

ADATA XPG 8200 is still better in low cost category. @Hilbert Hagedoorn : Are you keeping testing system up to date? If yes, it can pretty significantly affect results. Likely in negative way as newer patches for intel kind of degrade thing here and there. I am not saying that you should be testing on system without patches, that would be unfair for audience as they would see worse performance in their systems. I am saying that keeping all results perfectly fair with each other is quite some work for you.

I am really not seeing the point of these PCIe x2 drives if they continue to cost the same as budget x4 ones. The expensive part of an SSD is the NAND chips, so saving a couple bucks on a cheaper controller isn't going to make them cheaper.

My enthusiast side want to still get the 960 EVO at some point but my wallet says that this model is more worth it. 🙁 Also, HH, please, the copy and paste is extremely strong for the "Installation & Recommendations" part of the review. You give installation instructions for a SATA 6gbps SSD, not for an NVMe SSD. >.>

Lebon30:

e "Installation & Recommendations" part of the review

Ah yes, the correct NVMe snippet is now inserted. As to the other questions. Populated SSD testing is a good point, while I do not secure erase or format the SSD before each new benchmark run (as hey you do not do that in real life either) I do make sure the storage unit is empty. All SSDs and HDDs are prone to lose some performance once they get filled up. However keep in mind that there are more factors to consider, the biggest perf hits you see often are related to a boot drive, e.g. OS drive. Windows continuously do stuff these days and an SSD simply does need 20 minutes to recover. Also, power states have an effect on the SSD, e.g. if you proc is clocked down in desktop idle, your SSD can perform slower. I could consider loading the SSD 75%, but here again, if I would give the SSD time for trim, GC and recovery perf always has been back to normal levels in my experience.

Fox2232:

ADATA XPG 8200 is still better in low cost category. @Hilbert Hagedoorn : Are you keeping testing system up to date? If yes, it can pretty significantly affect results. Likely in negative way as newer patches for intel kind of degrade thing here and there. I am not saying that you should be testing on system without patches, that would be unfair for audience as they would see worse performance in their systems. I am saying that keeping all results perfectly fair with each other is quite some work for you.

I've answered that a couple of times already actually. Simply put: yes and no. Yes, but as little as possible. The SSD/NVME setup is not being firmware updated and automated windows patches on the rig - I patch as little as possible on it. However, the serious and mandatory microcode Spectre / Meltdown patches are in place (but show minimal perf decreases), and where it is measurable it is only in very specific workloads. Once hardware hardened processors are out I will update the storage test setup towards such a platform immediately. So really, the perf impact should be nominal and representable, however, I have seen some decreases in the 4K results. The latest round of L1TF gaps, I haven't patched these and am not planning to do so as really, I want to keep the test rig as balanced as possible. Both fortunately & unfortunately we tested over a hundred SSDs, most of them went back to the manufacturer, and cannot be retested. Neither would that be manageable time-wise as that is like a month of work. It's difficult to cope with, but yeah .. hey thank Intel there. I might jump with testing soon to a Ryzen platform though as it's less susceptible to the vulnerabilities and thus all patches, as recent each month there seems to be another vulnerability announced and patched at the Intel front. However, moving to Ryzen creates an issue with future possible Optane and Thunderbolt storage testing etc. I wish I had a better answer, it's just a clusterfuck series of dilemmas really.

The 480GB one would be perfect to use in my 2nd m.2 gen 2 slot. Already have a 960 evo 250gb in my gen 3 slot for OS and 3 SSD for game drives.

illriggerr And how many of those drives did you actually use yourself under real life conditions? There is a big difference between x4 and x4, and just because one gets faster benches, doesn't mean it actually will be. Most lower priced ones are actually getting so low read/writes that i question why someone would buy them (outside laptop or similar that is limited on connection/space. I rather have a x2 drive going full throttle the whole time, than an x4 that slows down to ssd speeds after 1s.