2TB SanDisk Extreme PRO M.2 PCIe 3.0 NVMe Solid State Drive

That PNY CS1030 is an entry level NVMe drive, while this one is more mid-range performance oriented.

2TB SanDisk Extreme PRO NVMe SSD vs. 2TB PNY CS1030 NVMe SSD
Max Seq. Read: 3400 MB/s vs. 2100 MB/s
Max Seq. Write: 2900 MB/s vs. 1900 MB/s
DRAM: DDR4 vs. DRAM-less w/ HMB (Host Memory Buffer)
Endurance: 1200TBW vs. None provided (kind of shady)
Warranty: 5 years vs. 5 years

The SanDisk drive is superior outside of it's price premium. The CS1030's lack of DRAM is the biggest consideration: it will use your system memory (HMB) to store mapping data instead, but this typically doesn't help with read caching or write buffering, so it will fall behind compared to the SanDisk Extreme PRO there - particularly with large write operations. However, I bet most laymen wouldn't notice this without access to a side-by-side comparison, some benchmarking tools, and a stop watch.

Unless you're working with lots of huge data files (e.g. real-time video editing, the 4k speeds make a much bigger difference than the sequential speeds. Say you need to read 1 GB of sequential data (large files) and 200 MB of 4k data (small files). Which will be faster:

a NVMe SSD with 3000 MB/s sequential speeds and 35 MB/s 4k speeds
a SATA SSD with 500 MB/s sequential speeds and 70 MB/s 4k speeds

Obviously the NVMe SSD right? It's 6x faster at sequential reads and there's 5x more sequential data. Whereas the SATA SSD is only 2x faster at 4k speeds. 6x * 5x is more than 2x, so of course the NVME SSD will be faster, right?

NVMe: (1000 MB) / (3000 MB/s) + (200 MB) / (35 MB/s) = 0.33 sec + 5.7 sec = 6.05 sec
SATA: (1000 MB) / (500 MB/s) + (200 MB) / (70 MB/s) = 2.0 sec + 2.86 sec = 4.86 sec

What the heck happened? You'll notice the 4k read times are a lot longer (5.7 and 2.86 sec, vs 0.33 and 2.0 sec) even though there's a lot less 4k data. This is the problem with measuring drive speeds in MB/s - it's the inverse of wait time. And since it's the inverse, it's actually the smaller MB/s number which makes the biggest difference, not the bigger number.

This is why you can't tell the difference between a SATA SSD and a NVMe SSD most of the time. The huge sequential speeds of the NVMe drive (where the SATA drive is speed capped) don't really contribute much wait time, while the 4k speeds contribute a lot of wait time. So the overall wait time depends more on the 4k speeds than on the sequential speeds. And SATA drives can still compete with NVMe drives at 4k speeds.

So when buying a SSD, what you really want is the drive whose slowest speeds are the fastest. And since 4k speeds are typically the slowest operation, you really want to be comparing drives by their 4k speeds, not their sequential speeds. Unfortunately, these are rarely reported. You usually have to dig through product reviews to find these buried, and have to find results from identical benchmarking tools to get numbers which are comparable between drives.

(MB/s would be the appropriate benchmark to use if you had only x seconds to transfer as much data as you can. Almost nothing with computers is done this way. Normally you have x MB of data and need to transfer all of it no matter how much time it takes, meaning sec/MB is the more appropriate benchmark.)

That said, PNY and ADATA have been caught submitting fast drives for review, then quietly changing the drive to use cheaper (slower) components and selling them under the same model number. I would just avoid them altogether unless you're prepared to extensively benchmark them yourself and return them if they don't perform as well as in the reviews.


99.9% of users will never come anywhere near 1200 TBW. Most users only write about 10 GB of data per day or less. I just checked the SSD on my main laptop (which I use for several hours every day), and it's racked up just 12.5 TBW in a bit less than 3 years, averaging about 14 GB per day. At that rate, it would take 235 years for me to reach 1200 TBW. And my use does involve a little video re-encoding (shrinking home videos shot with my phone), so my GB per day is probably a little high.

I had a 250 GB SSD which I used in a security camera computer. It experienced about 300 GB of writes per day. When I retired it after 3 years (replaced the computer), it had racked up roughly 300 TBW. Even under that strenuous a use case, it would take 12 years to hit 1200 TBW. And 300 TBW was double the 250 GB SSD's endurance rating of 150 TBW. So more than likely a 1200 TBW drive in my security camera use case would last 20+ years. By then, a replacement 100 TB SSD will probably only cost $100, making the endurance a moot point. I will want to replace it by that point.

For most users, if the drive is rated over 150 TBW, it will last longer than the computer you're putting it in. 300+ TBW and it will outlast the technology (at least I'm hopeful that in the next 50-80 years we'll develop something better than flash NAND drives). 1200 TBW and it will last longer than you will live.

I'm seeing $329.99 for that 2TB 970 Evo+ in your link. This one is $274.99 which is $55 cheaper than that 970 Evo+.

Nice! And I get a downvote from a guy who cannot do basic math.