Are young SSDs faster than advertised?

[Ertio]

My experience: An Intenso "Premium Edition" external SSD (128 GB) is advertised with a read speed of 320 MB/s.

 

(source: Amazon, but I got it from a local electronics store.)

 

After purchase (roughly two years ago), I tested it and it clocked in as high as 452 MB/s over USB 3.0.

 

 

After more than two years of regular usage, it has decreased to 320 MB/s, exactly as advertised.

 

Its average speed over the entire 128 GB is exactly 320 MB/s. But during the test (read using the badblocks tool), it fluctuated a bit between 318 and 322 MB/s (visible inside iotop -o), possibly due to caching or similar.  But it was originally faster than advertised.

 

Is this an isolated occurance only?

 

Could it be that parts that were never written to are read faster?

Edited 9 hours ago by Ertio
added screenshot, improved wording, mentioned USB 3.0

[Spakes]

Generally SSDs tend to get quite slower when they get full, not from old age.

[Ertio]

  7 hours ago, Spakes said:

Generally SSDs tend to get quite slower when they get full, not from old age.

Also for reading?

 

Normally, non-sequential reading as a result of fragmentation hardly makes almost no difference for reading. This is the major strength SSDs have over HDDs. Fragmented reading (random access) could make it slightly slower, but not much.

 

Almost full data storage is slower for writing (SSDs too) because the file system driver has to work harder to find free blocks, but I don't think reading should make a large difference.

 

But my block-level reading tests were fully sequential, so fragmentation should not have affected them.

[Lurking]

  6 hours ago, Ertio said:

Also for reading?

 

Normally, non-sequential reading as a result of fragmentation hardly makes almost no difference for reading. This is the major strength SSDs have over HDDs. Fragmented reading (random access) could make it slightly slower, but not much.

 

Almost full data storage is slower for writing (SSDs too) because the file system driver has to work harder to find free blocks, but I don't think reading should make a large difference.

 

But my block-level reading tests were fully sequential, so fragmentation should not have affected them.

Expand  

Writing should be impacted more. But I assume with fragments of data all over, reading could be impacted a bit (to lesser degree). 

SSD also use different NAND. They use some SLC NAD for chance etc. Once they get full, there is less of the fast NAND left. 

 

the performance hit also applies to HDD. So i don't know why people make a big deal for SSD. an HDD starts writing in continuous blocks on the outside (fastest) of the platter. As it moves to the inside (slower speed) and data are fragmented all over, HDD get significantly slower reading and writing. De-fragmentation tries to combat that a little. But it can't overcome the problem, and the different speeds outside and inside of the radius still is a problem. 

 

What you see should be normal. the data published should be based on a say 50% full SSD, or HDD. So, a fresh drive should be faster. I don't know if manufacturers actually are that honest. 

 

Best advice, buy larger drives. that covers you for more years, and gives you more speeds. Larger SSD of the same model are faster anyway due to parallelization of the chips