Page 1 of 1

HD vs SSD R-W test results

Posted: 2021.10.25. 03:45
by clif9710
This KingFast SSD is only 2 months old
This KingFast SSD is only 2 months old
KingFast HD Sentinel.jpg (401.1 KiB) Viewed 4577 times
I've used HDS for a long time on HD's and have come to expect uniform performance across the entire surface unless something is going wrong with a particular part. HD's have, for me, been very reliable despite long use.

Now that I have come to use SSD's I am surprised at their non-uniform test results. I have 3 1tB SSD's and each one of them has a unique pattern of response times. One of them has very substandard performance, about half of what one should normally expect from typical SSD specs.

See the three results attached. Can I expect to continue to see this variation from SSD to SSD? The T-Force SSD that performs even better than spec and fairly uniformly throughout is brand new.

Re: HD vs SSD R-W test results

Posted: 2021.10.25. 10:22
by hdsentinel
I can confirm that similar test result is completely normal and expected for many SSDs.

Generally SSDs may have caches and other methods used to "mimic" a very fast speed in quick benchmark tools.

As you can see, when a such intensive test used which writes high amount of data (during the complete write test) the SSD shows how long it can perform very fast (see the first area with light green blocks) but when its cache is full, then it constantly need to write/flush it - which is a much slower operation. This is why then you'll see many darker green blocks - as the SSD continuously need to perform writes to previous blocks too.

This is why you see the alternating pattern:
- light green blocks (where the SSD does not really perform writes, just shows that "I'm completed" but it only fills its cache)
- dark green blocks (where the cache is full and the SSD needs to perform the actual write of the previous blocks too)

Such image can show the real "power" (weakness...) of some SSDs, especially during high load - even if they seem lightning fast during a quick benchmark, because they usually designed to handle the traffic of the benchmark: their cache size is big enough to store what the benchmark tool may write (but the data is only written later, after the actual benchmark finished...)
You can click on the resulting speed/performance on the Temperature and transfer speed tab.