H2testw v1.4 or CrystalDiskMark 6.0.2 x64

photodgm

I bought 2 identical SanDisk SD cards and tested them both with H2testw.

Results for first:
Writing speed: 228 MByte/s
Reading speed: 249 MByte/s
Results for second:
Writing speed: 82.0 MByte/s
Reading speed: 86.6 MByte/s


I have been in contact with SanDisk (directed there by seller)
They wanted me to run CrystalDiskMark 6.0.2 x64


Results for both cards almost identical. These are for the 2nd card.
All MB/s
Seq Q32T1 Read 295.1 Write 217.9
4KiB Q8T8 Read 13.54 Write 2.741
4KiB Q32T1 Read 12.73 Write 2.668
4KiB Q1T1 Read 11.64 Write 2.641

Can anyone shed any light on this and suggest any way I can prove that the second card is faulty? (I presume it is as the H2testw took so much longer to finish on this card).

Neil_Jones

Well telling us which SD card, class and capacity you bought might help. Also what are you using it in?

But anyway SD card reading and writing speeds are usually listed as "up to" speeds.

photodgm

The cards are identical SanDisk Extreme Pro 64GB up to 300MB/s.
It wasn't the absolute values that confused me but the fact that H2testw showed large differences and CrystalDiskMark showed them to be almost identical.
I used a SanDisk ImageMate Pro reader connected to a Windows 10 PC using usb 3

Neil_Jones

Is this what you've bought?
https://www.amazon.co.uk/SanDisk-Extreme-300MB-UHS-II-Memory/dp/B01JYUU1ZU

But anyway H2TestW 1.4 is years out of date (it was last updated in 2008) and isn't optimised for modern big SD cards, so running that isn't going to prove anything.

If CrystalDiskMark (much newer and up to date) says the read/write values are near enough identical (within a reasonable margin) on both cards then I'd take that result over the 10 year old piece of software.

photodgm

That is the card in question. I know that H2testw is old software and takes time to run with large capacity cards however I have not found any information questioning its accuracy. It simply writes block after block of data to a card and times how long that takes. I don't understand the details of the CrystalDiskMark test.

Even if H2testw is suboptimal I would expect it to report similar results for identical cards. I have run the test several times on each card and the results are consistent.


I am thinking about trying some real world testing.

poppellerant

ATTO Disk Benchmark tends to be well regarded in terms of accuracy, so perhaps try that.

But I too wouldn't be too happy about two cards having such large differences in speeds. What will the cards be used for and is the speed an important factor?

[Deleted User]

poppellerant wrote: »

ATTO Disk Benchmark tends to be well regarded in terms of accuracy, so perhaps try that.

But I too wouldn't be too happy about two cards having such large differences in speeds. What will the cards be used for and is the speed an important factor?

That's presuming there is a difference in actuality as the very ancient H2testw v1.4 & the up to date CrystalDiskMark give differing results.

Paul_DNAP

How do they fair in a real-world test? Did you copy a bunch of files (various sizes) on them both and time it? Then copy them off again and time that? Any difference in the cards?

photodgm

poppellerant wrote: »

ATTO Disk Benchmark tends to be well regarded in terms of accuracy, so perhaps try that.

But I too wouldn't be too happy about two cards having such large differences in speeds. What will the cards be used for and is the speed an important factor?

Cards to be used in a camera capable of 11 frames per second. Since I often take photos of flying planes and birds in flight, the ability to take a number of long bursts in a short period of time is important to me. The camera needs to be able to empty its buffer to the card as fast as possible.

Paul_DNAP wrote: »

How do they fair in a real-world test? Did you copy a bunch of files (various sizes) on them both and time it? Then copy them off again and time that? Any difference in the cards?

I have tried some real world tests using the camera the cards will be used in. It is difficult to design a test that is fair, replicable and doesn't involve very short times such that reaction times are significant. I am beginning to think that the differences that the H2testw showed up are due to the size of the blocks of data used by the test. I'll do some more testing over the next few days.

poppellerant

photodgm wrote: »

Cards to be used in a camera capable of 11 frames per second. Since I often take photos of flying planes and birds in flight, the ability to take a number of long bursts in a short period of time is important to me. The camera needs to be able to empty its buffer to the card as fast as possible.

That makes sense and I too would want to make sure that I am using the best cards available to me.

Good luck with your testing and let us know about ATTO. Also, perhaps try formatting with various "allocation unit sizes" - there might be a sweet spot.