Does Passmark see ExpressCache on Samsung Series 7 Chronos?

[DaveD44]

I'm benchmarking my Windows devices with Passmark Performance Test 9.0 and the Samsung Laptop returns incredibly high disk performance numbers. The following numbers are the Samsung compared to my desktop with a (at the time I installed it several years ago) a high performance Samsung EVO 500GB SSD: sequential read 2270 to 490; Sequential Write 1769 to 440; and Random Seek + RW 1759 to 363. I also have a SSD main drive in the Samsung but it is even an older version than what is in the desktop. This Samsung laptop has an 8GB SSD soldered to the motherboard and uses something called ExpressCache to cache operations to the main drive. Device Manager identifies this SSD as a "Sandisk SSD i100 8GB".

Is Passmark aware of these kinds of front-end caches to the main drive and can it work around them? Is there something I can do to get it out of the picture for the test run? Or should I simply leave it there because PassMark is, after all, measuring performance as it ?exists? in the real world?. But it can't be measuring disk performance like in the real world in this case, because this laptop, even with its quad I7, simply is not that quick. And the desktop is incredibly fast, and shows it in the other tests. So what to do?

[David (PassMark)]

The Sandisk SSD i100 8GB doesn't seem very fast. So by today's standards it is pretty rubbish and probably slower than the actual drive it is meant to be a cache for.

More likely you are seeing caching in RAM. See also
https://www.passmark.com/forum/perfo...msung-evo-ssds

[DaveD44]

David, thanks for the response. So if I'm correctly reading the link you included, you turn off regular Windows disk caching when you run the disk tests but since Express Cache doesn't follow the rules with your request to turn off disk caching, your performance tests are not able to turn it off. So what I am seeing is the disk performance boost provided by Express Cache and, while it isn't very much of a boost as compared to Windows caching, it makes it look a lot better than disk operations without any cache at all.

So I went to the SSD i100 properties and disabled the driver and reran the tests. The results were about the same. I went back to the properties afterwards and the device was still disabled, so it apparently was not used during the tests. Whats happening? Is it that Windows caching is not being turned off since Express Cache was in the way of that to start with? Can I turn Windows caching off (along with Express Cache being off) to get a better idea of disk performance? If so, how?

[David (PassMark)]

Modern SSDs can in fact hit 2000MB/sec. So it isn't out of the question that you actually measuring the SSD speed. But it would need to be a good M2 NVME SSD.

Otherwise yes, the only conclusion is that you are seeing disk caching.

You can run some of the advanced disk tests to confirm this. If you run the disk test with a very large file (e.g. 4GB+) then there won't be enough RAM to cache it and then you'll see the real disk's performance.