Wondering if someone could explaine the differing benchmark numbering systems.

[tatermacandcheese]

Hello All,

I am getting ready to up grade my SDD from a 860 EVO to the new 870 EVO and wanted to run some benchmarks on the current SDD and I noticed the Magican software has what appears to be a way different bench-marking system than the hwbot.org website does.

I have attached snips of the two and wanted to know if someone could explain the numbering systems used in benchmarking, and what website should you use to keep it an apples to apples comparison.

I see the read/write is most common way to benchmark an SDD and the HWBOT uses the phrase "marks" along with other numbers is small boxes but I am just very confused with this process. It also seems to be different if you are benchmarking a processor instead of an SDD, and I am guessing they are benchmarked differently due to the different functions of each.

 

[StevenG]

Hello All,

I am getting ready to up grade my SDD from a 860 EVO to the new 870 EVO and wanted to run some benchmarks on the current SDD and I noticed the Magican software has what appears to be a way different bench-marking system than the hwbot.org website does.

I have attached snips of the two and wanted to know if someone could explain the numbering systems used in benchmarking, and what website should you use to keep it an apples to apples comparison.

I see the read/write is most common way to benchmark an SDD and the HWBOT uses the phrase "marks" along with other numbers is small boxes but I am just very confused with this process. It also seems to be different if you are benchmarking a processor instead of an SDD, and I am guessing they are benchmarked differently due to the different functions of each.

The Samsung magician is simple, Read = 4216 and Write = 3939 in MB/s then Random OPIS (Input/output operations per second) is Read = 116699 and Write = 57373 so, the IOPS value indicates how many different input or output operations a device can perform in one second, that is what all those figures mean.

I would use the "CrystalDiskMark 6" to do the SSD benchmarks, as its what all testers use testing SSD's speeds and is closer to the Samsung figures, the hwbot.org is a ranking comparison and is not a figure of how your SSD is performing per say, also install the TRIM command feature (through cmd prompt) that makes your SSD quicker and allows Windows 10 or any supported operating system, to notify an SSD which blocks of data are no longer in use and can be safely wiped out to be writable again. Having this operation done ahead of time improves performance, as the drive won't have to spend time erasing a particular block when space is needed to store new data, ensuring the SSD reaches its advertised lifespan.

Btw, I’ve done all the above and have used Samsung SSD’s both 2.5” EVO and NVMe M.2 for years and they are excellent and very reliable and quick SSD’s and I’ve had no problems and wouldn’t use any other drives.

 

[tatermacandcheese]

Hello StevenG !!

Thank you very much for responding, I almost forgot I made this thread it been like almost 10 days, lol!
I have two questions:

1) If the CrystalDiskMark 6 is similar to the Magician, that is "closer to the Samsung figures" why would you use it?

2) Is the Trim command the same Trim I have enabled in the Magician?

 

[StevenG]

Hello StevenG !!

Thank you very much for responding, I almost forgot I made this thread it been like almost 10 days, lol!
I have two questions:

1) If the CrystalDiskMark 6 is similar to the Magician, that is "closer to the Samsung figures" why would you use it?

2) Is the Trim command the same Trim I have enabled in the Magician?

These are the answers to your questions:

1) "CrystalDiskMark 7" and the other versions take eight measurements: the read and write speeds in megabytes per second (MBps) for sequential, 512KB, random 4KB and 4KB parallel disk operations to compare the true Read/Write performance figures with the vast data that testers use in their reviews, that is where its advantageous and helpful so you can compare the results, It works by reading and writing through the filesystem in a volume-dependent way. It generates read/write speeds in sequential and random positions with varying numbers of queues and threads. Solid-state drives tend to excel at random IO, as unlike hard drives it does not need to seek for the specific position to read from or write to.

2) Samsung’s "Rapid Mode in Magician" (with its so called TRIM) instead uses intelligence and other resources in your PC, namely the RAM, to cache files for quicker access and a more streamlined write process. While windows ‘TRIM CMD” notifies an SSD which blocks of data are no longer in use and can be safely wiped out to be writable again. Having this operation done ahead of time improves performance, as the drive won't have to spend time erasing a particular block when space is needed to store new data, ensuring the SSD reaches its advertised lifespan. As you can't physically make the drive faster, as the read/write speeds are bound by the limitations of the hardware.

The examples below is for one of my laptops an Aspire V3-571G (which is a pretty old laptop) with a 1x 2.5" SATA 6GB/sec 850EVO 500GB SSD drive and with 8GB memory (2x 4GB) at DDR3-1866 test done with "CrystalDiskMark 7" after enabling TRIM and disabling TRIM, note that when enabling/disabling TRIM do it with cmd prompt as "Administrator" and after each enabling and disabling of TRIM reboot computer after each CrystalDiskMark 7 test, that is how I did these tests with. You can see the massive improvement with TRIM enabled and Samsung Rapid Mode enabled also.

This is with TRIM for an EVO 850 enabled and Samsung Rapid Mode enabled

This is without TRIM for an EVO 850 disabled and Samsung Rapid Mode enabled

 

[tatermacandcheese]

Sorry for the looong delay in replying.

The issue and original motivation for this thread is that there are two sets of numbers, both three digit and four digit's long and depending on the person or company posting this information, either on a product review or a bench marking site is that they may use either the three or four digit sets of numbers.

And without the knowledge of what the difference is between the three and four digit numbers, which is the most meaningful pair in a real world application, (ie-the end user) and if the three digit numbers can can be somehow mathematically translated to the four digit, or visa versa, I feel it is useless to compare the benchmark numbers posted on sites or reviews with out having all the information presented as you have done here.

I feel I am heading in the right direction, in that I have more of the information I need to start doing the bench marking before I swap my drives and I appreciate the time you have spent helping me StevenG.