960 PRO 512GB (PCIE 3.0x4) & 970 EVO 1TB (PCIE 2.0x4) vs. 850 PRO 1TB (SATAIII & RAPID)

[Korporativ]

Lengthy title, I know. Can someone please make sense of these CrystalDiskMark results for me? RAPID or no, I find it hard to believe that the results I'm getting from a SATA SSD with RAPID are valid.

960 PRO (PCIe 3.0x4)
-----------------------------------------------------------------------
CrystalDiskMark 5.2.1 x64 (C) 2007-2017 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 bytes/s [SATA/600 = 600,000,000 bytes/s]
* KB = 1000 bytes, KiB = 1024 bytes

Sequential Read (Q= 32,T= 1) : 3394.625 MB/s
Sequential Write (Q= 32,T= 1) : 2043.884 MB/s
Random Read 4KiB (Q= 32,T= 4) : 1255.307 MB/s [306471.4 IOPS]
Random Write 4KiB (Q= 32,T= 4) : 1161.600 MB/s [283593.8 IOPS]
Sequential Read (T= 1) : 2624.401 MB/s
Sequential Write (T= 1) : 2022.181 MB/s
Random Read 4KiB (Q= 1,T= 1) : 56.754 MB/s [ 13856.0 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 195.778 MB/s [ 47797.4 IOPS]

Test : 1024 MiB [C: 36.5% (173.7/476.4 GiB)] (x5) [Interval=5 sec]
Date : 2018/12/08 19:37:12
OS : Windows 10 Professional [10.0 Build 17763] (x64)
=========================================================
970 EVO (PCIE 2.0x4) (Results to be expected due to PCIE2.0)

-----------------------------------------------------------------------
CrystalDiskMark 5.2.1 x64 (C) 2007-2017 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 bytes/s [SATA/600 = 600,000,000 bytes/s]
* KB = 1000 bytes, KiB = 1024 bytes

Sequential Read (Q= 32,T= 1) : 1622.235 MB/s
Sequential Write (Q= 32,T= 1) : 1604.716 MB/s
Random Read 4KiB (Q= 32,T= 4) : 1276.512 MB/s [311648.4 IOPS]
Random Write 4KiB (Q= 32,T= 4) : 1249.265 MB/s [304996.3 IOPS]
Sequential Read (T= 1) : 1528.196 MB/s
Sequential Write (T= 1) : 1527.987 MB/s
Random Read 4KiB (Q= 1,T= 1) : 48.049 MB/s [ 11730.7 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 167.890 MB/s [ 40988.8 IOPS]

Test : 1024 MiB [G: 0.0% (0.2/931.5 GiB)] (x5) [Interval=5 sec]
Date : 2018/12/08 19:57:03
OS : Windows 10 Professional [10.0 Build 17763] (x64)

==========================================================
850 PRO SSD - SATAIII with Samsung RAPID enabled

-----------------------------------------------------------------------
CrystalDiskMark 5.2.1 x64 (C) 2007-2017 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 bytes/s [SATA/600 = 600,000,000 bytes/s]
* KB = 1000 bytes, KiB = 1024 bytes

Sequential Read (Q= 32,T= 1) : 6289.936 MB/s
Sequential Write (Q= 32,T= 1) : 4906.678 MB/s
Random Read 4KiB (Q= 32,T= 4) : 1573.991 MB/s [384275.1 IOPS]
Random Write 4KiB (Q= 32,T= 4) : 1209.554 MB/s [295301.3 IOPS]
Sequential Read (T= 1) : 11346.504 MB/s
Sequential Write (T= 1) : 3789.692 MB/s
Random Read 4KiB (Q= 1,T= 1) : 537.205 MB/s [131153.6 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 332.991 MB/s [ 81296.6 IOPS]

Test : 1024 MiB [D: 78.0% (669.6/858.4 GiB)] (x5) [Interval=5 sec]
Date : 2018/12/08 20:16:37
OS : Windows 10 Professional [10.0 Build 17763] (x64)



I just fail to believe the results for the 850 Pro... is this some sort of artifact of the way that RAPID behaves making the results show false positive results?


Thank you again.

-Korporativ

 

[UsandThem]

I just fail to believe the results for the 850 Pro... is this some sort of artifact of the way that RAPID behaves making the results show false positive results?

That's normal with Rapid mode enabled. It makes synthetic benchmarks look amazing, and gives a small boost in performance for "real world" use. It basically uses RAM as a cache drive.

http://www.thessdreview.com/software-2/samsung-magician-4-5-rapid-mode-2-1-testing/

https://www.anandtech.com/show/7173...w-120gb-250gb-500gb-750gb-1tb-models-tested/5

 

[VirtualLarry]

With RAPID enabled, you are benchmarking your RAM, primarily, and not the physical SSD.

 

[Korporativ]

Ahhhh that makes so much more sense. So those are indeed falsely inflated numbers. Moving the data from my 1TB SSD should still theoretically see performance gains moving to my 1TB NVMe drive on PCIe 2.0x4, right?

 

[KentState]

The NVMe should double the write speed and quadruple the reads compared to a SSD if you are writing/reading large files. Once you get into random 4k reads, they start to equal out. The memory itself isn't that much different, but the controller and interface of PCIe is much faster than SATA, hence the benchmark improvements.