• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Is this a normal benchmark for an Samsung 960 Evo 1TB?

I just installed this new m.2 SSD into my computer and it doesn't seem to be any faster than my previous SSD RAID setup. I ran some benchmarks and all of them seemed about where you'd expect the 960 Evo to be except for the Random 4K read. I'm only getting 40MB/s read while I get 257MB/s write. Is this normal or could this be a defective SSD? Here are the other benchmarks for the drive:

-----------------------------------------------------------------------
CrystalDiskMark 5.2.0 x64 (UWP) (C) 2007-2016 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) : 3293.588 MB/s
Sequential Write (Q= 32,T= 1) : 1892.463 MB/s
Random Read 4KiB (Q= 32,T= 1) : 803.212 MB/s [196096.7 IOPS]
Random Write 4KiB (Q= 32,T= 1) : 698.754 MB/s [170594.2 IOPS]
Sequential Read (T= 1) : 1846.694 MB/s
Sequential Write (T= 1) : 1858.134 MB/s
Random Read 4KiB (Q= 1,T= 1) : 39.831 MB/s [ 9724.4 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 256.832 MB/s [ 62703.1 IOPS]

Test : 1024 MiB [C: 28.8% (267.9/930.8 GiB)] (x2) [Interval=5 sec]
Date : 2016/12/28 20:58:55
OS : Windows 10 [10.0 Build 14393] (x64)
 
Just off top of my head -- for something I may have read in recent days.

What would happen with a larger-than-4K block size? I assume the OP's default is 4K.

I myself am making this decision soon: to go pro or evo. The usually-quoted sequential read and write specs are just short of the pro.
 
Thanks for the help guys. I decided to return it. I think it's defective. When I cloned the original SSD to the m.2 SSD it took about 9 minutes. When I cloned the m.2 SSD back to the original SSD it took an hour. It shouldn't have taken that long. It was only about 200GB of data. I looked at the AT review of the m.2 drives and they look to be much higher than what I was getting. In fact, I used a different benchmark and it was showing 1.5MB/s for the random 4K read. I thought it wrong so I tried other benchmarks. They were all around the same except for the 1.5MB/s anomaly.
 
BTW, nice rigs BonzaiDuck!

I ordered a $25 Lycom PCI-E adapter after pulling the string for a 250GB 960 EVO M.2. The EVO cost me $130, no shipping or tax. I figure that this way, I can experiment with my SSD-caching from SATA SDDs and HDDs. But how to install, so that the SATA_1 and _2 ports don't encumber the bandwidth of the M.2? I'm even puzzled about the BIOS settings.

I was thinking "maybe a 1TB Pro or Evo" for boot-system and caching volumes. $620 or about $400 to have those boot-system benefits. I could even do it with a 512GB Pro (or EVO) with the Pro coming in at something over $350, but that would mean 2x 150GB volumes for dual-boot Win7/Win10, plus up to 200GB of 2x volumes for caching of either OS. A bit cramped, I anticipate.

Since I have a PCI-E x4 slot available, this $150 "investment" -- more of an "experiment" -- means minimum outlay to become familiar with these M.2 NVMe drives and -- with a little luck -- the experiment may be so much of a success that I can hold off for a year or longer spending that much on a 1TB drive.

As for the 6700K rig in general, I reviewed the trail-signs and indications. Board makers are typically volting their "Auto" vcore to run as high as 1.4V. It requires exactly 1.408V for stress-load and stability at 4.7 Ghz. Intel says their Skylake is "more resilient" to voltage damage. The 7700K will also fit in my motherboard with a BIOS update. So if I damage the 6700K, I could upgrade and exercise my usual care about voltages. But I think I'm right about this. If the worst happened, I'd simply order up a 7700K -- preferably binned @ something north of 4.8, and have it re-lidded with CLU.

77C (RA 77F) for affinitized LinX, max problem-size, max memory. I don't think I'll damage the 6700K from heat. So I found the sweet spot for 4.7 Ghz after only two detours to the BIOS menus.

With the GTX 1070 OC'd to run under Valley and other programs at 2,038 Mhz core 8,830 Mhz memory, this M.2 thing is the only major step remaining in my project.

Here's the LinX 25-iteration results for the second-pass voltage setting. Notice the GFLOPs column.

4_7%20Ghz%20target%201_408V_LinX-Loaded.jpg


Iteration 1 is usually an outlier, since it's already running when you set Affinity for the four cores. Other outliers could arise because of various system processes -- for instance, a scheduled AV scan or update. If you watch the system when these things occur, you can "pretty likely" identify the cause and mark the iteration as an outlier.

But here, the range of GFLOPs is 2 GFLOPs out of more than 200: 1 GFLOP on either side of the mean. This has to be . . . . "perfect."
 
Last edited:
I received my replacement drive yesterday and installed it. The random read 4K benchmark is better than the previous drive. Performance seems better although I could be imagining it.

-----------------------------------------------------------------------
CrystalDiskMark 5.2.1 Shizuku Edition x64 (UWP) (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) : 3310.793 MB/s
Sequential Write (Q= 32,T= 1) : 1880.042 MB/s
Random Read 4KiB (Q= 32,T= 1) : 784.903 MB/s [191626.7 IOPS]
Random Write 4KiB (Q= 32,T= 1) : 703.988 MB/s [171872.1 IOPS]
Sequential Read (T= 1) : 2147.933 MB/s
Sequential Write (T= 1) : 1820.191 MB/s
Random Read 4KiB (Q= 1,T= 1) : 55.149 MB/s [ 13464.1 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 253.494 MB/s [ 61888.2 IOPS]

Test : 1024 MiB [C: 29.4% (273.6/930.8 GiB)] (x5) [Interval=5 sec]
Date : 2017/01/11 9:48:24
OS : Windows 10 [10.0 Build 14393] (x64)
 
I received my replacement drive yesterday and installed it. The random read 4K benchmark is better than the previous drive. Performance seems better although I could be imagining it.

-----------------------------------------------------------------------
CrystalDiskMark 5.2.1 Shizuku Edition x64 (UWP) (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) : 3310.793 MB/s
Sequential Write (Q= 32,T= 1) : 1880.042 MB/s
Random Read 4KiB (Q= 32,T= 1) : 784.903 MB/s [191626.7 IOPS]
Random Write 4KiB (Q= 32,T= 1) : 703.988 MB/s [171872.1 IOPS]
Sequential Read (T= 1) : 2147.933 MB/s
Sequential Write (T= 1) : 1820.191 MB/s
Random Read 4KiB (Q= 1,T= 1) : 55.149 MB/s [ 13464.1 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 253.494 MB/s [ 61888.2 IOPS]

Test : 1024 MiB [C: 29.4% (273.6/930.8 GiB)] (x5) [Interval=5 sec]
Date : 2017/01/11 9:48:24
OS : Windows 10 [10.0 Build 14393] (x64)
Well, as another member points out, the benchmarks only tell you vaguely what you can expect with a particular usage pattern. It would all depend on whether your usage fits this or that combination of file activity. But the Random Read and Write on these drives is really phenomenal, when you think of what an SATA SSD would show. The Random Read 4KiB looks a tad low. But not that different from what you'd have with some SATA SSDs. IIRC . . .
 
Back
Top