How is the new 3D TLC NAND Intel 545s drive? Any reviews?

[VirtualLarry]

Just wondering if anyone knows much about this drive.

 

[Glaring_Mistake]

A few reviews of the Intel 545s: http://www.anandtech.com/show/11571...view-64layer-3d-tlc-nand-hits-retail#comments
https://www.pcper.com/reviews/Stora...-64-Layer-TLC/Internals-Testing-Methodology-a
http://www.tomshardware.com/reviews/intel-ssd-545s,5098.html

Seems to perform pretty good actually, trading some blows with the 850 EVO.

 

[cbn]

Very nice performance from the Silicon Motion SM2259 controller and the small die Intel-Micron Gen 2 (ie 64 layer) 256 Gb 3D TLC.

With the increased parallelism that small die brings I also wonder how well it would do with the Marvell 88SS1079 (or 88SS1074 controller)?

P.S. Also looking forward to seeing how Intel 545s (and other drives using Micron small die Gen 2 3D TLC NAND) compare to the WD Blue 3D and Sandisk Ultra 3D drives (which also use 64 layer 256 Gb TLC...but made by Sandisk, not Micron)

 

[Bouowmx]

So I presume higher capacity-to-price with 64-layer flash. So why is this product 512 GB/180 USD? Compare to Crucial MX300 525 GB/150 USD.

 

[VirtualLarry]

So I presume higher capacity-to-price with 64-layer flash. So why is this product 512 GB/180 USD? Compare to Crucial MX300 525 GB/150 USD.

I don't think that you "get" the NAND / SSD market.

Newer technologies are always bringing down the cost per bit of storage. But, at retail, to the consumer, they are "new", so they carry a premium.

Witness, on Newegg's ebay site, the Adata SU800 and SU700. The SU700 was introduced as a lower-cost 3D NAND-based SSD. But, it's "new". So, it carries a price premium to the consumer, that makes the SU700 (the supposedly lower-end model) MORE EXPENSIVE than the SU800 (a higher-end model).

Of course, it's not totally that simple, the SU700 also supports encryption, because it uses a different controller, and the SU800 does NOT. So that could be the reason too.

 

[Glaring_Mistake]

Very nice performance from the Silicon Motion SM2259 controller and the small die Intel-Micron Gen 2 (ie 64 layer) 256 Gb 3D TLC.

With the increased parallelism that small die brings I also wonder how well it would do with the Marvell 88SS1079 (or 88SS1074 controller)?

I don't think that's the only reason the MX300 fell a bit behind the 850 EVO.

First of all, their first gen 3D NAND was a bit slow but supposedly with the second gen their 3D TLC NAND is faster than their 2D MLC NAND.

Secondly some of the capacities contained NAND packages with different capacities (the 525GB consists of four packages, three with 144GB and one with 96GB).

Lastly I think they could have tuned the controller a bit better because sequential read speeds have trouble reaching even 500MB/s.
This I think to have been an issue with the controller since the UV400 also struggled with read speeds but the SanDisk X400/WD Blue (the UV400 and the X400 using the same 15nm TLC NAND) does not have such issues.
Thus MX300/UV400/X400 use the same controller (and the UV400/X400 the same NAND) but only one of them has normal sequential read speeds.

 

[Omar F1]

Unfortunate that Anand's review is missing the famous 'consistency' test for that drive.
Otherwise, I like its performance and definitely in my shopping list for a new drive.

 

[Billy Tallis]

Unfortunate that Anand's review is missing the famous 'consistency' test for that drive.
Otherwise, I like its performance and definitely in my shopping list for a new drive.

Of all the tests in our 2015 SSD test suite, that one is the least representative of real-world client use cases. It's a neat test for digging into some mostly academic details about how different controllers and firmware manage the flash, but it's not something that should be given any significant weight when considering which consumer SSD to buy (unless you're knowingly buying a consumer SSD for an enterprise workload). As such, it's my lowest priority to be updated for the 2017 test suite. The new performance consistency test will be added, but the 545s was just one of the many post-Computex releases to land on my desk in a short time span.

 

[Omar F1]

Of all the tests in our 2015 SSD test suite, that one is the least representative of real-world client use cases. It's a neat test for digging into some mostly academic details about how different controllers and firmware manage the flash, but it's not something that should be given any significant weight when considering which consumer SSD to buy (unless you're knowingly buying a consumer SSD for an enterprise workload). As such, it's my lowest priority to be updated for the 2017 test suite. The new performance consistency test will be added, but the 545s was just one of the many post-Computex releases to land on my desk in a short time span.

Thank you and appreciate your input.
Since first implemented and I had held that test as the holy grail of all, but no more.