Sign in to follow this  
Adam_a

Intel SSD 660p Series Review Discussion

Recommended Posts

Intel’s new value-driven SSD is part of a larger portfolio that consists of much faster drives; however, the 660p is designed specifically to replace HDDs and SATA SSDs as the most viable and least expensive option for client-based systems.  With its current price point, it certainly makes this case, though the 660p’s performance is low enough to cater to only those looking for a minimal upgrade from an HDD-based workstation. Intel’s new line does offer some pretty decent reliability features as well, with endurance numbers quoted at 100TBW for the 512TB drive (~400TBW for the 2TB capacity) bundled with a 5-year warranty.  Users also can leverage a boost mode using the Intel Toolbox software, which manually flushes the SLC cache so inbound workloads are prioritized. 

 

 

 

Intel SSD 660p Series Review

 

Share this post


Link to post
Share on other sites

How did you get the impossible graphs? Graphs #2, 4, 5, 6, 7 require multiple measured values for some levels of IOPs or throughput. Assuming you threw IOPs at it, and measured the latency.

There's some minor glitching in the comparison drives too. But the 660p is the main culprit, with squiggly plots.

Share this post


Link to post
Share on other sites

It's not just this review. I started checking older reviews in Firefox, Safari, Chrome, and Opera. It's the graphs, not the browsers. Here's a typical one:

StorageReview-Adata-XPG-480GB-SeqWrite-6

My personal favorite, resembling kindergarten scribbles:

StorageReview-Intel-660p-1TB-SeqWrite-64

Am I the only person who noticed the corruption? Checking reviews, it began happening around January 2018.

Dec 22, 2017. Crucial MX500 1TB review. Graph is clean.

crucial_mx500_1tb_iops_workstation.png

Jan 10, 2018. Crucial MX500 500GB review. Corruption present.

StorageReview-Crucial-MX500-500GB-VDI-Pa

It appears the graph data avoids an elliptical region in the upper right corner. But the ellipse size and proportions vary by graph.

Edited by reader50

Share this post


Link to post
Share on other sites

Its not corruption, its a function of the test. Prior charts as posted are graphed on IOPS vs queue depth. The new charting method charts IOPS versus latency. With consumer drives as they get overloaded, latency spikes up and IOPS will actually pull back, causing the swinging motions.  These are charted with VDbenches curve profile. unrestricted workload starts, measuring the "100%" performance of the drive, then it scales from 10-120%, plotting the bathtub profile up to and through peak saturation. Some drives peak more gracefully than others.

 

Drive models that have very smooth and predictable lines are generally the highest performing. Drives that get bogged down and go nuts are the ones that lag on performance. For QLC drives we are actually working on an even smaller subset of tested drive surface to allow the drive to better cope with the workload. Those will be coming out soon.

Share this post


Link to post
Share on other sites

Thanks, that makes sense. As opposed to people not noticing a glitch for 8 months.

I couldn't picture how you'd get such graphs by setting either axis, then measuring the other. You're actually setting values on a 3rd axis (% of capacity) then measuring both X and Y.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this