Intel Optane Memory H10 Review: Hybrid SSD Storage Acceleration

Due to the fact that the Intel Optane Memory H10 is destined to be an OEM part and won’t be available as a standalone retail purchase, we were seeded with an HP Ultrabook with the necessary BIOS and software installed to test the drive. The numbers posted in this review were not generated on our in-house storage test-bed, so they are not necessarily comparable to any of our previous storage-related review data.


That said, the HP Spectre X360 13” machine we were set up with was fairly high-end. The system featured an 8th Gen Core i7-8565U quad-core / eight-thread processor, with 16GB of RAM. For reference purposes, we compared the performance of the Optane Memory H10 – with and without Intel’s caching algorithm enabled – to an Intel SSD 760p standalone SSD that had the necessary OS image for the machine. We, however, installed all of the benchmarks and applications used for testing, except for Microsoft Office, which came pre-installed by HP.


We began our testing with some common benchmarks, just to set some baseline expectations. We’ve got three sets of numbers in each chart, one from the Intel SSD 760P and two for the Optane Memory H10 -- one with Optane Memory caching enabled, and another with it disabled. The reason we wanted to show you the performance of the H10 with its Optane Memory disabled is to present the “worst” cache scenario, where nothing is cached on the 3D Xpoint Optane memory.

Let’s start with the ATTO Disk Benchmark. ATTO is a "quick and dirty" type of disk benchmark that measures transfer speeds across a specific volume length. It measures raw transfer rates for both reads and writes and graphs them out in an easily interpreted chart. We chose .5KB through 64MB transfer sizes and a queue depth of 6 over a total max volume length of 256MB. ATTO's workloads are sequential in nature and measure bandwidth, rather than I/O response time, access latency, etc.

ATTO Disk Benchmark More Information Here: http://bit.ly/btuV6w

Intel's specifications claim up to 2400MB/s reads and 1800MB/s writes for the Optane Memory H10, but it doesn't come close to those numbers according to ATTO. Now, keep in mind, those specifications are for scenarios where both the Optane Memory cache and QLC NAND SSD are transferring simultaneously, at full tilt. Although we re-ran this test multiple times before recording our "final" result, to ensure the Optane Memory caching algo had seen the workload, the nature of the benchmark means it was essentially running entirely on only one of the drives and hence at about 1/2 peak theoretical bandwidth.

CrystalDiskMark x64 Benchmarks Synthetic File Transfer Tests

CrystalDiskMark provided some interesting results. In CrystalDiskMark's sequential test, both sides of the Optane Memory H10 are working in conjunction and we see sequential transfers in-line with Intel's  2400MB/s / 1800MB/s read / write specs. In the higher queue depth 4K transfer tests, the H10 exhibits relatively strong read performance, but actually trails in writes versus the SSD 760P and even the H10 when caching is disabled. In the QD1 4K test, the H10 shows just how strong it can be in the read test, which more than doubles the performance of the SSD 760P.

Futuremark's PCMark Storage 2.0 http://www.futuremark.com

We like PCMark's Storage 2.0 benchmark module for its real-world application measurement approach to testing. PCMark offers a trace-based measurement of system response times under various scripted workloads of traditional client / desktop system operation. From simple application start-up performance, to data streaming from a drive in a game engine, and video and photo editing with Adobe applications, we are comfortable that these tests reasonably illustrate the performance profile of SSDs in an end-user / consumer PC usage model, more so than a purely synthetic transfer test.