Saturated IOPS Performance – 4KB, 8KB Random, 128K Sequential
*note* much of the writing on this page is a repeat of the 375GB review. Results for the P4800X are now the 750GB model tested in-house, and we re-tested the P3700 as a newer firmware had been released since the last review. Micron 9100 MAX results remain unchanged from the last review as there was insufficient time to get that drive re-tested for this piece (and we have no reason to believe those results have changed regardless). Spec 'X's have changed on sequential results as Intel have updated their specs.
I'm carrying over the IOPS vs. % Read charts from my P3608 review. The former IOPS vs. Latency plots also used in that review has been superseded by the far superior Percentile method (on the next page). With sweeps of R/W in 10% increments and all Queue Depths covered, there's a lot of data on each chart, so here I have listed the charts sequentially but matched the scales of each pair for easier A/B comparison.
Note that since we are plotting a Read/Write percentage spread, we no longer need to include other specific workloads (OLTP, Database, etc), as those workloads are included as a part of the below charts. For reference, here is the IO distribution of typical purpose-specific workloads:
- Database / OLTP: 8KB 67/33 (or 70/30)
- Email Server: 8KB 50/50
- File Server: 80/20 of the following:
- 10% 512B, 5% 1KB, 5% 2KB *
- 60% 4KB, 2% 8KB, 4% 16KB, 4% 32KB, 10% 64KB
- Web Server: 100/0 (read only) of the following:
- 22% 512B, 15% 1KB, 8% 2KB *
- 23% 4KB, 15% 8KB, 2% 16KB, 6% 32KB, 7% 64KB, 1% 128KB, 1% 512KB
* We have discontinued the File Server and Web Server tests currently used by many other sites, as they employ legacy workloads that are 16 years old (yes, in the year 2000) and are simply no longer representative of modern technology. Specifically, modern enterprise SSDs are no longer optimized for <4KB random, yet the outdated Web Server workload applies nearly half (45%) of its workload at those 'wrong' sizes. While it makes for an interesting spread in the results showing artificial penalties with SSDs optimized for 4KB, those results are just no longer meaningful in modern day enterprise use.
4KB Random
Alright, starting out with 4KB random performance, we see a very linear response between some very impressive numbers here. Anything over QD=8 turns into that saturation blob at the top line there. I've never seen anything ramp up on IOPS so quickly. Lets put this in a bit more perspective by adding in the Intel P3700 and Micron 9100 MAX:
Holy crap! The P3700 (green) and 9100 MAX (gold) are literally wiping up the floor at these lower queue depths, while the P4800X just walks all over them – even its QD=1 performance is higher than the other two at QD=4 nearly across the board! Let's look out to the longer queue depths here to see if they can catch up:
Ok, so singling out reads, writes, and a 70/30 mix, only the Micron 9100 MAX is able to beat the P4800X in 4KB random performance, but in order to do so it must operate at a QD of nearly 128 to reach the same level seen by the P4800X at 1/10th the Queue Depth!
8KB Random
8KB random performance is very much the same story as it was with 4KB, the only exception being the P3700 gaining a tad more ground but still falling short overall.
128KB Sequential
This chart now has the official specs noted (2GB/s writes / 2.4GB/s reads). Note that the P4800X takes just a single step at QD=1 before reaching its saturation throughput at QD=2. Insanity! (The second data line you see is actually all other QDs results overlapped.)
Even though Micron's 9100 MAX can reach higher sequentials, it requires very high queue depths to do so. While the P4800X may not climb as high as the others, it gets there at the far lower queue depths, which is where it really counts.
the endurance and performance
the endurance and performance are impressive, and those prices are impressively high too!
Is it possible to get optane drives with slower speeds and same endurance? I mean, it seems like it would be cheaper and I’d be ok with SSD speeds we have now, just that endurance is really nice. I would literallly never replace the drive due to endurance.
Why would making it slower
Why would making it slower make it cheaper?
They make Optane drives that
They make Optane drives that are significantly cheaper at a slightly reduced endurance. They are called 900P.
Those are significantly
Those are significantly cheaper as compared to the new optane drives but are still WAY more expensive than sata ssds.
I think the idea is if the optane drive is much slower and still really good endurance that because it is slower it would mean even cheaper pricing.
Think about it, the faster devices are faster because hardware is more expensive to drive those devices faster.
Any real world testing ?
Like
Any real world testing ?
Like is this worth using in compile servers and workstation ?
If this save me 10 minutes a day in compile time, I would buy it.
But IOPS numbers doesn’t say much…
It really is workload
It really is workload dependent, and as we've found in our other research on Optane, it varies wildly by application. No specific real-world test would give you your answer unless we just happened to test your exact application on your exact hardware configuration. That said, we did note significant performance increases in similar applications – they are documented in this white paper.
Further, you should be able to monitor storage activity for your particular workload on your particular platform. If access times are totaling 10+ minutes for what you are doing, there's a good chance Optane will bring that number down significantly.
Many thanks again,
Many thanks again, Allyn.
It’s very gratifying to see Optane graduate
from questionable promises to production devices.
Guys trust me. Intel is
Guys trust me. Intel is making leaps and bound progress in making Optane win. I work for them. This is just the beginning. Prod Spec will only get better from here. End of 2018 there will be a Optane memory product along with storage.