Intel Optane 900P SSD

The Intel Optane SSD 900p 480GB Review: Diving Deeper Into 3D XPoint - AnandTech.com ( December 15, 2017)

"Our first look at the Intel Optane SSD 900p only included the smaller 280GB capacity. We now have added the 480GB model to our collection, and have started analyzing the power consumption of the fastest SSDs on the market."

"This second look at the Optane SSD 900p doesn't change the overall picture much. As we speculated in our initial review, the design of the Optane SSD and its 3D XPoint memory means that performance does not scale with capacity the way most flash-based SSD designs do. The Optane SSD 900p uses a controller with seven channels for communicating with the 3D XPoint memory. The difference between the 280GB and 480GB models is merely a difference of three or five 3D XPoint dies per channel."

Conclusions:
For the most part, the Optane SSDs are holding to their MSRPs, leaving them more than twice as expensive per GB as the fastest NAND flash based SSDs. They're a niche product in the same vein as the extreme capacity models like Samsung's 2TB 960 PRO and 4TB 850 EVO. But where the benefits of expanded capacity are easy to assess, the performance benefits of the Optane SSD are more subtle. For most ordinary and even relatively heavy desktop workloads, high-end flash storage is fast enough that further improvements are barely noticeable.

 

im gonna have a hardtime doing mod to make it work in 870tm-g man. no 12v rail on sata so i'll have to pull the 12v and ground from slave GPU connector from the mobo donno how thats going to work..

 

Happy Holidays and Merry Christmas!

It's not that high end storage has magically become 'fast enough' late into 2017...

It's that the PCIe connection used to that flash is too slow to exploit it further today. 3


All things being equal; I'd still take the 480GB 900P.

Take care.

 

Not exactly 900P but review of Intel Optane 32GB HardwareInside.de (de)

 

From the review of the Optane 900P SSD posted by @hmscott, @tilleroftheearth, and @Papusan...

"Both drives have a sequential read speed rating of 2.5GBps and a sequential write speed of 2GBps. Random 4K read speed is rated at 550,000 IOPS, while random 4K write speeds are rated at 500,000 IOPs. Latency is rated at less than 10 microseconds."

A quick injection on my part here... My new system is in the specs, and I'm getting some results with NVMe drives for comparison, so you can be the judge based on actual numbers:

Spoiler: Magician Benches

I also ran a quick test moving ~350GB of virtual machines from drive to drive on Linux. The copy took just under 3 minutes (about 160 seconds or so for ~2.2GB/s). Happy as a clam with the NVMe performance.

Now, I'm not going to run any tests like tiller would, so I don't know if thermal throttling will be an issue in my use case. However, with the M.2's mounted on the back side of my systemboard, and a small fan able to run against them if needed, I'm set. FWIW, without any extra cooling except what's in the case already, I haven't seen anything about mid to high 30C on the drives.

 

I don't know if your primary use is moving huge files. But I don't think this is the normal daily computer use for most people.
You could compare 4K speed and post your results CrystalDiskMark - AS SSD Benchmark

Intel Optane SSD 900P Review (480GB) – Understanding Disruptive Technology-thessdreview.com

 

AS SSD v2.0 is better than v1.9. It supports NVMe drives and possibly Optane as well.
CDM scores are out of this world. Even Samsung has Z NAND to play competitively with Intel/Micron.

 

I can't change the the review But although @jclausius now test with newest v2.0 it won't help much.

 

Yes. My use case is different than most here at NBR. While assisting our tech support, I can be handling 500 to 800 GB files in databases, virtual machines, etc. While it's not something I do 8 hours a day, 5 days a week, it comes up often enough that I was able to convince the powers that the ole x7200 is past its prime and to let me build The Mighty Muscular Mini.

NP. What would you like me to run Crystal Disk Mark, AS SSD or ATTO? What versions?

 

Use latest version if possible.

 

Download links in my previous post

 

I couldn't find AS SSD 1.9.x.x, but used the latest AS SSD (2.0.6485.19676). Some of the numbers were good, and some not so good:

Spoiler: Single Samsung 960 Pro 512 GB (Jan 3, 2018)

Spoiler: Single Samsung 960 Pro 1 TB (Jan 3, 2018)

---
Edit, I installed Intel Turbo Boost Max onto Win 10, and re-ran just on the 1TB 960 Pro. Slightly better scores -

Spoiler: Single 1TB 960 Pro - Run 2 (Jan 4, 2018)

 

Is this with Windows OS and Samsung Drivers? Check out your 4K scores vs. reviewed 960 Pro. Way too low. Maybe Phoenix can show up ASSSD tests with single 960 Pro.

Finally passed 2890 with 950 Pro


Damn @Phoenix I wanted 2900 Maybe another time?

 

I have not the scores for a single 960 Pro as my MSI BGA-Book comes with Super RAID 4 from the factory

But I can still smoke j00 bro

 

I can't see your scores on that crappy Photobucket. It never loads.

Use a proper hosting site like imgur or Postimage

 

You are smoked by your own old 950 Pro bruh Another stripe size, but still
AS SSD Benchmark: @Phoenix / Score: 4330


And my old slow single 950 Pro beat you in 4 and 8 KB in ATTO Benchmark. At least something good bruh

 

Use AS SSD 2.x to get some extra 100-200MB/s. BTW, how much TBs of writes as per CDInfo or HD Sentinel or HWINFO?

 

Papusan hunting old Phoenix scores from old threads >>>



Phoenix >>>

 

And you haven't said Amazing on my 2896 score above from the old 950 Pro Damn you bruh Sorry I haven't my old 950 Pro score from Raid0. Yoo would love it!!

 

bear in mind those 950 Pro scores was on my P870DM3 probably wish a much faster CPU that's probably the reason

 

BGA KILL ssd benchmarks? Put 4.7GHz and you are back on track

 

yes off-course, even when I had my P870DM3 the more I overclock it, the more the SSD Scores

Also, I have a crippled 960 Pro Firmware wait till Samsung releases a fix and let me take new benchmarks

 

Most likely will Samsung release a even more crippled firmware before they push out next refresh of their NVMe Pro series. Be careful bru, j00 can be screwed You are now warned by the Great Papusan. Your best friend in tech world

 

yes you are my best virtual friend who knows about me more than I do

This is the last time I upgrade a Samsung SSD Firmware. I never had anything but bad results with it and the ugly thing is you can't flash back to the original firmware!

Remember I told you that I get random full system freezes if I use IRST later than 15.2.0.1020? I tried every IRST after that, they all have this problem, turns out to be its from the firmware update I did a few months ago

 

YEES bruh.
Samsung and firmware is a lottery. You can at least use your ssd's. Better than bricked

 

read this:
Any way to revert back 960 Pro firmware to stock?

so I was the

 

Good luck bruh

Spoiler: I'm going to cheer you

 

Mr. Phoenix Since you have a high end system with RAID, I'd suggest you an experiment that I found out myself after looking at Samsung nvme driver inf.
Just make a quick diff. backup or full backup and try this app and change the IRST value to 2048(Samsung) or 4096(experimental but pushing the limits of the CPU and DMI3) https://github.com/CHEF-KOCH/MSI-utility/releases/download/2.0/MSI_util_v2.zip

 

Definitely Windows 10 x64. According to Samsung Magician, the Firmware is the latest, and each drive is using the Samsung driver. I installed RST, but that doesn't have any bearing on these NVMe drives. Nothing else really installed on the OS except normal drivers for the components in the system.

In regards to the other numbers on the benches, they are 'in the ballpark' to @Phoenix 's RAID... Again this is a single drive.

I couldn't tell you what is up, but not too concerned as I am mostly using the drives in Linux-land.


---

Edit - Was looking at some other reviews, and it seems like my numbers are in-line with what is reported here for the 960 Pro....

http://www.guru3d.com/articles-pages/samsung-960-pro-m-2-1tb-nvme-ssd-review,13.html

 

Bruh... First I've heard of photobucket (PB) causing any problems with thumbnails.

What is the problem? Do you see the thumbnails and the links don't expand? I could put in the larger images? Is your browser blocking something? Or perhaps PB's internet connection isn't 'global' friendly?

 

I do see the thumbnails bro but if I click on it it takes me to Photobucket and then shows a blank screen.

Edit: nevermind, it's working now.

 

On a whim, installed Intel Turbo Boost Max onto the otherwise clean system... Some slightly better scores...

Edited my OP - http://forum.notebookreview.com/threads/intel-optane-900p-ssd.810226/page-7#post-10657782

 

dont woryr phoenix, single SSD usually net you better "real" performance numbers in almost every aspect other than raw bandwidth because of lower latency.

think of ram single channel vs dual channel vs quad channel. latency goes up and the benefit is no longer 1x 2x 4x, but bandwidth still reaches pretty damn high. in storage, raid would increase overhead so higher latency and only real benefit is higher performance at lower queue depth past like 4-8, as well as good benchmark sequential numbers.

anyone that is interested should try changing these settings to help boost performance, below is stripe 16k but the tweaks will also work even at other stripe. reason i use smaller stripe is to get higher 4k performance. also note that i have purposely turned off cache write back so my random write isnt like 300-350MB/s+, this is to get an accurate comparison between difference in drivers.

for graph 4 (graph 1 is left most), i'd recommend Jon's review at tweaktown, https://www.tweaktown.com/guides/68...sd-performance-installation-guide/index8.html

 

Yes, you can.

The Samsung iso's contain plain binaries, even the new NVMe models. Extract the 1MB *.enc file from the earlier firmware and you can use that to flash the firmware eeprom with a programmer or, long shot, Intel's FPT.

Used to be you could simply use Samsung's own flash tool. There's also a tiny DSRD.ENC file that has the name of the ssd + fw-version-to-flash and the Samsung tool checks this against the current firmware version: if that is newer then it'll refuse to proceed. Those earlier dsrd.enc's were 'encrypted' with a substitution cipher (Caesar used one!) and had a lot of extraneous lines, so a little value counting and you'd know which byte to replace with which different one, yielding a plain xml. Edit the fw-version-to-flash to some future version, re-substitute and the tool would happily write the older firmware, thinking it was upgrading. Samsung seems to have noticed; the 960 PRO version is as minimalistic as possible and has hardly any repetition:

Code:

1x...96
2x...54
3x...21
4x....0
5x....5
....176

 

Understanding SSD Advertised Performance and Its Purchase Implications – An SSD Primer

"In the end, it confirms something we always thought but just didn’t really understand. Large sequential read and write access (high sequential and 512kb) are utilized by the average user less than 1% of the time yet the most used method of access is smaller random write access as shown by the 4k write at over 50%."

"Manufacturers showcase high sequential disk transfer speeds in selling their SSDs solely for their lightning fast appearance yet, these high sequential speeds are actually the disk transfer method that is used the least at less than 1% in total. As much as we would like to lay blame solely on the SSD companies, the majority of blame actually goes right back to the new SSD consumer who knows of no other way to differentiate between the myriad of SSDs available other than through high sequential performance."

All I can see...

 

@Papusan - I think I ran out of bandwidth. Moved these over to imgur and re-edited the OP. Can you see the benches in that post now?

 

They are back

 

nice find, i think it's rather biased to say sequential for 512kb blocksize and above is 1%. also, we first need tools to be able to tell us what file is at what block size. most video/audio/picture are incompressible files are likely sequential and larged blocked, maybe, same with big game files but could easily be a mix of bunch depending on how game loads its files, sequential can easily turn into random 4k, 8k etc etc.

i have plenty of use for sequential read, because of ram disk and i run a lot of things in parallel, which means sequential stacks into multi QD. a 1MB sequential block size file would break into 8QD 128kb for windows to read it natively so of course something over 512kb isnt very often. image restore likely sequential, some random too.

 

@Phoenix

i did a bit more digging for raid stripe, this is from tomshardware.

• Performance
  Conventional hard drives deliver their best transfer performance when they read or write sequentially, repositioning the heads as little as possible. From this standpoint, it makes the most sense to select the largest stripe size available, especially if your hard drives are good at providing high throughput. However, this only works if the files stored or read are at least as large as an entire stripe. If you will end of storing millions of text files, Word documents, small spreadsheets or similar small files, small stripe sizes will help to distribute all files across multiple drives to keep throughput high.
• 
  
• Capacity Used
  The stripe size also defines the amount of storage capacity that will at least be occupied on a RAID partition when you write a file. For example, if you selected a 64 kB stripe size and you store a 2 kB text file, this file will occupy 64 kB. Obviously, the stripe size defines the minimum amount of data that a RAID controller will distribute files across its hard drives, in my example, 64 kB. As long as a file can be written onto a single stripe or a single drive, you won’t have any advantage from running this particular RAID array.

from reading this i got a few ideas, first is OS environment and most consumer usage like browsing, doing regular work are beneficial when stripe size is small. from the 2nd paragraph perhaps because when larger stripe size is used, more data needs to be read thus more latency, just a guess but as long as it make sense we can make educational speculation.

 

@Phoenix Have j00 seen ASSSD scores much higher than this from single 950 Pro ? Passed well over 2900 as I promised with big margin

 

https://webcache.googleusercontent....review/index10.html+&cd=2&hl=en&ct=clnk&gl=ca

theres some weird info in the benchmark, jon's review shown he got close to 2900 with much higher 4k read/write but this guy got a better sequential read only, not sequential write, and higher 4K 64QD read/write which doesn't make sense because of his lower 4k QD1 results than tweaktown.

also look at his latency, much MUCH higher than what jon's shown, i'd seriously doubt it since single drive has no raid involved means no stripe size involved, everything is down to latency. when a drive latency result is almost half as the other while score isnt up to par, it makes no sense unless one of them is wrong. that or ADSSD is a messed up software.

 

From the review.
Have you seen similar 4K numbers from any single Samsung 950/960 NVMe before? And I have checked a lot of benchmarks from 950/960 Pro. None of them is near such a 4K score

 

yes, if you run it paired up with OS server it'll be very, very high. plus performance tweak with registry to ensure efficient cstate off, as well as disable ASM for DMI/PCIE devices would make it possible. Jon has an issue that he knows a lot of thing and test a lot of drivers, and done a lot of thing but dont necessary write in his review so i do understand people get suspicious. i've been following the guy since rwlab was still open like 7 yrs ago and let me tell you, when i saw his review at that time i thought it was bullsh*t numbers.

you also have to know his numbers are maxed out performance which majority of users wont bother tweaking, thats why he is a storage enthusiast. remember i showed you my CDM result of a single/raid PM961? 4k random I got 65 MB/s and 225 MB/s for read/write. well PM961 is probably on par with 950 pro in terms of performance except it uses TLC so when big write hits i get hit hard. im on server 2012 and if he uses 2008 r2 he'll get better numbers than i do for sure, cause u know win7 > 8 >10 and 2008 r2 is same as win7, except all server has much better storage performance than their consumer counterpart OS.

 

@tilleroftheearth @ajc9988 @Papusan @Phoenix

if optane 900p mod in laptop becomes a success, i am patching windows.
https://www.pcper.com/news/Storage/Meltdowns-Impact-Storage-Performance-Really-Issue

 

Same Q bro @Phoenix seen over 2920 in ASSSD with single 950 Pro? Where is my virtual cookie for this score?

 

My 960 pro

 

lmao ADSSD nice consistency. no more trusting on that software

 

I mean @Raiderman run AMD. Can this be one of the explanations for lower than normal ASSSD score for 960 Pro?

From the link "Well there you have it. Some gain and some lose. Given that a far lower latency device (900P) sees zero performance hit (actually gaining speed), I suspect that whatever penalty associated with Meltdown could be easily optimized out via updates to the Windows Inbox and Samsung NVMe drivers."

 

Your 4K R/W speeds are good. Did you try latest chipset drivers? 960 evo and PRO reviews https://www.notebookcheck.net/Samsung-960-Evo-and-Samsung-960-Pro-SSD-Review.189479.0.html
Samsung admits that SSD will be slower on AMD than intel.
I scored 2053 on PM951.

 

@tilleroftheearth @Papusan @Phoenix @camberman3000 @Meaker@Sager @Donald@HIDevolution

https://www.pcper.com/news/Storage/...s-Optane-800p-M2-2280-SSDs-60120GB-Capacities

CAN"T WAIT MAN