CRYSTAL DISK BENCHMARK VER. 4.0.3 X64
Crystal Disk Benchmark is used to measure read and write performance through sampling of random data which is, for the most part, incompressible. Performance is virtually identical, regardless of data sample so we have included only that using random data samples.
In Crystal Disk Mark we see reads reaching up to 3.3GB/s and writes reaching over 1GB/s again. The 3.3GB/s read performance could be a fluke as this is a newer version of Crystal Disk Mark than we have used in the past and the previous 1.2TB Intel 750 only reached 1.6GB/s. 4K performance is also very strong reaching almost 49MB/s for 4K read and nearly 350MB/s for 4K write. Overall, the performance is quite impressive, let’s see how it compares in our other benchmarks.
The toughest benchmark available for solid state drives is AS SSD as it relies solely on incompressible data samples when testing performance. For the most part, AS SSD tests can be considered the ‘worst case scenario’ in obtaining data transfer speeds and many enthusiasts like AS SSD for their needs. Transfer speeds are displayed on the left with IOPS results on the right.
Our next benchmark, AS SSD, shows a very good score. A Total Score of 3403 points was achieved, about 400 points below the 1.2TB model. The Intel 750 400GB reached 2,041MB/s read and 1,000MB/s write speeds. 4K performance is a bit less than shown in Crystal Disk Mark, now at 43MB/s read and 203MB/s write. During the 4K-64Thrd test it hit 357K IOPS read, just as its 1.2TB bother and 171K IOPS write. Latency is also within spec at 0.016ms read and 0.020ms write.
The AS SSD Copy benchmark shows some impressive scores also, though a bit less than its 1.2TB bother.
ANVIL STORAGE UTILITIES PROFESSIONAL
Anvil’s Storage Utilities (ASU) are the most complete test bed available for the solid state drive today. The benchmark displays test results for, not only throughput but also, IOPS and Disk Access Times. Not only does it have a preset SSD benchmark, but also, it has included such things as endurance testing and threaded I/O read, write and mixed tests, all of which are very simple to understand and use in our benchmark testing.
The 400GB Intel 750 showed very similar performance in Anvil Storage Utilities as well, with a total overall score of 11,492 points. Sequential reads are a bit shy of spec at 1,956MB/s and sequential writes are a bit higher at 1,008MB/s. Random reads reached over 152K IOPS and writes up to 234K IOPS.
Iometer is an I/O subsystem measurement and characterization tool for single and clustered systems. It was originally developed by the Intel Corporation however, they discontinued work on the program. In 2003 it was re-launched by an international group of individuals who are now continuously improving, porting and extend the product that is now widely used within the industry.
In order to attain the max IOPS we set 4KB random read and write workloads at a QD of 128 and tested for a 10 minute span. We can see that read IOPS averaged 443K while write averaged 36K. After about 30 seconds performance dropped off from 250K down to 20-25K IOPS, which is extremely good for a consumer SSD. For comparison most other SSDs will drop to around 10K.
Next we measured sequential performance and got an average read of about 2.4GB/s and average write speed of 1GB/s. The consistency of both random and sequential write performance is very good.
Thanks very much for this review – especially for also covering boot times.
Booting slower after post than expected and as samsung’s nvme drive is really a bit strange.
Maybe it simply depends on the driver? Which driver did you use for booting
test for the 750? Intel or Windows integrated one? Regarding
performance after boot there are some differences in speed depending on
the driver – maybe during boot too? Which driver was used for the
samsung? Many thanks again!
Hi, we used both the Windows and Intel NVMe drivers and both showed similar boot times after multiple trials.
thanks for this information. Sounds still strange to me – could you imagine any technical explanation for this? Why should the drive be slower during booting but faster after boot-process? The driver should be loaded right at the beginning so ist should work with full performance right from the start….do you have any contacts direct at intel to ask for?
I have reached out through my contact and will be sure to update you when I hear anything.
thank you – I’m excited 🙂
btw: have you experienced the same starting from sleep/hibernate?
More thorough analysis of boot times from various states should have been performed instead of just passing this solution off as a “workstation” platform and covering it in cursory fashion.
Well, it took me a few hours, but I just updated the last page to reflect hibernation and sleep resume times. Hibernation shows similar results to boot times while sleep is similar, yet still longer, compared to other SSDs. 🙂
awsome! Many many thanks! I think you are the only one in the web having analysed that!
(maybe you should add sleep and hibernate to the headline – so your site could be even better found for these results…)
I have received a reply: “it is
a known condition with the 750 Series. The firmware was optimized for reliability in the event of an unexpected power loss event. The boot delay was a side-effect of that change. Intel is currently
exploring options to make the boot time shorter in a safe way.”
Many thanks for this piece of information! So if they know it they can work on it 🙂
Would be awesome to have an eye on this if there is an update on Intel’s SSD Data Center Tool from https://downloadcenter.intel.com/download/23931 which would bring new versions of the firmware..
Hi Sean Webster, do you think, improving boot speed can be improved by upgrading the firmware?
Possibly.
Intel released a new version of firmware improves load time and the ability to initialize the device, whether the comparison at least for a short time after loading a firmware update?
source: https://communities.intel.com/message/335029#335029
We would like to inform that the Intel® Solid-State Drive Data Center Tool contains a firmware update for the Intel® SSD 750, the new firmware improves the Boot time of this drive significantly.
=> should be available soon
from https://communities.intel.com/message/335454#335454: “Just updated the firmware using the Intel SSD Data Center and boot time
went down from 15 seconds from the moment I see the Windows Loading icon
to 8 seconds!!”
sounds like this is the thing we were waiting for 🙂
@Sean: maybe you could add this as the fourth device to https://www.thessdreview.com/our-reviews/intel-750-pcie-ssd-review-400gb/5/ ?
I’d be interested in seeing latency tests done between NVMe SSD connected directly to CPU PCIe Gen3 lanes vs PCH Gen3 lanes on the upcoming z170 chipset. Do you have any plans on testing it out or do you think the differences would be so minor it isn’t worth it?
When we get a new test bench we will be sure to take that into consideration.
how about booting in non uefi systems like x58?
Not possible.
Are their motherboards/adapters that could put 4 of these into a RAID 10 and see even further speed benefits? I’d like to make a database server with Windows Server 2012. Not sure where to start.
You can use any motherboard that has support for 4x PCIe 3.0 x4 slots or more. Our X99 system we tested this drive in has 7 slots. Your RAID options are limited to software RAID. You can trick windows into doing a double software RAID set up like seen here: https://www.sgvulcan.com/2014/10/31/trick-windows-8-into-creating-a-raid10-stripped-mirrors-array/ I wouldn’t suggest it though. Too much overhead most likely. PCIe SSDs are not as flexible with RAID as SATA and SAS drives.
Hello,
I was reading this review and that of the Kingston HyperX Predator 480GB PCIe 2.0 M.2 SSD and I couldn’t fail to notice the huge difference between the writing endurance capabilities, while Intel 750 delivers 127TB writes (70GB/day), on the course of its 5 y warranty, the HyperX delivers a “staggering” 882TB writes (with an 1.7/day). I’m in the point of choosing on of them for my x99 system. My final objective would be: video/photo editing, gaming and last, but not least, running a couple of VMs (at least 5 VMs that are intended to simulate a Linux Lab and Exchange/AD environment). The main question is if this storage solution will last for at least 8-10 years if it only has a 70GB write/day?
Thank you in advance for your advice and response Sean.
Could you please provide a full Linux Kernel source build time ?
such as “time makepkg” with ArchLinux x64_64
As of 9/12/15 the Kingston Predator PciE ssd does not upgrade to windows 10 at all.
If boot times are important I guess it would be better to use PCIe SSDs for storage only, while having the OS installed on a SATA SSD.
You mention this SSD would be perfect for replacing a raid of SSDs used for media editing or virutal machine storage. Would it make a huge difference in terms of load times in certain games as well? Or would a high-end SATA SSD do the job just as good?
Just bought this drive and not seeing the speeds shown in this article. https://imgur.com/G2eBmYu is the speeds I am getting. Drive is plugged into a pcie3.0 x4 slot on a Dell T5600 workstation on a Windows 10 machine. Any pointers on what you needed to do to get the speeds mentioned in the article.
I have done the following attempting to get the speeds you mentioned
1. Installed latest intel nvme driver.
2. Installed intel ssd toolbox and verified that drive is using pcie 3.0 x4 channel.
3. Installed latest firmware.
4. Reboot after the above.
5. Switched to performance power plan.
Speeds I get are shown in following image – https://imgur.com/G2eBmYu
Any pointers on things I could try?
Maybee you just got a lemon
Here is an Video overview of this :
https://youtu.be/lH1Lexy61CM