The SSD Review The Worlds Dedicated SSD Education and Review Resource | 
TSSDR TEST BENCH AND PROTOCOL
For today’s testing we will be using our current test system that has been optimized with our SSD Optimization Guide. To see the best performance possible the CPU C states have been disabled, C1E support has been disabled, Enhanced Intel SpeedStep Technology (EIST) has been disabled. Benchmarks for consumer testing are also benchmarks with a fresh drive so, not only can we verify that manufacturer specifications are in line but also, so the consumer can replicate our tests to confirm that they have an SSD that is top-notch. We even provide links to most of the benchmarks used in the report.
We would love to thank those who jumped in specifically to help the cause. Key contributors to this build are our friends at ASRock for the motherboard and CPU and be quiet! for the PSU and cooling fans. Also, a big thank you to Thermaltake for the case and Kingston for the RAM. We have detailed all components in the table below and they are all linked should you wish to make a duplicate of our system as so many seem to do, or check out the price of any single component. As always, we appreciate your support in any purchase through our links!
SYSTEM COMPONENTS
This Test Bench build was the result of some great relationships and purchase; our appreciation goes to the mentioned manufacturers for their support in our project. All of the components we use for testing and evaluation can be easily purchased at a relatively affordable price. The links provided below can assist in pricing, as well as availability for those of you who may find interest in our equipment.
| PC CHASSIS: | Thermaltake Urban T81 |
| MOTHERBOARD: | ASRock Z97 Extreme6 |
| CPU: | Intel Core i5-4670K |
| CPU COOLER: | Thermaltake Water 3.0 Ultimate |
| POWER SUPPLY: | be quiet! Dark Power Pro 10 850W |
| SYSTEM COOLING: | be quiet! Silent Wings 2 |
| GRAPHICS CARD: | MSI GTX 660 Ti PE OC |
| MEMORY: | Kingston HyperX Beast 2400Mhz |
| SSD: | Samsung 850 EVO |
BENCHMARK SOFTWARE
To test RAPID the software we will be using for today’s analysis include Samsung Magician, BootTimer, PassMark AppTimer, and PCMark 8. We prefer to test with easily accessible software that the consumer can obtain, and in many cases, we even provide links. Our selection of software allows each to build on the last and to provide validation to results already obtained.
To start off our testing we chose to use Samsung Magician’s built in benchmarking tool to compare the performance of the drives. The darker blue bars are the results before RAPID was enabled on the system. The light blue bars are the results after RAPID was enabled.
Based on this synthetic test, performance seems to be night and day when comparing the drive with synthetic tests. Random read IOPS shot up to over 2x and write IOPS increased by 50%. Sequential reads and writes improved to 10-11x. From these results, impressive is an understatement. Now, let’s move forward to startup and application load times.
In order to test the startup time we used a simple program called BootTimer. BootTimer gave us consistent results on how long each system boot took over the course of testing.
Numbers don’t lie. Over the course of 5 system reboots, enabling RAPID results in slower boot time…while only slightly, it does have a detrimental effect.
APPLICATION LOAD TIMES WITH PASSMARK APPTIMER
For testing application load times we used PassMark AppTimer. AppTimer can measure the time it takes to open an application and automatically run multiple iterations of opening the application one after another. It then creates a log file with the results. Each application was launched for 5 iterations. Below are the results from launching popular software applications.
Based on these results it seems like Window’s default caching system already caches the programs into the cache after they are first launched. There is essentially no difference in application load time whether RAPID mode is enabled or not. So, we then rebooted the system and tested again, RAPID mode is supposed to retain the previously cached files after the system is rebooted, unlike the default Windows behavior which is to flush them.
We also noted when testing that RAPID will work 45 seconds after the system is booted and waited thus to begin testing again. This delay for RAPID mode can be seen in the memory tab of the task manager under Non-paged pool. Without RAPID the value is around 125MB in our system. Once RAPID begins working, the value changes to 1.3GB.
After testing multiple times, the results above bring us to the conclusion that RAPID does not in fact speed up application load times. Windows already does a good job at caching these files itself. However, one thing to notes is that while Windows already caches program files, Samsung’s algorithm is supposed to do it more aggressively to frequently used files. Let’s put that to the test.
For this we re-ran the iterations over 3 system reboots to ensure RAPID would have enough time to cache the files. The final reboot’s results are what are displayed. We can see that RAPID was able to load the Photoshop file initially after reboot a little faster than when RAPID is disabled, however after the first load time, it actually takes longer for consecutive load times.
Overall, the averages were practically the same. We are not seeing this claimed “almost 1.8x performance gains at Windows start up and application loading.” This is a bit of a disappointment as that is what they are essentially marketing RAPID to do. Startup times are slower and application load times were unaffected. Now, there is still more testing left, hopefully RAM caching will help in our next segment. Let’s move on to running light and heavy workloads to see if it really does increase performance in real world tasks.
Took 1.5GB I guess out of my 8GB pc. RAPID is really nice but I think I can live without it too. All that matters is that one has to move from a HDD to SSD for atleast the boot drive. Have the 850 Pro myself. Team RED we are 🙂
Amen to that muthafucka!
Hi Sean, what are your thoughts on RAID 0 Mode? I have 2 Samsung 120GB 850 EVOs. Should i leave them separate, first one being the system disk with RAPID enabled and second one for programs or should I RAID 0 them? I don’t care for the safety of the data, just speed. I do some prepress design and gaming. When in RAID 0 mode, Samsung Magician doesn’t seem to recognize them. Thanks for your help.
It is my understanding that the EVOs were never meant to be used in RAID configurations, which is probaby why the MAgician software is not recognizing them. For that reason, and better reliability, I chose to stick with the 840 Pro and 850 PROs.
owner of a 840 Pro, if you raid them you can’t see them in magician and trim is always disabled(stripe 0,1 etc) <<—no matter what model you have, same applies to 850 PRO
Trim has been supported for ssd raid arrays since 7 series motherboards (z77).. download the intel raid driver for your os and your golden. (driver can only be installed once in raid mode)
NOT in raid 0,1,5 ..ONLY raid mode but as single disks (TRIM is useless if you use RAID 0(stripe), or raid 1(mirror))
Do your homework and come back 😉
P.S. IDE, AHCI, RAID modes but nowdays only AHCI and RAID modes available
https://www.anandtech.com/show/6161/intel-brings-trim-to-raid0-ssd-arrays-on-7series-motherboards-we-test-it
This
article clearly states that trim is indeed supported for raid 0 arrays
on 7 series motherboards. Ive already done my “homework” sir.
Well that’s great, in the past the only thing intel did was to manage single drives in “raid mode” if they manage raid0 in 2 drives then kuddos to intel. Too bad you need specific motherboards to do that..
For me, I would set up which ever way is easiest for you. I would personally set them up in RAID 0 and store both the OS and programs on the same volume. Just make sure you do routine backups and you should be golden. Samsung Magician does not detect the drives as you have noticed, it is normal. I personally don’t bother with installing Magician in my own system.
Thanks a lot! RAID 0 it is! 🙂
you cant use RAPID in RAID…tried it
It might work if you use Windows ‘spanned volume’ ? So that’s not RAID, but essentially the same as RAID 0 (combining two disks without redundancy).
https://www.cnet.com/how-to/how-to-turn-old-hard-drives-into-one-large-drive-in-windows/
Spanned Volume is not the same as RAID 0. Spanning just writes to one drive at a time. Raid 0 uses two or more drives as one, so reads and writes are faster. A good two drive RAID 0 setup should increase throughput by 60%. So 540 transfer with one becomes 864 with two in RAID 0. A real good setup should almost double reads, writes, and transfer speeds. Spanning will not increase speeds.
my friend has a gaming pc with 2 1tb ssds in raid0 he said its the fastest storage he has seen (and he is a professional it technician). but it all depends on the type of raid controller. if its virtual raid I don’t think you should do it unless you really want all your space to be unified. but if its hardware raid you would probably get a faster write and read times.
just my opinion
Unless you are in an enterprise scenario. The raid 0 doesn’t affect real life performance, as most the time the SSD’s bandwidth isn’t your bottleneck. I have done a raid 0 with 850 pros and looked into upgrading my 850 pro to a 950 pro. Benchmark wise yeah the speed performance is way higher. But when i comes to boot time and application speed there is very little difference. Plus with raid 0 your data isn’t safe, and with Rapid Mode’s low latency, it becomes more of a want then a should. shabammm
Hi, so if the RAPID tech is ON, then I can get some performance gain in PC games for example or not really? 🙂 (I mean if these games are installed to this SSD & also this games are very often loading textures & other datas from this SSD where the RAPID tech is ON.)
Definitely not if you’re using a gaming computer. Textures are loaded into VRAM (memory on the video card itself) which is much faster than system RAM or the SSD.
Even machines without a video card will cache as much texture data as possible in System RAM.
I have 28 GB RAM on a HP Envy 700 430 qe. When I use the rapid drive, I get scores of 3700+ MB Read and 2400+ MB Write. Is that real? The computer is very fast with the 850 Pro 512.
In RAPID mode you are basically using a RAM drive that gets synchronized to disk as quickly as possible. Your RAM is very much faster than the SSD. But you risk losing the data in RAM if your computer loses power.
That’s what a laptop battery is for. To have no interruption when mains power is lost. I have a business class HP Workstation with two batteries on it for double security. On the workstation there is a lot going on to insure data loss is avoided. It is the perfect environment for the RAPID Mode, especially with 32GB RAM installed and a second SSD in the Optical Drive Bay. The way I work on a laptop makes data loss minimal too. Any data I lose will likely be progress since the last save. I am contantly saving my progress and so it will not be much. I also copy and paste so there is always a the original on the drive if the copy is interrupted – unlike with cut and paste when all the info is in RAM. It is tempting to cut a 29GB (Windows has 3GB reserved) chunk of info and instantly paste it and I don’t. RAPID mode pretty much does this and in a safer way. With all my interfaces SATA III running at 6MB/Sec and with 2 USB3.0 ports and a Sata port, transfering data is nifty, especially if I pack the info into a single zip instead of transferring individual files.
Backup transfers from the main SSD C: drive to the secondary SSD in the optical drive bay is is not noticed.
Where I see a big difference is any batch file processing, any repetitive process and during Photoshop CS6 Suite tasks which use only two Cores instead of all 4 because of sloppy programming and design, RAPID mode improves the speed. I have four 50gb scratch disk partitions on my 2TB C: Drive to double the processing speed of CS6 and any for any other program which needs to cache lots of temporay info on a separate partition.
The claim it can help boot up times is rubbish. Rapid Mode loads after windows has started up.
.
I have only 4GB RAM (DDR2) on a 6+ years old Lenovo Laptop (Core2Duo). The SSD is connected via SATA2 port. Even I got 1800 MB/s read & 1300 MB/s write speed, when rapid mode is enabled. But practically there was no other benefit other than the benchmark results (Other benchmarking tool revealed similar result too). So, finally, I disabled it as it took too much RAM space from my 4GB RAM.
You don’t really ‘lose’ that RAM, it’s actually a small RAM disk speeding along your computer. Even one with a SATA-1 MB would benefit, that is, it’s it’s an AHCI enabled one. Samsung tends to be picky, Intel SSD’s less so.
The RAPID app only uses 20-25% of your available RAM, depending on if it’s the 840 or 850 series, and in real world testing, I’ve found the RAPID app help SATA-2 computers more so than SATA-3, which usually has other built in features to boost performance.
With 4GB, at the most, RAPID will use 1GB RAM, and take my word for it, the RAM is NOT wasted. Your browsers, open tabs, everything you do is being boosted by RAPID. Plus your SSD gets less wear & tear, which doesn’t mean a lot these days as long as it’s aligned & not defragged, making it work less & your lower cost RAM do it’s job.
Cat
RAPID mode causes crashes–Windows 7,SP1, Samsung 850EVO. I enabled RAPID mode two months ago and ever since I have seen weird hangs and crashes every three or four days. Disabled RAPID two weeks ago and have seen no crashes since.
So with Rapid mode for the average user gain is…none, the same time you lose more than 1GB of RAM which is far more valuable for Gaming or Multitasking. Nice feature but I wont ever use it.
Good job I have 16gb of ram. Wooohooo I will use it all the time!!!!
32GB of memory here – this is really good for me.
Me too, so it doesn’t bother us at all.
you don´t lose anything.. if the memory is needed it will be released by the rapid software.
rapid only uses FREE resources.
if your game needs all the system memory it can get…then the rapid cache will be reduced to zero.
but it´s a useless feature anyway.
only good for PR.
I feel that’s a bit of an untrue statement, it does not release it. running games on very high settings with a gtx 1060 6gb of ram paired with 16gb of system and the games simply crash when trying to open up other programs while alt tabbed giving an out of memory error, upon inspection of win10 task manager it shows around 7 gigs inside the cache at the time of crash when it said it was out.
So if that was the case it would of released whatever was inside the 7 gigs of cache and allowed more memory to run whatever i was alt tabbed trying to do.
Thus what led me here in search of do I really want samsung to take up around 25% of my ram for—-? what gain in performance? and where. Far as I can see it’s a compromise if you don’t have TONS of ram, (32-64gigs), In this case trading multitasking for speed.
don’t care if the performance gain is minimal, I have 32gb ram 😛
Same here bro. No point of letting it go to waste right?
Wouldn’t those with SATA-2 based computers see as much as, if not more benefit, than one on a fast SATA-3 computer? Intel’s Rapid Storage Technology App is already built into many of these fast SATA-3 PC’s, so that’s why I see SATA-2 based computers as potentially benefitting the most from RAPID. Modern computers with fast controllers should need this as much as slower ones, which needs all of the help available.
Just a thought. I use the RAPID feature on my three SATA-2 based computers and see & feel a difference, whereas on my more powerful SATA-3 based PC with a 500GB Samsung 850 Pro, I really can’t tell the difference. In fact, seems to make the PC act funky more than anything, so disabled it & it’s naturally fast again. It’s not that way on my SATA-2 computers, the speed difference is there.
Cat
Rapid mode uses a different algorithm than Windows cache. It caches on block level, while Windows caches on file level. File level usually has more information, that’s why in most scenarios it’s better than block level hashing. But there are a few use cases where block level caching could be beneficial. For example, you have a virtual machine. From Windows point of view, it’s one monolithic 30GB file and Windows won’t cache it. But in fact, data moves within that file, and there is some pattern to these movements. A good block caching software can pick up that pattern and improve performance. Some other specific scenarios involving big data files can benefit too. But most Windows users won’t see any benefit.
still works perfectly on windows 10. just needs a little workaround to get it running
the fazit is pretty much nonsense.
you may can benchmark latency differences but the performance gain of RAPID (only under some circumstances) are so small that you won´t notice them EVER.
i would call myself a poweruser.
i am using premiere, nuke, 3d max, photoshop, after effects, solidworks.
i am always looking to make my system faster.. but rapid is not doing anything for me.
my own testing showed no benefits of rapid mode at all.
beside making low level benchmarks look better.
But what does this mean for games? there was BF3 and WoW listed but nothing was tested, is it more beneficial inside the games load times to have it enabled?, i know this is an old post just was a good read so had to ask.
Rapid can cause latencies of up to 31ms, which is a potential disaster for real-time usage, such as audio recording.
Most applications benefit from having MAXIMAL MB per second transfers speeds, but real time applications rely upon MINIMAL kB per millisecond deliveries to be responsive.
Maximum throughput is only achieved by running buffers full most of the time, which means EVERTHING is delayed.
It’s basic queuing theory. You get served more quickly at a bank if there is no queue, but the bank wants better teller utilisation which it gets by ensuring that there are always queues.
The drive manufacturers are like the banks in wanting greatest throughput for impressive specs, whereas your usage may require the fastest response. You need to make sure you choose the right drive criteria for your usage.