Enabling RAPID Mode in Samsung Magician?

[Ronin13]

Running Samsung Magician's own Performance Benchmark on a 850 EVO, here's how it affected my scores (all with 'OS Optimization' set for 'Maximum Reliability'):


Without RAPID Mode:

Sequential Read: 549 Sequential Write: 528

Random Read: 96839 Random Write: 88940


With RAPID Mode:

Sequential Read: 5879 Sequential Write: 5166

Random Read: 206342 Random Write: 161155


I'm just wondering if there are any reasons not to keep it enabled? Specifically with regards to drive longevity?

 

[Puffnstuff]

Rapid mode is ram cashing your data. Just check the next time it's running and see how much system memory it's taking up. As for whether or not you should run it well that's up to you. I stopped using it because I only have 8gb in my laptop and need it for programs.

 

[Berryracer]

If you turn on CRAPID mode, you are not benchmarking your SSD anymore, you are benchmarking your RAM.

Translate that intro real world performance and it means absolutely nothing.

For example, test this.....

Copy a huge 10GB + video file or whatever large file you want from your C: partition to another partition on another disk......

you will notice that the Windows file copy progress finishes insanely fast.....the moment it finishes the copy, I want you to restart your system

then check that file you copied, it would be corrupt, reason is, yes the file copy progress finished fast, but it didn't finish really, all it was doing is copying the file from your SSD to the RAM Cache and not the actual 2nd SSD or HDD you were intending to copy to, then after it goes to your RAM Cache using CRAPID, it is supposed to copy from the RAM Cache onto the actual disk in the background which didn't happen in this test I did since I restart immediately after the fake file transfer progress was finished.

so it's just cheating + placebo effect

And after you read this, you will never enable RAPID again.....it will actually make your performance worse not better

A Closer look at the crappy CRAPID

 

[Ronin13]

Researching RAPID Mode I did come across that piece you linked to. But in the updated look at RM that Techreport links to at the end, it seems that their view on it is rather positive, no?

I only have the single SSD in this system, so I can't do the test you suggest. I'm aware that RM runs the risk of losing data if you lose power while there's still stuff to be written in the RAM.

I've seen similar arguments to yours, that RM is mostly just fooling (some) benchmarks, with no real, noticeable benefit, so that whatever risk it comes with isn't worth it.

Is that the general consensus here? To leave it disabled?

 

[redzo]

It has the potential to impact performance quite a lot, but I believe that you need to get very picky with your work loads:
http://www.anandtech.com/show/8216/samsung-ssd-850-pro-128gb-256gb-1tb-review-enter-the-3d-era/6

At the end of the day, for the average user, it's just a fancy free feature that looks too good on paper with 0 user experience improvement.
I myself am on 32GB of RAM and I can't tell whether RAPID is on or off. For me It is not worth the additional bugs/annoyances that enabling RAPID might introduce, as I don't perceive any improvement at all.

 

[coercitiv]

Is that the general consensus here? To leave it disabled?

First of all, since you probably require this context, berryracer is an ex Samsung SSD fan turned sour. At the moment he literally hates Samsung and loves SanDisk. He may be honest regarding his sentiments, and has every right to be so especially considering some of the more recent developments, but having strong feelings for brands does impede objective testing in the same way a broken compass always points North.

Moving on to RAPID, I'll quote one of my own replies regarding a similar software solution from Crucial:

It is a software solution similar to Samsung's RAPID. It does have it's perks and restrictions tough, and I wouldn't recommend it unless you have a specific workload that demands it. Another benefit other than those mentioned might be lowered write amplification.

To put it bluntly, for office work and gaming you will not notice a performance boost if you enable it.

If you really want to extend your drive's lifespan, the best solution in my book is to overprovision more. It will help keep write amplification low and also ensure better performance consistency.

 

[Coup27]

RAPID only helps in specific workloads. I am a light user these days and I tested RAPID for a few months a while back and saw no appreciable performance increase. These days I don't use it.

A user did post here once who said they ran a lot of VM's from a single SSD and RAPID increased the performance quite a lot.

berryracer's test of rebooting after copying a file is a valid test, although if you use RAPID and understand how it works and if it's worth it to you I think you can successfully navigate around using it without corrupting your data.

 

[Rubycon]

It's a compliment to Windows caching. Like "Fancycache" it will use more memory for delayed writes and other tasks to improve i/o performance in specific tasks.

There is no best test to use other than using your computer and timing how long tasks/batches/jobs take to complete. To be honest, prepare for disappointment with this method because the real world gains are seldom worth the trouble in the first place. Windows cache already does a good job and like in the DOS / Win 3.11 days (sigh!) one had alternatives such as PC Kwik, Norton Cache, etc. Smartdrv.sys seemed to just work and be left alone but the geeks and nerds (like myself haha!) would plug away for that holy grail number in whatever to magically believe the system was faster.

Well until local bus hardware caching controllers came along.

To see RAPID make a difference, disable Windows cache first then install and compare the two. TBH I'm not even sure this is do-able in the first place and you can bet it's not supported. When dealing with storage optimizations like this *always* follow the cardinal rules such as:

1) Never do this on a production system
2) Make sure your computer has a reliable power supply and UPS
3) If you have any overclocking applied, over-spec RAM timings, etc. Return these to default!

Or expect troubles to appear, blue screens, and data to go bye-bye.

But if you have a test system and no fear/consequence of reloading your OS, etc. Go right ahead, full speed!

 

[Berryracer]

Researching RAPID Mode I did come across that piece you linked to. But in the updated look at RM that Techreport links to at the end, it seems that their view on it is rather positive, no?

I only have the single SSD in this system, so I can't do the test you suggest. I'm aware that RM runs the risk of losing data if you lose power while there's still stuff to be written in the RAM.

I've seen similar arguments to yours, that RM is mostly just fooling (some) benchmarks, with no real, noticeable benefit, so that whatever risk it comes with isn't worth it.

Is that the general consensus here? To leave it disabled?

you can run the test / example I mentiond by copying a large 10GB file from your SSD to an external USB HDD and when you see the progress bar finish quickly due to RAPID, reboot right away and then check that file on the external HDD, you will see it isn't complete and is corrupt because all RAPID did was lie to the Windows progress bar to make it seem as it finished but all it did really was copy it to the RAM and from then it needs to slowly copy it at the normal speed from RAM to the external HDD in the background. Just a cheating placebo effect.

 

[Berryracer]

It has the potential to impact performance quite a lot, but I believe that you need to get very picky with your work loads:
http://www.anandtech.com/show/8216/samsung-ssd-850-pro-128gb-256gb-1tb-review-enter-the-3d-era/6

At the end of the day, for the average user, it's just a fancy free feature that looks too good on paper with 0 user experience improvement.
I myself am on 32GB of RAM and I can't tell whether RAPID is on or off. For me It is not worth the additional bugs/annoyances that enabling RAPID might introduce, as I don't perceive any improvement at all.

ditto! my thoughts exactly@ RAPID on or off have nothing to do with how my system's performance is. In fact, when I see a progress bar finish copying a file or something, I want to be rest assured that it has really finished not just copied to RAM and then sit and guess when the actual file has been copied back from RAM to the destination disk.

 

[Ronin13]

Thanks for your replies everyone, I think I'll leave it off for now.

 

[Rubycon]

you can run the test / example I mentiond by copying a large 10GB file from your SSD to an external USB HDD and when you see the progress bar finish quickly due to RAPID, reboot right away and then check that file on the external HDD, you will see it isn't complete and is corrupt because all RAPID did was lie to the Windows progress bar to make it seem as it finished but all it did really was copy it to the RAM and from then it needs to slowly copy it at the normal speed from RAM to the external HDD in the background. Just a cheating placebo effect.

And how does this relate to your everyday work?

It's called deferred write. Something that's been around for years. And all Admins know if you use it you better have a BBU on your controller. In this case a UPS on your PC. And as I mentioned earlier be absolutely sure if you have any overclock or memory tweaks to be sure it's stable.

Of course having 10GB worth of deferred writes is going to be risky IF proper precautions aren't taken. Even with a proper BBU/UPS in place if your RAM isn't error free or your system is unstable and freezes that un-committed buffer data is toast...

But that is the price one pays for speed. Not that the SSD isn't fast enough already.

Or willing to pay for it, get an intelligent host and strap in a bucketload of NVME devices and stripe them to digital oblivion. Hello Mr. 8GB/Sec R/W speeds! :biggrin: