Samsung RAPID mode

[Charlie98]

Does Magician need to be running for RAPID mode to work?

 

[Carfax83]

I don't believe that Magician needs to be running for RAPID mode to work. That said, I don't recommend enabling RAPID mode. It does nothing for real world performance, and only increases benchmark scores..

 

[Railgun]

That's a rather broad statement. It all depends on your use case.

 

[G73S]

No reason for Magician to be running at startup. Once RAPID is enabled, it creates its own startup entry. I even called Samsung to confirm and they said no need to run it at startup.

I love RAPID! Gives meh the benefit of RAID without wasting 1 TB of my storage since I have two 1 TB Samsung 840 Evos. they fast as hell

 

[Coup27]

No reason for Magician to be running at startup. Once RAPID is enabled, it creates its own startup entry. I even called Samsung to confirm and they said no need to run it at startup.

+1.

It creates it's own startup entry called "SamsungRapidApp" which in turn starts two background processes called "SamsungRapidApp.exe" and "SamsungRapidSvc.exe" but once enabled and rebooted, you are free to remove the Magician start up entry like I have done.

 

[996GT2]

RAPID honestly doesn't do much for real-world performance. It's more of a way for Samsung to boost benchmark #s to look better in reviews.

I have an 840 Pro and I've tried it with and without RAPID. No difference in normal use.

 

[Burner27]

I have RAPID enabled on my 500GB EVO and that drive is used for my VMs. It makes a world of difference when firing up machines when it is enabled.

 

[BonzaiDuck]

I have RAPID enabled on my 500GB EVO and that drive is used for my VMs. It makes a world of difference when firing up machines when it is enabled.

I'd want to double-check, but I think it needs about 2GB of free RAM for Rapid-Mode. It's another caching strategy -- this time between SSD and RAM.

 

[h9826790]

I'd want to double-check, but I think it needs about 2GB of free RAM for Rapid-Mode. It's another caching strategy -- this time between SSD and RAM.

Yes it is, but for those who have plenty of RAM, it's always good to fully utilise the hardware. It's free anyway, even though just a little bit virtual performance boost. There is no harm to use it (unless you remove the power straight away after the data write into the RAM but not yet write into the SSD).

 

[Charlie98]

No reason for Magician to be running at startup. Once RAPID is enabled, it creates its own startup entry. I even called Samsung to confirm and they said no need to run it at startup.

I love RAPID! Gives meh the benefit of RAID without wasting 1 TB of my storage since I have two 1 TB Samsung 840 Evos. they fast as hell

Well, that's why I axe, I don't care for Magician to be running all the time and I have it disabled on startup.

As far as 'real world' performance, my 840Pro is pretty solid out of the box, but I've got it on a Pentium G3220... so any slight gain in performance is... a plus!

 

[BonzaiDuck]

Yes it is, but for those who have plenty of RAM, it's always good to fully utilise the hardware. It's free anyway, even though just a little bit virtual performance boost. There is no harm to use it (unless you remove the power straight away after the data write into the RAM but not yet write into the SSD).

If it proves STABLE, then it can only be a blessing.

I think I was living in denial about a problem I was having, and refused to consider or attribute it to a SandForce SSD, ISRT, my BIOS and IRST version -- which is also a "caching arrangement."

There's probably more real promise for the Rapid-Mode feature. It's Samsung's own hardware -- exclusively; it's supposed to work well with Intel controllers. With ISRT, you pair a chosen SSD "for better or for worse" -- wedding it together with your chipset and BIOS version and the IRST software -- come what may. It's likely I could relieve myself of the ISRT glitches by upgrading my BIOS and IRST version. I may plan to do that, too . .

But it is unlikely I would have a problem with a single Sammy SSD in AHCI-mode with "Rapid-Mode" and the existing BIOS and IRST software.

So I'm looking forward to it. Very soon -- after I tweak my Windows "Indexing" and see how my event logs behave for another day or so.

 

[CakeMonster]

Does it in any way reduce the actual amount writing being done to the disk?

Is there any benchmarks that tell us anything about actual performance? Normal benchmarks seem meaningless here.

I've tried googling reviews but this feature seems to be largely ignored and hardly explored at all by the pc hardware community. That's strange considering how we usually obsess about these things..

 

[BonzaiDuck]

Does it in any way reduce the actual amount writing being done to the disk?

Is there any benchmarks that tell us anything about actual performance? Normal benchmarks seem meaningless here.

I've tried googling reviews but this feature seems to be largely ignored and hardly explored at all by the pc hardware community. That's strange considering how we usually obsess about these things..

I found CrystalDiskMark screenies posted at the Samsung web-site in either a promotion or "white paper" they made available for Rapid-Mode. Of course, this seems like getting your "FDA approval" from the snake-oil salesman, but they'd have no reason to fudge the results. Better -- would be a benchtest from TechReport, CDRLabs, BitTech or some established source.

This is just my recollection, but it looked as though RAPIDMode could garner as much as 1,200 MB/s in sequential read tests.

If it's any way better than the regular SSD spec ~500 MB/s, and if it's stable, I'm gonna do it . . . . With 16GB of RAM, I'm usually using << 50%. 2GB or so is not much to give up for an SSD speed enhancement.

 

[CakeMonster]

I found CrystalDiskMark screenies posted at the Samsung web-site in either a promotion or "white paper" they made available for Rapid-Mode. Of course, this seems like getting your "FDA approval" from the snake-oil salesman, but they'd have no reason to fudge the results. Better -- would be a benchtest from TechReport, CDRLabs, BitTech or some established source.

This is just my recollection, but it looked as though RAPIDMode could garner as much as 1,200 MB/s in sequential read tests.

If it's any way better than the regular SSD spec ~500 MB/s, and if it's stable, I'm gonna do it . . . . With 16GB of RAM, I'm usually using << 50%. 2GB or so is not much to give up for an SSD speed enhancement.

Yep but again those tests are the ones that are giving us misleading results since they are constructed not to take advantage of windows caching. It seems that new test methods would need to be developed in order to tell us anything about what RAPID really provides (unless I've missed some proper test when I googled for RAPID reviews).

 

[Burner27]

I'd want to double-check, but I think it needs about 2GB of free RAM for Rapid-Mode. It's another caching strategy -- this time between SSD and RAM.

It only uses 1GB of RAM max. I have 32GB in my machine so I am not concerned about the performance hit (if any).

 

[Charlie98]

It only uses 1GB of RAM max. I have 32GB in my machine so I am not concerned about the performance hit (if any).

I only have a measly 8GB RAM... but hardly ever break 4GB in normal use. I'm OK with it...

 

[BonzaiDuck]

I didn't plan on posting the following image on this forum until after I had implemented "RAPID." You can take it on faith, though only reasonable, that when I ran Magician's "Performance Benchmark" initially, it showed Seq Reads of about 540 MB/s and writes around 490 MB/s.

Then I was able to Enable -- Rapid Mode . . . and it gives the following result. I try not to "imagine things," and I was a bit stunned at the improvement it seems to provide -- loading programs, doing this, doing that . . . So if it's STABLE, it means a couple things: First, skepticism about bench vs real-world performance could be diminished by the "look and feel" of it. Second: those folks who chose to keep "RAID-Mode" in their UEFI-BIOS to allow other disks off the same controller to serve as RAID arrays -- they may be "missing out." You don't need a RAID0 of SSDs with this feature. If you need RAID with HDDs, you may want to consider getting a PCI-E controller card. You can't have RAPID without AHCI, and you can't have AHCI if you want to RAID other disks with the onboard controller.

 

[BonzaiDuck]

PS -- When I set up the Magician "optimization" features, there were three possibilities: Optimize for "performance," optimize for "capacity," and optimize for SSD longevity and "reliability." I chose "Reliability," knowing that it might depress my RAPID bench.

I'm good with it!!

 

[Charlie98]

I got results very similar to that. You can deselect RAPID in MSCONFIG, restart, and bench the regular mode if you want a direct comparison.

PS -- When I set up the Magician "optimization" features, there were three possibilities: Optimize for "performance," optimize for "capacity," and optimize for SSD longevity and "reliability." I chose "Reliability," knowing that it might depress my RAPID bench.

I'd be careful with Magician... I tried the Optimization utility, and it reset some of the things I had changed manually (outside of Magician.) I don't use Magician in that manner anymore.

 

[CakeMonster]

Those numbers tell us very little or nothing, as I've already said a couple of times. You need a benchmark utility that will compare the windows caching to Samsung RAPID. Not regular benchmark mode (which is NO windows caching) compared to Samsung RAPID.

 

[Charlie98]

Those numbers tell us very little or nothing, as I've already said a couple of times. You need a benchmark utility that will compare the windows caching to Samsung RAPID. Not regular benchmark mode (which is NO windows caching) compared to Samsung RAPID.

Do you know of such a utility? Curiosity overwhelms me...

 

[BonzaiDuck]

Those numbers tell us very little or nothing, as I've already said a couple of times. You need a benchmark utility that will compare the windows caching to Samsung RAPID. Not regular benchmark mode (which is NO windows caching) compared to Samsung RAPID.

If it helps, Samsung has a PDF white-paper on their web-site for Magician which presents scores that I'm pretty sure were CrystalDiskMark. I can only say there is a noticeable "feel" of improvement or responsiveness.

Caching schemes are an established way of bridging gaps between the different levels of the hardware architectural pyramid. With SSDs, the technology has opened up a bottleneck that was long-standing and traditional -- the gap between RAM speed and HDD response. The ISRT configuration and now RAPID tries to bridge this gap in one way or the other. But it's likely that RAPID could be more reliable than ISRT, since it doesn't admit the uncertainty of SSD choice and some other things. It is solely a Samsung product, and Samsung acknowledges they'd made an effort making their products work well with the Intel controllers.

It would seem logical that the Samsung RAPID feature takes the speed bump in the opposite direction from ISRT, which nevertheless gives you about 80% of SSD performance with a hard disk.

The only thing that occurs to me: I have to wait and see. Right now, I like what I see, but it's only been set up this way for a couple days.

 

[Essence_of_War]

I am extremely skeptical that it looks and feels faster outside of the placebo effect.

You're posting sequential read-writes that exceed the limits of the SATA controller, unless I have misunderstood something, I don't think you can actually be reading/writing that fast.

 

[BonzaiDuck]

I am extremely skeptical that it looks and feels faster outside of the placebo effect.

You're posting sequential read-writes that exceed the limits of the SATA controller, unless I have misunderstood something, I don't think you can actually be reading/writing that fast.

Actually, if you read the white-paper or other material as to what it does and how, this is no different than ISRT showing sequential reads three times what the accelerated disk is otherwise capable.

RAPID uses system RAM for caching the Samsung SSD.

On "placebo effect," I might wonder about that myself. But it isn't just my imagination.

Even so, I'd often said that a RAID0 of two SSDs showing seq-reads of +800 or 900 MB/s would not be that noticeably different from the 400 to 500+ speeds you'd see with a single SSD.

But in any event, the controller bottleneck is "still there." It is just "superseded" by a data cache in RAM. And again -- this was always a common feature of caching strategies.

 

[Essence_of_War]

Doesn't the OS already have to cache all reads and writes in RAM anyway?

It sounds to me like what is going on here is RAPID is doing something that the OS does anyway, then claiming credit for it by measuring R/W speed on the wrong side of the SATA bus.