SM951 as OS boot device information

Ohh   Sorry, I forgot about that, I had the same problem creating the Recovery Media.

Are you just asking about the Recovery Media creation process, or actually using it? I created it but did not try it yet.

Here's what I did during the creation process.  It gets to a point where it states no drivers found for the Intel SATA controller, right? I also tried to get it to load the IRST 14.6 F6 driver, used during the Windows installation, but it kept saying no drivers found like you did.

So I decided to ignore the missing driver warning, and it will continue creating the Recovery Media on a USB flash drive. When it was done I looked at the folders on the USB flash drive. Guess what I found? A folder with the name f6flpy-x64 among the other folders on the drive.

I can't recall what F6 driver version it was (maybe 14.5?), but I just deleted that folder and copied the 14.6 F6 driver folder in its place. You have a copy of that, right?

You can also see that there is a Drivers folder on the flash drive. Opening that folder shows Disk, Network, and USB folders. All of those folder were empty for me. I'm wondering if those are the folder locations where the drivers that failed to load during the creation process are meant to be.

I don't think copying the 14.6 F6 driver into the Disk folder too would hurt anything. In the same folder is a SearchPaths file. I wondered if that needed to have a line added for the Disk, etc folders. I think that file is used to search for drivers on other drives besides the Recovery media drive, so not needed.

Does this make sense? If you have questions, please ask.

I thought of something else today, about the failure of the RAID 0 array when the RST PCIE Remapping option is Disabled. Recall that I thought removing the SM951s before starting the PC after a UEFI clear or update might save the RAID 0 array, once we could set RST PCIE Remapping to Enabled?

I don't think that will work at all. Do you remember that the RST PCIE Remapping option did not exist in the Storage Configuration screen until after the SM951s were installed in the board (with CSM either Disabled, or the Launch Storage OpROM policy was set to UEFI Only, and SATA mode set to RAID) and the PC was restarted at least once?

That's how it worked for me, and I could swear you wrote the same thing in an earlier post. If the RST PCIE Remapping options do not appear until an SM951 is installed in an M.2 slot, my whole idea is worthless.   See what I mean?

Newegg has 11/1/15 as the availability date for the Samsung 950. Usually Microcenter is slower to stock new products than Newegg, hopefully this time that won't happen. My local Microcenter is my second home...

RAPID is no big deal IMO. It is a Windows process with a delayed startup, so not running when the PC boots. Benchmarks with RAPID are simply that of your memory, since it is a DRAM cache as you know.

The other reality about RAPID is while benchmarks show super fast write speeds, that is the DRAM cache. Any saved data must still be written to the SSD, which is done at the speed of the SSD of course. Nothing can change that.

The 850 EVO is the fastest SATA SSD in 4K/small file read speed performance, which makes it great for loading an OS. PCIe SSDs have better 4K performance, but there seems to be a limit to loading Windows, no less than three seconds from POST beep to desktop. Even my PCIe SSDs can't seem to reduce that time.

For many things, you won't see any difference with a PCIe SSD. But some things are faster. For example, doing backups or cloning to and from a PCIe SSD is truly faster.

Curiosity got the best of me.  Microcenter had them in stock today so I grabbed the 250 gig version.  I started another thread and posted some quick benchmarks if you're curious.

Anyone got different results? Or a similar setup to check against?

TimH wrote: Here are two Crystal Mark Runs in Win7 64x, first is a SSD Boot Samsung 840 250ish meg holding the OS. The second is two Samsung 951 256mb M2 cards RAID 0. Empty Not sure why the SSD is kicking the RAID 0's numbers so badly. Shouldn't it be the other way around?

No, and it's very simple why that is.

Benchmark tests first write data to the device being tested, and measure that speed while it is being written. Then they read back the data it just wrote to measure the read speed. That cycle repeats depending upon the type of test being done. Such as, large file (128K+) sequential, or small file (4K) "random", meaning small files in different locations on the drive, or "Q32" tests, meaning 32 IO requests in a line, sent to the drive one after the other, instead of just one IO request.

Your 840 EVO is in RAPID mode, which uses your PC's memory as a RAM cache for the SSD. If you checked the amount of memory being used by your PC about one minute after it boots, you would find at least 2GB being used by the RAPID cache. That memory now receives ALL data that is "written to" the Samsung SSD first, which is then later written to the SSD itself.

The data written to the RAPID cache stays there until enough new data is written to the Samsung RAPID cache to fill up the RAPID RAM cache, and the older data is written over by the new data. That is, unless the older data has not yet completely been actually written to the Samsung SSD, at its actual write speed capability.

When the benchmark runs on a Samsung SSD in RAPID mode, all its test data is written to and read from the RAPID RAM cache. So in reality you've just ran a benchmark on your PC's memory, not the 840 EVO at all. If you used some RAM disk software to create a RAM drive, and ran a benchmark test on it, you would get the same kind of results.

In reality, your 840 EVO cannot write a large file at over 5,000MB/s. It can write at ~500MB/s, which is what happens when the RAPID RAM cache actually copies the data to the 840 EVO.

All benchmark tests run on a Samsung SSD that is using RAPID are not testing the SSD itself, but just the RAM cache created by RAPID.

So your PC's DDR3 or DDR4 memory is faster than two PCIe SSDs in RAID 0, which is not a surprise. Turn off RAPID mode on your 840 EVO, and run the benchmark again to see the actual speed of that SSD.

You can't use RAPID with SSDs in RAID arrays. But if you could, you would again be just seeing the speed of your PC's memory and not the RAID 0 array itself.

I know of at least one website that compiles benchmark test scores, that includes RAPID RAM cache results as if they were the true speed of an SSD!   Truly embarrassing!