MP All Models PCIe SSDs - NVMe & AHCI

[Earl Urley]

Speaking of firmware, there's a 12.3 firmware for Phison E12 based SSDs now out in the wild.

It's supposed to increase speeds.

Just ran it on a Sabrent 1 TB and a MicroCenter Inland Premium, non-destructive for me. Was already at 12.2 on both of them.

crjackson, want the link?

 

[jbarley]

Speaking of firmware, there's a 12.3 firmware for Phison E12 based SSDs now out in the wild.

It's supposed to increase speeds.

Just ran it on a Sabrent 1 TB and a MicroCenter Inland Premium, non-destructive for me. Was already at 12.2 on both of them.

crjackson, want the link?

Did you experience any speed increase?

 

[crjackson2134]

Speaking of firmware, there's a 12.3 firmware for Phison E12 based SSDs now out in the wild.

It's supposed to increase speeds.

Just ran it on a Sabrent 1 TB and a MicroCenter Inland Premium, non-destructive for me. Was already at 12.2 on both of them.

crjackson, want the link?

YES!

Send it please...

Is this the same link here

 

[crjackson2134]

Did you experience any speed increase?

I went ahead and ran the update from the link I posted earlier. It went flawless and yes, it did speed it up a little bit. I only ran one quick benchmark, and it was faster in AmorphousDiskMark.

 

[AlexMaximus]

Hi guys,

I noticed some minor delays in my system recently. My conclusion is some remaining spinner HDD's somehow clog up the system.
The Theory I came up with is to have no spinner in the system left, otherwise, there will be always delays because some data always remains on spinner HDD's, and they have to spin up to even read it, - hence waiting. I think the system should be a lot faster once any spinners have been eliminated.
Here is the question for the ones that have done this. Does a pure SSD System run faster compared to a mixed HDD/SSD system? Can you confirm this theory? Did you experience a speed boost once you kicked out the last spinner?

 

[crjackson2134]

Hi guys,

I noticed some minor delays in my system recently. My conclusion is some remaining spinner HDD's somehow clog up the system.
The Theory I came up with is to have no spinner in the system left, otherwise, there will be always delays because some data always remains on spinner HDD's, and they have to spin up to even read it, - hence waiting. I think the system should be a lot faster once any spinners have been eliminated.
Here is the question for the ones that have done this. Does a pure SSD System run faster compared to a mixed HDD/SSD system? Can you confirm this theory? Did you experience a speed boost once you kicked out the last spinner?

Unless you have your system set to put your hard drives to sleep, you shouldn’t see any delays attributable to waiting for HDDs to spin up.

I still have 2 HDDs in my system and experience no such delays. Pulling the drives from the system does not enhance the performance in the manner you are describing. Of course SSDs are faster when there are file transfers involved, but it doesn’t affect the responsiveness of the rest of the system in a perceptible manner when accessing a HDD. At least that’s how it works here.

That said, if your HDD is sleeping, and you initialize an access operation to that drive, there will be a delay as the drive spins up. All you have to do is go to your power management settings and turn off the option to put your drives into to sleep mode.

 

[Earl Urley]

YES!

Sent it please...

Is this the same link here

Yeah, it's the same file, but from another file service, so I won't bother..

 

[jbarley]

Is there any route for updating this firmware on a Mac computer?
The links I find are all .exe files, not much use on a mac.

edit:
which is the pcie slot of choice?
I have 2,3 and 4 available to me.

 

[crjackson2134]

Is there any route for updating this firmware on a Mac computer?
The links I find are all .exe files, not much use on a mac.

edit:
which is the pcie slot of choice?
I have 2,3 and 4 available to me.

One of the reasons I keep an install of WIN10 on a spare SSD because MANY hardware updaters only work in Windows. I know of no other way to update an NVMe SSD with a Phison controller.

SPECIAL DISCLAIMER:
This part of my answer is intended for jbarley running a cMP 5,1 and using an APFS non-Raided, NVMe setup only. It does not apply to all MP’s of other models or configurations. Excluded models are the 1,1 , 2,1, 3,1 , 6,1

jbarley (and only jbarley), if you aren't using a "Switched" PCIe adapter, it really doesn't matter which slot you use. In your case, I would just keep it away from the GPU (for cooling reasons). If I were using a x4 non-switched connection, I'd place it in slot 3.

 

[Earl Urley]

Sometimes I use a Win7 BootCamp but I also have access to a bonafide HP PC running Windows 7, so I use that and the M2X USB to NVMe adapter or the Plugable USB to NVMe enclosure.

Take the NVMe SSD, put it in the enclosure, boot to Win7 Bootcamp or boot up the Win 7 PC, plug in the USB to NVMe enclosure, run the updater. If the SSD is formatted for Mac it won't mount, but the updater will still 'see' the SSD on the USB bus. Click Power On and it will be recognized. Then if you have an older firmware, click on RUN.

The updater will work in Windows 7 and up.

 

[jbarley]

My new sabrent blade looked really good at first, until I installed a system on it.

newly partitioned..............................................................................new Mojave install...

 

[joevt]

If you aren't using a "Switched" PCIe adapter, it really doesn't matter which slot you use.

That's true for all Mac Pro's except the MacPro3,1 which has different speed slots. In the MacPro3,1, slots 1 and 2 are PCIe 2.0. Slots 3 and 4 are slower PCIe 1.0.

In MacPro4,1 and MacPro5,1, slots 3 and 4 use the same PCIe 2.0 x4 bandwidth so raiding them together won't give more performance unless the devices could not saturate a single PCIe 2.0 x4 link.

 

[crjackson2134]

That's true for all Mac Pro's except the MacPro3,1 which has different speed slots. In the MacPro3,1, slots 1 and 2 are PCIe 2.0. Slots 3 and 4 are slower PCIe 1.0.

In MacPro4,1 and MacPro5,1, slots 3 and 4 use the same PCIe 2.0 x4 bandwidth so raiding them together won't give more performance unless the devices could not saturate a single PCIe 2.0 x4 link.

Yes, I know this. I tend to forget about the 3,1s and older. Also, I wasn’t aware anyone was planning to run a RAID Array so I didn’t mention it. Thanks for clearing that up. However if you are going to target the post for being inadequately inclusive of All MP’s, then your post is also less than precise. You didn’t exclude the 1,1 , 2,1 or 6,1

Sorry @jbarley, I thought you were on a 5,1 with APFS. MY apologies for misleading you.
[doublepost=1563841783][/doublepost]

My new sabrent blade looked really good at first, until I installed a system on it.

newly partitioned..............................................................................new Mojave install...
View attachment 849470 View attachment 849471

Let the drive idle a bit. Sometimes a fresh install needs time to index. Also, make sure you have Trim enabled and showing for the new SSD.

After all that rerun the test. Also, download and run AmorphousDiskMark instead.

 

[jbarley]

Sorry @jbarley, I thought you were on a 5,1 with APFS. MY apologies for misleading you.

You're forgiven, I am running a 5,1, don't know where this 3,1 idea came from.

 

[joevt]

You're forgiven, I am running a 5,1, don't know where this 3,1 idea came from.

The thread says "MP All Models" so I had to qualify the answer for 3,1 (some of us still exist). And the raid thing is just in case someone wanted to try that but the limitation of slot 3 and slot 4 also applies if you were going to use two nvme devices as non-raid and wanted to read or write data to both devices at the same time. You should be able to read from one slot and write to the other slot at full speed though since there are separate lines for send and receive.

 

[lowendlinux]

Is there a real day to day difference between an SSUBX Apple Drive and an NVME drive. I know there’s a huge difference in benchmarks but those don’t always translate into the real world.

I think I’d prefer the Apple drive and I have an adapter sitting on the shelf but a new adapter and Samsung blade are about the same price as the Apple drive on EBay

 

[Earl Urley]

My new sabrent blade looked really good at first, until I installed a system on it.

newly partitioned..............................................................................new Mojave install...
View attachment 849470 View attachment 849471

Might be just Spotlight settling in.. it sometimes can take a day or two to index everything

 

[jbarley]

Might be just Spotlight settling in.. it sometimes can take a day or two to index everything

Unless the blade is not included in Spotlight searches.
I've removed it from indexing as a test and the results were still less then stellar.

 

[bsbeamer]

Initial cache/indexing of NVMe can take awhile. Even from clone it is usually at least partially rebuilt. Regardless of spotlight enable/disable, there is indexing that takes place within the OS. Leave running for 2-3 hours and then re-benchmark. Every single primary OS SSD I've setup (SATA, PCIe SATA and PCIe NVME) all show better results after whatever background indexing takes place.

 

[jbarley]

Initial cache/indexing of NVMe can take awhile. Even from clone it is usually at least partially rebuilt. Regardless of spotlight enable/disable, there is indexing that takes place within the OS. Leave running for 2-3 hours and then re-benchmark. Every single primary OS SSD I've setup (SATA, PCIe SATA and PCIe NVME) all show better results after whatever background indexing takes place.

2-3 hours, how about almost 24?
Write speed is still piss-poor compared to the first few tests.

 

[bsbeamer]

Install DriveDx and post the report.

 

[jbarley]

Done, here is a link to the report.
https://www.dropbox.com/s/tb9syji92u0qvd0/Log.rtf?dl=0

 

[crjackson2134]

[Deleted by Author]

 

[jbarley]

[Deleted by Author]

tried it with 2 different adapters, exact same results.

 

[joevt]

I have Samsung 960 Pro 1TB and 950 Pro 512MB in Thunderbolt 3 enclosures. The 960 Pro has slower write speed than the 950 Pro. I believe they're all using the latest firmware. Is this normal?

950 Pro: 1159/2386
960 Pro: 864/2508
960 Pro#2: 870/2499


#1000 woot!