Upgrading 2013-2015 Macbook Pro SSD to M.2 NVMe

[scorpic]

Just updated my MacBook Pro (Retina, 15-inch, Mid 2015) macOS Mojave 10.14.5
with Inland Premium 1TB SSD 3D NAND M.2 2280 PCIe NVMe 3.0 x4 Internal Solid State Drive and black long sintech adapter. Run the test and this is the result. Is this good or not? Thanks.

 

[lemonade09]

Hi guys I have 13inch 2017 MacBook Air and I want a NVMe SSD that has good power efficiency as I care about battery life than performance. Any suggestions and or experiences would be appreciated.

 

[Allistah]

Just updated my MacBook Pro (Retina, 15-inch, Mid 2015) macOS Mojave 10.14.5
with Inland Premium 1TB SSD 3D NAND M.2 2280 PCIe NVMe 3.0 x4 Internal Solid State Drive and black long sintech adapter. Run the test and this is the result. Is this good or not? Thanks.

From the research I’ve done that looks perfect!

 

[prof.philia]

why you consider upgrade? your laptop has a nearly perfect ssd for example in battery battle every ssd will be defeated for sure since ahci protocol is more optimized for mbp
as well in cpu temperature battle again for nearly 99% this is the best
and as I check speeds for this drive is enough fast at all and for this mbp for sure since it has pci-e x2 so your bottleneck is ~1500 in both ways higher is unreachable for this laptop

I guess you're right.
I was wondering how is the battery life compared to the SSUBX AHCI drive.

 

[Allistah]

I want to see if I can determine power consumption of a 2015 MacBook Pro before and after I upgrade to Sabrent 2TB NVMe drive.

I have Kill-a-Watt tool that can easily measure power usage at the wall plug in current and watts with some other stats. Was thinking of taking readings when the drive was idle, reading, and writing to see if I can tell how much of a difference it is after I put the new drive in. Anyone else try that with any luck? Would be interesting data if it works.

 

[Ovaq]

MBA 2013 mid 11 inch
SiliconPower P34A80 1TB
Sintech's black short

Both read and write are more than 1500mb / s, and it has come out to the limit of PCIe 2.0 x4.
But・・・ the lifespan of the battery has been reduced by about 20%.

I researched and found the app 「Volta（Ultimate power control for Mac）」, so I tried it,
The lifespan of the battery has improved a little bit.(About 30 minutes)

Are there any other good solutions?
I want to share information.

 

[Gunmoule]

I want to see if I can determine power consumption of a 2015 MacBook Pro before and after I upgrade to Sabrent 2TB NVMe drive.

I have Kill-a-Watt tool that can easily measure power usage at the wall plug in current and watts with some other stats. Was thinking of taking readings when the drive was idle, reading, and writing to see if I can tell how much of a difference it is after I put the new drive in. Anyone else try that with any luck? Would be interesting data if it works.

The Macbook Pro being a laptop I guess the only thing you'd achieve this way is measure what the charger pulls from the wall to charge the battery. It would give no clue about your components consumption, after the charger it is DC. I don't know exactly what a Kill a Watt could be, but if it's meant to measure at the wall plug it is measuring AC.

Could work to some extent if the battery is fully charged, but I'm not sure you would get any useful data.

 

[Allistah]

Yeah, the tool measures voltage, amperage, and calculates watts over time (kWh) or watts. Pretty cool tool. Power is power and if you have a reference point (before) and switch a drive out (and a full battery), in theory you could see how much the reading has changed.

You're right though - not the most optima point of measurement but it's a lot better than nothing and can give some sort of gauge whether it seems to be a lot, or not. Could turn out to be completely worthless data in which that's fine too. Nothing lost really but it's a guarantee loss of data if I don't try at all. Will see what happens!

 

[Trippy Bear]

Hibernate 0 as 25 will cause kernel panics.

idk why I changed the hibernate mode to 25 on my MBP after doing the SSD upgrade, but i was def getting panics/restarts as soon as it went to sleep. I just changed it to 0 and no problems, just like when I first installed the SSD.

Thanks for your post, just saved me a headache.

 

[Allistah]

Got the 2TB Sabrent installed. Getting some weird speeds though..

Mid-2015 MBP 2.8
Link Width: x4
Link Speed: 8.0 GT/s

Consistently get ~2300 MB on sequential reads and ~3000 on sequential writes. I would think both would be close to 3000 as I've seen from other users and also that I would expect writes to be a bit slower than reads. If I set it to 1, 3, or 5, doesn't seem to make a difference.

Any ideas?

Edit: Pulled the drive out, re-seated the adapter and the drive. Same result.

 

[BombKilla]

From the excellent comparison test conducted by @gilles_polysoft on the page 1 header of this thread, the 512GB Sabrent Rocket demonstrated the same characteristic you observed with the 2TB SSD. While is unusual for an SSD to write faster than it reads, this appears to be consistent behavior resulting from Sabrent design choices; it doesn't seem that you made any installation errors:

Right now, the choices in 2TB are limited. If you want capacity > 1TB, you may just have to "suffer" with 2300MB/s unless you're willing to spend more $ and try a different brand.

"Consistently get ~2300 MB on sequential reads and ~3000 on sequential writes. I would think both would be close to 3000 as I've seen from other users and also that I would expect writes to be a bit slower than reads...


View attachment 842420

[doublepost=1560321933][/doublepost]One more thing, Allistah... unless you are doing heavy multi-user server duties with your MacBook Pro, you probably won't ever get anywhere near a queue depth of 32 for reads or writes. QD32 inflates the benchmark numbers, but much more realistic "workstation" use-case results are obtained with a queue depth of 4 for both reads and writes (the setting is in the menu options).
Attached are two runs from a Mac Pro 5,1 (link speed 5.0GT/s, PCIe slot 2 on a generic 4-lane M.2 adapter card, OSX 10.13.6) with a Sabrent 2TB (fw12.2) SSD at QD4 r/w. The first used an HFS+J format, the second used APFS. Note:
1. Slower but balanced sequential r/w performance was evident (yes, QD4 throughput is bus-limited and the results with an 8.0GT/s bus may not be as balanced. That said, on an 8.0GT/s bus QD4 numbers should still be significantly lower than QD32 numbers).
2. 4K writes with NVMe and either format were an order of magnitude faster than writes on AHCI SSDs. Latency is a big deal, especially on real-world use-cases.
3. 4K write performance was 2.3-2.4x faster with HFS than APFS. I know APFS brings modern technological improvements to Mac disk usage, but I'm _at least_ keeping a small partition formatted HFS+ as a scratch drive for database writes, interpreted code languages. etc.

 

[Allistah]

Heh, I thought I saw someone with one of these drives and they got ~3000 reads and writes.

I did an erase, GUID partition map, APFS, and no disk encryption. I’ll have to go back and find that post/user.

Found the post.. top post of this thread but on a different page: https://forums.macrumors.com/thread...sd-to-m-2-nvme.2034976/page-157#post-27265322

I’m wondering if firmware versions on the drive has anything to do with it. Will investigate further today.

 

[rockstarjoe]

Just updated my 2015 MBP 15" 11,4 to the 1TB Sabrent Rocket using the Sintech long adapter. Everything went smoothly... installed Mojave clean and then migrated my old account from a TM backup.

System seems fast and scores look good, with the exception of 4K writes. I'm not sure why they are so low. Link speed is 8.0 and link width is 4.0.

I formatted my drive as APFS encrypted when I booted into the Mojave installer... maybe that is why?

 

[GerardDirks]

• Corsair NVMe SSD : MP500, MP510

?

Hello

I can confirm that the Corsair MP510 will work fine with both the short & long-adapter. At the end I decide to use the short Adapter because the MP510 has also Chips on the backside of the SSD. Therefore it doesn't fit smoothly if you use the screw at the end.

I will tried the long adapter to use with an SSD without chips on the backside

Regards
Gérard

 

[Allistah]

4K write performance was 2.3-2.4x faster with HFS than APFS. I know APFS brings modern technological improvements to Mac disk usage, but I'm _at least_ keeping a small partition formatted HFS+ as a scratch drive for database writes, interpreted code languages. etc.

I’m going to format a partition with HFS today and try again and see what happens. Will post back.

UPDATE: I partitioned the drive and made a Mac OS Extended (Journaled) partition and it didn't appear to change anything when I used that partition as the target. Speeds pretty much the same. Now I don't know if I need to just format this drive and completely start over with that partition type or not. If anyone with one of these drives can confirm what speeds they get on a 2015 MBP with link speeds of x4 and 8.0 I'd appreciate it. Going to drive me crazy until I can confirm some consistency between machines. I also am assuming that Mac OS Extended (Journaled) is an HFS partition isn't it or am I not remembering that right?

 

[djangoreinhardt442]

From the excellent comparison test conducted by @gilles_polysoft on the page 1 header of this thread, the 512GB Sabrent Rocket demonstrated the same characteristic you observed with the 2TB SSD. While is unusual for an SSD to write faster than it reads, this appears to be consistent behavior resulting from Sabrent design choices; it doesn't seem that you made any installation errors:

View attachment 842426

Right now, the choices in 2TB are limited. If you want capacity > 1TB, you may just have to "suffer" with 2300MB/s unless you're willing to spend more $ and try a different brand.



[doublepost=1560321933][/doublepost]One more thing, Allistah... unless you are doing heavy multi-user server duties with your MacBook Pro, you probably won't ever get anywhere near a queue depth of 32 for reads or writes. QD32 inflates the benchmark numbers, but much more realistic "workstation" use-case results are obtained with a queue depth of 4 for both reads and writes (the setting is in the menu options).
Attached are two runs from a Mac Pro 5,1 (link speed 5.0GT/s, PCIe slot 2 on a generic 4-lane M.2 adapter card, OSX 10.13.6) with a Sabrent 2TB (fw12.2) SSD at QD4 r/w. The first used an HFS+J format, the second used APFS. Note:
1. Slower but balanced sequential r/w performance was evident (yes, QD4 throughput is bus-limited and the results with an 8.0GT/s bus may not be as balanced. That said, on an 8.0GT/s bus QD4 numbers should still be significantly lower than QD32 numbers).
2. 4K writes with NVMe and either format were an order of magnitude faster than writes on AHCI SSDs. Latency is a big deal, especially on real-world use-cases.
3. 4K write performance was 2.3-2.4x faster with HFS than APFS. I know APFS brings modern technological improvements to Mac disk usage, but I'm _at least_ keeping a small partition formatted HFS+ as a scratch drive for database writes, interpreted code languages. etc.

View attachment 842434 View attachment 842435

What HE (she?) said

@Allistah I think your numbers are totally fine and within line. Nothing wrong with your ssd
Stop using qd32 numbers. They are useless for real world use for a client computer
[doublepost=1560348620][/doublepost]

Just updated my 2015 MBP 15" 11,4 to the 1TB Sabrent Rocket using the Sintech long adapter. Everything went smoothly... installed Mojave clean and then migrated my old account from a TM backup.

System seems fast and scores look good, with the exception of 4K writes. I'm not sure why they are so low. Link speed is 8.0 and link width is 4.0.

I formatted my drive as APFS encrypted when I booted into the Mojave installer... maybe that is why?

Redo the test: run it 5x and use 1gb file size and that 4k write will be over 200MB/s

1TB sabrent benches better than 2TB sabrent.

 

[cmwebdev]

Has anybody that purchased a Sabrent Rocket Drive never had a kernel panic (due to the drive)?

 

[Allistah]

Has anybody that purchased a Sabrent Rocket Drive never had a kernel panic (due to the drive)?

I just installed the Sabrent 2TB drive last night in a mid-2015 15” MBP and so far have not had any issues. I’ve read in this thread that pre-2015 machines have sleep/hibernation issues and you have to change some settings around sleep and hibernation to make those issues go away. Search in this thread if that applies to you.

 

[cmwebdev]

I just installed the Sabrent 2TB drive last night in a mid-2015 15” MBP and so far have not had any issues. I’ve read in this thread that pre-2015 machines have sleep/hibernation issues and you have to change some settings around sleep and hibernation to make those issues go away. Search in this thread if that applies to you.

Thanks. I have a mid 2015 11,5 15” MBP. I’m mostly concerned about kernel panics when you’re actually using the device, as opposed to when the lid is closed or when it goes to sleep/hibernation.

 

[Allistah]

Thanks. I have a mid 2015 11,5 15” MBP. I’m mostly concerned about kernel panics when you’re actually using the device, as opposed to when the lid is closed or when it goes to sleep/hibernation.

I haven’t had any issues so far. I did an erase and GUID partition and formatted with APFS. Also did a fresh install of Mojave.

Prior to that, the system had an old BootROM on Yosemite, so I installed High Sierra which updated the BootROM, and then when I installed Mojave, it updated the BootROM again. Not sure if any of that matters but that’s what I did.

 

[rockstarjoe]

Redo the test: run it 5x and use 1gb file size and that 4k write will be over 200MB/s

1TB sabrent benches better than 2TB sabrent.

Awwww yeah, you were right! Thanks

 

[Allistah]

Awwww yeah, you were right! Thanks

You have the 1TB Sabrent, the 2TB, or you have both of them? What is your machine that you've got it in?

 

[rockstarjoe]

You have the 1TB Sabrent, the 2TB, or you have both of them? What is your machine that you've got it in?

1 TB Sabrent w/ the long Sintech adapter in a 2015 11,4 MBP.

I also have a 1 TB Sabrent w/ the short Sintech adapter that I will be installing in my wife's 2017 MBA... I'll update on how that goes too.

 

[Allistah]

I'm curious if you get any different results with the two different adapters in the same computer. Not sure if that would be possible or not, just trying to identify if there is a speed difference between the two adapters alone in the same machine.

 

[rockstarjoe]

I'm curious if you get any different results with the two different adapters in the same computer. Not sure if that would be possible or not, just trying to identify if there is a speed difference between the two adapters alone in the same machine.

I'm sure it is possible. It seems like people run into more issues with the short adapter due to it not being electronically shielded as well from the MB. The long adapter in my opinion is superior in this regard. But I could be wrong.

But unfortunately I am too lazy to open my machine back up again to swap the adapter