Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

OWC Reveals Thunderbolt 5 Envoy Ultra SSD With 6000MB/s Speeds

Great, but largely pointless as most workloads don't require those speeds. Also, it's very likely that this drive won't do sustained 6000MB/s or even half of that.
 
Thanks for posting this! I've been eager to see a brand name TB 5 external SSD offer to get a sense of whether the price would go way up over TB 3 options. From the link, a couple of snippets:

"Pricing & Availability: The new Envoy Ultra Thunderbolt 5 SSDwill be available in 2.0TB and 4.0TB capacities for $399.99 and $599.99 respectively. It is available now for pre-order and will start shipping in late October. Meanwhile, the OWC Thunderbolt 5 (USB-C) Cables are available immediately in lengths of 0.3m, 0.8m, and 1.0m for $22.99, $27.99 and $39.99, respectively."

"The rugged Envoy Ultra is water and dust-resistant plus it's crushproof. As well as being bus-powered using its built-in Thunderbolt cable, The Envoy Ultra is silent thanks to its fanless, heat-dissipating, aluminum design."

I wondered whether we'd need to buy a cable separately, but sounds like it's built-in and the drive is bus-powered.

$600 for a 4-terabyte TB external SSD sounds great, a tad cheaper than the TB 3 options I've been looking at. At these prices + backward compatibility it might push TB 5 into prevalence over TB 3 SSDs going forward, which would provide critical mass to justify TB 5 ports in computers, and so on.

Wonder how hot these things are going to run? With Apples very high SSD upgrade prices, it's tempting to get a Thunderbolt external SSD and make it the startup disc, but I read of some running quite hot, and the speeds are roughly half (or less) what Apple's internal SSDs (except the 256-gig size) run.

But here's the question; how much is that evident in daily use? I'm not talking about Black Magic speed test results, or professionals moving 1/2-gig. files back and forth. What about regular users booting up their Mac, launching Word or Apple Photos with a very large photo library? If someone set up 3 Macs, one with 4-terabyte internal SSD, one with TB 3 4-terabyte external SSD and one with TB 5 4-terabyte external SSD, each of those the start up disc, and someone sat down to book up the system, launch Word and Photos, and spend the day doing regular home user things, would there be an observable difference in 'snappiness?'

Those of us looking forward to TB 5 sometimes get told people aren't even straining TB 3, but I wonder...if you're using a TB 3 instead of Apple's internal SSD for your start up disk, and you aren't a video-processing professional moving enormous files around, can you tell? Is there an observable difference, whether from throughput speeds or latency?
 
One big difference in speed one can see is whether we do a Time Machine backup from scratch fast. I did this while sitting in an Apple Store (while I was traveling) with a brand new SSD, so they could fix the suddenly-broken laptop. Another is if you have multiple SSDs, e.g. you have two 4TB SSDs instead of an expensive 8TB SSD. A multiple SSD enclosure could benefit from Thunderbolt 5.
 
  • Like
Reactions: drrich2
I keep coming back to that $600 for 4-terabyte thing. I've been following prices on some brand name 4-terabyte Thunderbolt 3 external SSD drives and it seems to be around $650 was roughly the price, give or take a bit. If OWC's Thunderbolt 5 offering hits the market at $600 and is backward compatible with TB 3 and 4, is there any reason going forward someone would choose to buy one of the TB 3 external SSDs? Even if they can't use TB 5 now, the OWC will work the same as a TB 3 and if the user ever switches to a TB 5 computer, the OWC will be ready to take advantage.

I haven't been following the DIY 'buy an empty enclosure and an SSD separately' - different brands, people posting about how hot some get, the complexities and hassles of adding a heat pad, compatibility between this brand SSD and that brand enclosure, etc... Not sure how much savings people see vs. the ready made options.

So the question becomes, will this lead to a bunch of Thunderbolt 3 external SSD products going on sale in the near future?
 
“…will this lead to a bunch of Thunderbolt 3 external SSD products going on sale in the near future?”

They already have.
My WD D50 TB3 2TB gaming dock was the same price as a Samsung T7 2TB USB 3.2 SSD, plus I got a daisy chainable dock so my monitor can be connected to it.
 
  • Like
Reactions: drrich2
we'll definitely see drops in prices over time for Thunderbolt 5, but it's interseting to see them coning out, before the launch of any Apple machines capable of Thunderbolt 5.

I'm interested in seeing an upgraded OWC Thunderbolt 5 hub, plugging in 4 of these drives and RAIDing them for redundancy and seeing what speeds you get!
 
Last edited:
  • Like
Reactions: drrich2
Has anyone here preordered this? From what I was told, the first batch shipped out. Just wondering what your experience is and how long it takes to get it.
 
  • Like
Reactions: Dutch60
Yes, I have the OWC 4TB Envoy Ultra TB5. I have it connected to my M4 Pro mini's backside TB5 port.

One can write to this device at a good 5 GB/s for the first 60GB, and then it will slow down to around 2 GB/s.
Also one can read from this device at close to 6 GB/s for ever.

It appears there's a write cache size of around 60 GB.

I used the dd command in Terminal to gather these data rates.

I have 2nd Envoy Ultra on its way to me and should have it in my hot sticky hands by end of February. I will be testing having 2x 4TB Envoy Ultra's setup as RAID-0 to understand this configs write & read data rates.

I also have the OWC TB5 Hub and shall test having the Envoy Ultras connected directly to the M4 Pro mini vs. having then connected to the OWC TB5 Hub to see if there's any significant difference. So far using just one Envoy Ultra connected directly to the M4 Pro mini vs. connected to the TB5 Hub I see no difference in the data rates.

It will be interesting to see if a pair of these Ultras setup as RAID-0 can deliver a read rate of at least 10 GB/s or better.
 
  • Like
Reactions: Brian33 and drrich2
this Thunderbolt 5 Envoy Ultra SSD is so durable….
(How durable is that?)
That drive is so durable, OWC wont package or bubble wrap the item when shipped!
They will just slap an address label on the device!
 
bxs said:
One can write to this device at a good 5 GB/s for the first 60GB, and then it will slow down to around 2 GB/s.
Also one can read from this device at close to 6 GB/s for ever.
Click to expand...
Thank you for this. I imagine a number of people will consider this and similar products to use as an external startup disc for their Mac, due to Apple's eye watering internal SSD upgrade pricing.

So basically, it seems as long as one isn't moving over 60 gigabyte file sizes in short order, it'll maintain speeds similar to what an internal SSD provides. And I don't think most home users do that often.

I've been wondering what the practical impact of the OWC 4TB Envoy Ultra TB5 having 'cache-limited' performance, and this is good to know.
 
MBAir2010 said:
That drive is so durable, OWC wont package or bubble wrap the item when shipped!
They will just slap an address label on the device!
Click to expand...
Ha we have actually reduced plastics and excessive packaging significantly in our shipping. I believe the Ultra is all paper based packaging and maybe a biodegradable wrap. So we don't quite slap a label on it and send it in the post. Though its strong enough that I don't see why we couldn't!
 
  • Like
Reactions: Alameda, MBAir2010 and drrich2
drrich2 said:
I've been wondering what the practical impact of the OWC 4TB Envoy Ultra TB5 having 'cache-limited' performance, and this is good to know.
Click to expand...
Likely in the real world not much at all for the majority of users. The vast majority of operations are read and there Ultra will sustain the read speed indefinitely. So in most cases, the limited cache doesn't affect real world operation. Plus even the speeds after cache is filled are still wicked fast.
 
  • Like
Reactions: drrich2
OWC_TAL said:
Ha we have actually reduced plastics and excessive packaging significantly in our shipping. I believe the Ultra is all paper based packaging and maybe a biodegradable wrap. So we don't quite slap a label on it and send it in the post. Though its strong enough that I don't see why we couldn't!
Click to expand...
I know, my OWC mac inini was perfectly packed!

usually I recommend products from OWC mac sales here for many years,
keep up the great services!
 
OWC_TAL said:
Likely in the real world not much at all for the majority of users. The vast majority of operations are read and there Ultra will sustain the read speed indefinitely. So in most cases, the limited cache doesn't affect real world operation. Plus even the speeds after cache is filled are still wicked fast.
Click to expand...
I concur with your assessment. In my specific scenario, I utilize applications that generate checkpoint files every two hours, potentially exceeding 60 GB in size (with some instances reaching up to 160 GB). These applications can operate for extended periods, spanning days and even weeks, depending on the complexity of the problem being addressed. Consequently, the speed of checkpoint file generation becomes crucial in optimizing the overall execution time of the application. The checkpoint file serves as a checkpoint, enabling the application to resume its execution from the state recorded at the time of its creation. In the event of an involuntary interruption that causes a long-running application to fail, the checkpoint file can significantly reduce the overall execution time.

Consider a checkpoint file of size 160 GB. It could be written out at a consistent 4.5 GB/s in 38 seconds or written out at an average 2 GB/s in 88 seconds due to the 60 GB write cache limit, which causes the write rate to decrease to approximately 1.5 GB/s beyond the 60 GB. The Ultra is likely to deliver an average write rate of approximately 2.3 GB/s when writing a 160 GB file to it. For the initial 60 GB, the rate will be around 4.5 GB/s, then it will fall off to approximately 1.5 GB/s, resulting in a total completion time of approximately 88 seconds.

Consequently, there is a penalty of approximately one minute of wall clock time required to write a 160 GB subject to the 60 GB write cache. While this may not appear significant, consider an application that operates for 30 days, necessitating the creation of a 160 GB checkpoint file every two hours. This results in an additional 6 hours of involuntary wait time over the 30-day period.

Here are some examples for writing then reading large files (160 GB and 50 GB) using my OWC Envoy Ultra.

Writing
dd if=/dev/zero of=<name of 160GB_file> bs=1g count=160 oflag=direct
160+0 records in
160+0 records out
171798691840 bytes transferred in 86.203039 secs (1992954005 bytes/sec)

Reading
dd if=/Volumes/OWC\ Envoy\ Ultra\ -\ Photos\ Libraries/160GB_file of=/dev/null bs=1g count=160 oflag=direct
160+0 records in
160+0 records out
171798691840 bytes transferred in 30.703018 secs (5595498522 bytes/sec)

Here's an example of writing and then reading a 50 GB file using my OWC Envoy Ultra that avoids the 60 GB write cache

Writing
dd if=/dev/zero of=<name of 50GB_file> bs=1g count=50 oflag=direct
50+0 records in
50+0 records out
53687091200 bytes transferred in 14.399542 secs (3728388806 bytes/sec)

Reading
dd if=<name of 50GB_file> of=/dev/null bs=1g count=50 oflag=direct
50+0 records in
50+0 records out
53687091200 bytes transferred in 9.933621 secs (5404584210 bytes/sec)
 
  • Like
Reactions: gregre and drrich2
bxs said:
I concur with your assessment. In my specific scenario, I utilize applications that generate checkpoint files every two hours, potentially exceeding 60 GB in size (with some instances reaching up to 160 GB). These applications can operate for extended periods, spanning days and even weeks, depending on the complexity of the problem being addressed. Consequently, the speed of checkpoint file generation becomes crucial in optimizing the overall execution time of the application. The checkpoint file serves as a checkpoint, enabling the application to resume its execution from the state recorded at the time of its creation. In the event of an involuntary interruption that causes a long-running application to fail, the checkpoint file can significantly reduce the overall execution time.

Consider a checkpoint file of size 160 GB. It could be written out at a consistent 4.5 GB/s in 38 seconds or written out at an average 2 GB/s in 88 seconds due to the 60 GB write cache limit, which causes the write rate to decrease to approximately 1.5 GB/s beyond the 60 GB. The Ultra is likely to deliver an average write rate of approximately 2.3 GB/s when writing a 160 GB file to it. For the initial 60 GB, the rate will be around 4.5 GB/s, then it will fall off to approximately 1.5 GB/s, resulting in a total completion time of approximately 88 seconds.

Consequently, there is a penalty of approximately one minute of wall clock time required to write a 160 GB subject to the 60 GB write cache. While this may not appear significant, consider an application that operates for 30 days, necessitating the creation of a 160 GB checkpoint file every two hours. This results in an additional 6 hours of involuntary wait time over the 30-day period.

Here are some examples for writing then reading large files (160 GB and 50 GB) using my OWC Envoy Ultra.

Writing
dd if=/dev/zero of=<name of 160GB_file> bs=1g count=160 oflag=direct
160+0 records in
160+0 records out
171798691840 bytes transferred in 86.203039 secs (1992954005 bytes/sec)

Reading
dd if=/Volumes/OWC\ Envoy\ Ultra\ -\ Photos\ Libraries/160GB_file of=/dev/null bs=1g count=160 oflag=direct
160+0 records in
160+0 records out
171798691840 bytes transferred in 30.703018 secs (5595498522 bytes/sec)

Here's an example of writing and then reading a 50 GB file using my OWC Envoy Ultra that avoids the 60 GB write cache

Writing
dd if=/dev/zero of=<name of 50GB_file> bs=1g count=50 oflag=direct
50+0 records in
50+0 records out
53687091200 bytes transferred in 14.399542 secs (3728388806 bytes/sec)

Reading
dd if=<name of 50GB_file> of=/dev/null bs=1g count=50 oflag=direct
50+0 records in
50+0 records out
53687091200 bytes transferred in 9.933621 secs (5404584210 bytes/sec)
Click to expand...
I think you would be a good target for the upcoming Thunderblade X12. It will sustain the full speed for the entire capacity. In fact, one Thunderblade x12 is faster than two Thunderblade X8's that are on independent buses and RAID'd together..
 
OWC_TAL said:
I think you would be a good target for the upcoming Thunderblade X12. It will sustain the full speed for the entire capacity. In fact, one Thunderblade x12 is faster than two Thunderblade X8's that are on independent buses and RAID'd together..
Click to expand...
Will they/it use TB5 protocol?
 
I bought an OWC Thunderbolt 5 Envoy Ultra 2TB and, following this video, decided to replace the internal SSD with a Samsung 9100 Pro 4TB PCIe 5.0 SSD.

I must report that this did not work: the OWC Envoy Ultra controller itself is detected in macOS, but the drive is not. It is simply not visible.

So, it is likely that PCI-Express 5.0 drives are not compatible with the OWC Envoy Ultra enclosure.

IMG_4241.png
 
Last edited:
  • Like
Reactions: drrich2
jsts said:
I bought an OWC Thunderbolt 5 Envoy Ultra 2TB and, following this video, decided to replace the internal SSD with a Samsung 9100 Pro 4TB PCIe 5.0 SSD.

I must report that this did not work: the OWC Envoy Ultra controller itself is detected in macOS, but the drive is not. It is simply not visible.

So, it is likely that PCI-Express 5.0 drives are not compatible with the OWC Envoy Ultra enclosure.
Click to expand...
Not entirely correct- likely that that specific controller is not compatible with the Barlow ridge chipset (likely any TB5 device, not just the Ultra that is). There is some info online about "stepping" of chips. This is also why you don't see any TB5 devices with open PCIe slots yet...

The drive chosen for the Ultra is very specifically chosen for both A) compatibility and B) power draw.
 
  • Like
Reactions: drrich2 and jsts
OWC_TAL said:
Not entirely correct- likely that that specific controller is not compatible with the Barlow ridge chipset (likely any TB5 device, not just the Ultra that is). There is some info online about "stepping" of chips. This is also why you don't see any TB5 devices with open PCIe slots yet...

The drive chosen for the Ultra is very specifically chosen for both A) compatibility and B) power draw.
Click to expand...

OWC_TAL, thank you for the clarifications. I hope this information will be useful for other users.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back