An open community 
of Macintosh users,
for Macintosh users.

FineTunedMac Dashboard widget now available! Download Here

Previous Thread
Next Thread
Print Thread
67% I/O Performance boost on M1 Macs
#58639 05/08/21 05:38 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
QUESTION: How do you improve the Write speed to an external SSD on an M1 Mac?

ANSWER:


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58640 05/08/21 05:40 PM
Joined: Aug 2009
Likes: 15
Online

Joined: Aug 2009
Likes: 15
I wonder if even Apple fully understands what they've unleashed on the world?

Subsequent testing on an M1 MacBook Pro and MacBook Air revealed the same speed increase.
So hooking a monitor up to a machine that's already got one built in results in the same speed boost?
As bizarre as the core finding is, that makes it seem run of the mill.


The new Great Equalizer is the SEND button.

In Memory of Harv: Those who can make you believe absurdities can make you commit atrocities. ~Voltaire
Re: 67% I/O Performance boost on M1 Macs
artie505 #58641 05/08/21 09:45 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
Originally Posted by artie505
I wonder if even Apple fully understands what they've unleashed on the world?

So hooking a monitor up to a machine that's already got one built in results in the same speed boost?
As bizarre as the core finding is, that makes it seem run of the mill.
It makes complete sense when you realize this is all a function of Thunderbolt 4 and the builtin displays on the M1 MacBooks and iMacs are connected via the HDMI interface and not through either of the Thunderbolt 4 ports. The Mac mini is unique among M1 Macs in even having an external HDMI port. NOTE: All of the M1 Macs can drive an external HDMI or display port display using the appropriate Thunderbolt adapter, but whether an HDMI or Display Port monitor connected via Thunderbolt port and adaptor counts as a Thunderbolt or HDMI, ¿quién sabe? 🤷‍♂️. What I know is, the M1 can only drive ONE Thunderbolt display, but it can drive up to fourteen or maybe more, HDMI/Display Port drives. I will leave testing all the various permutations of displays, ports, and adaptors to OWC. They have the necessary hardware and lab space, I don't have enough of either.


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58642 05/09/21 02:38 AM
Joined: Aug 2009
Likes: 15
Online

Joined: Aug 2009
Likes: 15
Originally Posted by joemikeb
Originally Posted by artie505
I wonder if even Apple fully understands what they've unleashed on the world?

So hooking a monitor up to a machine that's already got one built in results in the same speed boost?
As bizarre as the core finding is, that makes it seem run of the mill.
It makes complete sense when you realize this is all a function of Thunderbolt 4 and the builtin displays on the M1 MacBooks and iMacs are connected via the HDMI interface and not through either of the Thunderbolt 4 ports.
Leave it to Apple to come up with a way to force owners of laptops with built in monitors to buy external monitors. tongue

I'm curious to see how this plays out.


The new Great Equalizer is the SEND button.

In Memory of Harv: Those who can make you believe absurdities can make you commit atrocities. ~Voltaire
Re: 67% I/O Performance boost on M1 Macs
artie505 #58645 05/09/21 06:35 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
Originally Posted by artie505
Leave it to Apple to come up with a way to force owners of laptops with built in monitors to buy external monitors. tongue

I'm curious to see how this plays out.
As mentioned in the OWC blog post they have a device, intended for use on Mac minis running headless that may fool the system into believing there is a Thunderbolt monitor attached. If it works you know we will all hear about it.

And before. you go blaming Apple for this quirk in Thunderbolt, remember Intel is the developer behind both USB and Thunderbolt. Apple is an early adopter of the technology and sits on the standards committee, but Intel is unquestionably the driver.

Last edited by MacManiac; 05/09/21 08:27 PM. Reason: formatting fix...

If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58674 05/13/21 12:48 AM
Joined: Aug 2009
Likes: 15
Online

Joined: Aug 2009
Likes: 15
Originally Posted by joemikeb
Originally Posted by artie505
Leave it to Apple to come up with a way to force owners of laptops with built in monitors to buy external monitors. tongue

I'm curious to see how this plays out.
And before. you go blaming Apple for this quirk in Thunderbolt, remember Intel is the developer behind both USB and Thunderbolt. Apple is an early adopter of the technology and sits on the standards committee, but Intel is unquestionably the driver.
Is that a fair statement?

I'll guess that the Thunderbolt standard was developed before Apple silicon was around to be a consideration in its development, in which case, Intel couldn't be "at fault" for this glitch.

Isn't it more likely that the underlying issue here is either an unforeseen - and, of course, unforeseeable - incompatibility between the Thunderbolt standard and Apple silicon or an implementation failure on Apple's part?


The new Great Equalizer is the SEND button.

In Memory of Harv: Those who can make you believe absurdities can make you commit atrocities. ~Voltaire
Re: 67% I/O Performance boost on M1 Macs
artie505 #58717 05/18/21 11:34 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
I have performed a series of drive speed tests with different drives, different connection ports and I have been unable to replicate the reported improvement in performance with or without a Thunderbolt monitor. Tests were conducted using Stibium from The Eclectic Light Company (If you haven't checked them out they have an amazing number of useful free utilities.) Admittedly I could have run more trials in each configuration and there are innumerable other potential configurations that might have made a difference but here are my test results. Contrary to the reported results that started this thread, the only performance improvement I found was a modest 10% increase in the write speed by disconnecting the Thunderbolt Monitor which, of course, contradicts the findings reported in the article and seem more intuitive to me.

I was disappointed in the performance of one of my OWC drives and puzzled by the miserable performance of another when connected via a USB 3.1 gen 2 port. Note also that despite the reports of poor performance of HDs formatted APFS these tests found no significant difference. My belief is the USB 3.1 gen 2 port is the limiting factor not the disk format. I am adding a To Do to my list to check the connections and configuration of that particular hub. One final note: the performance of the Samsung Thunderbolt 3 SSD is exemplary (exceeding the theoretical max throughput of Thunderbolt 3 or 4) but under load if I used it as a coaster my coffee would be nicely warmed all day long — those drives do run hot. The enclosure is solid plastic and does not make a good heat sink where as the OWC SSDs have a metal enclosure which acts as a heat sink to keep them cool. I don't know what the long term implications of heat are on SSDs, but intuitively I am happier with the cooler drives.


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58725 05/19/21 07:52 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
I THINK I MISSED SOMETHING🤔

I DID MISS SOMETHING🤦🏻‍♂️

During my drive speed tests I noted the same drive with a USB 3.1 gen 2 port I/O was over 20 times faster when connected to a Thunderbolt 3 port than when connected to a USB 3.1 gen 2 hub. What I missed was the fact I had tried connecting that drive to the same Thunderbolt 3 port twice before and each time Big Sur refused the connection reporting the drive must be connected to a USB port. The difference in this latest trial I had used a Thunderbolt 4 cable and the two previous trials used USB type C and Thunderbolt 3 cables.

I have contended Thunderbolt 3 cables rated at 40Gb/s were equivalent to Thunderbolt 4 cables. Apparently this is not always true!


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58727 05/20/21 04:17 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
Further Thunderbolt 4 Cable Testing

Any lingering thoughts I might have harbored that Thunderbolt 3 cables are essentially the same as Thunderbolt 4 cables were removed this morning. I decided to backup my iPad Pro to my Mac mini in addition to the regular iCloud backup. The only conveniently available port is a USB 3.0 hub stuck to the edge of my desktop and since the port on the iPad pro is USB type C I grabbed a Thunderbolt 3 from my drawer and plugged that into a USB type A to Type C adaptor and connected the iPad. The iPad chimed and indicated it was charging but did not connect. I rummaged through my collection of USB, Firewire, HDMI, Ethernet, and heaven only knows what other cables and found the Thunderbolt 4 certified cable I had forgotten I bought and replaced the Thunderbolt 3 cable with it. 🛎 bing! the iPad chimed, and appeared in the Finder sidebar. I have retested using other Thunderbolt 3 cables (all certified and all rated a 40Gb/s) with the same results.

CONCLUSIONS:
  • Thunderbolt 4 > Thunderbolt 3
  • Thunderbolt 3 ≨ Thunderbolt 4
  • Thunderbolt 4 is downward compatible with Thunderbolt 2, Thunderbolt 3, and all USB protocols with proper connection type adaptors
  • it is going to get expensive replacing all my Thunderbolt 3 cables with Thunderbolt 4
  • I have an excuse to clean out that rat's nest of outdated USB and Thunderbolt 3 cabling, hubs, etc.


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58729 05/21/21 06:38 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
This OWC blog followup explains why I was unable to duplicate their speed improvement results. I have an OWC Thunderbolt 4 hub attached to the second Thunderbolt 4 port on my M1 Mac mini and was already enjoying the wrote speed increase. Who knew? My findings with Thunderbolt 4 cables still stand.


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58736 05/23/21 07:08 AM
Joined: Aug 2009
Likes: 15
Online

Joined: Aug 2009
Likes: 15
Originally Posted by joemikeb
This OWC blog followup explains why I was unable to duplicate their speed improvement results.
1. Is an M1 a computer or a Tinker Toy? tongue

And 2. Has Apple weighed in on the M1/Thunderbolt issues yet, even if only to acknowledge them?


The new Great Equalizer is the SEND button.

In Memory of Harv: Those who can make you believe absurdities can make you commit atrocities. ~Voltaire
Re: 67% I/O Performance boost on M1 Macs
artie505 #58742 05/24/21 05:28 PM
Joined: Aug 2009
Likes: 16
Moderator
OP Online
Moderator

Joined: Aug 2009
Likes: 16
Originally Posted by artie505
1. Is an M1 a computer or a Tinker Toy? tongue

And 2. Has Apple weighed in on the M1/Thunderbolt issues yet, even if only to acknowledge them?
1. It is a powerful, flexible computer. (Sidebar: The new iPad Pro with the same guts actually has higher Geekbench numbers but I think that is a function of MacOS vs iOS. The latter has far fewer background tasks running.)

2. The issue is is the standards for the interface and those are driven by a committee headed by Intel. As far as I know Intel has never admitted to an error of any type. By-the-way, my recent testing indicate Thunderbolt 4 cables deliver performance and functionality superior to Thunderbolt 3 cables supposedly certified at the same performance levels. I have discovered that certified Thunderbolt 4 cables up to 0.8M in length can be passive but to go beyond that length active connectors are required. At least that explains why the longer Thunderbolt 4 cables are so much more expensive and it is not just more wire.


If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein
Re: 67% I/O Performance boost on M1 Macs
joemikeb #58752 05/25/21 08:17 AM
Joined: Aug 2009
Likes: 15
Online

Joined: Aug 2009
Likes: 15
Originally Posted by joemikeb
Originally Posted by artie505
1. Is an M1 a computer or a Tinker Toy? tongue

And 2. Has Apple weighed in on the M1/Thunderbolt issues yet, even if only to acknowledge them?
1. It is a powerful, flexible computer.
You're too serious.

That was, of course, a facetious question, but considering the many configurations of ports, cables, and devices you've reported having had to experiment with before achieving your desired results (or lack thereof), "Tinker Toy" isn't all that far from being an appropriate analogy.

Originally Posted by joemikeb
2. The issue is is the standards for the interface and those are driven by a committee headed by Intel. As far as I know Intel has never admitted to an error of any type.
My original question remains unanswered: Judging from the timing, the basic Thunderbolt standard was developed before Apple Silicon was off the drawing board, maybe even before it was even on the drawing board, so how could that committee, even with Apple sitting on it, have considered it as a factor in the development of the standard?

Intel couldn't have known its ins and outs, so even if the development of the standard and Apple Silicon were concurrent, the ball was in Apple's court to "put the pieces together."

And under any circumstances, isn't it curious that Apple hasn't even acknowledged the issue yet? It's certainly not an inconsequential one.


The new Great Equalizer is the SEND button.

In Memory of Harv: Those who can make you believe absurdities can make you commit atrocities. ~Voltaire

Moderated by  alternaut, dianne, MacManiac 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.4
(Release build 20200307)
Responsive Width:

PHP: 7.4.33 Page Time: 0.029s Queries: 40 (0.022s) Memory: 0.6485 MB (Peak: 0.7641 MB) Data Comp: Zlib Server Time: 2024-03-28 17:58:08 UTC
Valid HTML 5 and Valid CSS