Відео

in the old days where years started with 19 we used hubs, not switches. with a hub you message was not shielded from collisions, one computer knew it was blocked by hearing anotherone while is was sending. the 100m cable length came from this origin too. a message would be sent in full before others knew the line was busy and so on. now with switches doing store and forward every message is on its own network in every cable, colissions do not occure in the same way = less trouble. the original rule I was tought was max 5 segmets and only 3 of those whas allowed to have client endpoints connected.

Professor Snape teaching Network Studies this year.

I'm guessing that most Australians will be more interested in signal availability rather than max possible speeds?

for this to be at all valid you need to load each switch up with clients and traffic then look and see if there were errata, maybe like network cameras that automatically send data traffic.

Not sure if you're still looking for ways to copy a bunch of small files from external media, but if you have access to the physical device and it's mostly empty, one thing you can do is ignore the files and copy the entire filesystem directly. You would then need to mount the resulting image to access the files, but you wouldn't need the drive at that point. This approach should give you the theoretical maximum, aka minimum of portable drive read, connection interface and host drive write speed, without any file creation overhead.

I was only getting 90 mbs on my 2010 CAT 5e unshielded outside line. I tried everything to get faster speeds but no luck. In reality, what is the average homeowner going to get on a 70 foot line to their router with 5e? When I switched to a CAT 6 cable I'm getting 300-320 mbs now.

How much packet loss?

The USB C connector does not work with an Android phone. it will not charge

"no problems at all" showing less than 5GbE connection speeds. :P

You do not demonstrate any ability if you want to connect a hdmi tv for live streaming or you tube video. Should do this in my opinion

GREAT VIDEO! Short, well done & super interesting :)

Did you turn on the jumbo packets?

to copy many files faster, try and make it one big file such as a zip or pkzip or similar. Each file copy has an overhead, such as file name write the FAT (or equivalent). One big file is always faster than several hundred small ones, even in a directory tree.

I thought you’d need a crossover cable to join two compters? Perhaps doesn’t apply to macs?

Great information ! Thanks!

There’s definitely overhead and a slight increase in latency.

Alex Ziskind is getting around 60gbps with TB5, so I dont expect 4 to be full bandwidth either. Did you test with iperf?

JUMBO FRAMES

This isn't a method of proving speed of networking. This was a method of proving speed of copy. To test speed of networking you need to do something equivalent to iperf and leave it at that.

Don't confuse the physical layer data rate of 10 Gbit/sec per lane for Thunderbolt 4 and the protocol data rate over those lanes. Thunderbolt transports several different packetized protocols over the same physical layer which have different priorities. Thunderbolt 4 mandates four bidirectional lanes with approximately 10 Gbit/sec of independent bandwidth. Video gets transported with embedded DisplayPort protocol with highest priority (as its traffic must arrive in time for your display to work correctly). Other protocols include embedded include USB and PCIe signaling which are bursty and use what bandwidth is remains. Thunderbolt 4 implements the embedded PCIe 3.0 protocol which is limited to 32 Gbits/sec and can't use the remaining 8 Gbits/sec which is reserved for video and other protocol overhead. Note: in most computers a physical PCIe 3.0 data lane on the mother board is attached to the thunderbolt controllers PCIe. A single data streams maxes out at 32 Gbit/sec before encoding overhead, 8/10 removes 20%, 64/66 3%, or 128/130 1.5% depending not the exact protocol and context. Network adapters attached to Thunderbolt can come in two types USB controllers that connect to the embedded USB protocol, and the faster lower overhead PCIe controllers that connect to the PCIe protocol. "Thunderbolt Networking" emulates a PCIe controller network card without physically having real network controllers. Thus the theoretical maximum with Thunderbolt Networking is 32 Gbits/sec and the practical limit is less than that with both the physical protocol overhead, the logical protocol overhead, and any other traffic. Also effective bandwidth is highly dependent on the network packet size... I always configure Thunderbolt networking with Jumbo (9000 byte) packets rather than the default (1500 byte) packets which for me is a 30-40% performance gain over the default. Connecting my MacBook Pro M1 Max -> CalDigit TS4 -> MacBook Pro M2 Max I constantly get 14.7 Gbits/sec iPerf bandwidth with my high resolution wide screen monitor also attached to the docking station. When I turn off the monitor it jumps to 20.4 Gbits/sec iPerf bandwidth. I see these as respectable numbers given the overheads of the protocols and other shared hardware on the docking station. (Note: I get the about the same effective speed with or without the docking station in between the computers). With Thunderbolt 5 in the default symmetric configuration the embedded PCIe 4.0 and is twice the speed at 64.0 Gbits/sec. In addition some of the overheads are fixed in size or optimized so I would expect (but have not yet been able to test) that Thunderbolt Networking properly configured would be a little more that 2x faster.

The transfer speed of ~470MB/s you show is (I guess) the write limit of your SSD's. In all fairness, that is about 4 gbps considering some overhead. That doesn't proof cat 5e can do 10 gbps over that 25m. Iperf or OpenSpeedTest local network tests would be able to go faster, right. Or am I seeing it wrong?

How many times did you run these tests? The numbers look similar enough that I’d just wonder about noise. You might want to rerun each test a larger number of times and compare the averages.

finally, chanced upon ur video. I have Cat5e in my walls. U answered my question. Thanks a lot

You might try changing to a smaller packet size. If you don't want to do that, you could try zipping the files before you send them. Then, it would only be one big file, not lots of smaller ones. In addition, you would get some benefit from compression. The question is, would compressing and decompressing the files take longer than the transfer time. I don't know. One silly question on the Thunderbolt side, are you sure you are running TB4 cables? It would be kind of sad if you were running cables for an older version of USB C. (How far are the cables running?)

In my experience, copying 1000s of smaller files takes longer than 1 large files. I believe it has to do with checking the file has been copied on the other end + your are actually doing random writes instead of 1 potential sequential write. Random writes are slower than sequential writes no matter what, even with SSD (ask ChatGPT for clarification). That’s why you larger files size but fewer are much faster to write (copy) I would probably compress all the 1000 files into 1 file copy over and decompress.

so even in worst case scenario the thunderbolt 5 connections will be faster then 10gbps ethernet

The short answer is yes. 1:50

There is actually difference in architecture within Thubderbolt3 and Thubderbolt4 you may test those variants as well

Informative video Matthew thanks!

Ok I just watched your earlier thunderbolt video and replied before watching this one. Here’s an idea for you…partition the SSD, add a ZFS partition, and then add a (16 GB max) RAM disk as an intent log. That way you save lots of RAM and still get some persistence across reboots. In either case, the performance gains should be more pronounced on shared filesystems.

Synchronous writes may be the bottleneck in your situation. One fast way to copy lots of small files very very fast would be to use ZFS SSD RAID with a RAM disk intent log. If that’s too insane you could try a file on a fast SSD for the intent log.

I think that a lot of the concerns about how many switch hops you can have come from much further back...there WAS a limit to how many HUBs you could daisy chain before it just plain broke. This was before switches were readily accessible or affordable. Nice test scenario, though!

As others mentioned you can’t use disk read writes to test network speeds. Please read up on iperf.

You will never go faster than your hard drive speed. It has nothing to do with the thunderbolt. You cannot reach 40 gigabits because your drives can’t go that fast.

This video is 1.5 year old but still awesome! @Matthew Daley Thank you! I've read some amazing comments here especially one from @James_Knott who is 100% on his, simply put I like seeing the number visually. ~3.7Gb over 25 meters of Cat5 on a 10Gb port is absolutely awesome for the cable itself. Speaks loads about 'older technologies' holding up to the newer. Granted, you are getting 37% out of a theoretical 100% but still that is great!

Yep, it has a lot of overhead, unless you're on a Linux machine where you can do unholy things to thunderbolt to eek out more networking performance. The 10Gb limit comes from the fact that that is all Thunderbold 4 reserves for PCIe bandwidth, which it uses here for networking. (not the usual display / data applications which should reach near 40G)

Had Mac M1 mini connected to a PC via Ethernet to share files. Replaced the PC with a Mac M2 mini and moved the Ethernet connection from PC to M2. I am anything but computer literate so, almost inevitably, no connection between the two Macs. Thought the previous connection was getting in the way as the M1 Ethernet said connected while the M2 Ethernet said not connected, after I tried everything that I could think of to convince them to speak to one another. So, I took option two on the M1 and deleted the service. Now, the M1 does not have Ethernet showing as a service in network, although there is a numbered service with an Ethernet symbol that is the same as the symbol accompanying the Ethernet service on the M2. Both show unconnected. Now, there is a description in Apple support files for the Mini explains how to set up two minis with an Ethernet cable between them, but when I follow the steps I come up against one that doesn’t show and I can proceed no further. Contacted Apple support via chat, was told that Mac OS does not support direct Ethernet connection. As I just entered my fourth quarter, I am having a difficult time dealing with tech that seems to have been set up by monkeys for chipmunks.

RAM disks are also an artifact of an era where programs tried or were forced to (x86 comes to mind) to use little memory. Most of the time, today, you are just removing memory from the program and making it simulate storage, basically nonsense. Exceptions are, for example, the physical RAM disk that sale for a fortune.

10 Gbps but you only showed 400-500MiB/s transfers... not even 5Gbps then?

maybe on ai?

Roberts Route

Man, excellent video!!! I had forgotten about RAM disks. This is an interesting way of configuring the hardware on these machines. I believe the only way to see the full usage of the memory bandwidth is to run multiple apps concurrently. For instance start a large render project, and encrypt a large file at the same time.

Don't use file copy to test connection speed but rather iperf/3

What's the point when SSDs have gone beyond 12000 GB/s write ... on a PC ? Imagine striping that ...

6:09 no. 400GB/s and 800GB/s arent avaiable for CPU cores to use. Only iGPU can take all that memory BW.

Use a cloth tonstop condensation

You probably need multiple cores accessing memory simultaneously to see the bandwidth increase. Not sure how that translates to a RAM disk test.

Yes today its mostly not of interest anymore, only in the case that the software heavily relies on temporary stored data written many times it is interesting to get rid of the wear and tear of the nand chips to shift that part to the non-degradable RAM. We had that issue in the past with TeX and Metafont in the times as DOS was limited to 1 GB of Ram and the rest was transformed to a RAMDisk instead of using HDD this was lightning fast. But today i think the RAM is always better used as RAM. Perhaps very rare cases where have usecase for that.

In pre-SSD days, the lots of small files issue and its associated "keep adding small bits to a file" performance was dominated by the head movement. (And the wait for the drive to rotate to the required position.) The need to access file system data, handle the actual data, and update the file system data. And, in the keep adding case, sometimes having to follow chains of entries to reach the last block - with each added fragment potentially being elsewhere on the drive. Plus, even if the problem file were deleted, the free space on the drive would be fragmented and affect everything else. (Some OS log files were among the worst for continually adding allocation unit by allocation unit spread across the drive.) If you had the option of putting a heavily hit "keep adding" file on a separate, newly formatted, drive then performance could be massively better. The difference between writing file A, then file B versus alternating blocks to A then B could be dramatic. Hence, I am not surprised if the difference between SSD and RAM disk is relatively small. The switch to SSD eliminated head movement! Allocation unit size (whatever approach was used) also had a huge impact. The bigger the AU, the fewer accesses required for file management. What allocation unit size was used for your RAM disks? And is there any way of increasing the size? (The above based on experience of numerous file systems but NOT the native macOS ones.)

At 2:20, you said you created a 128GB RAM Disk, but the visual graphic says 128TB. Maybe one day we’ll have a system with 128TB, but today is not that day 😂