Will that hub or dock slow your SSDs, or even make them faster?
eclecticlight.co
eclecticlight.co
They honestly just need to kill off USB at this point and just let Thunderbolt supersede it. Thunderbolt 4 and 5 are literally just implementations of USB4, except the Thunderbolt standards group is doing a hell of a lot better job at naming things and certifying cables than the USB group is.
If it's in my backpack or used for a dock/monitor, it's going to be a Thunderbolt cable.
Expensive? Absolutely. Unnecessary? Almost certainly. But I haven't had any issues with them whatsoever.
The only "issue" I had, was that you often ended up without the proper one when you needed it. Almost all the cables came with the device which needed it. Only bought 2 which were longer.
What issues do you experience so frequently that it would justify investing more money into it?
Q: What issues do you experience so frequently that it would justify investing more money into it?
A: ... often ended up without the proper one when you needed it
That's precisely the issue GP solves by only carrying one type of cable, Thunderbolt. If you want to transfer data at top speed, render high resolution high refresh rate video, or charge at maximum wattage, TB4 handles it. One cable to rule them all, no surprises.
If you are on anything different and want to hook up, say a Studio Display, you can't use USB-C (below 4) and that is not USB-IFs fault.
Besides that, what I meant was the connector standards mix before USB-C has become mandatory here, and I think you knew that. Other than that, all what you wrote is also determined by the devices connected, so this may be why most of the people out there don't even realize there might be a reason to buy Thunderbold.
Perhaps it's more useful to people who are constantly traveling, but for someone who isn't, I guess I just don't see a point in it. Would I turn it down? No. Would I pay more for it? If it's greater than 2$ more, no. Slow charge is "good enough" in my eyes.
I don't use it a lot, probably once a month on average. But the times I do it's invaluable.
Basically for any cheap use cases, you just have to buy a random "USB-C" cable with unknown capabilities, while for specific data use cases you have to buy a "USB-C" cable that also supports a specific data rate, either USB 3.1, USB 3.2, USB4 v1.0, USB4 v2.0, or Thunderbolt 3/4/5 (and most cables will support multiple of these, for example USB 3.2 Gen2x2 is the same speed as USB4 v1.0 and TB3).
Disagree on the replacement with thunderbolt, though. USB historically is very different, and it's USB4 that's a clone of TB3. Agreed the naming is better but a lot of micros have USB and thunderbolt would be ridiculous for them.
I think it matters a great deal, but not as they or you intend. Having technical and marketing terms sucks! I use Ubuntu on my home server. And whenever I need to troubleshoot some issues with packages or want to upgrade to a later release I am confronted with those awful names. I DONT KNOW WHAT JAMMY IS! JUST CALL IT 22.04, 22.10, 23.04, 24.04.......! I don't want to memorize random names and remember when they where released. And name your apt repos in the same version scheme goddamnit!
Same with USB! Just give it a sensible name and stop changing it for marketing reasons. Such a effing stupid thing to do!
Is the price difference really that high or are you comparing them to cheaper SSDs?
That's why external SSDs are so common in Mac setups, even accounting for the additional cost of a Thunderbolt enclosure it's usually still significantly cheaper than getting a bigger internal SSD from Apple.
Which isn’t anything against the SN850x, it’s a great fit for the intended use case it’s just many people assume there’s zero trade-offs involved beyond speed/price/capacity.
Apple is definitely raising storage prices to milk their customers and promote their iCloud cash cow, but it’s still worth considering when looking at ‘gaming’ parts in different situations.
Apple is overcharging for storage. You get a lot of compute for cheap though :/
SSD performance and power efficiency are significantly affected by the choice of controller. Apple's Macs have the controller built-in to the SoC, so it's a sunk cost that doesn't really factor in to upgrade pricing.
Anyway, what you might consider insignificant differences are things companies do consider these worth paying for. You not caring isn’t the same thing as nobody caring.
> significantly affected by the choice of controller
Aka it is more complicated than just slapping different controller on the same chips and calling it a day.
Maybe that doesn't make it a bad comparison, but the SN850X is def intended to be a gaming part.
[1]: https://www.westerndigital.com/brand/wd-black
[2]: https://www.westerndigital.com/en-in/products/internal-drive...
Gamers are the ones buying expensive parts so it makes sense to market to that. The next tier after this is basically server-class 10-20k machines which Apple is definitely not competing with (and SSDs are not really that much better in that class anyway). Dismissing SSDs as “gaming” parts as if it’s diminishing the quality misunderstands what’s happening here. It would be one thing if WD was ignoring fsyncs to achieve this performance but gamers don’t care about writes so much anyway and there’s no indication WD did that.
Source: I have the WD and Samsung parts as well as cheapo random SSDs.
The point I was trying to make by pointing out that the SN850x isn't a "gaming part" is that the SN850x is literally the top-of-the-line, most expensive consumer SSD sold by WD, and has practically the same specs as other top-of-the-line, most expensive competing parts like the Samsung 990 Pro. Being one of the most expensive SSDs on the market means that saying that the SN850x is a bad comparison because it's supposedly "lower price" is just false on its face.
To be more clear, getting equal performance without sacrificing anything would raise costs even further.
It’s easy to say moderate premium over normal business grade SSD’s but that doesn’t mean any specific number is correct. I’d say the equivalent to a 130$ to 220$ SSD assuming a stand alone equivalent exited, but the actual number depending on info Apple isn’t sharing. And yes the range is both above and below the specific part suggested.
But anyway, in what world isn’t this a gaming product: https://shop.sandisk.com/products/ssd/internal-ssd/wd-black-...
“Built for elite gaming.
Crush load times and slash throttling, lagging, and texture pop-ins with the WD_BLACK SN850X NVMe™ SSD. …
Do more with WD_BLACK Dashboard The downloadable WD_BLACK Dashboard (Windows® only) monitors your drive’s health, lets you customize your RGB lighting, and, exclusively on the SN850X SSD, enables Game Mode 2.0 to transform your gaming experience.”
There really isn't. Apple is reported to use SanDisk 3D TLC NAND chips. SanDisk is owned by Western Digital, and the WD SSDs use SanDisk chips. They're literally the same chips.
Hell WD chips could be of higher quality as I am not suggesting I know their internal processes. I am saying things are optimized differently.
It could just come down to different binning of the same part, and it would still make a difference.
The why I’m still talking is because people seem to think buying a gaming SSD is a good idea when they also want longevity / low risk of future. The parts can last 10+ years but they’re designed with something else in mind.
One hardware guy mentioned internal defects can cause differences is the amount of reserve sectors that a final product ends up with. That’s exactly the kind of arbitrary cutoff that lets companies charge different prices for the same part.
Lower-grade flash with higher initial defect rates is what gets used in USB flash drives and SD cards, and some bargain-bin SSDs with lower usable capacities (ie. 960GB rather than 1TB).
The stuff used in a WD Black or WD Blue branded consumer SSD is not a different quality grade from the stuff used in any other mainstream consumer SSD, Apple's included.
At what grade? Plus, how much extra endurance is baked in to Apple's drives, i.e. how over-provisioned are they?
My MacBook Air M1 reports 99% health after being daily driven (and some 26TB written to it) at work since 2020 (we got these as soon as they introduced), and I don't baby its drive in any way.
Any flash storage has two main components: A controller and a set of flash chips, and a third component enters to the picture when connecting these two: number & nature of channels.
Starting from the controller (and channels), beside the obvious PCIe generation, there are some other factors. DRAM support, NAND support (not all NAND is the same!), number of channels, and the speed of these channels. A DRAMless SSD will suffer after its "pseudo-SLC" cache runs out, and the performance of the drive will generally suffer if the number of channels can't absorb the traffic coming from the PCIe side. Here, to have a top notch SSD, you need to have a good/fast controller with DRAM support, and enough channels with enough speed to absorb all traffic requests, so you can get use of the premium NAND chips you bought.
Next, we have the flash layout. Flash chips vary in speed, density and drive. A high density flash chip might be slower, or a flash chip might require higher drive, resulting in higher temperatures in general. In some cases, instead of populating all channels, a manufacturer might decide to populate a few channels and leave the rest unpopulated, creating a big but slower SSD.
Beyond that, there are other considerations like over-provisioning at flash level, "soft SLC cache" size, wear leveling capabilities under sustained load, etc. etc.
For example, an enterprise SSD comes with the "same" TLC chips, but over-provisioned 5:1 or 10:1 (10TB flash for 1TB capacity)
Now, let's see some real-world examples:
- Kingston NV2, NV3: A budget SSD with great capacity and price. DRAMless, no channel number guarantee, and might come in with TLC or QLC chips. Burst speeds are OK, will make 90% of the people happy, but slows down in long transfers and under heavy load. Runs cooking hot in both controller and flash side.
- Kingston KC3000: A higher end drive with part/channel guarantees, handles sustained load better, runs way cooler, ironically.
- Samsung 980/990 Pro: Samsung's higher end drive. Runs cool, sustains speed all over due to DRAM and tons of channels and vertical integration of controller + NAND.
- Samsung T7 Shield: Looks like an bulky 1.8" drive, but it's selling point is it can sustain 1050MB/s writes wihtout even slowing down until it's full. Never gets warm.
So, flash drives comes in all shapes and sizes and with specifications and capabilities all over the place. A WD Blue and WD Black won't perform the same. Same for Sandisk's Plus, Extreme, Extreme Pro series.
This is why OWC was/is the go-to 3rd party SSD provier for Macs for quite some time. They tune their drives similar to Apple's and very similar to what OS expects as behavior. It's not slap some controllers and flash chips on a PCB, change three fields in a firmware and sell.
Flash storage is black magic at this point, and thinking every box is the same is a big mistake.
This comment can be easily 3x longer, but I want to keep it readable.
That's just marketing language for "this is expensive af but you'll buy it because otherwise you're not an elite gamer!".
Their manufacturers provide 5 years warranties unlike Apple. AFAIK Apple doesn’t even disclose endurance ratings. Wouldn’t you expect the opposite?
Yep and it's possible to get much more out of it.
I've been running a Crucial MX100 256 GB SSD for 10 years. It's at 63% health from a S.M.A.R.T. readout. It's been powered on 125 times over ~10 years and transferred 56 TB in that time. It's my main Windows partition and runs WSL 2 where I've built and ran thousands of Docker images. Basically, it hasn't been sitting here unused.
At scale failures are more than just endurance ratings. Gaming laptops for example often cook their components due to prioritizing performance over long term stability. That doesn’t guarantee early failure, but it reduces the likelihood the system is working in 4 years.
(Not sure if Apple SSDs have power-loss protection. Not using sockets probably eliminates one source of accidental power loss.)
Samsung is a much better example of a manufacturer that Apple would be emulating, investing in their own controllers, etc, and certainly not leaving out security features with no plan.
Intel leaving the SSD market probably has more relation to Apple than WD or Samsung.
I'm not asserting that Apple succeeds in making high quality, I'm saying their hardware trust makes them uncompetitive with the entire PC market where some brands will deliver high enough quality and all brands have access to low quality dumping to reach scale, etc.
Ok, I think your first mistake here is assuming that retail SSDs are a priority for Samsung. What they actually care about are the SSDs they sell to PC OEMs, and the enterprise SSDs. They don't really need the retail product line to serve as some kind of experimental dumping grounds. The low-end models that make it to the retail product line exist because they already have similar models in mass production for PC OEMs, and they can make a retail version by changing from green solder mask to black solder mask and printing a new sticker. (The high-end retail models exist for much the same reason, but sometimes have more meaningful hardware differences because PC OEMs aren't so obsessive about maximizing scores on bad benchmarks.)
I don't really understand how everyone on this thread can explain the nuances that should be what is interesting while claiming everything is the same.
Either technology/politics has changed and organizational structures we claimed in the 1980s to be both unethical and inefficient are now only unethical, or they have not changed and we are in the same cycle with these organizations as before but somehow elongated..
But no, what is interesting to HN is that there's a lot of fine detail in the physical market, but somehow when searching for any interpretation, a widget is a widget and Apple is WD and nothing interesting will ever happen again.
I bought a 4xM.2 card and 4x2TB Samsung Pro SSDs, cost me $1,300, I got to keep the 1TB "system" SSD, and was faster, at 6.8GBps versus the system drive at 5.5.
Similar with memory. OWC literally sells the same memory as Apple (same manufacturer, same specifications. Apple also wanted $3,000 for 160GB of memory (going from 32 to 192). I paid $1,000.
But at least the tech press wrote a bit about it, for example here[1], including a link on how Samsung implements it using journaling on consumer SSDs. I would expect WD to do something similar given that multiple WD drives passed the test.
[1]: https://hothardware.com/news/heads-up-nvme-ssds-lose-data-po...
M4 Pro read/write: ~5.4/6.7 GB/s
Samsung 990 Pro 2TB: $170
990 Pro R/W: ~7.1/6.2 GB/s
Samsung 990 Evo 2TB: $130
990 Evo R/W: ~4.8/3.9 GB/s
Sorry no links, looked at tomshardware, Amazon, and macrumors forums for numbers.
I have an older 2TB one. USB 3.2. Plenty of speed for putting lots of media, large software packages (e.g. Xplane 12 with a lot of scenery) etc.
So it’s probably even considerable more than 500%.
Unless you believe that Apple only buys “magic” components like the 8GB=16GB crowd there is nothing particularly special about their storage or memory.
Their SSDs even aren’t that fast. You can get faster ones for $200 (EXCEPT they are 2TB instead of 256GB)
https://www.amazon.com/gp/aw/d/B0CMT7WMVM
Note that as all docks it requires displaylink software on mac to use more than one screen, and the software is a bit buggy - needs to be restarted every couple of times i reconnect to the dock for the displays to work right. TBH this is the first dock i experience this with but then again os and software updates, and ive seen weird flaky behavior before, just not specifically this.
> On Mac systems, dual display is only supported on M1 Pro/Max, M2 Pro/Max, M3 Pro/Max, and M4/Pro/Max systems
So if you're still on an OG M1 it won't work for you.
That requires DP.
Still incorrect.
What you actually have to use is two thunderbolt ports on the MacBook. Then you can power not just two but four 6K screens, as each TB port can push 2 externals (in addition to built-in screen).
So this iVANKY dock is native TB quad screen, for example:
https://www.amazon.com/FusionDock-Thunderbolt-Monitor-Dockin...
I have this device and don't have this problem? I have a couple of self-powered hubs downstream of this thing and have plugged them into the USB-A-shaped ports on the back.
I don't have any downstream Thunderbolt devices plugged into this thing. Maybe that's the major difference between my setup and the author's? (Or maybe I'm running better firmware on this thing than he is?)
I found this blog post, switched to one of the ASMedia ports (the rear 10 Gbps USB-C port or the most left-hand USB-A port) and both of the Ethernet controllers are rock-solid now.
I now have it in the Thunderbolt port which ekes out another 100 Mbps or so compared to the ASMedia USB ports.
The blog post is probably a bit sensationalistic but I still can't recommend the dock to anyone when half the ports on it are flakey, especially at that price.
> ...I still can't recommend the dock to anyone when half the ports on it are flakey, especially at that price.
This is understandable.
However, I'm pleased with my purchase and I'm only using the built-in Ethernet port, the built-in DisplayPort port, and a couple of USB ports. If it lasts me five years, I will have paid ~60 USD per year, which I think is a bargain when you compare it to fighting with something that's flaky no matter what you do, or just fails after a year or two... requiring you to go on the hunt for a new "dock".
Honestly, I think it's pretty damn stupid to have so many USB ports on the thing to begin with. Decent USB hubs are inexpensive; remove all but one or two of the USB ports and use now-spare bandwidth to put additional DisplayPort ports on the thing!
I had two expensive caldigit docks (TS3 i believe). One was warranty replacement for the first one. Each died after about a year.
After that I had an expensive alogic dock. Seemed great when it was working, but after just over a year it went dead (warranty is for two years).
The dock i had is out of stock. After weeks of “checking” I was offered an exchange for a cheaper version (whatever) that doesn’t include features i rely on (3 screens).
As “compensation” I was offered about 30% of the original price, since “the item was used”.
Thanks but no thanks. Standard amortization time for computer equipment is 5 years. And in either case, who has patience to go for weeks without their familiar computing environment? And my cost is replacement of the item, i did not rent it.
I got a Kensington dock sold by amazon at slightly less than the original price of that one, with better features and a brand name that is worth more than the piece of paper it is written on.
We’ll see how it lasts
So, I've been using a CalDigit TS3 Plus device for the last two or three years. I have USB 3, Ethernet, and DisplayPort going out from it, and a Thunderbolt cable going into it. Other than sometimes having to unplug and replug the DisplayPort cable to get the screen to wake [0], it works fine.
[0] To make this easy, I have an F<->F coupler near the display that doesn't have latches. I just slip out one end of the cable from the coupler and slip it back in. Quick and easy, if slightly annoying that I have to do it at all.
There's the proliferation of both "manager speak" and "self-help coach vocabulary", but that's (sadly) not unique to HN.
Nah, there's no need to apologize. If folks don't want to read and/or participate in the subthread that changes the topic, then they can just fold up the subthread using the handy "fold up the subthread" button.
And yeah, with the exception of the fellow who called it "HN slang" (or whatever) the folks who replied to you all have good answers to your question.
Given that i had a third unrelated dock fail recently it wouldn’t be unreasonable to suspect something on my end might be causing this, but then again that’s the only hardware that failed on me in a while and i don’t do anything that unusual besides having 3x4k screens plugged into it and the occasional mouse charging/flash drive drawing power off it.
In all cases the PSU of the dock died with it (but also the docks themselves) so i suspect current DL chipsets overheat and eventually burn out when pushing clost to their max resolution.
I don't push it too often, but when I do, it's fast enough to play six or more concurrent 1080p video layers in Resolume from a single Gen4x4 NVMe. It's not as fast as my M1's internal storage, but it does the job.
I tried several NVMe/TB4 enclosures. Some of them were junk, some were just okay, and this one (which I now have two of) is great: https://www.amazon.com/gp/product/B0CSFFMQWF
I now also have three of these 8TB NVMe SSDs: https://www.amazon.com/gp/product/B0D9WT512W (I also have a few others that are slower and/or smaller.)
I've tried a few docks, and this is the one I'm using now: https://www.amazon.com/gp/product/B0DBHG7486 It's a good dock, but unfortunately needs the DisplayLink driver to use the DisplayPort output. This works okay, but the monitor stays dark until the desktop is booted, and cannot be used for switching between Linux virtual consoles.
First off, because my ThinkPad has only one TB4 port that also serves as the USB-C PD input, I needed a dock if I wanted to use a TB4 accessory while not running from the battery. The dock is also quite valuable for using an external (4K/60Hz) display when using another TB4 peripheral.
The SSDs work fine when plugged into the non-TB4 port, but they operate at less than half of their potential speed, and are enumerated as /dev/sdX instead of /dev/nvmeXnX.
Operating the SSDs from the TB4 port gives variable performance depending upon what else is connected, and when it was connected. The Linux PCI+bridge enumeration has some issues with hotplugable devices. Various combinations of pci=assign-busses, realloc, native, hpbussize=XX, lastbus=XX, hpmmiosize=XXXM, hpmmioprefsize=XG will all give varying results. At best, with my 4K monitor operating, I can get 20gbps on one external SSD, or some division of that speed distributed amongst multiple other SSDs.
Leaving the PCIe enumeration to the kernel with no additional boot arguments did not go well with 23.10, but works better with 24.04. Hot-plugging performance is always a compromise, depending upon kernel parameters, and the order in which things are plugged in.
I have a fancy pluggable dock and after my Mac goes to sleep it sometimes stops working when the Mac wakes up. Often, it appears to go into a loop where it detects an external monitor for 5 seconds then disconnects. Pretty annoying and guess away after a reboot.
It's bad enough that you have to have a box hanging off your machine, but with most docks you have to have several boxes.
It does do the job including display passthrough.
For me some things worked better than expected (5K + 4K monitor at 60Hz, even though it states only one 5K or two 4K monitors), some things don't (work laptop detects the hub, but the displays stay blank.