riobard

> What do these devices do that can't be accomplished by an OpenWrt One + an external AP for less money and fully FOSS?

Nice UI (as the company is best known for https://ui.com)

The assumed mentality of “being flexible” is the very reason WireGuard was created to fight against in the first place, otherwise why bother? IPSec is already standardized and with wide-spread hardware implementation (both FPGA and ASIC) and flexible.

Those pesky WD bridges usually support USB Bulk Storage only but not UASP, resulting in worse performance and higher CPU usage.

Also HDD power management is often complicated by the bridge chip sometimes intervening.

Not recommended for long-term use.

Yeah but those are pretty dated. I was under the impression those old Ampere servers are not efficient compared to modern EPYC anymore. So I’m wondering what their current generation of arm64 servers look like :p

What ARM64 machines are you using and what are they used for? Last year you were announcing Gen 12 servers on AMD EPYC (https://blog.cloudflare.com/gen-12-servers/), but IIRC there weren’t any mentions of ARM64. But now it seems you’re running ARM64 in full production.

Thank you very much for the clarification! :)

Thanks! The Cloudflare blog article explained GSO pretty well: application must send a contiguous data buffer with a fixed segment size (except for the tail of the buffer) for GSO to split into smaller packets. But how does GRO work on the receiving side?

For example GSO might split a 3.5KB data buffer into 4 UDP datagrams: U1, U2, U3, and U4, with U1/U2/U3 being 1KB and U4 being 512B. When U1~4 arrives on the receiving host, how does GRO deal with the different permutations of orderingof the four packets (assuming no loss) and pass them to the QUIC layer? Like if U1/U2/U3/U4 come in the original sending order GRO can batch nicely. But what if they come in the order U1/U4/U3/U2? How does GRO deal with the fact that U4 is shorter?

What mail clients do you recommend that make mailing list UI/UX tolerable?

Can someone explain how UDP GSO/GRO works in detail? Since UDP packets can arrive out-or-order, how does a single large QUIC packet be split into multiple smaller UDP packets without any header sequence number, and how does the receiving side know the order of the UDP packets to merge?

Unfortunately ads and tracking go hand in hand: without tracking, targeted ads are much less effective. If everything else stays the same, you'll just get more, bigger, and flashing ads to compensate for the inefficiency, like what you see today on smaller sites without sophisticated user data.

I'd certainly hope ISP follow your reasoning. BTW that was why I put my WireGuard server on port 443, hoping it would avoid the tragic QoS…

The paper is somewhat title-bait: most of the data flow and privacy concerns (section 3) isn't caused by eSIM but by MVNO and business practices which applies to physical SIM as well. Additionally, it's expecting too much from _travel_ SIM cards (physical or embedded) where the primary consumer concern is cost.

> Never had a single problem with it (but increased latency because of weird routings around the world).

UDP (which WireGuard uses to encapsulate your data) traffic is often de-prioritized. You won't notice it when the network load is low, but it will seriously degrade experience during high load periods.

Apple’s pricing structure is: X SE < X < X Air < X Pro.

Ah yes! Port multiplier is usually the source of most evils (after flaky bridge chips). Unfortunately enclosure makers seldom reveal the internal topology and usually only test again Windows, and Linux kernel has a long blacklist of bad bridge chips…

The issue with flaky bridge chips wasn't usually about data integrity——it works fine most of the time, i.e. data written got read back correctly.

But often after extensive use, `dmesg` would complain about problems when talking to the drives, e.g. drive not responding or other strange error messages (forgot the exact text but very irritating and google-fu didn't help). There were also problems with SMART commands passthrough and drive power management e.g. sleep/standby adjustment which wasn't reliable when talking via bridge chips.

I use only disks directly connected to SATA controllers afterwards and no such issues ever happened again.

Sounds great! Do you mind share the model of the enclosures and what bridge chip is used inside?

Have you researched the USB-SATA bridge chips in the enclosures? Reliability of those chips/drivers on Linux used to be very questionable a few years ago when I looked around. Not sure if the situation has improved recently given the popularity of NAS devices.

Not all Intel Macs. Only those without T2 chip, as T2 acts as the storage controller.

Would four 2TB SSD be more or less expensive than one 8TB SSD? And also counting power efficiency and RAID complexity?

I’ve been always puzzled by the strange choice of raiding multiple small capacity M.2 NVMe in these tiny low-end Intel boxes with severely limited PCIe lanes using only one lane per SSD.

Why not a single large capacity M.2 SSD using 4 full lanes and proper backup with a cheaper , larger capacity and more reliable spinning disk?

If they do trust their hardware, APFS won’t need to checksum fs metadata either, so I guess they don’t trust it well enough? Also I have external drives that is not Apple sanctioned to store files and I don’t trust them enough either, and there’s no choice of user data checksumming at all.

APFS checksums only fs metadata not user data which is a pita. Presumably because APFS is used on single drive systems and there’s no redundancy to recover from anyway. Still, not ideal.

Can we please stop using request/day as a perf metric?

Stop spreading misinformation.

> You still need a stateful server to assign IPv6 addresses for most use cases, through DHCPv6. SLAAC doesn't even give you a DNS server yet.

DNS now comes in Router Advertisement per RFC 8106. No need for DHCPv6 anymore.

> And even if it did, many ISPs assign too small address spaces for SLAAC, or your networks isn't so simple that you can just auto-negotiate some address.

Most residential ISPs allocate in /48, /52, /56, or /60. Even if they allocate in the smallest /64, it's still perfectly fine for SLAAC for most home users utilizing a single subnet.

The 72 PPI number was used back in the days when typical resolution was like 640/800/1024 horizontal pixels and display less than 20 inch diagonal. It was the Golden Age of pixel art.

Apple had been experimenting with the idea of Resolution Independence during the transition to “high resolution” LCD displays (1280/1440/1680/1920 horizontal pixels) before giving up to integer 2x (and later 3x on iPhone to overcome deficiency of OLED sub pixels) scaling on what they now call Retina displays. This was the rapid transit to 100~110PPI for non-retina and 220 PPI for retina on desktop and over 300PPI on mobile.

Modern Mac can also do non-integer scaling (eg 1.2x 1.5x) well enough on retina displays if you need extra working area. For example all current Apple Silicon MacBook Air default to non-integer scaling on its internal screen.

I don’t know any Mac support that.

That's the tricky question. It doesn't have one (or it has many???)

It's the same LG panel (LM315STA-SSA1) used by Dell's ugly-as-hell U3224KB, but in an all-aluminum case and stand weighing about 9KG. Assembled in China at half the cost of U3224KB and various Chinese brands are selling it with their logos etched on the back.

That's only one of the reasons. 8K 10-bit RGB requires ~60Gbps while HDMI 2.1 can carry at most 42Gbps. So video has to be compressed somehow and it's mostly likely YCbCr with subsampling (coz that's how most video is encoded anyway). HDMI 2.1 also supports DSC so it's an option but I doubt any non-computer-monitor-oriented TV will support that.

Another reason is PPI. macOS works best ~220PPI for 2x HiDPI mode so 8K would be about 40". But current batch of 8K TV comes in sizes over 60", which is too big for most desks.

There're also convenience factors like DDC control (where you can control brightness and volume using software on your computer) which most modern computer monitors support but I've never seen any TV supporting that. Without HDMI CEC on the computer side, you can't even auto wakeup the TV when you wake up the computer.

At ~220PPI and 2x HiDPI mode it works as properly as Apple's Pro XDR 6K display: everything is crisply clear and the working area is vast. You basically get two and a half 4K screens seamlessly in one rectangle.

Actually it has 4% more pixels than 6K Pro Display XDR (6144x3456 vs 6016x3384) and 44% more pixels than 5K Studio Display (5120x2880). It is the highest resolution computer monitor I can get at the moment. (8K and higher resolutions are mostly TV sets or cinema equipment, not optimized for computer use.)

I was even trying to attach two of these 6K babies to the M4 Mac mini. According to Apple spec it should work, but no matter how I tried, the second attached display only gets max 4K signal. Maybe it needs M4 Pro.