> On 18 Sep, 2023, at 11:24 pm, David P. Reed via Cake > <cake@lists.bufferbloat.net> wrote: > > I'm curious, too! We know that on older home routers, with really slow MIPS > processors, Cake struggles with GigE. As these old MIPS designs get phased > out and replaced by ARM designs, it will matter.
All qdiscs struggle with gigabit speeds on these SoCs. It's because their networking architecture is optimised for passing packets straight through on the switch fabric, not pulling them into the CPU on the way. The bandwidth to the CPU is comparable to an old PCI 32b 33MHz bus. At 100Mbps or so each way, they can keep up just fine. > Raspberry Pi 4's just aren't very good at networking because of their I/O > architecture on the board, just as they are slow at USB in general. That's > why the CM4 is interesting. It's interesting that the PiHole has gotten so > popular - it would run better on an Pi with a better network architecture. On the contrary, the Pi 4 has an excellent I/O architecture compared to most of its peers, and especially compared to the previous Pis. The built-in NIC is internal to the SoC and *NOT* attached via USB any more, so it can genuinely support gigabit speeds. The USB interface is also fast enough to support a second GigE NIC, though the latency wouldn't be as good as one attached over PCIe. That's with a standard, off-the-shelf Pi 4B. Here are some single-armed (just the internal NIC) results we ran a while ago. Note that Cake is actually faster than fq_codel - or rather, Cake's internal shaper is faster than HTB shaping with an fq_codel leaf queue - and (not shown here) both are faster on the Pi 4 than on an x86-based APU2. - Jonathan Morton
_______________________________________________ Cake mailing list Cake@lists.bufferbloat.net https://lists.bufferbloat.net/listinfo/cake
