Richard
@messaroundery.net
Co-developer @86box.net, compulsive storyteller with the most random interests.
The switch turns 16x 2.0 lanes from 990FX into 32x 3.0 lanes for the slots, as weird as it sounds. There is no 16x+16x uplink, the switch does support it but AFAIK you can't bond them together to speed up a single card (the 2019 Mac Pro has this but only lets you select one uplink for each slot).
October 9, 2025 at 2:05 PM
The switch turns 16x 2.0 lanes from 990FX into 32x 3.0 lanes for the slots, as weird as it sounds. There is no 16x+16x uplink, the switch does support it but AFAIK you can't bond them together to speed up a single card (the 2019 Mac Pro has this but only lets you select one uplink for each slot).
Piecing together all available info:
October 9, 2025 at 1:34 PM
Piecing together all available info:
Interesting yet deceptive. They put a PCIe 3.0 switch on a single 2.0 x16 uplink (the manual proves the article wrong) and called it a 3.0 board. It's useful for corner cases at best, like running 3.0 x4 SSDs from years later.
October 9, 2025 at 1:15 PM
Interesting yet deceptive. They put a PCIe 3.0 switch on a single 2.0 x16 uplink (the manual proves the article wrong) and called it a 3.0 board. It's useful for corner cases at best, like running 3.0 x4 SSDs from years later.
The STPC Client was only specified in 66x1 and 75x1 bins, there is a later Consumer II at 66x2 with the same package and feature set but it might not be pin compatible.
These chips are fascinating, there's a final 686-class (Rise mP6 based) one that apparently never shipped in volume.
These chips are fascinating, there's a final 686-class (Rise mP6 based) one that apparently never shipped in volume.
September 17, 2025 at 1:25 AM
The STPC Client was only specified in 66x1 and 75x1 bins, there is a later Consumer II at 66x2 with the same package and feature set but it might not be pin compatible.
These chips are fascinating, there's a final 686-class (Rise mP6 based) one that apparently never shipped in volume.
These chips are fascinating, there's a final 686-class (Rise mP6 based) one that apparently never shipped in volume.
Right up there with the PORNITOR
August 26, 2025 at 10:44 PM
Right up there with the PORNITOR
The underlying plastic plays a role. On this laptop, isopropyl just stripped the coating to reveal translucent plastic underneath, but degreaser cleaned the coating without stripping it: bsky.app/profile/mess...
In trying to clean the sticky rubberized finish on this old laptop, I ended up stripping a chunk of the coating, then asked myself: why not go the rest of the way? It looks way cooler than I expected. Happy little accident.
May 23, 2025 at 2:17 PM
The underlying plastic plays a role. On this laptop, isopropyl just stripped the coating to reveal translucent plastic underneath, but degreaser cleaned the coating without stripping it: bsky.app/profile/mess...
The active USB extension cable has failed. Hub no longer enumerates. I (literally) cracked it open and found the chip is heating up quite a bit on one end. Very minimal design, too bad the chip markings were sanded off.
March 12, 2025 at 9:31 PM
The active USB extension cable has failed. Hub no longer enumerates. I (literally) cracked it open and found the chip is heating up quite a bit on one end. Very minimal design, too bad the chip markings were sanded off.