Or lay a motherboard on top of the cable running flat out, this would represent a real world scenario.
What do you even do with a 1m riser cable?
3.0 has effectively double the bandwidth. For gaming, the difference is only slight, around 10% for the 1080 at 1920x1080, but that gap will reduce as framerate reduces (unlikely to be a problem with a 1060).Is there any big difference between PCIe x4 2.0 and PCIe x4 3.0 ?
The tricky part here is thin-ITX's PCIe slot is not intended to provide the 75W a purely bus-powered card expects to draw. For the 1060, this should* not be an issue as it has external PCIe power connectors which the card will draw from preferentially. Your only remaining problem would be that you now need a second PSU to power the GPU, separately from the thin-ITX board which will be powering itself from the 19V input. People have mostly addressed this by using a single large 19V brick connected to the thin ITX board, then using the 19V output on the thin ITX board (note, some boards lack this, or just have pads rather than a convenient connector) to power an internal DC-DC PSU that powers the GPU. This runs them from the same ground, so avoids ground plane mismatch issues.Is there a real risk to plug a GTX 1060 into a PCIe x4 slot (thin mini-itx board) ?
The tricky part here is thin-ITX's PCIe slot is not intended to provide the 75W a purely bus-powered card expects to draw. For the 1060, this should* not be an issue as it has external PCIe power connectors which the card will draw from preferentially. Your only remaining problem would be that you now need a second PSU to power the GPU, separately from the thin-ITX board which will be powering itself from the 19V input. People have mostly addressed this by using a single large 19V brick connected to the thin ITX board, then using the 19V output on the thin ITX board (note, some boards lack this, or just have pads rather than a convenient connector) to power an internal DC-DC PSU that powers the GPU. This runs them from the same ground, so avoids ground plane mismatch issues.
The RX 480 modification just slightly pulled the current draw back within the 75W (or rather, 66W from 12V alone) spec limit. The power draw distribution is mostly fixed by the physical connection of the VRMs to power planes on the card PCB.- Just in case, would it possible to mod the GPU bios to force it to draw only 25W from the PCIe slot ?
Is it not this way (sofwtare solution, not hardware) AMD solved the RX480 issue ?
200W at 19V is 10.5A, which should be fine.- By using the onboard 7.4mm barrel connector, is there any risk to get very close to the maximum wattage ratings and burn the connector ?
I expect my system to draw 200W maximum.
A single barrel connector should be fine for 200WThe idea was to use a modded Dell brick with two 7.4 barrel connectors, one for the motherboard, the other for the DC-DC board.
This may be a way to avoid overloading the single 7.4 barrel connector of the motherboard ? What do yout think ? Is it necessary ?
If the 2pin connector is present, then it is linked to the same power planes as the DC barrel. This is part of the Thin ITX specifications (it;s intended to be a DC in plug, but in practice it's just a different way to connect to the same power components that the DC barrel connector links to. There are no diodes to enforce directionality.- If I were to use a single 19V brick plugged in the sole motherboard barrel connector, how can I be sure that the 19V 2Pin inlet can also be used as passthrough ? This was my initial idea as well but it is not specified in the mobo manual.
http://asrock.pc.cdn.bitgravity.com/Manual/H110TM-ITX.pdf (p. 20)
The Add2PSU works fine, it's just excessively large.- In any case, the DC-DC board is synced with the motherboard using an ADD2PSU pcb, thanks to a Sata power to molex adapter.
This has been reported to work.