I wonder if we would see GTX 1060 Max Q for low profile video cards?

[cbn]

According to the spec on this page Max Q GTX 1060 (at 1280sp) is only 60W to 70W with either 3GB or 6GB VRAM.

However, due to the only slightly reduced clocks (but higher amount of stream processors) the performance of a 60W to 70W Max Q GTX 1060 3GB should be in the same ballpark as the full desktop GTX 1060 3GB which only has 1152sp. (and 60W to 70W Max Q GTX 1060 with 6GB should be even faster than a full desktop GTX 1060 3GB)

GTX 1060 Max Q = 1280sp @1063 - 1265MHz/1341 - 1480MHz

GTX 1060 3GB desktop = 1152sp @ 1506 MHz/1708 MHz

That could make for a very interesting GPU for SFF desktop users looking for the maximum amount they could get out of their machines.

 

[Guru]

You can already downclock and limit the power of the 1060's, it will consume 60-70w with low clocks.

 

[cbn]

You can already downclock and limit the power of the 1060's, it will consume 60-70w with low clocks.

Those would be full size cards though, I was thinking of a low profile GTX 1060.

With that mentioned, there have been high watt low profile cards in the past (see below 106W Palit GTS 450 low profile with 6 pin power connector)

However, I think the 120W TDP of a full strength low profile GTX 1060 would be too much to even allow booting in some Pre-built SFF systems. This assuming the user even has two free SATA or two free molex for a 6 pin power adapter.

 

[nathanddrews]

If you had asked me 5 years ago when NUCs were first announced, I would have said something like, "This probably means that MXM is finally on its way to the retail channel!" But here we are, all this time later, and nothing to show for it... unless you want an external GPU enclosure that's the size of five NUCs!

I think it just comes down to the fact that nearly every AIB partner wants to: overclock, whisper-quiet, or both. That requires large HSF setups. There are some passively cooled models, but then the heatsink is 2.5 slots wide. Whenever someone breaks the mold (literally?), then end up charging a lot more to get single-slot and/or low-profile, so that steers people away. "Why would I want something more expensive that runs slower and hotter?"

GPUs are headed toward miniaturization though, just look at all the "mini" GPUs that have come out in recent years. Smaller PCBs and stubby designs on mid-high end products, more low-profile options on the sub-$150 cards. Pascal was kinda the first generation to drag and drop a desktop GPU into a laptop without changing nomenclature, so we have evidence it can be done.

 

[cbn]

Whenever someone breaks the mold (literally?), then end up charging a lot more to get single-slot and/or low-profile, so that steers people away. "Why would I want something more expensive that runs slower and hotter?"

That is true that the low profile GTX 1050 and low profile GTX 1050 Ti are a bit more expensive.

However, some people want (or need) a relatively small PC. (Eg, Dell Optiplex 9010 SFF is only 8.5 liters)

Also, I've noticed over the years now that used SFF PCs are cheaper than the full size counterparts......so buying one of the GTX 1050 Ti or GTX 1050 low profile cards + used SFF PC works out to be about the same cost as full size GTX 1050 Ti or GTX 1050 + used Tower.

 

[cbn]

AMD is also applying a low power treatment to their RX 580 in the form of a 65W TDP. This used in the ASUS ROG STRIX GL702ZC Notebook.

http://wccftech.com/asus-rog-strix-notebook-amd-ryzen-7-8-core-cpu/