Memory Chip Swap Mod SUPERcharges an RTX 2080 Ti

[xrror]

This is some old school awesome. I really hope they put the slower memory they took from the 2080ti and at least tried to re-surrected the two 2080 doner cards though!

Also, does anyone know if they had added the missing 12th chip to the 2080ti if that would have been detected? Or is that memory channel hard fused off?

I'm assuming it wouldn't be that easy, because surely some OEM would try and ship that as a product much to the ire of nVidia.

 

[Keullo-e]

Would be interesting, never heard that anyone had tried adding the 12th chip to 1080 Ti or 2080 Ti.

 

[Prima.Vera]

How much a Titan RTX pull on that bench?

 

[xrror]

Would be interesting, never heard that anyone had tried adding the 12th chip to 1080 Ti or 2080 Ti.

Well it's like... surely it can't be "that easy" because IF the card picked up the 12th chip.... to even talk to it means the memory controller to that pad is still alive and not fused off.

Suddenly you've jumped from 352 to 384-bit memory interface. What else might be still intact? Would you get the full 96 ROPS? Are the SM's still there? Tensor? Caches? etc etc
As much as I'd love that the "12th block" simply "stayed off" due to a simple lack of a memory chip there, I'd have to assume nVidia would be wise to what would happen if all it took to make a budget Titan/Quatro was just take any TU102 and fully populate all 12 spots with the larger DDR6 modules.

Actually as I type that... I have to wonder... has anyone tried flashing the Titan or Quadro bios to a 2080Ti? Because even if the 12th channel is laserburnt/blown disabled ... even a 11/12'ths 2080ti hacked to a titan/quadro might still be a compelling option for a shade-tree compute cluster.

(note: please don't brick your 2080ti following my aimless rambling!)

 

[Parn]

Great craftsmanship, but not very practical (or creative). I doubt any 2080 Ti owners have the guts to try this out except for the richest of them.

 

[Keullo-e]

Well it's like... surely it can't be "that easy" because IF the card picked up the 12th chip.... to even talk to it means the memory controller to that pad is still alive and not fused off.

Suddenly you've jumped from 352 to 384-bit memory interface. What else might be still intact? Would you get the full 96 ROPS? Are the SM's still there? Tensor? Caches? etc etc
As much as I'd love that the "12th block" simply "stayed off" due to a simple lack of a memory chip there, I'd have to assume nVidia would be wise to what would happen if all it took to make a budget Titan/Quatro was just take any TU102 and fully populate all 12 spots with the larger DDR6 modules.

Actually as I type that... I have to wonder... has anyone tried flashing the Titan or Quadro bios to a 2080Ti? Because even if the 12th channel is laserburnt/blown disabled ... even a 11/12'ths 2080ti hacked to a titan/quadro might still be a compelling option for a shade-tree compute cluster.

(note: please don't brick your 2080ti following my aimless rambling!)

You have good points there, but I just guess that Nvidia didn't think so far that someone just slaps a memory chip there.

It would be hella interesting to see someone trying a 1080 Ti / 2080 Ti with that 12th chip, just like these crazy dudes making their super 2080 Ti.