Enclosure & Hardware Discussions
settle an argument over 1080 v 1080ti TB3 performance egpu
 

settle an argument over 1080 v 1080ti TB3 performance egpu  

  RSS

milkyjoe2
(@milkyjoe2)
Eminent Member
Joined: 3 years ago
 

this guy says, that a 1080ti will perform much better than a 1080 over TB3 egpu, and that there will be no issue with bandwidth bottlenecking at higher resolutions

but i think the 1080ti will be bottlenecked and not perform much better than a 1080

who is right

part of his argument says "Bandwidth bottleneck does not come into play at higher resolutions, "

i asked someone on the egpu discord who agrees with me saying "not worth using a 1080ti, minor performance upgrade"

the guy i'm arguing with wants some kind of proof, so if anyone has the maths / definitive theory or benchmarks, that would help

 

 

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
acantril
(@acantril)
Eminent Member
Joined: 2 years ago
 

I have a 1080 and 1080ti - both of which have been used in a Mantiz Venus eGPU case.

the 1080ti achieves a significant bump in FPS - along the lines of what you would expect with a 1080->1080ti upgrade.

Both cards suffer a slight performance hit - which is minimised if you are using external displays. I wouldn't say there is more or less impact with the 1080ti ... its the same.

Generally you see more limit on the ti's when you increase GFX settings that impact CPU ... and most people don't check that as a cause and blame the TB3 limit.

 

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


itsage liked
ReplyQuote
drax392
(@dieseldrax)
Estimable Member
Joined: 2 years ago
 

During most testing, it seems the eGPUs connected to an external display run at about 80% of the performance of what the card is capable of when used over TB3.  I think it was 80%, it's been a while since I looked at the numbers.  I haven't seen anything to indicate there is a wall that is hit by the high-end cards that ends up making them perform the same as a lesser card and thus makes them a waste of money for an eGPU.

As such, the difference between a 1080 and a 1080Ti in eGPU configuration should be basically the same % difference as between a 1080 and 1080Ti in a proper desktop.

Intel Skull Canyon NUC (NUC6i7KYK) - 1TB Samsung 960 EVO NMVe SSD, 32GB memory
Apple Mac Pro 2013 - Deca-core 3GHz Xeon, 32GB memory, 1TB Apple SSD, D700 dGPU, eGPU shared with above NUC.
AKiTiO Node - EVGA GTX 1080Ti SC2 Gaming, Corsair SF600 PSU, Cougar Vortex HDB 120 case fan
2013 17" ASUS G750JS (TB2) + [email protected] (AKiTiO Node via Apple TB3->TB adapter) + Win10


ppm liked
ReplyQuote
Dude
 Dude
(@dude)
Active Member
Joined: 2 years ago
 

I you'll permit a side question, I recall reading on these forums that Nvidia won't work under High Sierra? Something to do with Apple and the way it handles drivers?

I'm leaning towards a Vega card in December for this reason, when drivers are out for the iMac Pro.

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
acantril
(@acantril)
Eminent Member
Joined: 2 years ago
 

Good question re. nvidia ... i think the initial issue is the NVIDIA web driver availability. 

Once we have those, we'll see i guess.

I'd hold off if you are making a buying decision until there is some clarity. I have a VEGA64 and the drivers are pretty bad for macOS tbh.

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote
milkyjoe2
(@milkyjoe2)
Eminent Member
Joined: 3 years ago
 

does this mean that a 1080 will also perform, against a 1070 over TB2 egpu, roughly the same comparative improvement you might expect for a 1080 over 1070 in a desktop build?

Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts


ReplyQuote