Gigabyte Aero 15 X9 Laptop Review: RTX Inside

Unfortunately there was never going to be significant GPU gains until Nvidia transition onto a much better process. 12nm TSMC isn't exactly a huge gain over 16nm and we see exactly that in laptops because they are especially constrained by performance per watt.

This 2070 Max Q is only 10 percent faster than the Max Q 1070. Which is now virtually a two year old part- a 10 percent gain in two years is not getting anyone excited.

For laptop graphics to take a step up across the board we need some 7nm parts. It's likely at least another year away.
 
My biggest gripe with my Aero 2018 is that the manager software Gigabyte uses runs as an application instead of a service so when you're tabbing through open apps you have a blank window that says "GBSOD". If you close that window you can no longer use the function keys change the screen brightness until you reboot. It's maddening.
 
Unfortunately there was never going to be significant GPU gains until Nvidia transition onto a much better process. 12nm TSMC isn't exactly a huge gain over 16nm and we see exactly that in laptops because they are especially constrained by performance per watt.

This 2070 Max Q is only 10 percent faster than the Max Q 1070. Which is now virtually a two year old part- a 10 percent gain in two years is not getting anyone excited.

For laptop graphics to take a step up across the board we need some 7nm parts. It's likely at least another year away.

It will be interesting to see Nvidia's 7nm parts, mostly because their die size is so massive. That'll be a real good indicator of how mature the process is.
 
It will be interesting to see Nvidia's 7nm parts, mostly because their die size is so massive. That'll be a real good indicator of how mature the process is.

I would say that an RTX2080Ti die that is massive on 12nm TSMC would be comfortably manageable on 7nm Samsung or TSMC. Anything bigger might take a bit longer to arrive, more maturity required. Most indicators are Nvidia won't launch 7nm parts until early 2020. The process should have had an entire year of mass production by the time Nvidia need stuff ~500mm² churned out.

Nvidia could gain 25-30 percent performance across the board with their current RTX lineup for the same power envelope without even bothering to tweak the design specifically for 7nm. Radeon 7 is evidence of that.

One assumes Nvidia will also go for more architecture advances, realistically we could see an uplift of 40 percent on Nvidia's GPU range. 7nm will be very welcome.

After Nvidia's 2020 7nm GPUs we could be in for a seriously long generation, surpassing even Pascal's lifespan. Moore's law is long dead. I plan to buy the fastest 7nm card I can next year, and I bet it won't be significantly surpassed for at least 3 years after that.....
 
I would say that an RTX2080Ti die that is massive on 12nm TSMC would be comfortably manageable on 7nm Samsung or TSMC. Anything bigger might take a bit longer to arrive, more maturity required. Most indicators are Nvidia won't launch 7nm parts until early 2020. The process should have had an entire year of mass production by the time Nvidia need stuff ~500mm² churned out.

Nvidia could gain 25-30 percent performance across the board with their current RTX lineup for the same power envelope without even bothering to tweak the design specifically for 7nm. Radeon 7 is evidence of that.

One assumes Nvidia will also go for more architecture advances, realistically we could see an uplift of 40 percent on Nvidia's GPU range. 7nm will be very welcome.

After Nvidia's 2020 7nm GPUs we could be in for a seriously long generation, surpassing even Pascal's lifespan. Moore's law is long dead. I plan to buy the fastest 7nm card I can next year, and I bet it won't be significantly surpassed for at least 3 years after that.....

I say, yes 40% is a very realistic number for Nvidia's 7nm parts. That might depend on what Nvidia does with RT cores though. If Nvidia opt to use more die space for RT cores, which would be required if they wish to push RT past only doing a single effect at a time, it's going to eat away at some of that rasterization performance.

I can't predict the future so I have no idea what's going to happen. All I can say is 2020 will still be very early for real time ray tracing and we'll see if Nvidia is going to fully commit by beefing RT performance or if they are fine keeping it in line with the likes of other GameWorks tech.
 
Back