The 5080 strategy is quite similar. The core count increases mildly (the original 4080 increases by 10.5 % and only 5 % at 4080 supermarkets). In addition, the same 256 -bit memory interface has the same 16GB RAM. However, if you switch from GDDR6X to GDDR7, you can obtain appropriately bumps in the memory bandwidth in the order of 3080 and 4080 SUPER.
Like the 5090, the 12 -pin power connector is slightly embedded and angled, so you are not far away.
Credit: Andrew Caningham
However, all of them are very calm compared to the core count, memory bass width, memory size, and the width of the memory bass, the width of the memory bass, and the width of the memory band that can be obtained from RTX 4090 to RTX 5090. It is gradual. The core count and memory bandwidth are below RTX 4090. This has the advantage that 5080 power requirements do not explode at 360 W. It was 40 ° higher than the 4080, and in the actual test, 5080 did not actually consume so many power on average, which is also meaningful. The 4090 and 5090 continue to leave other lineups that cannot be accessed to those who are less than tens of dollars on GPUs.
Regarding the 5080 NVIDIA Founders Edition version of physical design, it is the same as 5090, as 4080 looks like 4090. I feel 5080 is lighter than 5090. Power consumption and half CUDA core do not require the same type of cooling device. However, the dimensions and connectors of both cards are exactly the same.
GPU temperature measured while the benchmark is running.
Credit: Andrew Caningham
In our test, 5080 ran a little warmer than 4080 loaded, but was about 5 °. The 5080 Founders Edition should be a little easier to install one of these and keep the system cool, to execute 11 ° or 12 ° cooler than the 5090 Founders Edition.
