Nvidia RTX 4080 Graphics Card Review
Peter Donnell / 1 year ago
Nvidia RTX 4080 Hardware
When looking at the main specs, it’s a hard one to compare to, because, what do we compare the 4080 against? The 3080 10GB and 12GB because it is the natural successor in name, or against the 3080 Ti as NVIDIA claim the 4080 to be twice as fast, while using around 10% less power. On top of that, the RTX 4080 is also launching for the same $1199 price tag that we saw the 3080 Ti launching for, so I guess that seems like the obvious card to focus on in our comparisons.
Third-Generation RT Cores & Fourth-Generation Tensor Cores
When looking at the specs like for like, we do see some interesting points, as the RTX 4080 actually comes in with less CUDA cores, less stream processors, less texture mapping units, tensor cores, and RT cores, while the ROPs or Render Output Units stay the same at 112. Luckily, it’s not as clear cut as that, as NVIDIA are now claiming that their new streaming multiprocessors can offer up twice the performance and power efficiency, while the new third generation RT cores claim to give up to twice the Ray Tracing performance, and the new fourth generation tensor cores are touted to give up to twice the AI performance, and if history has taught us anything, it’s that these jumps have been fairly significant generation to generation and something we’ll definitely be putting to the test a little later on including DLSS3, especially now with more titles supporting it than ever and a lot more coming in the very near future, and I feel that this is where the 4080 is going to completely outshine the previous generation cards, and of course, the competition from AMD, where NVIDIA have always been the leader.
DLSS 3
A couple of areas when comparing the 3080 Ti and 4080 that do differ comes down to the memory, where the 4080 now comes with 16GB, opposed to 12GB, though it still uses the same GDDR6X memory, though it does operate at a lower 256-bit bus, giving us bandwidth of 716.8GB/s, and of course, power which sees the 4080 coming in with a TDP of 320 Watts, so a small reduction from the 350 Watt TDP of the 3080 Ti. Due to this reduction in power, and the new efficiency of the core, we’re now also seeing a slight reduction in the maximum GPU temperature from 93 degrees on the 3080 Ti to 90 degrees on the 4080. NVIDIA also recommends a 750 Watt PSU for the FE but obviously AIC cards with potential higher power limits may require more so definitely keep that in mind.
The other thing worth noting, is just like the 4090, it still uses the PCI-Express 4.0 x16 interface, even though we do have boards that support PCIe 5.0 on the market. Even though the 40 series is stupidly fast as a whole, if the 4090 is anything to go by, we’re not even close to maximising the throughput of bandwidth of PCI-Express 4 just yet hence the decision on that one.
In terms of the reference specs, the 4080 as standard and therefore the Founders Edition, comes with a GPU clock of 2205MHz, a boost clock of 2505MHz and a memory clock of 1400MHz or 22.4 Gbps Effective.
Now I did mention that I have a total of FIVE cards with me today and while we’re be focussing on the performance of the Founders card for the majority, we will be looking at all of the cards and how they compare performance wise a little later on, alongside the performance of the coolers, thermals, power and of course overclocking.
