NVIDIA RTX 4080 Benchmarked In Geekbench, Scores 25% Better Than The RTX 3090

The GeForce RTX 4080 is team green’s latest addition to their Ada Lovelace lineup. The RTX 4080 features the AD103 GPU along with 16GB of GDDR6X memory. While we still have around 10 days before this GPU retails, benchmarks have already surfaced showcasing what sort of performance you should expect. Over at Geekbench, the RTX 4080 has been tested in the CUDA, VULKAN, and OpenCL APIs.

The Test Bench

The same test setup was used for all these benchmarks. Therefore, it is safe to assume that they are all from the same source. Going into the specifics, we see AMD’s Ryzen 9 7950X paired with 32GB of RAM. As for the performance metrics, let us go over them one by one.


Starting off with CUDA, which is exclusive to NVIDIA GPUs, the RTX 4080 scores 300728 points. This is a nice 45% improvement as compared to the RTX 3080. Across the board, it remains ahead of the pack because the only real competitor here is the RTX 4090.

CUDA Benchmarks


Moving over to VULKAN, the RTX 4080 manages to score 289014 points. The RX 7900 XTX showcased here is assumed to be 1.7x faster than the 6900XT. That may not be the final result, but it can give us a rough idea of where RDNA3 stands. Interestingly, the RTX 4080 here loses to the 7900XTX despite being more expensive.

VULKAN Benchmarks


OpenCL sees the RTX 4080 manage to obtain 148838 points, marginally beating the RTX 3090. Interestingly, AMD’s 7900XTX is faster than even the RTX 4090 in this test. As for the RTX 4080, it offers a 7% improvement generation on generation against the RTX 3080. That’s kind of disappointing though the final reviews will tell the complete story.

OpenCL Benchmarks

RTX 4080 

The RTX 4080 16GB is packed with 16GB of GDDR6X memory over a 256-bit bus where the effective bandwidth stands at a whopping 720GB/s. Such insane bandwidths can be achieved thanks to the 22.5Gb/s memory which is 20% faster than last-gen. The 4080 16GB is equipped with the state-of-the-art AD103 GPU and 9,728 CUDA Cores for that extra compute performance. To top it off, it consumes just 320W of power.

SKUChipFP32/CUDASMsMax ClockCacheMemory BusVRAMMemory SpecSpeed (Gbps)TDP
RTX 4000 TitanAD102-450184321443.0GHz+?96MB?384/382-bit48GBGDDR6X24~800W
RTX 4090 TiAD102-350181761423.0 GHz?96MB382-bit24GBGDDR6X24600W?
RTX 4090AD102-300-A1163841282.52GHz96MB384-bit24GBGDDR6X21
450W+ (TGP) / 660W (Max TGP)
RTX 4080 TiAD102148481162.7 GHz?80MB?320-bit20GBGDDR6X23420W
RTX 4080 (16GB)AD103-300-A19728762.505GHz64MB256-bit16GBGDDR6X22.5
320W(TGP)/ 516W (Max TGP)
RTX 4070 Ti (4080 12GB Revamped) (Could Be Different)AD104-400-A17680602.61GHz48MB192-bit12GBGDDR6X21
285W (TGP) /366W (Max TGP)?
RTX 4070 AD104-300?7168562.61GHz?48MB?192-bit/160-bit12GB?GDDR6X21?250W
RTX 4060 TiAD1046656523.2 GHz?40MB?160-bit10GBGDDR6X (?)18200W
RTX 4060AD1064352343.0 GHz?32MB128-bit8GBGDDR6X (?)18180W


Source : CUDA, VULKAN, OpenCL and Videocardz


Abdullah Faisal

With a love for computers since the age of five, Abdullah has always sought to delve into the depths of information, and uses it as his guiding light. He believes success is of utmost importance as history is written by the victor.