“Build it, and they will come” must be NVIDIA’s thinking behind their latest consumer-focused GPU: the RTX 2080 Ti, which has been released alongside the RTX 2080. Following on from the Pascal architecture of the 1080 series, the 2080 series is based on a new Turing GPU architecture which features Tensor cores for AI (thereby potentially reducing GPU usage during machine learning workloads) and RT cores for ray tracing (rendering more realistic images). Unfortunately, there aren’t (m)any games that make use of these capabilities so the $1200 price tag on the RTX 2080 Ti Founders Edition is difficult to justify. The 2080 Ti also features Turing NVENC which is far more efficient than CPU encoding and alleviates the need for casual streamers to use a dedicated stream PC. On paper the 2080 Ti has 4352 CUDA cores, a base/boost clock of 1350/1545 MHz, 11GB of GDRR6 memory and a memory bandwidth of 616GB/s. The upshot is that it has around a 30% faster effective speed than the 1080 Ti, which at 18 months old continues to offer comparable value for money and currently dominates the high-end gaming market. Professional users such as game developers or 4K gamers may find value in the 2080 Ti but for typical users (@1080p), prices need to drop substantially before the 2080 Ti has much chance of widespread adoption. [Sep '18GPUPro]
The RTX 4090 is based on Nvidia’s Ada Lovelace architecture. It features 16,384 cores with base / boost clocks of 2.2 / 2.5 GHz, 24 GB of memory, a 384-bit memory bus, 128 3rd gen RT cores, 512 4th gen Tensor cores, DLSS 3 and a TDP of 450W. Performance gains will vary depending on the specific game and resolution. With a 4090 tier card 1080p in-game fps will often get CPU bottlenecked which prevents the GPU from delivering higher fps. At higher (often sub-optimal) resolutions (1440p, 4K etc) the 4090 will show increasing improvements compared to lesser cards. When fps are not CPU bottlenecked at all, such as during GPU benchmarks, the 4090 is around 75% faster than the 3090 and 60% faster than the 3090-Ti, these figures are approximate upper bounds for in-game fps improvements. The fact that the 4090 is currently the fastest performing graphics card available is reflected in its jaw-dropping MSRP of $1,600. Since PC gamers rarely buy AMD GPUs, Nvidia only have themselves to compete with. Surprisingly, $1,600 is actually reasonable value for money, when compared to previous gen. flagship cards. Prospective buyers will need a top of the line system to extract maximum performance from the 4090 and because of its monstrous dimensions, many will also need a new PC case. Consumers that demand value for money, should wait a few more months for the 4060 / 4070 models by which time AMD's 7900 series will also probably be heavily discounted. Alternatively, shoppers looking to buy in the near term should consider the last gen. 3060-Ti, which offers excellent real-world (1080p) performance at a fraction of the price ($400 USD). [Oct '22GPUPro]
We calculate effective 3D speed which estimates gaming performance for the top 12 games. Effective speed is adjusted by current prices to yield value for money. Our figures are checked against thousands of individual user ratings. The customizable table below combines these factors to bring you the definitive list of top GPUs. [GPUPro]
Welcome to our PC speed test tool. UserBenchmark will test your PC and compare the results to other users with the same components. You can quickly size up your PC, identify hardware problems and explore the best value for money upgrades.