There are three new products in the RTX40 SUPER series, namely RTX 4070 SUPER, RTX 4070 Ti SUPER and RTX 4080 SUPER. The RTX 4080 SUPER is the only model with increased quantity and reduced price. The specifications are stronger than the RTX 4080, but the starting price is lower. If we start from this perspective, then the RTX 4080 SUPER can be said to be the most conscientious model among the three. What I bring to you today is a review of the TUF GeForce RTX 4080 SUPER O16G GAMING (hereinafter referred to as TUF RTX 4080 SUPER OC) graphics card from ASUS .
Appearance design
The packaging box of TUF RTX 4080 SUPER OC is consistent with its own RTX 4080. All four corners are “cut”. The octagonal design is still very distinctive among new cards.
Coming to the graphics card itself, the TUF RTX 4080 SUPER OC comes in matte black color with a frosted surface. The whole thing is not particularly high-profile and has a high quality. The air deflector remains the same as the previous generation, with sharp lines and sharp edges.
TUF RTX 4080 SUPER OC
RTX 4080 SUPER OC is designed with three fans, all of which are axial fans. The orientation and direction of the left and right fan blades are different from those in the middle. The left and right fans rotate counterclockwise, and the middle fan rotates clockwise. The reason for this design is to reduce turbulence and increase the air volume of the radiator. The center of the left and right fans has the TUF logo, and the center fan has the ASUS logo. All three fans have 7 blades and support intelligent stop.
The back of TUF 4080 SUPER OC uses an aluminum backplate, and the outermost position corresponding to the fan is a hollow design to improve heat dissipation capacity. Corresponding to the core, there is also a performance mode and quiet mode switching lever near the top. This graphics card takes up 3.65 slots. You can see very thick cooling fins from the top, and there are 8 heat pipes running through it. The cooling specifications are quite high-end. There is a 12VHPWR power supply interface in the center, a luminous TUF logo on one side, and a luminous light strip next to it. Both support ARGB.
In terms of interfaces, TUF RTX 4080 SUPER OC provides 3 DP1.4a and 2 HDMI2.1a, with a maximum supported resolution of 7680×4320. Overall, the appearance design of TUF 4080 SUPER OC has the feel of a thug in a suit, a character who is ruthless and doesn’t talk much.
Brief analysis of NVIDIA GeForce RTX 4080 SUPER architecture
The three newly released GeForce RTX 40 SUPER graphics cards still use NVIDIA Ada Lovelace architecture and TSMC4N NVIDIA custom process. Compared with the previous generation NVIDIA Ampere, NVIDIA Ada Lovelace has more than 2 times the performance improvement under the same power, reaching up to 90-TFLOPS shader data throughput.
The core code of RTX 4080 SUPER is AD103-400-A1, with a total of 10240 CUDA cores, providing 52-TFLOPS computing power; 80 third-generation Ada RT Cores have 121RT-TFLOPS; 320 fourth-generation Tensor Cores can provide 836Tensor -TFLOPS(AITOPS).
Cache capacity of RTX 4080 SUPER
The L2 Cache capacity of RTX 4080 SUPER is 64MB and the bandwidth is 736GB/s. NVIDIA officials also emphasized the importance of L2 Cache. We use two pictures to briefly explain the role of L2 cache.
The path for the GPU to obtain data is L1 cache, L2 cache, video memory and system memory. The L1 cache is located right next to the processing core and has very low latency. The first choice for the GPU to find data is the L1 cache. If the target data is not in the L1 cache, the GPU will look for it in the L2 cache. Compared with the L1 cache, although the latency of the L2 cache cannot be as low as the L1 cache, the capacity can be increased. If the data is not in the L2 cache either, the GPU will look for it in video memory. If it is not found again, it will be obtained from the system memory.
In fact, it can be see that the earlier the GPU obtains data, the higher the performance and efficiency will be. Since the core and L1 cache are located too close, the capacity of the L1 cache cannot be very large. Therefore, a larger L2 cache will effectively improve the data hit rate, thus improving the overall operating efficiency. RTX 4080 SUPER has a larger L2 cache capacity, so the probability of hitting data is higher.
Introduction to test platform
In order to ensure the performance of RTX 4080 SUPER OC, the processor selected is Intel 14th generation Core i7-14700K. And the energy efficiency core is turn off in the motherboard BIOS. The motherboard is ROG MAXIMUS Z790 DAKR HERO from ASUS; the memory is two 16GB DDR5 7200 memories, forming a 32GB capacity. The power supply is ASUS ROG Thor 2nd generation 1600W titanium. The running system is Windows 11 23H2.
The specs of the RTX 4080 SUPER OC can be see via GPU-Z (2.57.0). It has 10240 CUDA cores with a base frequency of 2295MHz and a Boost frequency of 2610MHz. There are 112 and 320 raster and texture units respectively. It uses 16GB GDDR6X video memory with a memory width of 256bit and a bandwidth of 736.3GB/s. The driver version is 551.22.
Theoretical performance test
This test of TUF RTX 4080 SUPER OC mainly compares two graphics cards, RTX 4080 and RTX 3080 Ti. Since some test items are relatively new, comparative graphics card data may be missing.
The following is the 3DMARK FS suite use to measure the theoretical performance of the graphics card DX11 : FS, FSE. And FSU respectively correspond to the theoretical performance of the graphics card at 1080P, 2K and 4K.
[…] […]
[…] […]