ASUS GeForce RTX 3080 TUF Gaming OC Review 173

ASUS GeForce RTX 3080 TUF Gaming OC Review

Pictures & Teardown »



The ASUS TUF Gaming GeForce RTX 3080 OC is the company's biggest bet on the TUF Gaming brand since its transformation from a hyper-durable motherboard label to a gaming-centric name targeted at gamers looking for durable, value-conscious products a few years ago. ASUS typically positions TUF Gaming below the coveted ROG brand. Over the past few years, we've seen the company design TUF Gaming graphics cards based on entry-thru-mainstream GPUs, but this is the first outing with an enthusiast-segment GPU, NVIDIA's flagship GeForce RTX 3080 "Ampere," which debuted earlier this month. ASUS directed considerable engineering efforts into making this a premium product with several practical features.

The GeForce RTX 3080 by NVIDIA, based on the Ampere graphics architecture, is the green team's first new consumer graphics technology in two years, and an exercise at making real-time raytracing meet next-generation performance. NVIDIA took the bold step of introducing real-time raytracing to the gaming segment in 2018 with RTX—a technology that combines real-time raytraced elements, such as lighting, shadow, reflections, ambient occlusion, and global illumination, with conventional raster 3D scenes to significantly increase realism. With Ampere, NVIDIA is introducing its 2nd generation with even more RTX effects at better performance. The RTX 3080 is designed to make AAA gaming with raytracing at 4K UHD possible at 60 Hz, or with high refresh-rates at lower resolution.

The GeForce Ampere architecture introduces a new generation double-throughput CUDA core that can perform concurrent FP32+INT32 math. The new 2nd generation RT core doubles the BVH traversal and intersection performance over the previous generation and introduces new fixed-function hardware that enables newer RTX effects, such as raytraced motion blur. The 3rd generation Tensor core shares many design elements with its HPC cousin powering the A100 Tensor Core processor NVIDIA launched this spring, which leverages the sparsity phenomenon in deep-learning neural networks to double AI inference performance by an order of magnitude. NVIDIA heavily uses AI in its raytracing pipeline, and image-quality features such as DLSS. The new DLSS 8K feature takes a stab at 8K gaming by taking advantage of AI.

The new GeForce RTX 3080 features more than double the CUDA core counts than the previous-generation, with over 8,704 CUDA cores, 68 RT cores, 272 tensor cores, 272 TMUs, and 96 ROPs. All this compute muscle is fed data by an updated memory setup consisting of 10 GB of 19 Gbps GDDR6X memory across a 320-bit wide memory interface, which works out to 760 GB/s of bandwidth; that's 70% higher than the previous generation. There are several other "next generation" bits, such as support for PCI-Express 4.0 x16 bus, the latest HDMI 2.1 and DisplayPort 1.4a connectivity, and AV1 video acceleration. NVIDIA built the GA102 over the 8 nm FFN process Samsung designed specially for NVIDIA.

The ASUS TUF Gaming GeForce RTX 3080 OC retains the military-grade metal aesthetic characteristic of the brand. It uses a meaty cooling solution that features multiple fin stacks which are ventilated by a trio of ASUS's new-generation Axial-Tech fans. There's lavish use of metal on both the cooler shroud and the backplate. The cooler is longer than the PCB underneath. Much like NVIDIA's Dual Axial Flow-Through cooling solution on the Founders Edition card, ASUS used the extra length of the cooler to vent air through the card. ASUS gave the card factory-overclocked speeds of 1785 MHz GPU Boost (vs. 1710 MHz reference). The card draws power from a pair of conventional 8-pin PCIe power inputs. ASUS is asking $730 for the RTX 3080 TUF Gaming OC, a tiny $30 premium over the reference design. In this review, we pit the card against a large selection of graphics cards, across an equally large selection of games.

GeForce RTX 3080 Market Segment Analysis
GTX 1080 Ti$6503584881481 MHz1582 MHz1376 MHzGP10212000M11 GB, GDDR5X, 352-bit
RX 5700 XT$3702560641605 MHz1755 MHz1750 MHzNavi 1010300M8 GB, GDDR6, 256-bit
RTX 2070$3402304641410 MHz1620 MHz1750 MHzTU10610800M8 GB, GDDR6, 256-bit
RTX 2070 Super$4502560641605 MHz1770 MHz1750 MHzTU10413600M8 GB, GDDR6, 256-bit
Radeon VII$6803840641802 MHzN/A1000 MHzVega 2013230M16 GB, HBM2, 4096-bit
RTX 2080$6002944641515 MHz1710 MHz1750 MHzTU10413600M8 GB, GDDR6, 256-bit
RTX 2080 Super$6903072641650 MHz1815 MHz1940 MHzTU10413600M8 GB, GDDR6, 256-bit
RTX 2080 Ti$10004352881350 MHz1545 MHz1750 MHzTU10218600M11 GB, GDDR6, 352-bit
RTX 3070$5005888961500 MHz1725 MHz1750 MHzGA10417400M8 GB, GDDR6, 256-bit
RTX 3080$7008704961440 MHz1710 MHz1188 MHzGA10228000M10 GB, GDDR6X, 320-bit
TUF Gaming OC
$7308704961440 MHz1785 MHz1188 MHzGA10228000M10 GB, GDDR6X, 320-bit
RTX 3090$1500104961121395 MHz1695 MHz1219 MHzGA10228000M24 GB, GDDR6X, 384-bit
Next Page »Pictures & Teardown
View as single page