- Joined
- Aug 19, 2017
- Messages
- 3,002 (1.07/day)
Next week's planned launch of NVIDIA's GeForce RTX 5060 Ti series brings two SKUs differentiated by memory capacity and pricing. Both models leverage the GB206‑300 GPU, made on a 5 nm node from TSMC, and feature a 128‑bit memory interface paired with GDDR7 chips running at an effective 28 Gbps. According to IT Home, NVIDIA has communicated MSRP figures to its key AIC partners ahead of the mid‑April rollout. The entry‑level 8 GB variant is set at an MSRP of ¥3,199 (roughly $379), while the 16 GB version carries an MSRP of ¥3,599 (about $429). This is a reduction from the $399 and $499 prices anticipated earlier. NVIDIA is adjusting its pricing strategy for these mid-tier chips to align itself against the competition better and draw more new buyers.
Under the hood, the GB206‑300 core activates 36 streaming multiprocessors, delivering a total of 4,608 CUDA cores. The GPU operates at a base clock of 2,407 MHz, boosting to 2,572 MHz under load. Memory runs at 1,750 MHz (28 Gbps effective), routed through the 128‑bit bus to yield up to 448 GB/s of bandwidth. Graphics‑specific throughput is augmented by 144 texture mapping units and 48 render output units, while 36 dedicated ray‑tracing cores handle real‑time lighting calculations. Additionally, 144 tensor cores accelerate AI‑driven workloads such as DLSS upscaling and machine‑learning inference. Power delivery for both cards is managed via a single 16‑pin connector, with a total board power of 180 W. Display connectivity includes one HDMI 2.1b port alongside three DisplayPort 2.1b outputs, and the card interfaces with host systems over PCI Express 5.0 x16.
View at TechPowerUp Main Site | Source
Under the hood, the GB206‑300 core activates 36 streaming multiprocessors, delivering a total of 4,608 CUDA cores. The GPU operates at a base clock of 2,407 MHz, boosting to 2,572 MHz under load. Memory runs at 1,750 MHz (28 Gbps effective), routed through the 128‑bit bus to yield up to 448 GB/s of bandwidth. Graphics‑specific throughput is augmented by 144 texture mapping units and 48 render output units, while 36 dedicated ray‑tracing cores handle real‑time lighting calculations. Additionally, 144 tensor cores accelerate AI‑driven workloads such as DLSS upscaling and machine‑learning inference. Power delivery for both cards is managed via a single 16‑pin connector, with a total board power of 180 W. Display connectivity includes one HDMI 2.1b port alongside three DisplayPort 2.1b outputs, and the card interfaces with host systems over PCI Express 5.0 x16.
View at TechPowerUp Main Site | Source
)
