PNY has recently announced that the Quadro RTX 8000 and Quadro RTX 6000 GPUs can be passively cooled. Both of these cards are $6,000 and $4,000, respectively. PNY has also stated that these passively cooled versions only sacrifice a little bit of performance to achieve a passively cool design with a cooling capacity of up to 250W.
PNY Introduces Passively-Cooled, 250W TDP, Quadro RTX 8000, & RTX 6000 Workstation Graphics Cards
The CUDA cores remain precisely the same, being 4608 CUDA cores. The boost clock has been lowered by 8% down to a still very respectable 1,620 MHz, and the base clocks have been dropped by about 12%, the memory clock has also received a small performance hit of 7% to bring it from 14 Gbps to 13 Gbps. The memory bandwidth has also been reduced from 672 GB/s to 624 GB/s, the standard TDP of Quadro RTX 8000 is 295-watts while the passively cooled variants have a lowered TDP of 250-watts.
The comparison between the RTX 6000 and the RTX 6000 Passive offers a very similar story being constrained to the same TDP as the Quadro RTX 8000 Passive, that being just 250-watts. The base clock of this card is lowered to 1275 MHz from the original base clock of 1440 MHz from the non-passive version. This boost clock on the passive cooler has also been decreased by 150 MHz from the original Quadro RTX 6000's 1770 MHz boost clock.
The passively cooled RTX 8000, and RTX 6000 offer no display outputs
The standard Quadro RTX 8000, and Quadro RTX 6000 offer four DisplayPort 1.4 ports along with a USB-C used for Nvidia's VirtualLink. These passively-cooled variants offer no display outputs, as they are designed to be used mainly in a server where having a display output isn't necessary.
PNY Quadro RTX Passive-Cooled GPU Specifications
| Graphics Card | PNY Quadro RTX 6000 Passive | NVIDIA Quadro RTX 6000 | PNY Quadro RTX 8000 Passive | NVIDIA Quadro RTX 8000 |
|---|---|---|---|---|
| GPU | Turing GPU | Turing GPU | Turing GPU | Turing GPU |
| GPU Process | 12nm | 12nm | 12nm | 12nm |
| Die Size | 754mm² | 754mm² | 754mm² | 754mm² |
| GPU Cores | 4608 Cores | 4608 Cores | 5120 Cores | 5120 Cores |
| Tensor Cores | 576 Cores | 576 Cores | 640 Cores | 640 Cores |
| Base Clock | 1275 MHz | 1440 MHz | 1230 MHz | 1395 MHz |
| Boost Clock | 1620 MHz | 1770 MHz | 1620 MHz | 1770 MHz |
| Single Precision | 14.9 TFLOPs | 16.3 TFLOPs | 14.9 TFLOPs | 16.3 TFLOPs |
| Ray Tracing Spec | 10 GigaRays/Sec | 10 GigaRays/Sec | 10 GigaRays/Sec | 10 GigaRays/Sec |
| VRAM | 24 GB GDDR6 | 24 GB GDDR6 | 48 GB GDDR6 | 48 GB GDDR6 |
| NVLINK VRAM | 48 GB With NVLINK | 48 GB With NVLINK | 96 GB With NVLINK | 96 GB With NVLINK |
| Memory Bus | 384-bit | 384-bit | 384-bit | 384-bit |
| Memory Bandwidth | 624 GB/s | 672 GB/s | 624 GB/s | 672 GB/s |
| TDP | 250W | 295W | 250W | 295W |
| Price | TBD | $4000 US | TBD | $5500 US |
| Launch Date | Q1 2020 | Q4 2018 | Q4 2018 | Q1 2020 |
The demand for visualization, rendering, data science and simulation continues to grow as businesses tackle larger, more complex workloads than ever before. However, enterprises looking to scale up their visual compute infrastructure face mounting budget constraints and deployment requirements.
Meet your visual computing challenges with the power of NVIDIA Quadro RTX GPUs in the data center. Built on the NVIDIA Turing architecture and the NVIDIA RTX platorm, the RTX 6000 passively cooled graphics board features RT Cores and multi-precision Tensor Cores for real-time ray tracing, AI, and advanced graphics capabilities. Tackle graphics-intensive mixed workloads, complex design, photorealistic renters, and augmented and virtual environments at the edge with NVIDIA RTX, designed for enterprise data centers.
The passively-cooled RTX 8000, and RTX 6000 are backed by PNY with a limited three-year warranty. Along with the warranty, these cards come with Dedicated Quadro Field Applications Engineers and free dedicated phone and email tech support. PNY didn't stat any pricing for these variants. With the standard RTX 8000, and RTX 6000 costing just $5,500 and $4,000 respectively, so these variants are expected to cost in a similar range to these models.
