Along with at present’s NVIDIA Studio branding announcement, NVIDIA can be utilizing Computex to replace their lineup of Quadro GPUs for notebooks and cell workstations. Along with bringing among the present Quadro RTX desktop components to the cell house, the corporate can be launching a sub-series of components beneath the Quadro T sequence, and eventually a pair of latest Quadro P sequence graphics adapters for the low-end.
Starting issues off, we’ve got the cell Quadro RTX components, that are all new for the cell house. Like NVIDIA’s GeForce cell counterparts, these Quadro RTX cell components are primarily the identical chip configurations as their desktop siblings, however put right into a cell kind issue and with their TDPs and clockspeeds turned down accordingly. As a consequence the cell Quadro RTX components pack all of the options and VRAM of the desktop components that NVIDIA has beforehand launched, whereas retaining a great deal of their efficiency and the entire Turing structure’s performance.
| NVIDIA Mobile Quadro RTX Spec Comparison | ||||||
| RTX 5000 | RTX 4000 | RTX 3000 | P5200 | |||
| CUDA Cores | 3072 | 2560 | 2304 | 2560 | ||
| Boost Clock | ~1.53GHz | ~1.56GHz | ~1.39GHz | ~1.74GHz | ||
| Memory Clock | 14Gbps GDDR6 | 14Gbps GDDR6 | 14Gbps GDDR6 | 8Gbps GDDR5 | ||
| Memory Bus Width | 256-bit | 256-bit | 192-bit | 256-bit | ||
| VRAM | 16GB | 8GB | 6GB | 16GB | ||
| Single Precision Perf. | 9.four TFLOPs | Eight TFLOPs | 6.four TFLOPs | 8.9 TFLOPs | ||
| Tensor Perf. (FP16) | 75.2 TOPs | 63.9 TOPs | 51.four TOPs | N/A | ||
| TGP Max Power | 80-110W | 80-110W | 60-80W | 150W | ||
| GPU | TU104 | TU104 | TU106 | GP104 | ||
| Transistor Count | 13.6B | 13.6B | 10.8B | 7.2B | ||
| Architecture | Turing | Turing | Turing | Pascal | ||
| Manufacturing Process | TSMC 12nm “FFN” | TSMC 12nm “FFN” | TSMC 12nm “FFN” | TSMC 16nm | ||
| Launch Date | 05/27/2019 | 05/27/2019 | 05/27/2019 | N/A | ||
Owing to the tighter TDPs of cell, NVIDIA’s cell Quadro RTX stack doesn’t go fairly as excessive because it does on the desktop. For cell the quickest half is the Quadro RTX 5000, which relies on the identical TU104 GPU because the desktop model. This half replaces the Quadro P5200 as NVIDIA’s flagship cell Quadro half. Meanwhile beneath that we’ve got the Quadro RTX 4000 and RTX 3000, which seem like primarily based on a cut-down TU104 and full-fledged TU106 GPU respectively.
In phrases of efficiency, the RTX 5000 will prime out at 9.four TFLOPs, adopted by Eight TFLOPs for the RTX 4000 and 6.four TFLOPs for the RTX 3000. NVIDIA’s peak clockspeeds appear to differ a bit relying on the processor – we’re estimating wherever from 1.39GHz to 1.56GHz – although these are nonetheless pretty aggressive for a cell half. Sustained efficiency will probably be decrease, in fact, with that various with the cooling capabilities of the host laptop computer.
Meanwhile by way of reminiscence, the state of affairs is once more a mirror of the desktop. The RTX 5000 will get 16GB of GDDR6 – a full complement of reminiscence for a cell TU104 half – whereas RTX 4000 and RTX 3000 drop all the way down to 8GB and 6GB respectively. NVIDIA continues to deal with reminiscence capability as a characteristic differentiator between the Quadro and GeForce households and even amongst Quadro playing cards, so the 16GB RTX 5000 is a halo half on this respect. The flip facet, nonetheless, is that RTX 5000 doesn’t enhance on its predecessor right here so far as capability goes, as each the outdated and new playing cards are 16GB.
It is attention-grabbing to notice that whereas efficiency has gone up and reminiscence capacities have at the least held even, energy consumption is definitely down generation-over-generation. Starting with the cell Quadro RTX sequence, NVIDIA is offering a spread of max energy values as a substitute a single worth, however even on the prime of this vary, none of those playing cards passes 110W, properly beneath the 150W that the older P5200 peaked at. The RTX 4000 and RTX 3000 components don’t see fairly the identical financial savings as their very own predecessors, however the vary remains to be there. NVIDIA…
