When Nvidia announced its mobile RTX cards earlier this month, it didn’t reveal anything about core counts, clock speeds, or GPU features beyond those common to the RTX family. Now that this information is available, we can speak to the configurations the company will support with its RTX 2070 and 2060 GPUs.
In the old days, Nvidia would release mobile GPUs that shared product names with their desktop counterparts but rarely shared an identical core configuration. The mobile GPUs were always smaller, with fewer cores, texture units, and ROPs. With Pascal, Nvidia began selling the same silicon in both mobile and desktop, opting instead to reduce clock speeds. The company is continuing that trend with Turing — the RTX 2080 Max-Q and RTX 2070 Max-Q will offer all of the features of their desktop cousins. They’ll just do so at much lower frequencies:
This data, from Notebook Check, shows the sharp reduction. The laptop version of the RTX 2080 has a boost clock 60 percent as high as its desktop counterpart, while the base clock is 49 percent the frequency. Power consumption is significantly reduced as well, however, with a board power of 80W compared to 225W for the full-fat desktop flavor.
Nvidia will also sell a non-Max Q version that allows for even higher board power in the mobile form factor. This version of the RTX 2080 will be allowed to pull 150W or more, though whether the GPU will sustain the full performance of its desktop counterpart is highly questionable. This is more or less a question of binning, and it’s not clear if Nvidia’s best RTX 2080 GPUs can fit into a 150-180W power envelope and still maintain 100 percent of desktop speed. The Max-Q GPUs clearly won’t ever approach desktop performance, though that’s a necessary balance given the number of cores and overall capabilities of the card in question.
Between binning and the lower clocks, Turing should be capable of fitting into the listed mobile power envelopes. As we saw with the AMD Radeon Nano back in 2015, relatively small adjustments to a GPU’s operating frequency, combined with good binning, can lead to significant overall improvement in performance/watt. While the degree to which this improves a GPU depends on the architecture, we can expect the mobile RTX cards to also be more power efficient overall than the desktop versions.
- Nvidia Stock Smashed on Weak Earnings, Turing Woes
- Reports: TSMC Accident Destroys Tens of Thousands of Nvidia GPU Wafers
- Why Won’t Apple Support Nvidia Silicon With macOS Mojave?