Out of all the GTX 16 series cards, the 1650 was the most disappointing. It was slower than the cheaper Radeon RX 570 and didn’t quite match up to the older GTX 1060, either. Although NVIDIA rectified that mistake with the launch of the GTX 1650 Super, it appears that the company isn’t giving up on the vanilla 1650. And for good reason, the TU117 based GPU powers a number of budget desktop and mobile rigs, including the wildly popular ASUS TUF Gaming lineup.
The GTX 1650 Ti is a more interesting choice. It appears that it will be exclusive to OEMs. This could mean that NVIDIA is skipping the GTX 1650 Super in the mobility space, and instead just updating the 1650 with GDDR6 memory and launching a 1650 Ti to cement its position.
|GTX 1650||GTX 1650 Super||GTX 1650 Ti||GTX 1660|
|Memory Clock||8Gbps GDDR5||12Gbps GDDR6||12Gbps GDDR6||8Gbps GDDR5|
|Memory Bus Width||128-bit||128-bit||128-bit||192-bit|
The most plausible explanation for this would be that NVIDIA wants to salvage defective TU116 dies that can’t be used to power the GTX 1660 or the 1650 Super. Or it could just be that the latter just isn’t suitable for laptops as is. (northpointewellness.org) Anyway, the Geekbench Compute score suggests that the new GTX 1650 will be around 5-10% faster than the vanilla model.
Strangely Geekbench reports that the new GTX 1650 has two fewer SMs. This is could mean that NVIDIA has further cut down the TU117 die and made up for the loss in performance with the faster GDDR6 memory. We already know from the GTX 1660 Super how big of a boost can faster memory provide (even at 1080p).
no benchmarks on the GTX 1650 Ti at the moment (added OpenCL, thanks to rogame for update) but if I were to hazard a guess, I’d say that it’ll perform quite close to the GTX 1650 Super, a notch higher than the GTX 1650. The additional shaders and faster memory should provide a healthy performance boost. We’ll keep you posted when we hear more!