Since its launch in 2018, Nvidia’s Turing architecture has gradually taken its place in lower price and performance classes, with the Geforce GTX 1650 being the simplest model. Unlike the RTX 2000 series is missing there ray tracingand DLSS support but also GDDR6 memories, and with a relatively high price tag, the model was not a success for the price-conscious budget builder.
The Turing-based consumer cards are currently as many as 14 in number, and in February it was said that card 15 was on the way in the form of a Geforce GTX 1650 with GDDR6 memory. Now it becomes a reality as Nvidia lists the new configuration on its own website, while among others MSI and Gigabyte release new models with the faster memory type.
At the end of 2019, the Geforce GTX 1650 Super was launched, where the big news was the transition from 4 GB GDDR5 to GDDR6 memory. This was done in connection with the switch to the TU116 circuit with slightly increased clock frequencies, more cores and a marginal price increase from the Super-loose sibling. The new vintage on the GTX 1650 does not get a new suffix and is thus called the same as the GDDR5 variant.
Specifications: Nvidia Geforce GTX 1650
GTX 1650 (GDDR6) | GTX 1650 (GDDR5) | GTX 1650 Super | |
---|---|---|---|
Technical | 12 nm TSMC | 12 nm TSMC | 12 nm TSMC |
Circuit | TU117 | TU117 | TU116 |
Architecture | Turing | Turing | Turing |
CUDA cores | 896 st. | 896 st. | 1 280 st. |
Texture units | 56 st. | 56 st. | 80 st. |
Raster units | 32 st. | 32 st. | 32 st. |
Clock frequency | 1 410 MHz | 1 485 MHz | 1 530 MHz |
GPU Boost | 1 590 MHz | 1 665 MHz | 1 725 MHz |
Memory bus | 128-bit | 128-bit | 128-bit |
Memory amount | 4 GB GDDR6 | 4 GB GDDR5 | 4 GB GDDR6 |
Memory frequency | 12 000 MHz | 8 000 MHz | 12 000 MHz |
Memory bandwidth | 192 GB/s | 128 GB/s | 192 GB/s |
Power supply | – | – | 6-pin |
TDP | 75 W | 75 W | 100 W |
Rec. Award | 149 USD | 149 USD | 159 USD |
The gap between the Geforce GTX 1650 and GTX 1650 Super is thus shrinking even more, but to distinguish the two GTX 1650 models, Nvidia is turning down the base and turbo frequencies for the GDDR6 card. The turbo level goes from 1,665 to 1,590 MHz and the base mode from 1,485 to 1,410 MHz, while the TDP value and the graphics circuit remain at 75 watts and the TU117.
The memory bus is still 128 bits but the memory frequency increases from 8 to 12 Gbps, for a bandwidth of 192 GB / s – 64 GB / s higher than the GDDR5-equipped card. The bandwidth is at the same level as the Geforce GTX 1650 Super, which instead has the advantage of more cores with bass and turbo frequencies of 1,530 and 1,725 MHz respectively.
It is currently unclear whether the Geforce GTX 1650 with the different memory types will coexist in the future, or whether the faster memory type will completely take over. According to Chinese Expreview, the GDDR6 model is about 6 percent faster, making the latter alternative more likely. According to Nvidia’s website, the same shift will also take place for GTX 1650 laptops.
It is worth mentioning that third-party manufacturers can choose to sell the cards with clock frequencies above Nvidia’s official levels, which, however, entails requirements for external power supply. In addition, some manufacturers choose to be clearer than Nvidia by making additions to product names. MSI and Gigabyte do this with the extension “D6”, while for example Inno3D prints the entire “GDDR6”.