A bit over eight months after all of it started, the tail-end of NVIDIA’s GeForce Turing product stack launch is lastly in sight. This morning the corporate is rolling out its newest and most cost-effective GeForce Turing video card, the GeForce GTX 1650. Coming in at $149, the most recent member of the GeForce household is about to convey up the rear of the GeForce product stack, providing NVIDIA’s newest structure in a low-power, 1080p-with-compromises gaming video card with a budget-friendly worth to match.
In very conventional NVIDIA style, the Turing launch has been a top-to-bottom affair. After launching the 4 RTX 20 collection playing cards early within the cycle, NVIDIA’s efforts within the final two months have been targeted on filling within the again finish of their product stack. Central to it is a design variant of NVIDIA’s GPUs, the TU11x collection – what I’ve been dubbing Turing Minor – that are supposed to be smaller, simpler to supply chips that retain the all-important core Turing structure, however cast off the devoted ray tracing (RT) cores and the AI-focused tensor cores as nicely. The finish results of this bifurcation has been the GeForce GTX 16 collection, which is designed to be a leaner and meaner set of Turing GeForce playing cards.
To date the GTX 16 collection has been comprised of solely the GTX 1660 household of playing cards – the GTX 1660 (vanilla) and GTX 1660 Ti. Both of those have been based mostly on the TU116 GPU. However right this moment the GTX 16 collection household is increasing, with the introduction of the GTX 1650 and the brand new Turing GPU powering NVIDIA’s junior-sized card: TU117.
Unofficial TU117 Block Diagram
Whereas the GeForce GTX 1660 Ti and the underlying TU116 GPU served as our first glimpse at NVIDIA’s mainstream product plans, the GeForce GTX 1650 is a way more pedestrian affair. The TU117 GPU beneath it’s for all sensible functions a smaller model of the TU116 GPU, retaining the identical core Turing characteristic set, however with fewer assets throughout. Altogether, coming from the TU116 NVIDIA has shaved off one-third of the CUDA cores, one-third of the reminiscence channels, and one-third of the ROPs, leaving a GPU that’s smaller and simpler to fabricate for this low-margin market. Still, at 200mm2 in dimension and housing 4.7B transistors, TU117 is certainly not a easy chip. In reality, it’s precisely the identical die dimension as GP106 – the GPU on the coronary heart of the GeForce GTX 1060 collection – so that ought to offer you an concept of how efficiency and transistor counts have (slowly) cascaded all the way down to cheaper merchandise over the previous few years.
At any charge, TU117 will probably be going into quite a few NVIDIA merchandise over time. But for now, issues begin with the GeForce GTX 1650.
| NVIDIA GeForce Specification Comparison | ||||||
| GTX 1650 | GTX 1660 | GTX 1050 Ti | GTX 1050 | |||
| CUDA Cores | 896 | 1408 | 768 | 640 | ||
| ROPs | 32 | 48 | 32 | 32 | ||
| Core Clock | 1485MHz | 1530MHz | 1290MHz | 1354MHz | ||
| Boost Clock | 1665MHz | 1785MHz | 1392MHz | 1455MHz | ||
| Memory Clock | 8Gbps GDDR5 | 8Gbps GDDR5 | 7Gbps GDDR5 | 7Gbps GDDR5 | ||
| Memory Bus Width | 128-bit | 192-bit | 128-bit | 128-bit | ||
| VRAM | 4GB | 6GB | 4GB | 2GB | ||
| Single Precision Perf. | Three TFLOPS | 5 TFLOPS | 2.1 TFLOPS | 1.9 TFLOPS | ||
| TDP | 75W | 120W | 75W | 75W | ||
| GPU | TU117 (200 mm2) |
TU116 (284 mm2) |
GP107 (132 mm2) |
GP107 (132 mm2) |
||
| Transistor Count | 4.7B | 6.6B | 3.3B | 3.3B | ||
| Architecture | Turing | Turing | Pascal | Pascal | ||
| Manufacturing Process | TSMC 12nm “FFN” | TSMC 12nm “FFN” | Samsung 14nm | Samsung 14nm | ||
| Launch Date | 4/23/2019 | 3/14/2019 | 10/25/2016 | 10/25/2016 | ||
| Launch Price | $149 | $219 | $139 | $109 | ||
Right off the bat, it’s attention-grabbing to notice that the GTX 1650 is not utilizing a fully-enabled TU117 GPU. Relative to the complete chip, the model that’s going into the GTX 1650 has had a TPC fused off, which suggests the chip loses 2 SMs/64 CUDA cores. The internet result’s that the GTX 1650 is a really uncommon case the place NVIDIA doesn’t put their greatest foot…
