Nvidia has always manufactured an xx50 card to suit the needs of a budding gamer or a budget watcher. These cards have an extremely high price to performance ratios due to their low prices and can still kick out frame rates worth talking about. These cards have always been aimed at the entry-level to mid-range market to ensure they can cater to all performance needs. The latest xx50 series card, the GTX 1650 (there is a 1650 Super but that comes at a higher cost) was released with 2 iterations, we’ll be looking at the latest available model the GDDR6 1650 released in 2020 as it is more widely available.
Being entry-level cards we wouldn’t recommend using them in builds higher than 1080p if you’re planning on playing games at max settings at any decent frame rate. If you’re willing to sacrifice some graphical quality for those extra frames then the 1650 works well for a 1440p build. Depending on whether you’re looking to hit 60 fps or 144 fps you’ll have to inevitably forgo demanding settings like ambient occlusion and anti-aliasing especially in visually demanding games like Cyberpunk 2077.
These cards are a great way to get into PC gaming and can offer significantly improved gaming performance over console gaming but you will obviously need to add a few more components to your pc before you can game. The upgrade path is fairly easy with graphics cards, these are fairly new so any compatible components will stay that way for a few years as standards aren’t changing too rapidly. What I mean by that is that these are PCIe 3 cards that work in PCI 4 motherboards so you won’t need to worry about compatibility.
The only component that would need reconsidering if you are planning on upgrading your GPU down the line would be the power supply. Both of these cards are fairly low power; they will both work with a 300W power supply. Newer and beefier cards will require much more power, so invest in a power supply early to keep your options open.
To compare these 2 cards we’ll be looking at their architecture, their clock speeds and benchmarks, their cooling and overclocking capabilities, and their memory.
These cards are based on two separate Nvidia architectures that are sequential releases of each other. The GTX 1650 is the most direct upgrade of the previous generation’s GTX 1050. The Pascal-based architecture had its limits on the number of transistors and Cuda cores they could fit on the xx50 die because of the 14nm lithography. The Turing architecture implemented a 12nm lithography made it more efficient and physically smaller meaning that it was more cost-effective to produce.
Turing also introduced a new processor architecture called the Turing SM. The main benefits of this new architecture can be seen in the shading efficiency, delivering over a 50% boost in performance compared to the equivalent Pascal GPU. The new architecture essentially allowed for instructions to be executed simultaneously with floating-point instructions. Essentially doubling the bandwidth.
The Turing tensor cores are a new and specialized type of execution units that perform tensor or matrix operations that are prevalent in machine learning and neural network training. In terms of gaming, this means that AI based capabilities are unlocked for you. Things like DLSS can be used with Turing-based GPUs, unfortunately, this isn’t available for the 16 series cards yet but like RTX was enabled for the 16 series cards through a software update the same is likely for DLSS.
The 1650 also has 896 Cuda cores which is significantly more than the 768 in the 1050. Essentially Turing is more efficient, allows for higher memory bandwidth, has more transistors, and introduces new tensor cores to the consumer GPUs.
These cards don’t have hugely different clock speeds but the type of processors used in these cards contribute to the massive performance increases you’ll see in the benchmarks.
- Base clock: 1392 MHz
- Boost clock: 1506MHz
- Base clock: 1410 MHz
- Boost clock: 1695 MHz
We’ve run a few games at 1080p high settings to get a good idea of where these cards sit. They both max out at 1080p if you’re hoping to achieve higher than 60 fps. The GTX 1650 can hold its ground at 1440p at medium settings but only at 60 fps maximum. We’ll include a table for both sets of results so you can see the difference in performance. Both cards are running at stock speeds.
Average FPS (1080p High Settings)
|GTX 1050||GTX 1650|
|Rainbow Six Siege||100||164|
Average FPS (1440p medium settings)
|GTX 1050||GTX 1650|
|Rainbow Six Siege||88||140|
In all tests, the GTX 1650 performed at least 50% better than the GTX 1050 meaning that Nvidia’s claims weren’t unsubstantiated. The improvements made to the card offer drastically more performance and a much better gaming experience. As you can see though, the cards both struggled at 1440p in graphically demanding titles like Cyberpunk 2077 (mostly because of a lack of optimization in-game and a lack of RT cores).
Turning off demanding settings like anti-aliasing and RTX can massively improve performance with low-end graphics cards and these are no exception. If the frame count is what matters most to you then running your games at medium settings instead of high at 1080p will get you another 50% performance boost amongst most of the titles mentioned here.
GTX 1050: The reference card is very low power and so doesn’t actually have a dedicated power connector. This means that the card is powered by the motherboard and whatever power it can pass through the PCIe slot. Because of this, you have a fairly hard limit on overclocking abilities. There are plenty of aftermarket coolers and manufacturer overclocked 1050s available so we would recommend purchasing one of these rather than overclocking and cooling a reference card.
Many users have been able to achieve an overclock up to 1900MHz but because of the poor power delivery and limited cooling in-game speeds were closer to 1750MHz. This amounted to a roughly 10% performance boost which might seem like a significant boost. When you consider the benefits of purchasing an overclocked card from a different manufacturer, overclocking this card yourself just isn’t worth it.
GTX 1650: We can see a similar pattern with the 1650. Again because of its low TDP of 75W you don’t have much headroom to overclock the card yourself and it is fairly limited. The cooling potential of this card is very good however, water blocks and aftermarket coolers are plentiful with the Turing GPUs. Again, we would recommend purchasing a pre-overclocked GPU if your budget allows you to, as cooling and a higher-rated PSU can make up the difference in cost.
With the 4GB GDDR6 model however enthusiasts have been able to achieve clocks up to 1860MHz that are sustainable under load so your mileage may vary depending on your setup. This again nets you a performance boost of around 10% so if you need those extra frames buy an overclocked aftermarket card for a bit extra.
Both cards offer a decent performance boost with an overclock but at this price point, you’re better off purchasing an overclocked card with a decent cooling system from a different manufacturer.
When the GTX 1650 was originally released it came equipped with 4GB of GDDR5 memory with a speed of 8Gbps and a bandwidth of 128GB/s. The updated GDDR6 model was released with the same amount of memory but obviously GDDR6 instead of GDDR5. This update meant that the memory speed increased to 12Gbps and the bandwidth also increased up to 192GB/s. Moving to GDDR6 improved the performance of the GTX 1650 by 6%-10% depending on the application but it was overall a better card for the same price.
There were also 2 versions of the GTX 1050 released, one with 2GB of memory and the other with 3GB, this definitely boosted performance with higher resolutions but it was still GDDR5 memory meaning that the overall performance boost was negligible at native resolution.
GDDR6 is also more power-efficient which is why both cards have the same TDP despite the fact that the GTX 1650 has an increased memory bus. It is currently the industry standard for high-speed VRAM and will continue to be the main memory type for a few years now. Again meaning that your upgrade path is wide open if you go for one of these cards.
These cards aren’t made to run DOOM eternal or Cyberpunk 2077 at 4k 60fps, they’ve been designed and produced to be entry-level cards and they do that very well. Currently, for a very small price increase, the Turing-based GTX 1650 is much better than its predecessor and will do you just fine for most games running at native resolution with medium to high settings.
If you’re looking for that crisp 144fps gameplay at 1080p or 1440p then you’ll want to look at the upgraded Turing cards currently available. The GTX 1660 ti and RTX 2060 offer a significant performance boost over the GTX 1650 but they also come in at least double the cost of the GTX 1650. This card is a great starting place for pc building so we say get stuck and get one. You can always upgrade to a more modern card when finances allow.