First leaked benchmarks for the upcoming RTX 3050 desktop GPU have surfaced and, it turns out that the RTX 3050 is just a GTX 1660 Super in disguise, with ray tracing and DLSS support. The benchmakrs comes courtesy of Videocardz who managed to get a hold of an RTX 3050 and tested it in 3DMark, putting it through three different tests.
For the uninitiated, the RTX 3050 uses the same GPU die as the RTX 3060: GA106. But the RTX 3050 is built on a cut-down version of GA106 so it only has 2560 CUDA Cores, as compared to the RTX 3060’s 3584 CUDA Cores. The card also has 8GB of GDDR6 memory clocked at 14Gbps, resulting in a total bandwidth of 224GB/s. Moreover, the base clock here is 1550Mhz while the card can boost up to 1780Mhz.
A little context
NVIDIA announced the RTX 3050 at CES 2022 as its newest entry in the ever-growing lineup of RTX 30-series GPUs. The graphic card, which promised to lower the barrier of entry to current-gen gaming, was presented as an upgrade to people still running on generations-old cards like the GTX 1050. At the reveal event, NVIDIA showedf some benchmarks comapring the card with previous 50-class GPUs and, sure enough, it was faster than them all. But, we didn’t any more info on how the card would perform, especially against more recent GPUs.
Yesterday, NVIDIA put up a new benchmark for the RTX 3050. The benchmark appeared on the card’s product page where it was pitted against the GTX 1050 and GTX 1650 in a graph depicting ray tracing performance. In case you didn’t know, both GTX 1050 and GTX 1650 are cards that are incapable of hardware-accerlated ray tracing; they effectively cannot do ray tracing becuse they lack the dedicated Tensor cores to do so. That’s why they have the “GTX” moniker instead of “RTX” in their names. Hence, the comparision was bonkers and showed just how low the company would drop to show its product in a positive light.
Coming to the benchmark itself, the RTX 3050 managed to score 6,865 points in Time Spy. That’s actually lower than GTX 1660 Ti‘s score of 6,360 points, but above the 4,970 points acheived by AMD‘s RX 6500 XT. All in all, the card is pretty much on-par with the GTX 1660 Super and about 20% slower than the RX 6500 XT. When compared to the RTX 3060, the card falls short by roughly 43% due to RTX 3060’s score of 8,783 points in the same benchmark.
Moving on, we also have scores for Fire Strike (at 1080p) and TimeSpy Extreme. The RTX 3050 netted 15,843 points in the Fire Strike benchmark. That’s almost identical to the RX 6500 XT, and only 6% lower than the GTX 1660 Ti, both of which managed to score 15,260 and 16,774 points, respectively. When taking into account the GPU one notch above in the RTX 30-series portfolio, the RTX 3060, is about 24% faster than the RTX 3050.
Lastly, TimeSpy Extreme shows the most drastic difference between the RTX 3050 and RTX 3060 with the former being a full 34% slower than the latter. The RTX 3050 achieved 2,801 points in this benchmark. For comparision, the RX 6500 XT scored only 2,270 points which constitiutes a 19% decrease. On the other hand, the GTX 1660 Ti scored 2,882 points making it 3% faster than the RTX 3050, so essentially the same.
Now, it’s important to mention that 3DMark scores, or any other non in-game benchmark for that matter, isn’t as important or indicative of actual gaming perfortmance anymore. Technologies like DLSS, DLDSR, AMD‘s FSR, and more can all defy on-paper numbers to squeeze more power out of the GPU and boost performance, so keep that in mind. Still, if we use 3DMark as a medium acorss all SKUs to compare them, we can get a decent idea of what real-world performance might look like.
RTX 3050 is set to release on January 27th with a “suggested” retail price of $249. IF the card actually sells at that, it’s not too bad of a deal, considering that’s how much the GTX 1660 Ti roughly costs, and performance of both these are neck-in-neck. In fact, the addition of DLSS and ray tracing on the RTX 3050 are enough to justify the purchase at MSRP.
I am a bit more hopeful for this GPU launch because recent reports have made it clear that the RTX 3050 is not exactly a good card for crypto mining which can alleviate some of the scalping. And, NVIDIA has said they’re hoping to push a lot of volume for this SKU as they’ve physically made more units to fight the shortage. For now, all we have is hope, let’s see what happens at launch.
All benchmark data including the pictures were sourced verbatim from Videocardz, so the slight discrepancy between the results in the image and the text are from the source itself.