On the plus side, this might very well be the best product for those looking to upgrade OEM machines without external power connectors, on the other hand, fuck NVIDIA.
We’ve mentioned previously how NVIDIA were going to terminate the production of the RTX 3050 8GB model due to its sales somehow impeding upon NVIDIA’s RTX 4060 market share, given how both are similarly priced and yet offer performances that are worlds apart.
Rather than simply dropping the price of their overpriced entry level RTX 4000 series graphics card, NVIDIA has instead opted for more drastic measures to forcefully upsell consumers into buying the RTX 4060 or if they continued to purchase the RTX 3050, they would provide a vastly inferior product and make a greater profit from those that do manage to sell.
We specifically mentioned how NVIDIA were going to end production of the RTX 3050 8GB and provide consumers with an inferior 6GB variant instead, given how the standard graphics card comes packing a 128-bit memory interface, to drop the VRAM down to 6GB would result in either a 192-bit or 96-bit memory bus, and it comes as no surprise that NVIDIA have opted for the smaller memory interface.
To ensure maximum profitability, the upcoming RTX 3050 6GB model will feature a decreased core count, alongside the reduced memory capacity and severe bandwidth reductions. On the plus side the power consumption has dropped considerably.
These significant specifications reductions indicate that it should not bear the RTX 3050 nameplate, but rather this is ultimately a “3040” class product, but given NVIDIA’s usual deceptive tactics with differing specifications under the same SKU naming scheme, this comes as no surprise.
The anticipated RTX 3050 6GB is projected to have 2304 CUDA cores, marking a 10% decrease compared to the 8GB model and aligning with NVIDIA’s OEM version of the RTX 3050.
However, this card goes a step further by reducing its memory interface from 128-bit to a mere 96-bit memory bus, whereas the OEM variant retained the original 128-bit bus with 8GB of VRAM, while its mobile variant kept the 128-bit memory interface with just 4GB of VRAM.
This alteration significantly impacts the overall peak memory bandwidth. While the RTX 3050 8GB achieves a peak of 224 GB/s, the 6GB model is subjected to a substantial reduction, plummeting to a mere 168 GB/s, reflecting a 25% decrease in overall memory bandwidth.
As for base and boost frequency, the RTX 3050 6GB will have a severe reduction, to the sum of a 468MHz base frequency drop while its maximum boost clock has been reduced by 290MHz versus the 8GB model, such a sheer drop in core frequency combined with the reduced GDDR6 memory capacity has had a massive effect on power consumption.
The RTX 3050 8GB has a TDP value of 130W while this new 6GB model has ultimately shrunk down to a minimal TDP of just 70 watts, meaning it should be seamlessly compatible with barebones OEM machines that have limitations in terms of their power supply or have no additional power connectors available.
Making this the best graphics card available on the market without the requirement of an external power connector.
Which is downright fucking shameful considering how NVIDIA’s RTX 4050 Mobility unit, with 2560 CUDA cores and 6GB of faster 16Gbps GDDR6 memory across a 96-bit memory interface comes baring a TDP of just 50 watts. If NVIDIA were actually serious about providing consumers with a modern product that doesn’t actually need an external power connector they could very easily provide a discrete variant of the RTX 4050 with reduced frequencies.
Because the RTX 3050 6GB is a fraud, NVIDIA doesn’t actually give a fuck about its consumers, the only reason they’re resorting to such backhanded tactics such as to keep outdated and older generation hardware around is simply because consumers are dumb enough to keep paying $249 for the RTX 3050 8GB.
NVIDIA just want to recycle its excess stock or rather upselling consumers on the RTX 4060 by providing a massively inferior product as its replacement, the performance difference between the “RTX 3050” 6GB, which doesn’t deserve to bare the same SKU nameplate versus the 8GB will likely exceed 15% quite easily.
Given how there’s a 10% drop in CUDA cores, a massive frequency drop and vastly reduced memory bandwidth, the product itself is a fucking mockery and a testament to see exactly how retarded modern consumers are.
