Players often ridicule the "Huang's knife method", and they really don't believe it, and the various product combinations are dazzling. At present, NVIDIA is launching a series of new cards, but all of them are on the existing products.
The released RTX 2060 12GB is not simply a doubling of memory, core size, frequency, and power consumption have all changed.
The RTX 3080 12GB is a similar operation, opening up more cores, video memory, and mining performance has soared by 21%, but its release time is now a mystery.
The RTX 3070 Ti 16GB is set to be postponed until next year, but it should be the least changed, after all, it is a full-blood core.
The RTX 3090 Ti will be a full-blood version of ga102 big core, 10752 stream processors, and the power consumption of the whole card has soared to 450W, which is more of a symbol of strength.
For the entry-level market, NVIDIA prepared the RTX 3050, originally thinking that it would be very simple, and the result was also a moth.
The card was reported as early as last November, when it was said to be a GA107-300 small core, 2304 stream processors, and 90W of power consumption.
According to the latest exposure, the RTX 3050 actually uses the GA106 core, further castrated, and divided into two versions:
One is the GA106-150 core, 2560 stream processors (20 optical chasing units), 8GB GDDR6 video memory.
The second is GA106-140 core, 2304 stream processors (18 optical chasing units), 4GB GDDR6 video memory.
Why is it so subdivided? I don't know......
Why not use a smaller GA107 core? I don't know......
You know, the GA107 core was originally 2560 stream processors, which is actually just right. Is it that there are too many core defects in GA106 and should continue to be used?
In addition, regarding the RTX 3050 Ti, there is no news at the moment, either canceled or continue to be determined.