Buzz
Let's get straight to the point. If you happened to snag a 3070 Ti or 3080 at a good price, don't worry, at the very least, it will perform well for your gaming needs for the next 4 to 5 years.
Now, two years ago, the RTX 3080 graphics card was introduced with a recommended price of 699 USD. Currently, to own an RTX 4080, you need to prepare at least 899 USD for the AD-104 GPU-equipped version with 12GB VRAM. This figure doesn't even include electronic import taxes and VAT in Vietnam. The RTX 4090 is priced at a thousand six dollars, and in Vietnam, it's likely to cost around 50 million Dong, give or take.
This implies three possibilities. First, Nvidia is either highly confident in the graphics processing capabilities of the Ada Lovelace card. Second, the production cost of graphics cards has increased, and Nvidia is passing on some of the costs to consumers. The third possibility is that Nvidia is following the path of Sony and Apple, creating high-end products to earn higher profit margins.
RTX 4080 12GB, or a renamed RTX 4070?
Looking at Nvidia's recent business data, the third possibility suddenly becomes incredibly logical. It's unclear whether Nvidia's efforts to reduce prices and promotions to clear inventory are helping them fulfill orders for RTX 30 series graphics cards.
By the way, let's take a quick look at the product lineup Nvidia unveiled in a concise livestream during the recent GTC 2022 event. No need to dwell on the RTX 4090, as it's Nvidia's flagship offering for the consumer market. But why introduce two RTX 4080 models simultaneously, with varying CUDA cores and VRAM?A conspiracy theory gaining traction and seeming plausible is that the RTX 4080 12GB is essentially the product that should have been named the RTX 4070. The RTX 4080 16GB features the AD-103 GPU, while the 12GB variant is equipped with the AD-104. Do you still recall the GPU in the RTX 3070? It was the GA-104. With a price tag of 899 USD, if named the RTX 4070, it would undoubtedly stir up strong reactions in the community.So, here we have the 'budget' version of the RTX 4080. However, the term 'budget' here still comes with a figure higher than the recommended prices when the RTX 3080 and 3070 were launched. The prices of 699 and 499 USD are now just in the dreams of many enthusiasts.
This conspiracy theory becomes even more plausible when we look at the performance of the RTX 4080 12GB, approaching that of the RTX 3090, the flagship product of the Ampere architecture introduced at that time. Traditionally, every card with a '70' designation has matched or slightly outperformed the most powerful product of the previous generation: the 2070 on par with the 1080 Ti, the 3070 equal to the 2080 Ti, and so on.Perhaps Nvidia anticipated the fact that, regardless of the pricing strategy, the sales of the RTX 40 series would still be impacted by the millions of old-generation cards flooding the PC hardware market. Instead of compromising, they opted to create a premium product with exceptional performance. Those who can afford it will still buy, while the rest of us...Using an RTX 30 card is still fine, so don't worry, folks.
Gotta say, Nvidia this time around has been wiser than in 2018 when they introduced the Turing architecture, featuring the first graphics cards with real-time ray tracing support. Back then, the cryptocurrency craze of 2017 still haunted them, and wherever you looked, there was the GTX 1060, a favorite among gamers and miners alike. Interestingly, that's probably why the GTX 1060 still reigns as the king of the gaming PC industry, consistently holding the throne every time Steam Hardware Survey releases its survey results.Nvidia's wisdom lies in creating products at a higher tier than the RTX 30 series, with no real overlap into the segments of older-generation products. Not to mention, the RTX 4080 is set to hit the market in November. By then, according to Nvidia's calculations, those who want to buy older cards would have already made their purchases, clearing the way for the new-generation products.
But how powerful are these new-gen products? In short, if you're not gaming at 4K resolution, the RTX 3060, 3070, or 3080 will still meet your needs perfectly. Even if you're into 4K gaming, with the current prices of used 3090 Ti cards, it's still a solid choice at a range of under 30 million Dong.There's an interesting point that few have noticed - Nvidia is now introducing new products just like Apple. They refrain from disclosing precise frame rates, the most ideal metric for assessing GPU performance. All charts and comparisons are based on the previous generation of graphics cards as a reference point, measuring the performance of RTX 4090 and 4080 to yield figures that boast performance gains of multiple times, such as 2x or 3x, for instance:
At the very least, until October 12th, the RTX 3090 Ti remains the most robust consumer graphics card on the planet. After October 12th, with the launch of the RTX 4090, it's highly likely that many affluent individuals will release their old cards to embrace the new ones, simply because they can and want to. By then, the RTX 3090 Ti will naturally become a cost-effective gaming toy. Both the 3090 Ti and 4090 can handle 4K gaming, and if the frame rate difference isn't too crucial, the buying price of the 3090 Ti makes it a highly desirable product, a complete contrast to the time Nvidia introduced and priced it at $1,999.What about DLSS 3? Do we need the latest 4 series card for that?
It must be admitted that many are eagerly anticipating the RTX 40 series simply due to the power of the tensor processing unit in the Ada Lovelace GPU. While watching the livestream event, I was overwhelmed by the capability Nvidia introduced, utilizing the GPU to autonomously generate new frames without burdening the CPU to its full capacity. It's something unprecedented, and Nvidia pulled it off.Suddenly, DLSS 3 becomes a more significant selling point than the performance difference between Ampere and Ada Lovelace. This new technology enables the AI algorithm and the Optical Flow Accelerator processing unit to 'predict' and 'paint' an entirely new frame based on previously available data, without the need for the CPU to construct polygons for each frame as traditionally done. According to Nvidia, this capability will completely eliminate the bottleneck between the CPU and GPU, especially when the GPU is powerful, and the CPU struggles to handle multiple tasks simultaneously.
Here, two issues need addressing. First, we haven't experienced DLSS 3 hands-on to compare it with DLSS 2 or benchmark it against the processing performance of previous-generation graphics cards. And second, Nvidia's own engineers have stated that, theoretically, DLSS 3 could operate on older GPU hardware, even those with the Optical Flow Accelerator processing unit, but lacking the performance to autonomously generate entirely new frames:'OFA has been around since the Turing architecture GPU generation. However, on Ada, OFA performs significantly better, enough to incorporate the Optical Multi Frame Generation technology in DLSS 3. If users attempt to use DLSS 3 on the RTX 20 and 30 series, they may experience laggy visuals, lower image quality, and most importantly, no boost in frame rates - the primary objective of DLSS.'Those are the statements from Nvidia's Vice President of Applied Deep Learning Research, Bryan Catanzaro. In theory, the OFA of the RTX 3080 or 3090 Ti could potentially support the Optical Flow Accelerator technology in DLSS 3. However, Nvidia emphasizes that implementing this technology would require careful research and development. This opens up a slight possibility for gamers with Ampere cards to enjoy better frame rates while minimizing hardware stress, allowing resources to be allocated for handling ray tracing effects.
