After a month-long countdown, Nvidia just held their GeForce special event, where they reflected on their past in graphics technology. The limelight was their announcement of the upcoming skew of enthusiast graphics cards, which sure pack a punch. After a painstaking wait of 2 years – gamers and content creators alike can rejoice. This generation of graphics cards promises to be 2x better than the 20 series released in 2018. And paying heed to Nvidia’s pricing, we can see that they also undercut the previous generation in price.
Nvidia’s CEO – Jensen Huang conducted the event as a live stream from his kitchen. He unveiled 3 new enthusiast graphics cards. They were; a midrange RTX 3070, their flagship RTX 3080 and a replacement for their Titan lineup – The RTX 3090. He put great emphasis on how Nvidia is going to push the limits of computer graphics. Furthermore, he teased the release of Nvidia Reflex technology that aims to reduce latency and lag. He also highlighted how Nvidia is working to make sure streamers can conduct better streams while everyone works from home. Jensen showcased the new Machinima app that enables content creators to create 3D content easily.
Generational Improvements
The greatest generational improvement in the world of graphics cards, is seen in the second series of a new technology/ architecture. The 2nd lineup of graphics cards focused on Ray Tracing is the new RTX 30 series. As a result, the improvement over the now 2-year-old RTX 20 series is very apparent as shown as follows.
These new GPUs (Graphics Processing Units) use Nvidia’s Ampere architecture. Named after French Scientist André-Marie Ampère, this new architecture is the second generation RTX architecture. It boasts faster 2nd gen Ray tracing (RT) cores and 3rd gen AI Tensor cores and new streaming processors. All of which contribute to an overall 2x more throughput over last generation’s Turing Architecture. Built on Samsung’s 8nm silicon process that sports up to 28 billion transistors.
The RTX 3080 brings many improvements to the table over its predecessor RTX 2080. With 8704 CUDA cores, up from 2944 on the 2080, it features a similar clock speed, but greater performance due to more than double the CUDA cores. Nvidia is also using the fastest memory this time round, by partnering with Micron to produce GDDR6X memory, it has a theoretical bandwidth limit of 936 GB/s. And bridging the gap between the processor and the GPU is the latest PCIE Gen4 technology.
The fact that they use an 8nm fabricating process, means that the cards will be more efficient. This is down from last gen Turing’s 12nm process. According to metrics shown at the event, the performance per watt is expected to increase by 1.9 times. This is essentially a 2x more efficient GPU; which means a lower electricity bill and less heat generation. Speaking of heat, Jensen showed off the new cooling system of the cards. Having undergone drastic changes, the cooling solution now employs 2 fans. One fan blows heat directly out of the case and the other will blow heat towards the exhaust fans of the case.
Ray Tracing Improvements
A key improvement lies in the original function of the RTX series of cards – Ray Tracing. The RTX 20 series saw the debut of Ray Tracing (RT). Here Nvidia changed not just the name (from GTX to RTX) but also added specialized new hardware to support Ray Tracing. But it was not worth the expensive price tag of the 20 series for this advantage in RT related programs. Having learnt from flaws of the RTX 20 lineup, Nvidia have optimized ray tracing on the RTX 30 series. So much so that now, it is worth the cost in price and in game performance.
Nvidia RTX 3070
The midrange enthusiast gamer is the target audience of this graphics card. Many people speculate that this will be the most popular card out of the 3. After all while being priced lower, it still performs better than last generations top end card. It packs 5888 CUDA cores, and 8 Gigabytes of GDDR6 Vram (fast but not the blazing fast GDDR6x). It also supports a maximum resolution of 8k (7680x4320) and has support to up to 4 monitors. You will need a beefy power supply if this is in your bucket list as it draws 220 Watts of power.
Nvidia RTX 3080
Jensen dubbed this card the flagship of the 30 series, putting it in the realm of the high-end gamer/content creator. 8704 CUDA cores and 10 Gigabytes of fast GDDR6X Vram enables this card to easily tackle 4k high-refresh rate gaming. Competitive First-Person Shooter (FPS) titles like Call of Duty and Battlefield, greatly benefit from a higher refresh rate and resolution. Together they allow one to see an opponents every move, from far away; this is the promise of the RTX 3080. Its maximum resolution is also 8k and it supports up to 4 displays. This GPU draws a large 320 Watts of power.
Nvidia RTX 3090
The whole audience watched in stunned awe as Mr. Huang lifted the behemoth of a graphics card straight from the oven behind him. He called it a beast, a ferocious GPU and gave it the lovable title of the BFGPU (Big Ferocious GPU). And what a beast it was, both in size and numbers, this was what everyone was waiting for. It has a staggering 24 Gigabytes of Vram and a whopping 10496 CUDA cores. That makes it the most powerful graphics card in the world. It is also the only one capable of reaching the limit of nearly 1 TB/s of the memory.
Elite gamers will find this an appealing choice, even though it is targeted at a professional audience. This is a card that will finally enable games on 8k resolution to run at 60 frames per second! This has not been achieved before and will offer great competition to the upcoming consoles. This card also supports a max resolution of 8k and 4 monitors. It draws a mind boggling 350 Watts of power at peak usage. The office PC I’m typing this on uses less power than that!!
What was missing
While the event had thousands of fans all hyped up it was not without some disappointments though. Nevertheless, it must be stated that these are minor problems when considering the grand scheme of things. A technology that enables multiple GPU’s to function in parallel (called SLI) is now gone from their flagship RTX 3080. But everyone saw this coming as, multi GPU systems are slowly dying. And Nvidia has been removing support one card at a time since the GTX 10 series. However, the RTX 3090 still supports SLI, so there is some comfort there. Which makes one wonder how much power can be obtained by linking 2 of those already overpowered cards together.
Another concern was since these are the first PCIE Gen4 cards from Nvidia. Many people still have older motherboards with older processors that support only the older generation PCIE Gen3. Therefore, there is a chance that these new cards will not be able to be used optimally to their maximum. However, this is a very niche situation and most people will be fine. It is shown to effect only people using many PCIE devices (i.e.: SSD’s, network cards etc.).
The future of RTX
Later on in the year, sources say Nvidia will release a few more graphics cards; like the budget/entry level gamer focused RTX 3060 and maybe a RTX 3050. We also might see a reappearance of RTX 30 Super cards, priced equally as the RTX 30 series but better. And if the performance of the present cards is anything to go by, the supers will not disappoint. This all brings to question, how is AMD – Nvidia’s chief competitor going to answer this release and when. They are set to launch “Big Navi” – their new series of GPU’s, very soon. The PS5 and Xbox Series X, which are looming on the horizon are powered by Big Navi.
All this can be summed up by what Mr. Jensen Huang said, “If the last 20 years was amazing, the next 20 will seem nothing short of science fiction.”