Here’s why there’s little hard data on the RTX-40 series (report)
Igor’s laboratory recently released a detailed look at Nvidia’s estimated production and development schedule (opens in a new tab) for its new RTX 40-series GPUs. In it, it throws cold water on hot rumors of an imminent Nvidia launch, suggesting that these won’t be joining the top graphics cards anytime soon. . The current rumors surrounding the RTX 40 series are incredibly vague and contain very little hard data, other than the information that leaked via the $LAPSU hack earlier this year.
Igor’s opinion, based on his anonymous “sources”, is that Ada is not yet out of the engineering and/or design validation phase. This means that none of Nvidia’s AIB partners or Nvidia themselves have working technical samples of RTX 40-series cards. The only GPUs available right now are Ampere-based test products. That means it’s impossible for anyone to have anything approaching accurate performance data for Ada’s gaming or computational prowess. All we have are highly speculative rumors based purely on potential core counts and memory configurations.
If you’ve been following the GPU industry, that’s not particularly surprising. GPU prices are plummeting from their lofty highs of last year, and current indications indicate that we may be entering a phase of oversupply. Nvidia and AMD will want to clear as much current-gen inventory as possible before launching next-gen parts, so there’s no need to rush. And of course, China’s Covid-19 lockdowns aren’t entirely a thing of the past. Take that and then try building a whole new GPU architecture, with months of testing, and you can start to get a feel for how difficult it is to launch a new GPU.
Based on all of the above and the lack of reliable “sources” in the industry suggesting otherwise, Igor states that the RTX 40 series’ supposed launch window of July through early September is highly questionable. Again, he claims that there are currently no working Ada unit prototypes, which means that having the cards finalized and released within three months would be incredibly ambitious.
Nvidia and its partners have yet to obtain working samples, then conduct electromagnetic interference testing, production validation testing, driver and BIOS verification, and mass production. Given all of these factors, we’re probably looking more at a mid-Q4 launch window (around October or November) rather than a Q3 release. Tha, t or Nvidia has managed to crack down on leaks seriously for a change.
Nvidia is probably not alone. Intel has repeatedly pushed back the launch of its Arc GPU, which is now expected to hit desktops perhaps in Q3. Intel may be new to the dedicated GPU market, but it’s likely also facing similar issues, and AMD will do the same with RDNA 3. Hopefully things can stay on track for a 2022 launch of GPUs. next-gen, but we wouldn’t be shocked to see a relatively limited supply for whatever remains of 2022 when or if they launch.