There’s never been much love lost between AMD and Nvidia when it comes to GPUs, but Nvidia CEO Jen-Hsun Huang’s recent comments about AMD’s just-announced Radeon VII are rather cutting.
Huang blasted the GPU as “underwhelming,” according to PCWorld, noting “The performance is lousy and there’s nothing new…[There’s] no ray tracing, no AI. It’s 7nm with HBM memory that barely keeps up with a 2080. And if we turn on DLSS we’ll crush it. And if we turn on ray tracing we’ll crush it.”
As for FreeSync, according to Jen-Hsun, it’s not even in competition with AMD. “We never competed,” Huang told PCWorld concerning FreeSync support. “(FreeSync) was never proven to work. As you know, we invented the area of adaptive sync. The truth is most of the FreeSync monitors do not work. They do not even work with AMD’s graphics cards.”
Huang went on to brag that the RTX 2080 is sold by every vendor before saying “It’s a weird launch, maybe they thought of it this morning.” Huang didn’t confine himself to insulting AMD; he also went after Intel, saying: “Intel’s graphics team is basically AMD, right?” Huang asked. “I’m trying to figure out AMD’s graphics team.”
Fact Check: Radeon VII vs. Nvidia RTX
AMD’s Radeon VII has not yet launched and its performance has not been measured. We can, however, estimate how the card might perform based on public statements. Anandtech’s Bench has data on GPU performance and the Vega 64 and RTX 2080 can be compared.
If we assume a flat 1.25x performance improvement over Vega 64, the field is mixed. In some cases, AMD would now be in competition with the RTX 2080. In some cases, it would still be behind. But AMD has never said that Radeon VII is 25 percent faster than Vega 64; it said only that it picked up 25 percent more performance in the same power envelope. That’s typically a reference to process node improvements rather than architectural improvements, and we don’t know if AMD fixed any low-hanging fruit with Radeon VII. It’s possible that the chip is nothing but a straight 7nm port. It’s also possible it picks up some additional perf.
Regardless of how attractive Radeon VII is, Nvidia has no ground to stand on when it talks about offering lousy performance and underwhelming value, considering the entire Turing GPU family is nothing but a price increase married to promises of future gaming support for ray tracing features you can currently use in a single game and an admittedly nifty anti-aliasing feature available in just one other. In 2016, $ 350 bought you 8GB of Nvidia VRAM with the GTX 1070. In 2019, $ 350 buys you 6GB of Nvidia VRAM on the RTX 2060, despite the near-term arrival of new consoles and the inevitable increase in RAM demands that comes with each new gaming generation. Yes, more games will eventually feature RTX and DLSS support, but the state of access to these features today is very weak.
I’ll be the first to acknowledge that the Radeon VII could be underwhelming and the 6GB VRAM buffer on the RTX 2060 could be properly sized. But the company that’s spent four months trying to shove price increases, currently useless features, and reduced VRAM at the same price point down the throats of the gaming community doesn’t get to complain about the lack of features in its competitors’ product as if it had demonstrated some kind of amazing alternative value proposition.
Fact Check: FreeSync
Claiming that FreeSync was “never proven to work” is a flat lie. Furthermore, it’s an insult to the reviewers that have evaluated FreeSync GPUs, monitors, and performance. I’ve never written a specific FreeSync versus G-Sync eval myself, but I’ve tested FreeSync and G-Sync displays. I’ve seen the effect of both with my own eyes in games I played. I’ve used tools like OCAT to compare performance with FreeSync enabled versus disabled. It works. So does G-Sync.
Replacing thoughtful and nuanced commentary with fact-free hurling of insults may fit the current political climate of the United States, but it’s a poor way to run a company. I have tested FreeSync and G-Sync on a variety of displays. While I’ve seen the usual gamut of oddities and compatibility issues that can crop up, I’ve seen absolutely nothing to suggest that most FreeSync monitors “don’t work.” And it’s hard to argue that he intended to make a more specific statement, given that he claims FreeSync doesn’t even work with AMD GPUs.
Obviously, we can’t literally state that there are no FreeSync monitors that mis-implement the standard. There are literally hundreds of FreeSync displays. But what we can say is that FreeSync panels have been reviewed. The feature has been tested. The implication that the standard was “never proven to work” is claptrap of the highest order. There have also been no general user reports indicating that FreeSync displays suffer from catastrophic failures when the feature is engaged and FreeSync has been in market for several years now. If this problem has flown under the radar it needs to be addressed, but the standard works. If it didn’t, why would Nvidia be moving to support it?
Nvidia’s overall market position is exceptionally strong. Truth be told, I’ll be surprised if the Radeon VII is robust competition for the RTX 2080, because it’s clear that AMD’s major focus is on Navi and future GPU generations. 7nm Vega may be a high-end replacement, but it’s also something of a stop-gap. But these kinds of remarks aren’t professional and they aren’t accurate. And to be perfectly blunt, we expect better. No one expects Nvidia to be full of praise for AMD products, but there’s a difference between talking up one’s own solution and lying about a competitor’s. Claiming that most FreeSync displays don’t work, even with AMD cards, is the kind of bombshell statement that can only be made responsibly if backed up by specific model data and third-party demonstrations under neutral conditions. When it comes to claiming FreeSync has never been proven to work at all, we have just two words in response: