Almost two years ago, I published a review of the RTX 2080 and RTX 2080 Ti titled “You Can’t Polish a Turing.” The article was as much a deep dive into the history of GPU improvement introductions and the amount of time it takes for new technologies to be adopted as it was a direct review, and my conclusions followed the title: Turing, while innovative, was overpriced, unlikely to remain the top-end solution for as long as Pascal had, and a poor overall value. Nvidia would later fix their pricing problems with the launch of the RTX “Super” family once AMD had a competitive GPU in-market, but ray tracing remained a questionable investment for the entirety of Turing’s turn at the top of the stack.
Why open a discussion of Nvidia’s latest Ampere with a reminder of Turing’s launch flaws? Because Ampere, as near as I can tell, fixes every single one of them. Having taken such a negative stance on Turing at launch, it’s appropriate to revisit the question of whether I feel the same way about Ampere on its debut.
I do not.
ExtremeTech has not yet received an Ampere GPU for review, so I’m basing this opinion on coverage from our sister site PCMag, as well as HotHardware and Ars Technica. Sometimes, different reviewers have significantly different opinions on a product, but that’s not the case here. Everyone thinks Ampere is an amazing GPU.
Ars Technica writes: “With AMD finally nipping on its heels this year, Nvidia has arrived with a $699 beast of an RTX 3080 whose “Founders Edition,” without any overclocks added, [and] consistently beats an overclocked 2080 Ti Founders Edition in modern 3D games and software.”
PCMag: “In short, the GeForce RTX 3080 Founders Edition is among the most beautifully designed, well-supported, and uniquely powerful GPUs ever released. It’s a sea change in modern graphics-card manufacturing, and it represents a new era for a venerable company that shows no signs of slowing down anytime soon.”
Here’s Hot Hardware: “The bottom line is, NVIDIA’s got an absolutely stellar-performing GPU on its hands, and the GeForce RTX 3080 isn’t even the best Ampere has to offer — the upcoming GeForce RTX 3090 is bigger and burlier across the board.”
Image by Hot Hardware
There’s a lot more you can read about Ampere, from a discussion of its features and latency-reducing capabilities (Chris Stobing at PCMag is a one-time competitive gamer) to its jaw-dropping performance (all three sites serve up a healthy number of benchmarks in various categories). If you want to understand its workstation performance, I particularly recommend Rob Williams’ work at TechGage, where he puts the card through its paces in benchmarks like Blender and Optane. Spoiler: The numbers look great here, as well.
Ampere offers performance uplifts of 1.5x – 1.8x in 4K gaming, with the degree of improvement depending on the title. If the 1080 Ti was the first GPU to offer uncompromising 4K performance, the RTX 3080 is the first GPU to offer good 4K performance with a chance of turning ray tracing on. Features like DLSS 2.0, while not strictly introduced with Ampere, make DLSS far more interesting (and useful when it comes to boosting game performance). At $700, the GPU isn’t cheap — but it’s a vastly better deal than the RTX 2080 represented at launch. Comparing backward against Pascal, the launch price on the RTX 3080 is higher, but performance is far better. The eventual Ampere GPUs that hit the $500 price point will still significantly outperform any GTX 1080.
As for AMD, RDNA2 is going to have to be something very special to compete against Nvidia at the top of the product stack. Multiple reviewers also acknowledge that Big Navi needs to deliver more than 2x the performance of the top-end 5700 XT to match or exceed the new RTX 3080. AMD has promised a 1.5x performance-per-watt improvement over RDNA, but that doesn’t tell us anything about what the average performance improvement will be. We can assume that doubling up core counts would roughly double performance assuming the GPU is balanced, but that’s only an approximation — and AMD would need additional gains from clock or efficiency to bring Big Navi into the RTX 3080’s performance range.
AMD always had the option to position Big Navi as a strong option against the RTX 2070 and lower GPUs, similar to how the RX 5700 XT and 5700 were competitors for RTX 2060 and 2070. For now, however, Nvidia is decisively redefining high-end performance — and they’re doing it at better launch pricing this time around.
I haven’t had the ability to test one yet, but I trust the opinions of my colleagues. Ampere looks to be everything Turing wasn’t, I’m considering buying one out of my own pocket for AI upscaling work, and the GPU has no weak points that anyone has identified. Stellar work.
Now Read:
- Nvidia’s Ampere Looks to Deliver on Turing’s Promises, Smash Performance Records
- Nvidia Announces RTX 3070, 3080, 3090 GPUs: Ampere Crushes Turing
- AMD RDNA2 Launches Oct 28, But How Will It Respond to Nvidia’s Ampere?
No comments:
Post a Comment