![]() ![]() At one point in his presentation, Jensen said that Ampere delivers 1.9X the performance per watt as Turing. Let's also tackle the efficiency question quickly. Does the Ampere architecture justify the pricing? We'll have to wait a bit longer to actually test the hardware ourselves, but the specs at least look extremely promising. The RTX 3080 meanwhile costs $699, and the RTX 3070 will launch at $499, keeping the same pricing as the previous generation RTX 2080 Super and RTX 2070 Super. The GeForce RTX 3090 is set to debut at $1,499, which is a record for a single-GPU GeForce card, effectively replacing the Titan family. Thankfully, depending on how you want to compare pricing, pricing isn't going to be significantly worse than the previous generation GPUs. And if those workloads include ray tracing and/or DLSS, the gulf might be even wider. Combined with the architectural updates, which we'll get to in a moment, Nvidia says the RTX 3080 has double the performance of the RTX 2080. It's the largest single generation jump in performance I can recall seeing from Nvidia. ![]() What does that mean to the end users? Besides potentially requiring a power supply upgrade, and the use of a 12-pin power connector on Nvidia's own models, it means a metric truckload of performance. The RTX 3090 comes with an all-time high TDP for a single GPU of 350W (that doesn't count the A100, obviously), while the RTX 3080 has a 320W TDP. GeForce RTX 2080 Ti was a 250/260W part, and the Titan RTX was a 280W part. The V100 was a 300W part for the data center model, and the new Nvidia A100 pushes that to 400W. Nvidia is taking the middle route and offering even more performance at still higher power levels. While 7nm/8nm does allow for better efficiency at the same performance, it also allows for much higher performance at the same power. The consumer GPUs also increase in transistor counts while greatly reducing die sizes. That's a massive 156% increase in transistor count from the GV100, while the die size is only 1.3% larger. GA100 for example has 54 billion transistors and an 826mm square die size. Instead, Nvidia is taking all the extra transistors and efficiency and simply offering more, at least at the top of the product stack. With the shift from TSMC's 12nm FinFET node to TSMC N7 and Samsung 8N, many expected Ampere to deliver better performance at lower power levels. Of course, third party designs are free to deviate from Nvidia's designs. As an Nvidia video notes, "Whenever we talk about GPU performance, it all comes from the more power you can give and can dissipate, the more performance you can get." A reworked cooling solution, fans, and PCB (printed circuit board) are all part of improving the overall performance story of Nvidia's Ampere GPUs. That GPU also happens to offer Tensor and RT cores (edit: it does not have RT cores, we blanked!), as well as 32GB of HBM2, but it’s still hard to ignore that kind of price hike.Besides the underlying GPU architecture, Nvidia has revamped the core graphics card design, with a heavy focus on cooling and power. While you can potentially search used bins for older GPUs that happen to have decent double-precision performance, on the current GPU market, the only way you’re coming close to the Pro VII’s level of FP64 performance is to opt for the Quadro GV100, which costs about four-times as much. The Pro VII uncaps it the rest of the way, a la the Radeon Instinct cards, and ultimately delivers 6.5 TFLOPS of double-precision performance. The Radeon VII itself had partially unlocked FP64, at 1:4 the ratio of FP32. ![]() What truly sets the Pro VII apart from the crowd is its unlocking of double-precision (FP64) performance. AMD says that its Pro VII offers a bit more single-precision performance, and not to mention, PCIe 4.0 support – something last year’s Radeon VII did not. Its closest competitor is NVIDIA’s Quadro RTX 5000, which also includes 16GB of error correction memory (albeit with less bandwidth), but costs $400 more than the Pro VII. Vega 5th-gen GCN = WX 8200~9100 & Pro VIIĪMD is trying to make a really big statement with the Pro VII. It goes even further than that, though.įirst, here’s an overall look at AMD’s current lineup, with the Radeon Pro VII joined in:ĥ GDDR6 (ECC) 6 GDDR5X (ECC) 7 GDDR5 (ECC) 8 HBM2 (ECC) It also has 3,840 cores, which might lead you to believe this is realllly similar to the Radeon VII, just with pro unlocked features. This card is built on top of the compute-focused Vega architecture, not Navi, and like last year’s VII, it offers 16GB of HBM2 memory, which provides a healthy 1TB/s bandwidth. The fact that this is called Radeon Pro VII might give you some hints as to what’s under the hood.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |