Nvidia’s GeForce RTX 3080 graphics card symbolizes why we inform individuals to attend for the second era when bleeding-edge know-how seems.
The radical new-look Turing GPUs inside Nvidia’s GeForce RTX 20-series packed all kinds of cutting-edge applied sciences designed to usher in real-time ray tracing, a protracted sought-after objective for the gaming business. Not solely did Turing introduce specialised RT cores dedicated to processing ray tracing duties, it additionally debuted tensor cores, devoted {hardware} that makes use of machine studying to assist denoise ray traced visuals and allow AI-enhanced instruments just like the improbable Deep Learning Super Sampling (DLSS) know-how. Turing’s enhancements additionally prolonged to the standard shader cores, introducing an overhauled processing pipeline higher outfitted to deal with video games constructed utilizing the newer DirectX 12 and Vulkan graphics APIs. All of those had been enormous departures from the norm.
But new doesn’t all the time imply nice. While the RTX 20-series certainly birthed a brand new ray-traced period, video games that supported ray tracing or DLSS had been few and much between over its lifetime. Worse, the RTX 20-series playing cards supplied primarily the identical efficiency in conventional video games as their older GTX 10-series predecessors on the identical worth level. The preliminary reception (and gross sales) weren’t glowing.
The $699 GeForce RTX 3080 and Nvidia’s new Ampere GPU structure modifications all that. This factor freakin’ smokes. It’s a large improve over the older $699 GeForce RTX 2080, considerably quicker than the previous $1,200 RTX 2080 Ti flagship, and for those who’ve been holding onto your older GeForce GTX 1080? The GeForce RTX 3080 completely crushes it. This is a superb graphics card for 4K and ultra-fast 1440p gaming—for those who can afford it when it launches on September 17.
Editor’s word: This complete assessment of the GeForce RTX 3080 goes longer than most because it’s our first analysis of an Ampere-powered GeForce RTX 30-series graphics card. Check out Nvidia GeForce RTX 3080 tested: 5 key things you need to know for high-level takeaways of this in-depth information, or use this desk of contents to hop between the varied sections of the assessment.
GeForce RTX 3080: Specs, options, and Ampere
Before we dive into what’s new within the GeForce RTX 3080, right here’s a high-level take a look at the Founders Edition card’s specs. You can discover extra information about the way it stacks up towards the earlier era in our GeForce RTX 30-series vs. RTX 20-series spec comparison.
- CUDA cores: 8,704
- Boost clock: 1.71GHz
- Memory: 10GB GDDR6X at 9500MHz
- Memory bus: 320-bit
- Memory bandwidth: 760GB/s
- RT cores: 68 (2nd-gen)
- Tensor cores: 384 (Third-gen)
- NVLink SLI: No
- PCIe: Gen 4
- HDMI: 2.1
- HDCP: 2.3
- Display connectors: 1x HDMI 2.1, 3x DisplayPort 1.4
- Length: 11.2 inches
- Width: 4.4 inches
- Height: 2-slot
- Maximum GPU temp: 93
- Graphics card energy: 320W
- Recommended energy provide: 750W
- Power connectors: 2x 8-pin (with provided 12-pin adapter)
Got it? Good. Now we’re going to get geeky for a bit. Skip to the following part for those who aren’t desirous about some deeper particulars on how the tech contained in the GeForce RTX 3080 works.
NvidiaThe beating coronary heart contained in the GeForce RTX 3080 is Nvidia’s new GA102 “Ampere” GPU. Ampere is constructed utilizing Samsung’s 8nm processor node, transferring up from the TSMC 14nm and barely modified 12nm nodes used for the GTX 10- and RTX 20-series playing cards, so this can be a generational leap for Nvidia. The final time Nvidia leaped a node, the GeForce GTX 10-series demolished their direct predecessors. It’s no totally different this time round, as you’ll quickly see.
NvidiaInside an RTX 30-series Ampere SM.
While Turing shook up GPU design, Ampere builds on Turing’s foundations. Turing’s simultaneous multiprocessors (SMs)—the constructing blocks of the GPU—obtained a major overhaul, including a brand new integer pipeline (INT32) alongside the floating level pipeline (FP32) historically used to course of shading. The new pipeline let Nvidia’s GPU deal with integer directions similtaneously conventional FP directions, giving Turing-based graphics playing cards an enormous velocity enhance in video games that leaned closely on these duties—specifically, well-optimized Vulkan and DirectX 12 video games. Performance in conventional video games stayed largely stagnant within the RTX 20-series graphics playing cards, nevertheless, partially as a result of FP32 is mostly extra essential for gaming workloads.
Ampere each builds on Turing and scales again that design by doubling the variety of CUDA cores in every SM. The new structure introduces an information path devoted solely to these essential FP32 duties, whereas a second path can course of both INT or FP duties. That makes Ampere a lot quicker at conventional sport rendering.
Note that total sport efficiency doesn’t scale up completely with CUDA core counts, particularly as a result of INT directions can now doubtlessly eat into the full FP throughput on that second path. All in all, it’s , affordable tweak. To feed the beasts, Nvidia doubled the bandwidth and partition measurement of the L1 cache in every SM, together with including 33 % extra L1 capability.
Those specialised RT and tensor cores additionally obtained upgrades—the RT cores to a second-generation model, and the tensor cores to the third era. There are extra RT cores crammed into the RTX 3080, and fewer tensor cores. But these third-gen tensor cores are far more succesful than their predecessors, enabling increased efficiency regardless of the lowered rely.
The RT cores additionally gained the flexibility to interpolate triangle place. This significantly helps in scenes with movement blur, doubtlessly rushing up ray traversal by 8X. Overall, Nvidia says its new RT cores are 1.7X quicker than earlier than, whereas the tensor cores are 2.7X extra environment friendly—although once more, that doesn’t scale completely to precise gaming efficiency.
NvidiaThose features are boosted even additional by Ampere’s newfound means to run duties on the tensor and RT cores concurrently. In Turing, ray-traced duties ran alongside shader capabilities, however the GPU wanted to complete that ray-traced job and hand it to the tensor cores for denoising and DLSS earlier than spitting out the ultimate picture. In Ampere, tensor cores can run DLSS to upscale one body whereas the shader and RT cores work on the following one, all three course of duties operating concurrently. Paired with the improved shader capabilities, a body that took as much as 13 milliseconds to run on Turing with ray tracing and DLSS enabled can now be spit out in 6.7ms with Ampere. Hot rattling.
Bottom line? Improvements within the Ampere GPU structure give an enormous enhance to efficiency in each conventional video games in addition to video games that leverage Nvidia’s RTX capabilities.
But the GeForce RTX 3080’s features don’t come solely from the brand new Ampere structure. Like the GTX 1080 and RTX 2080 earlier than it, Nvidia’s newest flagship introduces cutting-edge reminiscence, too. This time round, the GeForce RTX 3080 (and forthcoming RTX 3090) faucet Micron’s blazing-fast GDDR6X reminiscence modules. Micron’s reminiscence rolls out “PAM4” signaling know-how, which lets them ship 4 attainable values per cycle, up from the standard two. That lets GDDR6X transfer information twice as quick as earlier incarnations. The 10GB of GDDR6X reminiscence within the RTX 3080 sends that information over a 320-bit bus, for a complete reminiscence bandwidth of 760GBps. By comparability, the GeForce RTX 2080 Super’s 8GB of non-X GDDR6 used a 256-bit bus for 496GBps of total bandwidth.
NvidiaTraditional GDDR6 vs. the RTX 3080’s GDDR6x reminiscence.
Nvidia additionally tweaked how the reminiscence works to enhance its effectivity— factor, because the 320W GeForce RTX 3080 attracts noticeably extra energy than even the previous-gen RTX 2080 Ti flagship. The firm added 250mV voltage steps and new “max transition avoidance coding” to benefit from the four-signal PAM4 reminiscence with out having the chips swing wildly from the bottom to highest values, which might trigger power arcs. Nvidia additionally gave the graphics core and reminiscence system their very own devoted energy rails, letting the GPU fine-tune the facility ranges of every individually for higher effectivity. Previous chip designs used a shared energy rail for the 2.
Now that we’re speaking energy, it’s a good time to shift gears to the design of the GeForce RTX 3080 Founders Edition.
Next web page: RTX 3080 FE design