NVIDIA RTX 3080 Founder's Edition Performance Review
New Post has been published on https://videogaming.newonline.help/nvidia-rtx-3080-founders-edition-performance-review/
NVIDIA RTX 3080 Founder's Edition Performance Review
Two weeks ago, NVIDIA announced its most anticipated GeForce RTX 30 series video cards. Due to the ongoing global pandemic, the event was hosted online, live from the kitchen of NVIDIA’s CEO. Jensen Huang took the viewers on a tour of the Ampere architecture at the heart of the new cards and the new features and technologies that it made available. He also presented to the world three video cards from the top of Ampere’s product stack, the RTX 3080 FE, RTX 3070 FE, and RTX 3090 FE.
During the presentation, it was made clear that NVIDIA is not just building bigger and faster video cards. Taking in consideration the evolution of the PC gaming industry, the California-based tech giant is continuing to refine its GeForce ecosystem. With the introduction of Ampere, NVIDIA also revealed many new features and technologies that will benefit gamers, streamers, and content creators.
To achieve this, NVIDIA built its RTX 30 series around the new Ampere architecture which is fabricated on a SAMSUNG 8 nm custom process and boasting 28.3 billion transistors on a 628.4 mm2 die. This new architecture features an improved Streaming Multiprocessor (SM) design with second-generation Ray Tracing cores for improved ray tracing hardware acceleration, and third-generation Tensor Cores for increased AI inference performance and DLSS improvements that allows for 8K gaming. NVIDIA also implemented some cool features that are first seen with the RTX 30 series such as HDMI 2.1 support, 8k HDR capture, and AV1 encode. Here are the deets straight from the Green Team:
HDMI 2.1: GeForce RTX 30 Series GPUs are the first available to feature HDMI 2.1 support, for single cable output on 8K HDR TVs. This is critical for 8K HDR 60 FPS gaming since you need at least 71.66 Gbps before compression. This was not possible on TVs using HDMI 2.0b. Previous generation GPUs had HDMI 2.0b (18.1 Gbps) or DisplayPort 1.4a (32.4 Gbps), so an 8K TV would require 4 HDMI 2.0 cables, or a specialized DP-to-HDMI 2.1 converter, making setup more complex and expensive. Now, one HDMI 2.1 cable simplifies connection for 8K HDR gaming.
8K HDR Capture: Available with the launch of RTX 3080, the latest update to GeForce Experience delivers the first ever software capture solution capable of 8K HDR recording. GeForce Experience can now capture up to 8K 30 FPS, and support HDR capture at all resolutions, so you no longer need to buy a separate and expensive capture card for 8K HDR. You can also simply select Manual or Instant Replay from within the GeForce Experience overlay, record 8K 30 FPS HDR content, and share directly to YouTube.
AV1 Decode: Traditionally, 8K streaming using H.264 required up to 140 mbps of internet bandwidth for smooth real-time playback. With new AV1 decoding, bandwidth is reduced by more than 50%. However, even high-end CPUs struggle to keep up with AV1 decode. With AV1 GPU hardware acceleration on RTX 30-Series GPUs, users can consume up to 8K60 HDR content in AV1 seamlessly from YouTube.
At the heart of the RTX 3090 and RTX 3080 cards is the GA102 GPU, which is the top dog of the new NVIDIA GA10x class of Ampere-based chips. The full GA102 GPU implementation incorporates seven Graphics Processing Clusters (GPCs), 42 Texture Processing Clusters (TPCs), and 84 Streaming Multiprocessors (SMs). Each SM in GA10x GPUs contain 128 CUDA Cores, four third-generation Tensor Cores, a 256 KB Register File, four Texture Units, one second-generation Ray Tracing Core, and 128 KB of L1/Shared Memory. This means that a full GA102 comes with 10752 CUDA Cores, 84 RT Cores, and 336 Tensor Cores. The RTX 3080 FE is powered by a scaled down GA102 chip featuring 64 SMs, which translates into 8704 CUDA Cores, 68 RT Cores, and 272 Tensor Cores.
The Ampere architecture brings three major improvements compared to the previous generation Turing -based GPUs. As per NVIDIA, “the Streaming Multiprocessor (SM) in the Ampere GA10x GPU Architecture has been designed to support double-speed processing for FP32 operations. In the Turing generation, each of the four SM processing blocks had two primary datapaths, but only one of the two could process FP32 operations. The other datapath was limited to integer operations. GA10x includes FP32 processing on both datapaths, doubling the peak processing rate for FP32 operations. As a result, GeForce RTX 3090 delivers over 35 FP32 TFLOPS, an improvement of over 2x compared to Turing GPUs”.
The GA10x GPUs also feature NVIDIA’s second-generation ray tracing technology. “Each new RT Core includes a number of enhancements, combined with improvements to caching subsystems, that effectively deliver up to 2x performance improvement over the RT Core in Turing GPUs. In addition, the GA10x SM allows RT Core and graphics, or RT Core and compute workloads to run concurrently, significantly accelerating many ray tracing operations”.
Next are NVIDIA’s new third-generation Tensor Cores, “which support many new data types for improved performance, efficiency, and programming flexibility. A new Sparsity feature can take advantage of fine-grained structured sparsity in deep learning networks to double the throughput of Tensor Core operations over the prior generation Turing Tensor Cores. The third-generation Tensor Cores accelerate AI features such as NVIDIA DLSS for AI super resolution now with support for up to 8K, the NVIDIA Broadcast app for AI enhanced video and voice communications, and the NVIDIA Canvas app for AI-powered painting”.
Pushing the limits even more, NVIDIA equipped the RTX 3090 and RTX 3080 with 24GB and 10GB, respectively, of Micron’s GDDR6X. Announced as the world’s fastest graphics memory, GDDR6X “is the next big advance in high-bandwidth GDDR DRAM memory design. GDDR6X preserves the same data access granularity and memory module size as the widely accepted and high-performance GDDR6 memory standard introduced in 2018 but improves data rate and transfer efficiency in many ways. GDDR6X is the first consumer GPU graphics memory to reach memory bandwidth over 900 GB/sec. To achieve this breakthrough, innovative signal transmission technology and four-level pulse amplitude modulation (PAM4) completely redefines how the memory subsystem moves data. By using PAM4 multilevel signaling techniques, GDDR6X transfers more data and at a much faster rate, moving two bits of information at a time, doubling the I/O data rate of the previous PAM2/NRZ signaling scheme. Data-hungry workloads, such as AI inference, game ray tracing, and 8K video rendering, can now be fed data at high rates, opening new opportunities for computing and new end-user experiences”. For more details about Micron’s GDDR6X discrete graphics memory solution, check the official announcement.
To put all this in perspective, the table below shows the specs of the RTX 3080 FE compared to the RTX 2080 Super FE. This bodes well for the levels of performance to be expected from the Ampere cards in general. Today however, our focus is on the RTX 3080 FE.
GeForce RTX 2080 Super FE
GeForce RTX 3080 FE
SMs
48
68
CUDA Cores
3072
8704
Tensor Cores
384 (2nd Gen)
272 (3rd Gen)
RT Cores
48 (1st Gen)
68 (2nd Gen)
Texture Units
192
272
ROPs
64
96
GPU Boost Clock
1815 MHz
1710 MHz
Memory Clock
7751 MHz
9500 MHz
Total Video Memory
8192 MB GDDR6
10240 MB GDDR6X
Memory Interface
256-bit
320-bit
Memory Bandwidth
496.1 GB/s
760 GB/s
TGP
250W
320W
With the announcement of the GeForce RTX 30 series, NVIDIA rehauled its Founder’s Edition design in a manner even more drastic than what we’ve seen going from Pascal to Turing. We already took a closer look at the RTX 3080 FE and explored the design elements and cooling solution. NVIDIA had to design a compact PCB that wouldn’t interfere with the airflow needed to make this cooling solution work. Rated at 320W TGP, the RTX 3080 FE is equipped with 18 power phases fed by NVIDIA’s new 12-pin power connector, which requires the included power adapter to work with standard 8-pin PCIe cables.
The new Flow-Through design, as NVIDIA calls it, seems to integrate nicely with the standard airflow paths in most chassis. The two fans of the RTX 3080 FE are independent from each other and are strategically positioned to achieve different objectives. The fan located underneath the PCB, the one on the front of the card if you will, pulls air from inside the chassis then pushes it out through the large vents in the PCIe bracket. This takes care of a good portion of the hot air that otherwise would’ve been dumped inside the chassis. The second fan, located on the back of the card, pulls air straight through the fin-stack and directs it towards the exhaust areas in the top and back of the chassis.
This combined approach to GPU cooling worked well during my testing, which was done in a CORSAIR Crystal Series 680X RGB chassis. The RTX 3080 FE was very quiet under load and the fans managed to keep the temps below 76C at just 1000RPM, or 30% speed. When card is idling, the max temp I recorded was 47C with the fans completely still. I was pleasantly surprised to see NVIDIA implement a 0dB fan technology in its reference design. The moment the GPU reaches 60C, the fans ramp up in response and keep going until the temps are down to 33C, where the 0dB feature kicks in again. The rest of the components within the main chamber of the 680X chassis were not adversely affected by the new NVIDIA Founder’s Edition cooler. I was a bit worried about the RAM temps initially, and I kept an eye on the modules while gaming/benchmarking. When the card is under load, the RAM showed a negligible 2C temp increase.
Source link neoseeker.com