GeForce RTX 3070, 3080 and 3090 preview & analysis

Graphics cards 1048 Page 1 of 5 Published by

teaser

NVIDIA Ampere and Introduction

GeForce RTX 3070, 3080 and 3090
NVIDIA Ampere GPUs coming to the consumer

In this technology preview and a bit of analysis, we'll be discussing the pending GeForce RTX 3070, 3080 and 3090 from NVIDIA as well as Ampere architecture. This article is preliminary based on what NVIDIA has been sharing (to date) as well as the stories that made the news in the past weeks. All that to get a grasp of what NVIDIA is releasing, with their new Ampere GPU architecture. During the presentation today, we'll try and fill in the gaps in specifications as anything you read hereafter this point, some are unconfirmed.


Untitled-11

Leather jacket man holding an RTX 3080, which he called the flagship, but really it isnt ..


Let's begin at the beginning; we need to go back to Q4 2of the year 2017 (!); here it was the first time the name Ampere for GPU was introduced by a German website, yes 2017. Of course, there also was chatter about other names, but again in 2017 Ampere got mentioned, and to date, NVIDIA has not listed this name in any of its roadmaps on the consumer side. It was with military-level secrecy that the Ampere consumer part was developed. Up-to a week or so ago, NVIDIA was able to keep everything under a very tight leash. Yes, it was unfortunate for them that the cooler design for the founders' cards leaked, but even now, to date, have you seen any benchmarks leak?

No benchmarks have leaked, how is that possible?

I'll let you in on that secret. The AIB partners have all been prepping their cards for months now. They have the products, engineering boards for a while. NVIDIA however, has not released a driver that works with anything other than the test software they supply. So get this, I am writing this article on September 1st, hours before the presentation, and still, the board partners have no idea what the performance is going to be like. We need to advance on that as the board partners even do not know the thermal capacity effect of their products. NVIDIA has provided them with test software that will work with the driver. Basically, these are DOS-like applications that run stress tests. No output is given other than PASS or FAIL. We know the names of these test applications: NVfulcrum test and NVUberstress test. For thermals, there is another unnamed stress test, but here again, the board partners can only see PASS or FAIL. Well, we assume they have tested with thermal probes. What this paragraph, well, to show you the secrecy that NVIDIA applied for this Ampere project. 

Ampere

First of all, the meaning of Ampere, of course, is the base unit of electric current in the international system of units. But the GPU is named after André-Marie Ampère, a French mathematician and physicist, considered the father of electrodynamics. NVIDIA has a track record of naming their GPU architectures after mathematicians and physicist or closely related fields, to name a few; Pascal, Fermi, Kepler, Maxwell and more recently Turing. While it was no secret that the new GPUs would be based on Ampere, we've seen much discussion about fabrication nodes, architecture, and specifications. Still, everybody seems to have forgotten that Ampere already launched earlier this year for the HPC market. The very first product based on Ampere was the NVIDIA Tesla A100, outfitted with a GA100 Ampere GPU based on 7nm fabricated at TSMC; that product holds 54 billion transistors and has 6912 shader cores. The GPU has a 7nm Ampere GA100 GPU with 6912 shader processors and 432 Tensor cores. That GPU has 108 Streaming Multiprocessors (x64 Shader processors for each SM). Mind you that A100 is not a fully enabled chip. This product was paired with 40GB of HBM2e memory. With the base design already placed into GA100, graphics cards like the GeForce RTX 3080 / 3090 the picture now should become more apparent. 

GPU Specifications

NVIDIA announced three Ampere based graphics cards. However, a week before announcements, specifications of the GeForce RTX 3080 and 3090 had leaked onto the web; however, it ended with a big twist in shader core counts as they doubled up from what everybody expected.  Interesting was to learn the GPUs are fabricated on an 8nm node derived from Samsung. This process is a further development of Samsung's 10nm process, which means that no EUV is applied in production just yet.

The first wave of announcements will see the GeForce RTX 3080 and 3090 being released first. These cards will be announced and then launched half September, and as a bit of a surprise, the GeForce RTX 3070 should be arriving as well. The initial launch thus entails the GeForce RTX 3070 8GB GDDR6, RTX 3080 10GB GDDR6X, and a 24GB GDDR6X based flagship, the GeForce RTX 3090. The lineup will see Gen2 ray-tracing cores and the 3rd iteration tensor cores. The cards will be PCIe 4.0 interface compatible and offer HDMI 2.1 and DisplayPort 1.4a. Specs then, below an overview. The number of RT and Tensor cores are not yet known, but shader count, memory volume, and clock frequencies we can report.  The NVIDIA GeForce RTX 3090 as a graphics card is going to be big on that GPU die size alright, the GA102-300-A1 GPU is armed with 10496 shader processors and 28 billion transistors.

    

Sdf4w

RTX 2080 Ti

RTX 3090RTX 3080RTX 3070
GPU 12nm TU102 8nm GA102-300 8nm GA102-200 8nn GA104-300
Transistors  18.6 Billion 28 Billion 28 Billion 17 Billion
Shader Cores 4352 10496 8704 5888
Raytracing Cores 68 82 68 46
Tensor Cores 544 328 272 184
ROPs 96 96 96
Texture Units 272 272
Base Clock 1350 MHz 1400 MHz 1440 MHz 1500 MHz
Boost Clock 1635 MHz 1695 MHz 1710 MHz 1730 MHz
Memory 11GB G6 24GB G6X 10GB G6X 8GB G6
Memory Clock 14 Gbps 19.5 Gbps 19 Gbps 14 Gbps
Memory Freq 7000 MHz 9750 MHz 9500 MHz 7000 MHz
Memory Bus 352-bit 384-bit 320-bit 256-bit
Bandwidth 616 GB/s 936 GB/s 760 GB/s 512 GB/s
Shader Perf TFLOP 13.4 35.6 29.8

20.3

RT Perf TFLOP 110 285 283

163

PCIe Gen 3.0 x16 4.0 x16 4.0 x16

4.0 x16

TGP 320W 350W 320W

220W

Price $1199 $1499 $699

$499

Released Out September 24 September 17

October

   

On the next pages an overview of the cards after which we'll dive into some architectural changes related to the shader processor count.

Share this content
Twitter Facebook Reddit WhatsApp Email Print