NVIDIA stirred up the graphics card market last week with the debut of its GeForce RTX Super family of graphics cards that increase performance at existing price points in a bid to preempt AMD's Radeon RX 5700-series "Navi" graphics cards. The first two of three RTX Super-series SKUs are the RTX 2060 Super and the RTX 2070 Super. NVIDIA displaced the RTX 2070 from its $499 price-point with the better-endowed RTX 2070 Super as AMD beat the RTX 2070 at $399.
The most interesting aspect about the RTX 2070 Super is that it's based on the 13.6 billion-transistor "TU104" silicon since NVIDIA had maxed out the "TU106" with the original RTX 2070. The "TU104" is at the heart of the much pricier RTX 2080 and the upcoming RTX 2080 Super graphics cards. What this means to consumers is that most custom-design add-in card (AIC) partners would rather reuse their existing RTX 2080 board designs with a little cost cutting on the VRM instead of spending money on developing and validating new PCBs. Another benefit is partners using heavy cooling solutions that were originally designed to handle the much hotter RTX 2080, and perhaps even the RTX 2080 Ti.
NVIDIA carved the RTX 2070 Super out of the "TU104" silicon by disabling an entire GPC worth CUDA cores, leaving the chip with 2,560 out of its 3,072 CUDA cores enabled, besides 160 TMUs, 64 ROPs, 320 Tensor cores, and 40 RT cores. The memory subsystem is untouched. 8 GB of memory ticks at 14 Gbps and sits across a 256-bit wide GDDR6 memory interface, churning out 448 GB/s of memory bandwidth. The GPU clock speeds are increased, too, with up to 1770 MHz GPU Boost frequency compared to 1620 MHz on the original RTX 2070. Another neat little perk of being based on the "TU104" silicon is NVLink support, which enables 2-way SLI.
We have with us the ASUS Republic of Gamers (ROG) STRIX GeForce RTX 2070 Super O8G graphics card, which is the company's top offering based on this chip. Since it's based on the larger "TU104" silicon, ASUS uses the same board design as for its RTX 2080 STRIX OC graphics card, which is priced a segment above. What's more interesting is that unlike other manufacturers, ASUS did not disable any power phases and reuses the RTX 2080 STRIX OC PCB cent per cent. The massive DirectCU III cooler is also practically unchanged and capable of handling much higher thermal loads than what the RTX 2070 Super can put out. You get factory-overclocked speeds of 1905 MHz GPU Boost. ASUS is pricing the RTX 2070 Super ROG STRIX OC at $590, a massive $90 premium over the $499 MSRP.
The ASUS RTX 2070 Super STRIX OC looks identical to the same model without the "Super", which of course makes economical sense. On the back, you'll find a high-quality metal backplate. Dimensions of the card are 30.0 x 13.5 cm.
Installation requires three slots in your system.
Display connectivity options include two standard DisplayPort 1.4a, two HDMI 2.0b, and a VirtualLink connector, which is basically USB-C with DisplayPort routing and USB-PD, so a single cable can power, display, and take input from your VR HMD.
NVIDIA has updated their display engine with the Turing microarchitecture, which now supports DisplayPort 1.4a with support for VESA's nearly lossless Display Stream Compression (DSC). Combined, this enables support for 8K@30Hz with a single cable or 8K@60Hz when DSC is turned on. For context, DisplayPort 1.4a is the latest version of the standard that was published in April, 2018.
At CES 2019, NVIDIA announced that all their graphics cards will now support VESA Adaptive Sync (aka FreeSync). While only a small number of FreeSync monitors have been fully qualified with G-SYNC, users can enable the feature in NVIDIA's control panel regardless of whether the monitor is certified or not.
These points let you manually test the various voltage domains of the card by using a multi-meter and possibly making some tweaks through soldering.
A little button lets you turn off the RGB lighting of the card without the use of any software.
ASUS includes a dual-BIOS feature on their STRIX OC, which lets you switch to a quiet BIOS with fan stop and a lower fan curve.
You also get two 4-pin PWM fan headers to sync your case fan to the graphics card's fans and an addressable RGB header other RGB components can be connected to.
The board uses two 8-pin power connectors. This input configuration is specified for up to 375 watts of power draw.
With Turing, NVIDIA is using NVLink as a physical layer for its next-generation SLI technology. NVLink provides sufficient bandwidth for multi-GPU rendering at 8K 60 Hz, 4K 120 Hz, and other such bandwidth-heavy display resolutions. It's a point-to-point link between your GPUs, so, latencies will be lower compared to pushing data through the PCI-Express bus.
Disassembly
ASUS is using an elaborate system of six heatpipes on their cooler.
Once the main heatsink is removed, a black baseplate becomes visible, which provides cooling for the VRM circuitry and memory chips.
The backplate is made out of metal and protects the card against damage during installation and handling.
On the next page, we dive deep into the PCB layout and VRM configuration.
High-resolution PCB Pictures
These pictures are for the convenience of volt-modders and people who would like to see all the finer details on the PCB. Feel free to link back to us and use them in your articles or forum posts.
High-res versions are also available (front, back).