Nvidia relies on real-time raytracing in the Turing graphics architecture. And as early as 2018, the company's skeptics wondered if this game was smart. Today, with significant performance enhancements in the first line of raytracing games and a growing number of Triple A companies committed to adding the technology, it seems pretty clear that many respected developers really want raytracing in their projects ,
Now it's all about making enough hardware available to a wider audience fast enough. The lowest Turing board with raytracing acceleration is the GeForce RTX 2060, which is available from $ 350. The GeForce RTX 2070 costs only $ 500, the $ 2080 700. Nvidia's own overclocked Founders Edition models have earned over $ 1
Today's GeForce RTX 2060 Super, GeForce RTX 2070 Super, and GeForce RTX 2080 Super Nvidia product launches offer RTX cards at more competitive prices, boosting performance across the board and eliminating the Founders Edition tax.
Still the Same
Although we are discussing three new products, the GPUs they focus on and their underlying architectures remain unchanged from what we have introduced in Nvidia's Turing Architecture Explored. We all know that the impetus for these supercards is renewed competition from the upcoming AMD Radeon RX 5700 and 5700 XT. However, Nvidia says it only uses the improved yields and process optimizations to balance resources and clock rates on-chip.
| Vintage GeForce RTX product line
|| New GeForce RTX product line
|| GeForce RTX 2080 Ti FE – 1200 USD
GeForce RTX 2080 Ti Partners – 1000 USD
|Now|| GeForce RTX 2080 Ti FE – $ 1200
GeForce RTX 2080 Ti Partners – $ 1000
| GeForce RTX 2080 FE – $ 800
GeForce RTX 2080 Partners – 700 US $
|GeForce RTX 2080 Super – $ 700||7/23|
| GeForce RTX 2070 FE – $ 600
GeForce RTX 2070 Partners – 500 US -Dollar
|Discontinued||GeForce RTX 2070 Super – $ 500||7/9|
|GeForce RTX 2060 Super – $ 400||7/9|
| GeForce RTX 2060 FE – $ 350
GeForce RTX 2060 Partners – $ 350
|Now||2060 – $ 350||Now|
The result is a slightly larger stac k, where the GeForce RTX 2070 and GeForce RTX 2080 are completely replaced by their super-successors. The GeForce RTX 2060 remains at its previous price, while the GeForce RTX 2060 Super is directly above anyone willing to pay an additional $ 50. GeForce RTX 2080 Ti also maintains its lead in Nvidia's GeForce RTX portfolio.
How do these maps compare to the upcoming Navi boards from AMD? The Radeon RX 5700 XT, which is expected to surpass or bind the original GeForce RTX 2070, costs $ 450. That's right between GeForce RTX 2070 Super and GeForce RTX 2060 Super. In a world without the supers, the AMD card would have been faster and cheaper from the start. However, if it turns out that Nvidia's claim that the GeForce RTX 2070 Super proves to be correct 16% faster than a 2070 vanilla, the Radeon loses its advantage and has to fight without raytracing support.
||GeForce RTX 2060 Super|| GeForce RTX 2070 Super
||GeForce RTX 2080 Super|
| Architecture (GPU)
19659013] Turing TU106)
|Turing (TU104)||Turing (TU104)|
| CUDA cores
||2176||2560|| 3072 [1965900FP32calculation
||7.2 TLFOPS||9.1 TFLOPS||11.2 TFLOPS|
| Tensor cores
||272||320|| 272 cores
||34  40||48|
| Texture units
||136||160||192|| ] Base clock
||1470 MHz||1605 MHz||1650 MHz|
| GPU Boost Rate
||1650 MHz||1770 MHz [196590111815MHz|
| Memory Capacity
||8 GB GDDR6  8 GB of GD DR6||8 GB of GDDR6|
| Memory Bus
||256 bits||256 bits||256 bits|
| Memory bandwidth
||44 GB / s||448 GB / s||496 GB / s|
| L2 Cache
||4 MB||4 MB||4 MB|
||175 W||215 W||250 W|
| Transistor count
|13.6 billion||13.6 billion|
| Die size
||445 mm²||545 mm²||545 mm²|
|SLI support ||No||Yes||Yes|
A Radeon RX 5700 for $ 380 splits the GeForce RTX 2060 Super and the GeForce RTX 2060 in a similar way. According to Nvidia, the average power of the 2060 Super is 15% higher than that of the standard 2060. It's slightly faster than the Radeon RX 5700, with the added benefit of having these fi RT cores with fixed function. Of course, a $ 20 higher price gives players something to think about when both cards strike a beat.
Nvidia does not try to advertise bargain-priced enthusiasts here. It calculates as much as possible at each credit point while addressing the competition of today's top titles. With the support of Ray Tracing in real time, the company wants to marginalize future-oriented gamers. And with a growing list of Raytrace games, it's becoming increasingly difficult for skeptics to call the technology a gimmick.
Get to know GeForce RTX 2070 Super.
Nvidia presents GeForce RTX 2070 Super as a faster version of Vanilla 2070, which outperforms the GeForce GTX 1080 Ti. However, the 2070 Super is based on a completely different GPU than its predecessor.
The card introduced in the Founder Edition of Nvidia GeForce RTX 2070: Replacing the GeForce GTX 1080 used a full-featured TU106 processor. There was nowhere to grow. Nvidia would probably have been able to improve its 1.710 MHz GPU boost rating a bit. However, when the company wanted to outperform the AMD-reported Radeon RX 5700 XT performance requirements, it needed something faster.
Instead of working with an already mature TU106, Nvidia began using the same TU104 chip in GeForce RTX 2080/2080 Super. In summary, a complete TU104 consists of six graphics processing cluster building blocks, each with four texture processing clusters. The TPC is further divided into two streaming multiprocessors (SMs). Each SM hosts 64 CUDA cores, eight tensor cores, one RT core to speed ray tracing, four texture units, 256 KB of register space, and 96 KB of L1 cache / shared memory. TSMC manufactures the second-largest Turing GPU on its 12 nm FinFET node and packages 13.6 billion transistors on a 545 mm² chip.
When eight of the chip's streaming multiprocessors are worked out, a GPU with 40 active SMs remains, spread across five or six Graphics Processing Cluster (GPC) devices are. This number varies depending on where the SMs are cut out. Multiply the contents of each SM by 40, and you get a front end with 2,560 CUDA cores, 320 tensor cores, 40 RT cores, and 160 texture units.
This 11 percent infusion of resources is then boosted by higher clock rates. While the GeForce RTX 2070 was launched with a base frequency of 1,410 MHz and a GPU boost frequency of 1,710 MHz, the 2070 Super starts at 1,605 MHz, but generally operates at 1,770 MHz, according to Nvidia. In our tests, the GeForce RTX 2070 Super reached an average of 1,870 MHz and reached the 1,806 MHz of the Vanilla 2070.
The TU104 and TU106 both feature 256-bit aggregate memory buses running at 8 GB with GDDR6 motion data up to 448 GB / s are equipped. Each of its 32-bit memory controllers has 512 KB of L2 cache and eight ROPs, so the two GPUs similarly have 4 MB of L2 and 64 ROPs. The clock rate is a significant performance penalty compared to the original GeForce RTX 2070 connected. The 215W power is 23 percent higher than the 175W older card specification, which puts power consumption above the average 2070 Super's performance. Efficiency seems to be struck in the interests of absolute frame rates. By the way, the 2070 Super appears under the GeForce RTX 2080 Founders Edition and the 225 W specification.
On the positive side, the acquisition of TU104 means that the GeForce RTX 2070 Super theoretically supports a single x8 NVLink connection with a capacity of 50 GB / s. pioneering. On the Founders Edition card that Nvidia sent us to test, the physical interface is exposed.
In fact, the 2070 Super looks just as different with the exception of the GeForce RTX 2080 Founders Edition from Nvidia Model identifier between his fans and on the back wall. That means you now need to attach a six-pin and eight-pin auxiliary plug instead of the single eight-pin plug on the back of the GeForce GTX 2070 Founders Edition. While Nvidia's own 2070 offers two DisplayPort 1.4 interfaces, HDMI 2.0b, a rare DVI-DL output, and VirtualLink support via USB Type C, the new 2070 Super DVI swaps against a third DP port, as does GeForce RTX 2080 Founders Output.
Meet GeForce RTX 2060 Super.
Nvidia had more flexibility with GeForce RTX 2060 Super. The original 2060 already used a slimmed-down version of TU106, which came from incomplete processors that could not be used for GeForce RTX 2070. A complete TU106 consists of three GPCs, each with six texture processing clusters. The TPCs each include two SMs, each containing up to 36 SMs per processor. By reducing six SMs for GeForce RTX 2060, Nvidia received 1,920 CUDA cores, 240 tensor cores, 30 RT cores, and 120 texture units. The 2060 also lost two of the available memory controllers of the TU106, 1 MB cache and 16 ROPs. 
With GeForce RTX 2060 Super, Nvidia enables most of the missing resources. Now only two SMs remain inactive. The resulting configuration comprises 34 SMs comprising 2,176 CUDA cores, 272 tensor cores, 34 RT cores and 136 texture units. The base clock of the 2060 at 1365 MHz is increased to 1470 MHz, while the GPU boost rating of the original 1680 MHz card is slightly reset to 1650 MHz.
The 2060 Super also benefits from a wider 256-bit memory bus. A complete TU106 GPU has eight 32-bit controllers, two of which are disabled in GeForce RTX 2060 to create an aggregated 192-bit path. Activation in GeForce RTX 2060 Super results in a significant increase in bandwidth from 336 GB / s to 448 GB / s when using the same 14 Gbit / s GDDR6 modules. Equipping the design with 33% more throughput has other pleasant side effects, such as increasing the storage capacity to 8 GB, adding an extra megabyte of shared L2 cache, and exposing two more ROP partitions for a total of 64 ROP units.
Nvidia photographed here for a very specific performance profile. With the GeForce RTX 2060 continuing to be sold, the company needed a much faster, but not-so-upgraded version that would outmaneuver existing GeForce RTX 2070 cards. The GeForce RTX 2060 Super fits perfectly for this purpose. We are not entirely convinced that the additional model was necessary. Someone there wanted to outfit AMD's upcoming Radeon RX 5700 with low-priced RT-core cards so they could say, "See, we're doing everything they do, plus ray tracing." The result was a $ 400 GeForce RTX 2060 Super ,
The output on Nvidia's own Founders Edition card looks exactly like the one we discussed Map in Nvidia GeForce RTX 2060 Review: Is the ray tracing feature for the mainstream really? You get two DisplayPort 1.4 interfaces, HDMI 2.0b and the rare DVI-DL port. VirtualLink continues to be supported via a USB Type-C port. And like the other RTX 20 Series cards, you can use all four display outputs simultaneously for multi-monitor arrays. 
A single eight-pin power connector on the back should not cause assembly problems. Since 2060 Super is just as short as the 2060 Founders Edition, a cable that comes out of the back does not hit low-hanging hard drive cages. And it definitely looks better than a plug that sticks out. The performance of the 2060 Super board jumps from the 160-W specification of the 2060 Super to 175W. However, nine percent more power is child's play when the average frame rate of our benchmark suite is 13.5 percent faster.
At the front, Nvidia uses the same pair of 8.5-cm axial fans with 13 blades found in all others Founders Edition models that blow through a vertically aligned fin stack underneath. While we still miss the radial fan and acrylic window of the previous Nvidia reference cards, there is no doubt that the 2060 is super rugged and made of high quality materials.
MORE: Best Graphics Cards
MORE: Desktop GPU Performance Hierarchy Table
MORE: All Graphics Content