GPU Architecture and specs

GPU architecture and specifications

You may expect several graphics cards within the new 6000 lineups. AMD strickly is focussing on high-end to enthusiast-class with this release. But we have no doubt that RDNA2 based architecture of course will find its way to the lower spectrum of the product line over time as well. For the initial Big Navi release, we'll see the two Radeon RX 6800 (XT) graphics cards and a Radeon RX 6900 XT; they all are based on the same GPU holding a whopping 26.8 billion transistors. Inevitably, of course, there will be a Radeon RX 6700 series released as well. AMD claims a 50% perf per watt improvement in performance, double that of last-gen performance. Let's first have a look and overview of the specifications per product. As you can see AMD has been able to nearly double up the shader/stream processor count for a full enabled GPU. However take a look at the difference in transistor count, yes that's a big chip alright, reasonably explained due to the new infinity cache, as well as additional RT cores. As you can observe, a fully enabled Big Navi GPU holds 80 CUs (Compute units); each CU holds 64 shading/stream cores. So when multiplied you'll end up at 80x64= 5120 shader cores. Each CU has 1 RT (Raytracing) core thus 80 cores for the 6900XT, 72 for the 6800 XT, and 60 for the 6800. Historically AMD has had 4 texture units per CU, so that's 320 units for the 6900 XT, 288 for 6800 XT, and 240 units for the 6800. The ROP count is 128 units for 6800 XT and 6900 XT, 96 for the 6800. 

Infinity Cache

One of the biggest changes from previous GPU architecture is Infinity Cache (IC). Why IC? well, the choice of GDDR6 memory is a far cheaper approach than what NVIDIA is doing with GDDRX. This will help in the bill of materials for a graphics card. However, gDDR6 tied to 16GB AMD faces the challenge that the memory bus is a bit limited at 256-bits. 512-bit is complicated signal and wires wise so they figured that adding a level of cache memory will take the load off the memory bus. and that helps tremendously in performance per watt really, but also greatly helps raytracing.  So very simply put, IC is cache memory, and that cache memory (128MB) is placed directly into the chip itself (on-die). This also is one of the reasons that the Navi 21 GPU is considerably larger than Navi 10. Normally a GPU has a few megabytes of cache memory (L1 and L2). Then there's a huge gap in between the many gigabytes of VRAM that the frame buffer has. This gap is bridged with Infinity Cache. Arbitrarily speaking you could look at IC as an L3 cache that is more capable to provide the GPU with sufficient and faster data in a faster manner and reduces frame buffer utilization. AMD injected 128MB Infinity Cache used in combination with a 256-bit memory bus provides more than twice the bandwidth of a 384-bit memory bus. So a lot of the transistor budget was used for this feature. The cache does not just help in performance, as the GPU now takes advantage of lower energy consumption because there is less utilization on the memory controllers. 128MB Infinity Cache is implemented and used on all the RX 6900 XT and 6800 XT and RX 6800 graphics cards. In our findings, this big 'L3' cache helps out greatly.  

Raytracing

So what is raytracing really? Well, with raytracing, you basically are mimicking the behavior, looks, and feel of a real-life environment in a computer-generated 3D scene. Wood looks like wood, however, the leaking resin will shine and refract its environment and lighting accurately. Glass and waves of water get refracted as glass based on the surroundings and lights/rays. Can true 100% raytracing be applied in games? Short-term answer, no, partially. As you have just read and hopefully remembered, Microsoft has released an extension to DirectX; DirectX Raytracing (DXR). AMD now has dedicated hardware built into their GPUs to accelerate certain raytracing features. You have seen these in current games mostly as Shadow optimization, but most commonly used are reflections (water, puddles, windows, tiles, and so on). Rasterization has been the default renderer for a long time and you can add to that a layer of raytracing. Combining rasterization and raytracing we like to call Hybrid raytracing. and offers the best of both worlds. But make no mistake, the RT cores inside Big Navi can do full scene raytracing, however, would never be fast enough for real-time rendering.