AMD today takes a swing at the lucrative performance segment of PC graphics with the Radeon RX 5700 series. This price band, ranging from $350 to $500, has graphics cards you'd want for serious gaming at resolutions of 1440p and above, or 1080p with high refresh rates. The lack of much serious competition from AMD for the past couple of years has given NVIDIA an unchallenged dominance with which it could sell the 6 GB GeForce RTX 2060 for as high as $400 on some partner models, and the RTX 2070 starting at $500, with some partner models even touching $600.
NVIDIA's generally high pricing for its RTX 20-series has been supported by the GPU's new real-time raytracing and AI acceleration capabilities, which do seem to take up billions of transistors on the silicon. NVIDIA was also the first adopter of the GDDR6 memory, which may have been pricey initially. AMD's official position on real-time raytracing for the client segment is that while it is one of the new innovations, it's not a killer feature that should influence your buying decisions. Today's 3D games are still rasterized and will be for the foreseeable future. You should hence base your buying on classic parameters, such as gaming performance, energy-efficiency, noise, future-proofing, and overclocking-headroom, or so believes AMD.
The AMD Radeon RX 5700 lineup consists of two models at this time: the RX 5700 XT and the RX 5700. Both cards are based on the swanky new "Navi 10" silicon built on the 7 nanometer silicon fabrication process at TSMC. This is AMD's second 7 nm GPU after "Vega 20." Unlike Vega, which used HBM2 memory, Navi is more purpose-built for the client segment and retains the conventional single-die package and external memory chips. This is AMD's first GPU to implement GDDR6, the new high-performance memory standard with a data rate as high as 14 Gbps, which translates to 448 GB/s of memory bandwidth across the 256-bit wide memory bus, which is similar to the bandwidth the "Vega 10" silicon achieved with 2048-bit HBM2 memory.
Navi also implements PCI-Express gen 4.0, a new bus standard that doubles host interface bandwidth over PCI-Express gen 3.0. AMD is also launching its "Valhalla" desktop platform today, which combines a 3rd generation Ryzen processor with a motherboard based on the AMD X570 chipset. This is the only platform with PCIe gen 4.0 support until Intel "Ice Lake" comes along. Of course, Navi Radeon RX 5700 fully supports older PCIe standards and should run at gen 3.0 or lower just fine. The blurbs "7 nm" and "PCIe gen 4.0" are extensively used in AMD's packaging of these products, as if to say that "Navi" is a generation ahead of the competition that's stuck with 12 nm and PCIe gen 3.
"Navi" isn't an optical shrink and upscaling of AMD's existing GPU IP to 7 nm (which is what "Vega 20" was). Instead, it introduces the first major update to the core number-crunching machinery of the GPU since 2013, when AMD introduced Graphics Core Next (GCN). In its place, AMD is debuting the new RDNA SIMD architecture, which has numerous innovations that increase IPC over GCN without losing the kind of parallelism that made GCN rock at general compute applications. On the following pages, we will dive deep into the nuts and bolts of RDNA.
AMD also used the opportunity to update its display engine, adding support for more display formats and 8K over a single cable due to DSC 1.2a, or 4K at higher refresh rates. Radeon RX 5700 also has updated multimedia hardware acceleration to support some of the newer consumer video formats, such as H.265 HEVC and VP9 at various 4K and 8K resolutions.
The Radeon RX 5700 XT we are examining in this review today leads the pack at $399, a whole $100 cheaper than the GeForce RTX 2070 and recently announced RTX 2070 Super. There is a slightly faster AMD 50th Anniversary Edition of this card at $449, but it is a very limited launch and won't be considered standard for this SKU. The RX 5700 XT maxes out the 7 nm "Navi 10" silicon featuring 2,560 stream processors, 160 TMUs, and 64 ROPs. That's right, AMD has finally broadened its render-backends to improve raster performance. Until now, only the larger Vega GPUs and their high-end predecessors had such a high ROP count. NVIDIA implemented 64 ROPs on its performance-segment GPUs since "Maxwell". The RX 5700 we are reviewing in a second review today is a slightly cut down sibling that's $50 cheaper at $349.
Our exhaustive coverage of AMD's 7/7 Launch Day includes the following content:
AMD Ryzen 9 3900X 12-core processor | AMD Ryzen 7 3700X 8-core processor | AMD Radeon RX 5700 XT graphics card | AMD Radeon RX 5700 graphics card | AMD Zen 2 Memory Performance Scaling | Ryzen 3900X and 3700 on X470 vs X570 platforms | Radeon RX 5700 XT Navi PCI-Express 4.0 Performance Scaling | ASRock X570 Taichi motherboard | ASUS Prime X570-Pro motherboard
|RX Vega 56||$300||3584||64||1156 MHz||1471 MHz||800 MHz||Vega 10||12500M||8 GB, HBM2, 2048-bit|
|GTX 1660 Ti||$280||1536||48||1500 MHz||1770 MHz||1500 MHz||TU116||6600M||6 GB, GDDR6, 192-bit|
|GTX 1070 Ti||$450||2432||64||1607 MHz||1683 MHz||2000 MHz||GP104||7200M||8 GB, GDDR5, 256-bit|
|RTX 2060||$340||1920||48||1365 MHz||1680 MHz||1750 MHz||TU106||10800M||6 GB, GDDR6, 192-bit|
|RX 5700||$350||2304||64||1465 MHz||1625 MHz||1750 MHz||Navi 10||10300M||8 GB, GDDR6, 256-bit|
|GTX 1080||$500||2560||64||1607 MHz||1733 MHz||1251 MHz||GP104||7200M||8 GB, GDDR5X, 256-bit|
|RTX 2060 Super||$400||2176||64||1470 MHz||1650 MHz||1750 MHz||TU106||10800M||8 GB, GDDR6, 256-bit|
|RX Vega 64||$500||4096||64||1247 MHz||1546 MHz||953 MHz||Vega 10||12500M||8 GB, HBM2, 2048-bit|
|GTX 1080 Ti||$700||3584||88||1481 MHz||1582 MHz||1376 MHz||GP102||12000M||11 GB, GDDR5X, 352-bit|
|RX 5700 XT||$400||2560||64||1605 MHz||1755 MHz||1750 MHz||Navi 10||10300M||8 GB, GDDR6, 256-bit|
|RTX 2070||$480||2304||64||1410 MHz||1620 MHz||1750 MHz||TU106||10800M||8 GB, GDDR6, 256-bit|
|RTX 2070 Super||$500||2560||64||1605 MHz||1770 MHz||1750 MHz||TU104||13600M||8 GB, GDDR6, 256-bit|
|Radeon VII||$680||3840||64||1802 MHz||N/A||1000 MHz||Vega 20||13230M||16 GB, HBM2, 4096-bit|
|RTX 2080||$700||2944||64||1515 MHz||1710 MHz||1750 MHz||TU104||13600M||8 GB, GDDR6, 256-bit|
|RTX 2080 Super||$700||3072||64||1650 MHz||1815 MHz||1940 MHz||TU104||13600M||8 GB, GDDR6, 256-bit|
|RTX 2080 Ti||$1100||4352||64||1350 MHz||1545 MHz||1750 MHz||TU102||18600M||11 GB, GDDR6, 352-bit|