Welcome to the Red Crew. That is the message that first greets you for folk who grab the lid on one of AMD’s original Radeon 6000 sequence graphics playing cards, and or no longer it’s becoming – at the least, AMD’s fans are some of basically the most die-laborious in gaming, and these ‘Tall Navi’ GPUs were anticipated with nearly messianic fervour by the community at colossal. Fortunately, all of that particular, memeable energy has discovered an implausible goal in the RX 6800 and RX 6800 XT, two of the strongest creations by AMD’s Radeon graphics team in some time.
Sooner than we discuss how well these playing cards plot in a enormous assortment of games – and do not anxiety, there can be hundreds of that – let’s like a flash location some expectations. At $580 (£530) and $650 (£600), respectively, the RX 6800 and RX 6800 XT slot in above the $500 (£450) RTX 3070 and under the $700 (£650) RTX 3080. That implies we’ll sight an RTX 3070-beating performance out of the RX 6800, whereas the RX 6800 will trust well to blueprint with the elevated-priced Nvidia card. More importantly, it also capability that AMD is at final competing at the high discontinuance of the graphics market yet again, something the firm hasn’t finished for the reason that in uncomfortable health-fated Radeon 7 nearly two years up to now – and something it hasn’t finished successfully for many more years than that.
Besides to great improved out-and-out performance, these Tall Navi playing cards also sign the PC debut of AMD’s RDNA2 architecture, which also kinds the root of both the PlayStation 5 and Xbox Series X|S consoles. Esteem those subsequent-gen machines, the RX 6000 sequence helps hardware-accelerated ray tracing, at final ending Nvidia’s RT monopoly, plus a raft of more than a few DirectX 12 Closing aspects indulge in DirectStorage, variable price shading and mesh shaders that can boost performance in myriad recommendations. And talking of performance boosting, AMD’s engineers hang also added aspects indulge in ‘Dapper Memory Entry’, ‘Rage Mode’ overclocking and an ‘Infinity Cache’ that can hang to all conspire to push these playing cards to their limits.
From this bewildering description on my own that you just would possibly sight that there is hundreds to veil right here, so let’s procure began with a nearer witness at every of these playing cards – how are they built, how does the hardware stack up and the scheme does it tie into the original aspects on offer right here?
|AMD GPU Specs||CUs||Increase Clock||VRAM||Mem Interface||TDP||Designate|
|RX 6900 XT||80||2250MHz||16GB GDDR6||256-bit + 128MB IC||300W||$999|
|RX 6800 XT||72||2250MHz||16GB GDDR6||256-bit + 128MB IC||300W||$649|
|RX 6800||60||2105MHz||16GB GDDR6||256-bit + 128MB IC||250W||$579|
|RX 5700 XT||40||1905MHz||8GB GDDR6||256-bit||225W||$399|
|RX 5700||36||1725MHz||8GB GDDR6||256-bit||180W||$349|
|RX 5600 XT||36||1560MHz||6GB GDDR6||192-bit||150W||$279|
|RX 5500 XT||22||1845MHz||8GB GDDR6||128-bit||130W||$199|
Whereas specs sheets once in a while is a little bit of tiresome, taking a take a study this one tells you rather loads this time spherical. Whereas AMD are working with the an identical 7nm TSMC process as the authentic ‘Diminutive’ Navi procure, their original architecture manages to wring out hundreds more performance. Here is pushed by considerably elevated clock speeds all around the board and a elevated complement of more in my opinion performant compute units. There would possibly be also devoted ray tracing hardware backed into every CU, which is ten instances sooner at calculating ray intersections than a purely system capability – and it will likely be fascinating to sight the scheme it compares with Nvidia’s 2nd-generation RT hardware.
These compute improvements are backed with an modern original memory answer. By pairing the 16GB of GDDR6 video memory with a 128MB cache, equal in procure to the L3 cache discovered on AMD’s Ryzen processors, AMD are ready to function some blistering results from the cardboard’s somewhat pedestrian 256-bit bus – of their slides, the machine affords 2.17x the bandwidth you’d quiz from a outmoded 256-bit interface. This boost in memory bandwidth helps offset the elevated compute unit (CU) depend and elevated clock frequencies, whereas conserving vitality consumption under care for a watch on. It is a suave transfer that implies that Nvidia’s advantage from its early adoption of (pricey) GDDR6X memory on the RTX 3080 is at the least partially ameliorated. It also capability that AMD can provide you the money for to provide 16GB of VRAM on both objects, something that can little doubt salve the nerves of avid gamers panicked about working out of video memory with a raft of subsequent-generation games in the pipeline. This 16GB of VRAM can even be accessed with out downside on Ryzen 5000 systems by AMD’s Dapper Entry Memory tech, a performance-enhancing characteristic we’ll veil in elevated depth on page six of this analysis.
That correct about wraps up the architectural improvements interior, so what about the business procure exterior?
Whereas custom RX 6800 and RX 6800 XT playing cards can be produced by a wide form of AMD companions, at the original time’s analysis centres on AMD’s reference objects. Whereas these GPUs are location to retail at the harmful RRP of every card – $580 for the RX 6800 and $650 for the RX 6800 XT – they feel indulge in anything else however entry-level designs, with a mighty three-fan axial cooler, hundreds of steel fins to dissipate heat and a more advanced aluminium masks than the boxy Radeon 7. Total, AMD’s Radeon 6000 playing cards are weighty and feel thoroughly constructed, without a give or flex evident wherever. The vanilla RX 6800 feels in particular dense, with a more compact two-slot procure, whereas the 2.5-slot procure of the RX 6800 XT affords more space for the elevated-powered card to breathe. Both designs are the an identical size and height, at 267mm (10.5″) long by 120mm (4.7″) high, so that they favor to suit into any identical outdated ATX case with out points.
These designs are a enormous step forward from the blower-fashion RX 5700 sequence playing cards, and will plot great better in identical outdated PC cases with adequate airflow. I in particular loved the zero RPM fan mode, which appears to settle between spherical 52 levels celsius and cuts the fans fully for affirm silence throughout identical outdated day to day computing duties. Here is much from the principle GPU to boost this kind of characteristic needless to claim, however it is uncommon to sight it on a first-birthday party procure and it speaks to the usual of AMD’s cooling answer that or no longer it’s even purposeful to incorporate. All the scheme via my sorting out, the playing cards ran nonetheless, with the ‘balanced’ vitality profile we historical for sorting out capping the fan velocity at correct 50 per cent of their ~2000RPM maximum. We are going to hurry away it to the Buildzoids and Avid gamers Nexuses of the sector to trust a deeper dive into the internals, however the 15 vitality phases and 14-layer PCB particular sound impressive – and we’ll procure a nearer witness at vitality usage later on this very page.
In terms of ports, we’re taking a take a study a rather assorted outfit as Nvidia’s RTX 30-sequence playing cards: one HDMI 2.1 port (supreme for those subsequent-gen 4K 120Hz or 8K 60Hz TVs that you just would possibly well possibly additionally want picked as much as employ with a brand original console) plus two DisplayPort 1.4 ports and one USB-C port (which needless to claim can be historical as a third DisplayPort, HDMI and even DVI-D with the supreme kind adapter). As with the Nvidia playing cards, AV1 decoding is incorporated which would possibly hang to permit websites indulge in Netflix and Twitch to movement top of the variety, high frame-price video articulate material to you whereas requiring severely lower bandwidth than newest codecs. Sooner or later, the cardboard is PCIe 4.0 like minded however works correct fine in PCIe 3.0 slots too. For vitality, you procure two eight-pin inputs on both card, with AMD recommending a 650W PSU for the 6800 and a 750W PSU for the 6800 XT – pretty sensible for mid-differ to high-discontinuance graphics hardware.
To procure a greater judgement of Tall Navi’s vitality efficiency whereas gaming, we historical Nvidia’s (agnostic) PCAT machine – an interposer board that sits between the graphics card and its vitality inputs to measure proper vitality consumption. When twinned with in-sport performance metrics, indulge in frame-price and frame-time, we can calculate exactly what number of watts are required to render every frame, and thereby procure a measure of proper-world efficiency. AMD cite a 54 per cent enchancment in performance per watt between RDNA and RDNA2, so is that borne out by sorting out? And how does Nvidia’s Ampere architecture, our newest efficiency champion, compare?
|Joules Per Frame||RX 6800 XT||RX 6800||RTX 3080 FE||RTX 3070 FE||RTX 2070||RX 5700XT|
|Joules Per Frame||RX 6800 XT||RX 6800||RTX 3080 FE||RTX 3070 FE||RTX 2070||RX 5700XT|
We sight that in Death Stranding, AMD’s RDNA2 architecture is a little bit more efficient than Nvidia’s Ampere, and severely better than first-gen RDNA (represented by the RX 5700 XT), coming shut to AMD’s claimed efficiency beneficial properties. Alternatively, this sport is terribly kind to AMD hardware, so what a pair of more balanced chance?
In Gears 5, the 3070 affords better competitors towards the RX 6800 and proves to be the a little bit more efficient card, although the margin is correct five per cent in Nvidia’s favour. Clearly, both original architectures are severely more vitality-efficient than their predecessors, which speaks to the engineering that has gone into every chance.
Interestingly, the RX 6800 XT requires between 15 and 20 per cent more joules per frame in the 2 games we examined, so the more compact 6800 is surely the sequence chief with regards to vitality efficiency. We seen a equal margin between the 3070 and 3080, so right here’s infrequently surprising.
With our transient vitality sorting out concluded, or no longer it is time to transfer onto the sport benchmarks. You’re going to sight a pair of original titles right here for folk who would possibly well hang no longer checked out our RTX 30-sequence critiques yet, besides a brand original test platform. If you happen to would possibly well hang gotten read those critiques – thanks! – then be at liberty to skip to the next page as there is nothing original right here.
We examined AMD’s Ryzen 3950X as a candidate for our GPU test bench plot, as this lets in procure correct of entry to to the elephantine PCIe 4.0 bandwidth supported by AMD’s 6000 sequence and Nvidia’s 30-sequence, however we at final settled on an Intel Core i9 10900Okay for our test machine, replacing the Core i7 8700Okay we historical for our prior generation assessments. Although the Intel machine is technically constrained to PCIe 3.0 bandwidth, thus doubtlessly limiting performance in GPU-toddle eventualities, we restful got elevated 4K frame-rates with the Intel chip than with the fastest Ryzen processor accessible at the time. The Intel machine showed a more valuable frame-price lead at 1080p and 1440p, thus making it our ultimate wager for conserving off CPU bottlenecking as great as imaginable to permit assorted graphics playing cards to differentiate themselves. We are going to take into yarn Ryzen 5000 or its successors the next time we change our test bench, however for now it is life like to stay with it so we don’t must retest dozens of graphics playing cards in basically the most busy length of the year (and throughout a brand original console cycle).
To develop particular that our results are connected, the Core i9 10900Okay is locked to a 5GHz all-core frequency and cooled by a 240mm Alpacool Eisbaer Aurora AiO which keeps the overclocked machine at spherical 75C under elephantine load. The 10900Okay is backed by an Asus Maximus 12 Wrong Z490 motherboard and two 8GB sticks of G.Ability Trident Z Royal 3600MHz CL16. Our games are breeze from a capacious 2TB Samsung 970 Evo Plus NVMe pressure equipped by Field. Your whole rig is powered by a 850W gold-rated Gamer Storm vitality provide.
Now that you just’re accustomed to our test mattress and hang a appropriate kind overview of the playing cards themselves, let’s procure into some gaming assessments to sight how they plot.
AMD Radeon RX 6800 and RX 6800 XT Diagnosis
Introduction, Hardware and Energy Diagnosis [This Page]
Doom Eternal, Retain a watch on, Borderlands 3, Shadow of the Tomb Raider – Game Benchmarks Share 1
Death Stranding, Some distance Notify 5, Hitman 2, Assassin’s Creed Odyssey – Game Benchmarks Share 2
Metro Exodus, Grime Rally 2, Assassin’s Creed Solidarity – Game Benchmarks Share 3
Retain a watch on, Metro Exodus, Battlefield 5 – RTX Game Benchmarks
Dapper Entry Memory benchmarks and requirements
RX 6800 and RX 6800 XT – the Digital Foundry verdict