Where Gaming Begins
I may be extremely late but the time is finally upon us, AMD have officially announced their range of RDNA 2 based graphics cards and boy are they something special.
The reveal showcases three individual GPUs in the RX “6000” series, those being the RX 6800, RX 6800XT and the RX 6900XT to tackle NVIDIA’s Ampere and would you know it they’re mighty successful at that task.
Based on the new RDNA 2 architecture or more specifically “Navi 2X” carries with it substantial architectural changes and improvements such as a serious uplift of both efficiency and performance while being on the same processing node as RDNA 1.
NVIDIA GPU’s for the past two generations now have offered a measly ~30% uplift in performance give or take between generations, this includes coming from Pascal to Turing and Turing to Ampere however the in the case of the most recent addition NVIDIA can’t seem to get any traction due to woeful production, on top of that is the utilization of the horrendous Samsung 8nm processing fabrication which is absolutely fucking horrible.
Turing to Ampere may only provide yet another measly 30% effective gaming performance gains but it does so at the cost its power efficiency of which Ampere is much worse than the older Turing cards based on TSMC’s 16nm + or simply “12nm”.
However in the case of RDNA 2, AMD have pulled off what seems like a miracle as the performance per watt uplift while using the exact same TSMC 7nm processing node from RDNA 1 to RDNA 2 is a staggering 54%, and that’s not all. AMD have actually managed to DOUBLE the performance of the 5700 XT with Navi 21. Truly marking RDNA 2 as a sensational architectural improvement over the last, certainly something NVIDIA wishes they could accomplish.
Along with the serious uplift in efficiency while utilizing the same processing node is the inclusion of the “Infinity Cache” technology. Yes! It’s real and it certainly makes its impact known.
Navi 21 based GPUs offer an incredible 128 MB of on-die cache which aids to free up unnecessary memory bandwidth and helps boost the effective bandwidth of the otherwise thought rather plain 16Gbps GDDR6 memory on a 256-bit memory interface to a staggering 1664 GB/s effectively.
It is also one of the more substantial factors as to how AMD have managed to accomplish the vast performance per watt improvement with RDNA 2 as the cache aids in total effective bandwidth with less effective power requirements of faster memory modules or a wider memory bus.
Lastly one of the new features with RDNA 2 that we’ll be hearing a lot more about in just a moment and months well after the release of these new graphics cards.
SAM (Smart Access Memory)
AMD Smart Access Memory allows for your Ryzen 5000 series processor to make full use of the onboard GDDR6 memory from your RX 6000 series GPU.
This is rather ingenious feature considering just how godawful the Windows kernel actually is at utilizing hardware, where the processor would generally have access to only a small portion of the VRAM of your graphics card however with SAM the processor can utilize all of it.
Having Smart Access Memory enabled typically provides over 6% better performance in various titles on average or in the best case up to a staggering 11%
However this feature does come with quite a serious catch, you may only enable Smart Access Memory if you’re obviously using an AMD RX 6000 series graphics card for starters but it also requires the usage of AMD’s brand new Ryzen 5000 series processors along with a 500 series motherboard such as the B550 or X570. Thus making the combination of Zen 3 and RDNA 2 a deadly one at that.
Now finally we can get into the good stuff, and that being the Radeon RX 6000 series.
Radeon RX 6800
$579 Launching Nov 18th
3840 Stream Processors – 16GB GDDR6 Memory
Faster than the RTX 2080Ti / 3070
Starting things off small with the RX 6800 which features a streamlined cut of the Navi 21 core only packing 60 CUs (Compute Units) totalling 3840 stream processors. However much like its bigger brothers that we’re about to see it too features 128MB of Infinity Cache along with 16GB of GDDR6 memory marking the RX 6800 as having no compromises despite being a cut down design.
With a TDP value of 250W it offers a game clockspeed of just 1815MHz and an effective boost of 2105MHz which is a little low in comparison to the higher end SKUs but I digress.
Performance of the RX 6800 shows it being considerably faster than NVIDIA’s 2080Ti at 1440p resolution which in and of itself is faster than the RTX 3070 in “games” that aren’t synthetics or rendering. It also however reigns superior when bumping the resolution to 4K.
However the tests conducted with the RX 6800 are with AMD’s Smart Access Memory enabled which will generally give the 6800 a performance boost of around 5% meaning the performance will vary depending on your system specifications if you’re planning on switching to AMD’s new Ryzen 5000 series combined with a 500-series motherboard.’
The Radeon RX 6800 is scheduled for launch starting at $579 on November 18th which I’ll admit is a little higher than the RTX 3070 being $80 more expensive however regarding the performance advantage the 6800 holds over the 2080Ti and 3070 the Radeon RX 6800 contains over double the VRAM capacity while most certainly providing a more efficient GPU despite the differences in advertised TDP values.
Considering the production woes for NVIDIA combined with hardly any store managing to claim sufficient stock, it’s unclear how the launch for RDNA 2 will be in comparison to Ampere however I can guarantee that we won’t see paper launch levels remotely on par with NVIDIA.
And unlike any Ampere based GPU ever you’ll most certainly be able to overclock the RX 6800 considerably well providing even superior performance than the charts would otherwise show.
Radeon RX 6800 XT
$649 Launching Nov 18th
4608 Stream Processors – Better than RTX 3080
The second card revealed during the event is the RX 6800 XT which offers much of the same 128MB of Infinity Cache and 16GB of GDDR6 memory however the main difference here is that AMD have finally broken the curse of the GCN architecture as the 6800 XT provides a total of 72 CUs totaling 4608 stream processors.
The RX 6800 XT has a TDP of 300W and game and boost frequencies of 2015MHz and 2250MHz and is scheduled for launch alongside the RX 6800 on November 18th for the price of $649 which undercuts the non-existent RTX 3080 by $50 or in reality probably around $200 considering the markups on those Ampere graphics cards.
The performance of the RX 6800 XT wasn’t conducted with either AMD’s Smart Access Memory RAGE mode which essentially is a higher power target allowing for possibly higher but certainly longer lasting and a more stable boost frequency. With this being completely indecisive out of box performance of the 6800 XT in comparison with NVIDIA’s own RTX 3080.
And as you’re clearly able to see the Radeon RX 6800 XT offers superior performance than the RTX 3080 at 1440p resolution and sways a tad at 4K resolution offering either superior or parity performance without even enabling Smart Access Memory.
When push comes to shove the 6800 XT is clearly the must superior option given that combined with AMD Ryzen 5000 series processors the performance gap widens anywhere from 5-11% and that doesn’t even account for if either card were overclocked in the hands of the consumer.
Because as you’re more than well aware RTX Ampere can’t overclock for shit, and with reports circulating from my own sources at AMD’s most respected exclusive AIB partner these GPUs can hold upwards of 2300MHz with relative ease.
OC for OC and the 6800 XT simply walks away from the RTX 3080, let alone enabling Smart Access Memory.
The RTX 3080 has been made completely redundant.
Radeon RX 6900 XT
$999 Launching December 8th
Tackling Giants – Performance Parity With RTX 3090
Here it is, the biggest baddest Navi of them all, the RX 6900 XT. Offering a full 80CUs equally 5120 Stream Processors, it comes with the same 128MB of Infinity Cache and 16GB of GDDR6 memory as the former two cards as well.
The launch for the RX 6900 XT will arrive a few weeks later than the other two Radeons arriving on December 8th for $999 which is substantially less than what this GPU is competing against, the RTX 3090.
The 6900XT carries most of the weight when it comes to efficiency offering a whopping 65% superior performance per watt comparative to RDNA 1 coming in with the same TDP as the 5800XT at 300W with the same game and boost frequencies. This is most certainly coming from more desired binned chips that make up the 6900 XT.
Now the performance of the RX 6900 XT is certainly dominate however the testing is conducted with the usage of both RAGE mode and Smart Access Memory being enabled being the best case scenario to the 6900 XT. And the RX 6900 XT certainly doesn’t disappoint offering superior performance or outright performance parity with NVIDIA’s $1499 offering.
For comparison sake especially when taking into account the price hike of NVIDIA Ampere GPUs you’re able to buy not only the 6900 XT but also a Ryzen 9 5950X for the same price as what can be found for the 3090 currently. Hell, drop the 5950X and instead opt for the 5900X and get yourself a free 500 series motherboard. Now that’s what I call value.
Once again if we take individual overclocking into account since Ampere has zero headroom I have no doubt in my mind that the 6900 XT will hold the performance crown over the RTX 3090 with a small margin but in convincing fashion.
AMD’s own performance data analyzed
This is even backed up by AMD themselves publishing actual framerate numbers of the games they’ve tested at 1440p and 4K resolutions on their website which can be found here.
Now AMD’s own website requires an individual to select the particular game at either 1440p or 4K resolution of which you’re only able to see the actual FPS figures if you hover over the specific card with your cursor.
Luckily for us Twitter user harukaze5719 has scraped the data off AMD’s website and provided some quick and dirty performance values and the average performance over the ten games tested.
It’s well obvious from the charts than AMD provides exceptional 4K and 1440p performance with the new RX 6000 series GPUs either eclipsing the competition or offering objectively superior hardware for a lower price point.
When averaging out the performance of all ten games showcases the superiority of the 6900 XT being 3.54% faster than the RTX 3090 which costs $500 more, or in actual retail figures it’s hard to find a RTX 3090 for less than $1800 in stock.
Comparing the 4K performance of the RDNA 2, the Radeon RX 6800 XT is only 4.54% behind the pace set by NVIDIA’s 3090 and keep in mind the performance testing for the 6800 XT was conducted without having AMD Smart Access Memory enabled which would otherwise bridge the gap to an even smaller figure if not actually beating the 3090 depending on the title.
Not only that but the 6800 XT is also 4.75% faster than the RTX 3080, as previously mentioned without having Smart Access Memory enabled which would otherwise increase the lead against the RTX 3080 favoring AMD convincingly.
The base RX 6800 shows to being 15.14% faster than the previous generation RTX 2080Ti which in and of itself proves to be superior in actual games than the RTX 3070 which recently just “launched” for $499. Greater than 15% performance benefit with double the memory capacity for just $80 more, if only NVIDIA were able to produce and supply the RTX 3070 in great enough quantities to stave off any potential price hikes but I seriously fucking doubt that. You’re better off simply buying the RX 6800 which would probably launch to be cheaper.
Bumping the resolution down to 1440p where these new cards seem to excel the most on the 6900XT actually extends its performance lead over the RTX 3090 to 6.36%, and even the 6800 XT gets in on the action with a 0.74% advantage over the 3090 as well. For $649 the Radeon RX 6800 XT seems to be a stellar option.
The RTX 2080Ti and 3070 for that matter fall short to being 22% behind that of the RX 6800 which in and of itself is only 6.12% slower than the RTX 3080.
AMD’s new Radeon RX 6000 series ultimately renders NVIDIA’s RTX Ampere obsolete. Both in performance which AMD now has a stronghold on in both the CPU and GPU markets but also for a better value too.
10GB isn’t sufficient enough for 4K gaming right now or for the next generation of games. All Navi 21 based graphics cards offer 16GB of VRAM on board, clearly AMD has gotten the memo that games actually evolved so why hadn’t NVIDIA?
Because NVIDIA is a company so stuck up its own arse they’ve prioritized profit margins over performance, this is proven by the failure to Jew out TSMC for 7nm orders and as a result they’ve made the switch to Samsung’s garbage 8nm mobile processing node.
More VRAM, better performance and better efficiency for much better value. There’s absolutely no reason to buy Ampere anymore, RDNA 2 is a game changer.
This is where gaming truly begins.
Mama said knock you out.