AMD Radeon RX 6000 Specifications Revealed / Leaked from 2 Separate Locations: Up to 5120-Cores & 2.5GHz Clock Speeds

[DildorTheDecent]

31 minutes ago, HenrySalayne said:

It's not too far off. The 3080's base clock is 1.44 GHz, but some cards are easily boosting beyond 1.9 GHz.

(RTX 3080)

 

Nearly 300MHz difference between the factory base and boost clock is perfectly normal. Some AIBs are bigger and some are smaller. Anything over the factory boost clock or the guaranteed boost clock (such as the 1900MHz boost clock seen by some 3080s) is the result of GPU Boost 4.0 or whatever it is now.

 

There's no way these NAVI cards have a 1000MHz difference between the factory base and factory boost clock. That 2500MHz clock has to be the maximum adjustment range reported by the driver. And the 1500MHz must be an ES card.

[Maticks]

58 minutes ago, Energycore said:

256-bit GDDR6 looks a bit scarce compared to Nvidia's 320-bit GDDR6X.

 

I'm gonna call (X) Doubt on that spec.

Also gddr6x is power hungry it's an odd choice for the 3080.

[RejZoR]

1 hour ago, Chris Pratt said:

The big sticking point for me currently is Nvidia's DLSS. AMD thus far has nothing to match it, and it frankly gives Nvidia a huge edge. Granted, not every game supports it, but more and more will over time. Even if AMD somehow came out with GPUs more powerful than Nvidia's, you'd still get better perf out of Nvidia's simply because of DLSS. I'm not sure AMD can even compete with that, since they don't have any AI chops.

When NVIDIA manages to make DLSS game agnostic, then it's really going to hurt AMD which will have to match them with raw horsepower unless they get an answer to DLSS of their own somehow.

 

The other problem I have ditching NVIDIA is "Fast V-Sync". I don't have FreeSync or G-Sync monitor and this mode gives me by far the best tear free performance. AMD has "Enhanced Sync",but as far as I know it works like NVIDIA's Adaptive V-Sync and there is some possibility for slight tearing in some rare cases where Fast never has it. I generally hate a lot of stupid NVIDIA things, especially retarded NVCP, but Fast V-Sync is one of those that I do really like. I'm not sure I can ditch this even if Radeon is much faster.

[leadeater]

5700 XT vs 6700 XT doesn't look all that great on paper, hopefully it's a bit better for cheaper but still odd it's drop VRAM and thus bandwidth too.

[S w a t s o n]

If the 80cu die can OC to 2500MHz on air/water I'm in

[RejZoR]

1 minute ago, S w a t s o n said:

If the 80cu die can OC to 2500MHz on air/water I'm in

Older chips were made for wide design with low clock. RDNA seems to be focused on really high clocks and narrower approach. Sort of like NVIDIA has been doing since Maxwell 2. And it's paying off for AMD. First RDNA cards have gone in the right direction and I'm really curious how they'll advance it further with RDNA2. I'm also wondering what kind of approach AMD is using for ray tracing. They clearly have a functional method since it's being advertised for consoles. Why does end of October have to be so far away...

[S w a t s o n]

4 minutes ago, RejZoR said:

I'm also wondering what kind of approach AMD is using for ray tracing.

The Xbox team basically explained RDNA2 raytracing at Hot Chips this year, watch the presentation

[Briggsy]

1 hour ago, leadeater said:

5700 XT vs 6700 XT doesn't look all that great on paper, hopefully it's a bit better for cheaper but still odd it's drop VRAM and thus bandwidth too.

The Road to PS5 presentation from March given by Mark Cerny is fairly interesting. His remarks about certain features from Sony and AMD's collaboration ending up in the Desktop GPU's are intriguing (around the 25:42 mark), as one of the apparent features is large cache on-die. Whether that makes its way into the Desktop cards is just rumour at this point, but it would explain why AMD don't need a larger bus width than 256bit on the top card.

 

 

[GatioH]

It’ll be impressive to see if the 6700XT is cheaper and faster than the 5700XT at only 150w

 

pls no more space-heater wars

[cj09beira]

2 hours ago, DildorTheDecent said:

Boost clocks can't be real. There's no way a 6700 XT has a 1500MHz base and 2500MHz boost clock. Guaranteed base and boost clocks are not that far apart. Maybe it's base and max boost assuming power, voltage and temperature are in check but certainly not the numbers they'll put on the box. Or it could be the max boost that the driver or program reports.

 

Even the 6900 XT has a 700MHz difference. No way, Jose.

that could have been the case except the same value for the navi 21 is 2050 or 2200, which are far too low for max allowed clocks, and considering that the ps5 will run at 2.23Ghz, its not that much of a stretch, still its some really high clocks, my guess here is that navi 2x had some sort of bug at high clocks (the ps5 dude talked about it) so they must have fixed it by the time the made navi 22.

its high clocks also kinda of explain the navi 23 which usually would not have nearly the same amount of Cus

[RejZoR]

42 minutes ago, S w a t s o n said:

The Xbox team basically explained RDNA2 raytracing at Hot Chips this year, watch the presentation

Hot Chips? All I get searching for it is hot chips challenges on youtube... With literal spicy chips...

[BabaGanuche]

5 minutes ago, RejZoR said:

Hot Chips? All I get searching for it is hot chips challenges on youtube... With literal spicy chips...

Hear is an article about the event.

 

https://www.anandtech.com/show/15994/hot-chips-2020-live-blog-microsoft-xbox-series-x-system-architecture-600pm-pt

[Belgarathian]

I guess the boost clocks aren't impossible... RDNA 2 PS5 clocks are 2.23GHz, so it's conceivable they might be able to push them higher.

[Trik'Stari]

Please beat the pants off a 1080ti so I can finally kick Nvidia.

[Energycore]

1 hour ago, GatioH said:

pls no more space-heater wars

Tell Nvidia that

 

Though I still don't fully believe these specs. Need more people to confirm them

[Energycore]

1 minute ago, Trik'Stari said:

Please beat the pants off a 1080ti so I can finally kick Nvidia.

*just bought a used 1080 ti LOL

 

(but it's more than I'll need for a while cuz I don't do AAA games, I just wanted it for 4K Skyrim-level of detail games)

[Trik'Stari]

19 minutes ago, Energycore said:

*just bought a used 1080 ti LOL

Don't get me wrong. It's a monster of a card. Especially water cooled.

 

Quote

(but it's more than I'll need for a while cuz I don't do AAA games, I just wanted it for 4K Skyrim-level of detail games)

If you mod Skyrim or Fallout 4, you will not be disappointed. It can handle a lot. Mind the god rays though.

[BiG StroOnZ]

Was bored, so I decided to play with the colors/colours of the supposed design of the upcoming Radeon RX 6000-series card's cooler. Came out pretty good if I say so myself: 

 

 

 

 

 

 

If this actually ended up happening (even with only these colors as alternatives to the original/standard Red/Scarlet option), I really think AMD would give NVIDIA a run for their money on reference cooler design. 

 

Of course this wouldn't be complete without proper related song referencing:

 

 

[AldiPrayogi]

Please be good AMD so that the price would drop down

[AlTech]

8 hours ago, DildorTheDecent said:

Boost clocks can't be real. There's no way a 6700 XT has a 1500MHz base and 2500MHz boost clock. Guaranteed base and boost clocks are not that far apart. Maybe it's base and max boost assuming power, voltage and temperature are in check but certainly not the numbers they'll put on the box. Or it could be the max boost that the driver or program reports.

 

Even the 6900 XT has a 700MHz difference. No way, Jose.

I have serious doubts about the 2nd table shown with the base clocks.

 

8 hours ago, Energycore said:

256-bit GDDR6 looks a bit scarce compared to Nvidia's 320-bit GDDR6X.

 

I'm gonna call (X) Doubt on that spec.

I would tend to agree but MLID has commented on G6 vs G6X

 

7 hours ago, BiG StroOnZ said:

snip

Well, AMD wasn't lying about the first 50% part from GCN to RDNA1 so why wouldn't they also be telling the truth about RDNA2?

 

A 60 CU Vega 2.0 card performs the same (ish) as a 40CU Navi 1.0 card. If we're right then that should perform like a 27 CU RDNA2 card. And hey presto, wouldn't you know it but PS5 has a 36CU 2.2GHz GPU. Giving us a bit more than 2080 perf. Presumably around 10-15% faster as optimistic even though 25% may not be out of the question in one or two titles.

 

What's 20-30% faster than a 2080? A 2080TI.

 

What else is gonna target 2080Ti performance? RTX 3070. So really I hope AMD will need only 40-44CUs to beat it but if they don't then I'm sure they could brute force it with 56 or 60.

[Patrik_]

8 hours ago, leadeater said:

5700 XT vs 6700 XT doesn't look all that great on paper, hopefully it's a bit better for cheaper but still odd it's drop VRAM and thus bandwidth too.

Yeah I dont get it, why would you downgrade ram on especialy now. I hope this is gonna be 170-210USd card to replace 5500XT because If its like 300 than Im kinda mad at AMD....  

[Patrik_]

I have really bad feeling that Nvidia will destroy AMD by having so many options 3060,3070,3070ti, 3080 10gb,3080 20gb and 3090. I know that TI cards are rumors but we know that Nvidia doesnt drop prices they just replace card or just make other price point. 

[LAwLz]

1 hour ago, AluminiumTech said:

I would tend to agree but MLID has commented on G6 vs G6X

MLID has pretty much never been correct so why does it matter what he says?

He is about as accurate as WCCFTech in his "leaks" and predictions.

 

You just have to go and look at his Ampere "leak" video to see how much things he got wrong. Even if you buy into the whole conspiarcy theory that Nvidia deliberately sent out incorrect info and MLID's insider source accidentally leaked this incorrect info, that still means MLID's sources have a terrible track record and therefore we can't trust MLID either.

[suicidalfranco]

hope it's not an over 90°C card when running the latest games. Cause if i wanted a jet engine in my room i'd get a console

[CephDigital]

2.5Ghz isn't that far fetched considering the PS5's GPU reaches 2.23Ghz and most likely that's power and thermal limited. I bet if you give it enough power and a decent cooler, it should hit it. The base clocks look wrong to me though.