Question Speculation: RDNA2 + CDNA Architectures thread

Page 227 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

uzzi38

Platinum Member
Oct 16, 2019
2,595
5,767
146
All die sizes are within 5mm^2. The poster here has been right on some things in the past afaik, and to his credit was the first to saying 505mm^2 for Navi21, which other people have backed up. Even still though, take the following with a pich of salt.

Navi21 - 505mm^2

Navi22 - 340mm^2

Navi23 - 240mm^2

Source is the following post: https://www.ptt.cc/bbs/PC_Shopping/M.1588075782.A.C1E.html
 

GodisanAtheist

Diamond Member
Nov 16, 2006
6,767
7,090
136
I think it's more related to the 6 nm refresh, which is looking more likely they are doing a full refresh. Mining specific SKUs wouldn't surprise me though.

-Indeed, moving to 6nm would allow AMD to produce more dies per wafer and they would not have to compete with 5nm supply with their CPUs and potentially console SOCs.

Given AMD (and NV) still have not even released their full current gen line-up, it doesn't look like anyone is in a big hurry to move on to next gen archs at the moment for any number of different reasons.
 

jpiniero

Lifer
Oct 1, 2010
14,562
5,195
136
-Indeed, moving to 6nm would allow AMD to produce more dies per wafer and they would not have to compete with 5nm supply with their CPUs and potentially console SOCs.

Given AMD (and NV) still have not even released their full current gen line-up, it doesn't look like anyone is in a big hurry to move on to next gen archs at the moment for any number of different reasons.

I would be surprised for AMD to make a further lower tier card for desktop. I think this is it, and what the 6 nm would be (which would make sense to me) is largely the same specs with some clock/memory/price (heh) changes.
 

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
They will as long as people still buy them...

A 6nm refresh of the current cards to replace the 700/600/500 level SKUs seems like a no-brainer. ITs already rumored that the 6800 chip is being bumped down to the 700, and the rest down as well. Might as well combine the two.
 

eek2121

Platinum Member
Aug 2, 2005
2,929
4,000
136
Yeah, but is it going to cost less than $280?

Part of the problem is that even if it can't mine, the market is so distorted that this will wind up getting scalped as well.

Do chickens have wings? NOTE that I can get a 6500xt for cheaper than $280 right now. The cheapest public price that I've found is $260. The cheapest bulk (non-reseller) price is $189.96.
 
  • Like
Reactions: scineram

maddie

Diamond Member
Jul 18, 2010
4,738
4,666
136
Do chickens have wings? NOTE that I can get a 6500xt for cheaper than $280 right now. The cheapest public price that I've found is $260. The cheapest bulk (non-reseller) price is $189.96.
Curious. What & where is the min order for bulk and why (non-reseller) description? Is reseller even cheaper?
 

DisEnchantment

Golden Member
Mar 3, 2017
1,599
5,765
136

Kaluan

Senior member
Jan 4, 2022
500
1,071
96
Some strange but maybe good performance uplift results on RDNA2 cards with the newest 23.2.1 driver in ray-tracing synthetics (and allegedly DOOM Eternal w/ RT on too), 3DMark RT Feature Test score (framerate) gone up by up to 40%, 6900XT getting close to 3080 in DXR pure RT:



Requires further investigation IMO
The only pure RT related driver performance boost claim AMD made is for Quake II RTX
So there might be some truth to it.
Also, unsure if any of this applies to RDNA3

Edit: What both of those games (Quake and DOOM) have in common is that they're running of Vulkan API. But then again, 3DMark's DXR test obviously doesn't. Really strange.

Edit2: Just checked and except for Borderlands 3 (and Hogwarts Legacy?), all performance improvement claims are made for games with various ray-tracing settings on (& at 4K, so unlikely to be a driver overhead thing). Which makes me lean towards this driver being legit about a RT optimization overhaul.

Big if true. RDNA2 FineWine(tm) still going it seems.
But it also ticks me off that they're silent on improvements for RDNA3.
 
Last edited: