• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

RDNA 3 rumours Q3/4 2022

Status
Not open for further replies.
Besty said:
If Avatar is any good, I will eat my Lisa Su portrait.
Click to expand...

I'm expecting it to be the standard run of the mill ubi open world game but being by massive and on the snowdrop engine (division 1 and 2), I'm expecting a somewhat good game, at the very least, it will look great :D
 
Last edited:
timorous said:
In the 3 games without DLSS3 the 4090 was ~1.6x faster than a 3090Ti. The 3090Ti is 1.17x faster than the 6900XT at 4K according to the TPU charts.

That puts the 4090 in raw raster as around 1.87x the 6900XT. If top N31 is 2x faster than it is ahead in raw raster by a margin. If it is 2.5x faster like some rumours state then it will be ahead by 2 performance tiers.

When it comes to RT RDNA2 loses 60-70% performance where as Ampere is closer to 40-50%. If RDNA3 is the same as RDNA2 then with a 2.5x boost in raster it will probably be competitive with the 4090. With an improvement to the relative raster performance to even 50% it is likely to be ahead.

Now rumours may be incorrect and RDNA3 may under perform but 2.5x the shaders with 50% more memory bus width and faster ram combined with the much higher clock speeds vs RDNA2 makes me think 2x performance is going to be the low end outside of CPU limited scenarios.

Really the numbers are there so unless there is a massive IPC and perf/shader regression (I expect some but not like 50%) then it does add up.

I don't think AMD have much to worry about and I really think that N33 is going to be nipping the heels (around 90% performance) of the 4070 (sorry 4080 12GB) which is using AD104.
Click to expand...
No DLSS 3 though :)
 
If it is true that RDNA 3 is getting close to 4ghz then that is not really good news. Once again, less hardware clocked to max from the factory. The least they can do is to sell it cheaper and i mean at most as expensive as RDNA2 was but i don't think there is any chance to see that from AMD.
 
timorous said:
Indeed. All frames will be the real deal rather than fake ones to boost numbers but do nothing for input latency.
Click to expand...

Forget frame prediction, 5000 series will just guess what game you want to play based on cloud algorithms. If it’s the ‘wrong’ game it’s not the fault of the GPU but because you didn’t want to play the right game in the first place.
 
Last edited:
timorous said:
In the 3 games without DLSS3 the 4090 was ~1.6x faster than a 3090Ti. The 3090Ti is 1.17x faster than the 6900XT at 4K according to the TPU charts.

That puts the 4090 in raw raster as around 1.87x the 6900XT. If top N31 is 2x faster than it is ahead in raw raster by a margin. If it is 2.5x faster like some rumours state then it will be ahead by 2 performance tiers.

When it comes to RT RDNA2 loses 60-70% performance where as Ampere is closer to 40-50%. If RDNA3 is the same as RDNA2 then with a 2.5x boost in raster it will probably be competitive with the 4090. With an improvement to the relative raster performance to even 50% it is likely to be ahead.

Now rumours may be incorrect and RDNA3 may under perform but 2.5x the shaders with 50% more memory bus width and faster ram combined with the much higher clock speeds vs RDNA2 makes me think 2x performance is going to be the low end outside of CPU limited scenarios.

Really the numbers are there so unless there is a massive IPC and perf/shader regression (I expect some but not like 50%) then it does add up.

I don't think AMD have much to worry about and I really think that N33 is going to be nipping the heels (around 90% performance) of the 4070 (sorry 4080 12GB) which is using AD104.
Click to expand...

Which charts are you looking at? The 3090Ti in the first 3 games , matches or is slower than a 6950XT
 
Yep, there is no way the 3090ti beats the 6900xt or 6950xt in AC Valhalla, it's well established that game heavily favours AMD GPUs - Im surprised Nvidia included it in their benchmarks since they knew it would make them look bad.

From what I can tell only the 16gb rtx4080 is matching the 6950xt's performance in that game, which is not good for Nvidia but that's still like 20 or 30% faster thanbthe 3090ti
 
Last edited:
Grim5 said:
Yep, there is no way the 3090ti beats the 6900xt or 6950xt in AC Valhalla, it's well established that game heavily favours AMD GPUs - Im surprised Nvidia included it in their benchmarks since they knew it would make them look bad.

From what I can tell only the 16gb rtx4080 is matching the 6950xt's performance in that game, which is not good for Nvidia but that's still like 20 or 30% faster thanbthe 3090ti
Click to expand...
Look at the Valhalla benchmark thread on this forum.
 
Last edited:
Status
Not open for further replies.
Back
Top Bottom