• 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.

NVIDIA 4000 Series

  • AD102: 608 mm² die, 76.3B transistors
  • AD103: 378.6 mm² die, 45.9B transistors
  • AD104: 294.5 mm² die, 35.8B transistors
videocardz.com

NVIDIA confirms Ada 102/103/104 GPU specs, AD104 has more transistors than GA102 - VideoCardz.com

NVIDIA Ada GPUs have significantly higher ROP count NVIDIA is clarifying the specs for RTX 40 series. The company released full info on the die sizes and transistor counts on AD102, AD103 and AD104 GPUs. All three are to launch in the following weeks. NVIDIA already provided important figures...
videocardz.com videocardz.com
 
kalniel said:
Lucky for you it doesn't do anything of the sort. It generates single frames inbetween two properly rendered frames - smoothing motion but not affecting any interaction. If you were shooting based on 50 frames a second before you're still shooting based on 50 frames a second now, only you also get a visual effect of almost 100 frames per second.
Click to expand...

Not quite, in order to insert this fabricated frame in-between two real frames; the graphics card has to delay showing you the oldest frame thus there is an inherent delay. This is why things like low latency settings exist at a driver level and it's why those same settings prevent a buffer of 2, 3 or 4 frames from being stored before being sent to the display - because those frames cause a delay as they haven't yet been sent to the display while the graphics card is already generating the next batch.

I don't think the general latency on DLSS 3 is going to bad as bad as people are making it out to be, the problems are likely going to come in other areas like HUD elements getting a bit ****** up on the AI generated frames. That being said, there will be extra latency.
 
Bill Turnip said:
With all this new DLSS 3 power, I'd have thought they'd have been pushing 8k gaming a lot more, a la 3090 marketing. Guess they know themselves that it's too much of a stretch? Or has there been mention of it?
Click to expand...

The coke bucket from the mining boom ran dry and now that nVidia execs are on a comedown they realise nobody cares about 8K gaming and have instead gone for a more realistic push for high FPS 4K gaming.
 
Ayahuasca said:
the-rock.gif
Click to expand...

That's the US dollar bending over the pound for a...pound
 
Halliphax said:
The coke bucket from the mining boom ran dry and now that nVidia execs are on a comedown they realise nobody cares about 8K gaming and have instead gone for a more realistic push for high FPS 4K gaming.
Click to expand...

Tell that to digital foundry, they will be doing 8k benchmarks on the 4090
 
Last edited:
Grim5 said:
Tell that to digital foundry, they will be doing 8k benchmarks on the 4090
Click to expand...
The internet sadly forms echo chambers, even ones we all think we are free of, DF will be the death of themselves eventually unless they come back to reality in the now.
 
I was watching some AIB product release videos and the 4090s come with silent mode by default which is 450w and you have to switch to gaming mode for better performance. Pretty incredible how power hungry these cards are
 
Iron Man said:
I was watching some AIB product release videos and the 4090s come with silent mode by default which is 450w and you have to switch to gaming mode for better performance. Pretty incredible how power hungry these cards are
Click to expand...

I'd assume silent mode is just standard (or 4 or 5% better since they'll be better binned chips) FE performance numbers/levels, performance vBIOS mode might be the one that pushes the card to the 600w limit for a big OC.

The recent Cyberpunk figures apparently show the card can go to 2.8Ghz on stock which somehow only ran at 55c but a leak stated the frequency ceiling was 3.4Ghz) - be interesting to see how far these cards overclock.
 
Halliphax said:
Not quite, in order to insert this fabricated frame in-between two real frames; the graphics card has to delay showing you the oldest frame thus there is an inherent delay.
Click to expand...
If you're talking about frame 1 the first time a game loads, sure - that will be held, but no reason that's any different from the usual 3 or so frame buffer that cards do unless you enable reflex/low latency/low-zero prerendered frames.
 
kalniel said:
If you're talking about frame 1 the first time a game loads, sure - that will be held, but no reason that's any different from the usual 3 or so frame buffer that cards do unless you enable reflex/low latency/low-zero prerendered frames.
Click to expand...

DLSS 3 will presumably take two pre-rendered frames and use AI to generate a middle frame between them, that combined with the latency of how long it takes the card to actually generate the frame will combine into the total added latency of DLSS 3 as an interpolation solution. Reflex will reduce the latency but it cannot eliminate it entirely because the graphics card still needs to generate that extra "fake" frame.

This is assuming a DLSS 3 level of Quality, I'm guessing Balanced, Performance and Ultra Performance will generate an extra 2, 3 and 4 frames respectively and further increase the latency.
 
I won't be buying these are current prices, my benchmark is 3070 cost me £370 so the 4070 is worth £400ish to me, when it's that price I'll entertain the idea. There isn't even anything my PC can't run at max settings so maybe if a most-have game comes out but they usually get delayed anyway.
 
You must log in or register to reply here.
Back
Top Bottom