• 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 RTX 50 SERIES - Technical/General Discussion

I've just been looking around the cards (MSI) and noticed they only have ONE 12v pwr connector on the top, split into 3 or 4 for the 5090
I'm surprised there isn't TWO
 
cliffy said:
I think you either got to get used to it or give up on gaming because it certainly seems like fake frames is teh way all manufacturers are going
Click to expand...
Every frame is technically fake, even rastered frames. Hardware Unboxed and DF covered this ages ago but people seem to have forgotten lol.

Anyway, we now come to a time where frame gen latency is even lower than before (with the extra help of the new Reflex 2 tech) so the only downside for framegen at a technical level has been reduced, Ray Reconstruction is now 40% odd more efficient, DLSS upscaling is now 40% odd more efficient by ditching CNN and using a Transport system instead, and both are also higher fidelity. All pros really that all RTX card owners can enjoy.
 
Last edited:
cliffy said:
I think you either got to get used to it or give up on gaming because it certainly seems like fake frames is teh way all manufacturers are going
Click to expand...
Ill just stick to indie games then and releases that isn't to focused on "Tripple A" BS graphics while still looking presentable.
 
Slam said:
As expected, there's a big premium between the 5090 and 5080 models. Do people think that the top product lasts twice as long, maybe because of the extra memory.
Is that how its turning out for the 4090?
Click to expand...
The 24Gb of VRAM on the 4090 sure is nice to have ... haven't bumped up against any of the out-of-memory VRAM issues that some people have experienced eg. Indiana Jones. Looking ahead I wouldn't want any GPU with less than 16Gb VRAM.
 
mrk said:
Every frame is technically fake, even rastered frames. Hardware Unboxed and DF covered this ages ago but people seem to have forgotten lol.

Anyway, we now come to a time where frame gen latency is even lower than before so the only downside for framegen at a technical level has been reduced, Ray Reconstruction is now 40% odd more efficient, DLSS upscaling is now 40% odd more efficient by ditching CNN and using a Transport system instead, and both are also higher fidelity. All pros really that all RTX card owners can enjoy.
Click to expand...


oh there you are

everybody @mrk has RECOVERED !
 
Guru said:
Do you think the Corsair rmx shift 1200 PSU would work with the 5090?
Click to expand...
Corsair have a great guide on their website on what PSU you need
www.corsair.com

Best PSU for RTX 5090

The biggest, baddest GPU is here, and it needs an equally big and bad PSU. Here are the best ones.
www.corsair.com www.corsair.com

TLDR Yes that psu would be fine, 12VHPWR / 12V-2x6 cables are identical and preferred but you can use other adaptors if you dont have 12VHPWR / 12V-2x6
 
arc@css said:
Another model, I'm interested in seeing how it performs, is the 5060Ti 16 GB.
I wonder how much of a boost above the 3070 it will deliver?
Click to expand...

The 5060Ti is the card that could be the most interesting if the price is right. The 3060Ti was such a brilliant card, especially at the FE price point.
 
Slam said:
As expected, there's a big premium between the 5090 and 5080 models. Do people think that the top product lasts twice as long, maybe because of the extra memory.
Is that how its turning out for the 4090?
Click to expand...
I think the question is more, will 16GB be enough for that individuals use case.
If they're running 4k 120hz+ with maxed settings, that game may want to use more than 16GB of VRAM.
Then there are the games that will be released within the next 2-4 years.
Personally I wouldn't expect many games to want more than 16GB for a while yet, since 8-16GB is still the norm, and will continue to be until the average GPU increases capacity beyond that (bit of a chicken and egg situation, especially if you throw consoles into the mix).

Maybe they also want to try some AI/ML for themselves and not just use it for gaming though.
 
Also remembered don't buy into the GDDR7 and texture compression PR claims, this is only usable in games that are updated for it, all other games will still use the same VRAM, so if 16GB isn't enough on 40 series, the same remains on 50 series. The only silver lining is that DLSS FG uses less VRAM now on DLSS4, but that's not exclusive to 50 series either...
 
Last edited:
mrk said:
Also remembered don't buy into the GDDR7 and texture compression PR claims, this is only usable in games that are updated for it, all other games will still use the same VRAM, so if 16GB isn't enough on 40 series, the same remains on 50 series. The only silver lining is that DLSS FG uses less VRAM now on DLSS4, but that's not exclusive to 50 series either...
Click to expand...

well, the 5090 is the only one with over 16Gb

i just don't understand why the 5080 is 16Gb not 20 or 24
 
Capitano said:
I didnt even say I was going to buy one? Im just saying having a 5090 would massively benefit me.

Regarding what I play, I play call of duty and getting high FPS in there is always a benefit. At the same time I like the game to look good - which it can at 4k and high settings.
All this time I've had to choose 1 or the other, but with a new card I can have both
Click to expand...

I'm playing everything except Flightsim 2024 maxed out or high/ultra settings on a 6800xt, 60fps + inc Indy, why is your pc so slow? 5700x3D cpu too.
 
mrk said:
Every frame is technically fake, even rastered frames. Hardware Unboxed and DF covered this ages ago but people seem to have forgotten lol.

Anyway, we now come to a time where frame gen latency is even lower than before (with the extra help of the new Reflex 2 tech) so the only downside for framegen at a technical level has been reduced, Ray Reconstruction is now 40% odd more efficient, DLSS upscaling is now 40% odd more efficient by ditching CNN and using a Transport system instead, and both are also higher fidelity. All pros really that all RTX card owners can enjoy.
Click to expand...
Mean all games are fake if you want to get technical.

But if a frame does not register an input because that's not what the game generated then it's fake and contributing to lower or literally fake response or non existent in that frame then you aren't getting more performance.

I'm not playing a movie where the game is played for me, I'm playing the game so input matters.
Again if Disney dreamlight valley is your thing then that's fine, I'm not judging taste but I play games where input registry matters.
 
mrk said:
Every frame is technically fake, even rastered frames. Hardware Unboxed and DF covered this ages ago but people seem to have forgotten lol.
Click to expand...



Depends how you look at it. Are there fake elements in raster games? Sure, fake lights, fake reflections etc it's been around for decades right. However what is real is that the GPU render pipeline has been rendering every single one of those pixels and your CPU is accepting mouse inputs on every single one of those frames. The more frames your generate, the more times per second your CPU is accepting inputs and therefore your latency and lag is reduced.

With frame generation, most frames and pixels are not rendered through the pipeline, they are simply displayed. This means your CPU is not accepting mouse inputs when these fake frames are displayed - this increases latency and lag for the user and the more fake frames you display, the more latency and lag.
 
Last edited:
mrk said:
Every frame is technically fake, even rastered frames. Hardware Unboxed and DF covered this ages ago but people seem to have forgotten lol.
Click to expand...

nOMpSBd.jpeg
 
ShiWarrior said:
well, the 5090 is the only one with over 16Gb

i just don't understand why the 5080 is 16Gb not 20 or 24
Click to expand...

Because of the 256-bit bus.

The bus dictates the number of memory modules. Each module has a 32-bit interface, so 256/32 = 8 modules. With 2GB modules that gives you 16GB.

When 3GB modules come on stream a 24GB model would be technically possible, maybe for the 5080 SUPER.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom