NVIDIA 5000 SERIES

[Absolutely Clueless]

Wouldn't latency still be pretty good, with a base of 120 fps going to 480?

It's awful. No matter my base FPS, engaging DLSS3 frame gen on my 4090 is like using a mouse routed connected to my pc via a satalitte. It is the worst of all Nvidia's DLSS/RTX/software features.

 

[[timko]]

That Asus tax is seriously getting out of hand lol

They are (want to be) the Apple of PC components with their ROG Strix line. I mentioned elsewhere last week that they would be the company to get closer to the 3k line than any other and it's currently holding true.

 

[muon]

I’m not sure I’ve seen this clarified: will the 40 series cards also be getting the new version of frame gen albeit locked to x2? Or will they be getting DLSS4 upscaling only with the older version of frame gen?

The only thing they won't be getting is frame gen greater than 2x. Although note frame at 2x might be better on 50 series cards, as it is different.

They will otherwise get the better standard DLSS without frame gen.

 

[LeBeast]

I’m not sure I’ve seen this clarified: will the 40 series cards also be getting the new version of frame gen albeit locked to x2? Or will they be getting DLSS4 upscaling only with the older version of frame gen?

40 series will get the more efficient frame gen, but still only up to 2X.

 

[markymoo]

Undervolting will be your friend this generation

This review tested undervolting in one game, so more needed please but results are quite decent

RTX5090 default power profile: 133fps and 575w used

Manual undervolt 0.970v: 133fps, 490w used

Manual undervolt 0.900v: 125fps 410w used

Manual undervolt 0.875v: 115fps 350w used


So at least in this one game, you can cut power consumption by 85watts and lose no performance and if you're happy to lose 5% performance you can cut power by 165watts. But remember with the 0.9v undervolt, the performance lost now means the 5090 is only 25% faster than the 4090

I wonder if the A.I can get to a point to auto adjust volts.
If the machine learning can work out frames so quickly, maybe it's possible that it can drop the volts when needed.
Maybe even if they machine learn together and create an online data base for what games require what power at certain stages in the game, so that it avoids any stutter from quick changes.

For instance I game 4k and 120fps is my max.
Maybe on those games if it can generate more, it can just lower the volts to save on energy, heat and the lifetime of the components.
Also Auto frame generation whereby if you max a set power requirement the card can dynamically set if Grame generation is required and by how much in order for best quality and energy consumption based on the users fps requirements.

 

[P.B]

Wait until you see the price for Asus' ROG Astral OC at 2.8k from another reddit post https://old.reddit.com/r/nvidia/comments/1i7ojvo/rtx_50805090_aib_prices_leaked_via_pc_partpicker/

RTX 5090

• Asus ROG Astral OC: $2799.99
• MSI GAMING TRIO OC: $2349.99
• MSI SUPRIM LIQUID SOC: $2499.99
• MSI SUPRIM SOC: $2399.99
• Asus TUF GAMING: $2449.99
• MSI VENTUS 3X OC: $2199.99
• MSI VANGUARD SOC LAUNCH EDITION: $2379.99
• MSI VANGUARD SOC: $2379.99

RTX 5080:

• Asus TUF GAMING OC: $1699.99
• Asus ROG Astral OC: $1899.99
• MSI GAMING TRIO OC: $1199.99
• MSI Inspire 3X OC (Gold) - $1169.99
• MSI GAMING TRIO OC (white): $1199.99
• Asus PRIME: $1399.99
• Asus PRIME OC: $1499.99
• MSI VENTUS 3X OC PLUS: $1139.99
• Gigabyte GAMING OC: $1199.99
• Gigabyte WINDFORCE SFF: $1369.99
• MSI SUPRIM SOC: $1249.99
• MSI SUPRIM LIQUID SOC: $1299.99
• MSI VANGUARD SOC: $1229.99

Ay, no more strix ?

 

[Sparky]

That Asus tax is seriously getting out of hand lol

Be cheaper to fly to the States and pick one up

 

[Absolutely Clueless]

They are (want to be) the Apple of PC components with their ROG Strix line. I mentioned elsewhere last week that they would be the company to get closer to the 3k line than any other and it's currently holding true.

They need to work on their customer support and warranty if they aim to be like Apple. They've got a long long long way to go.

 

[LeBeast]

They are (want to be) the Apple of PC components with their ROG Strix line. I mentioned elsewhere last week that they would be the company to get closer to the 3k line than any other and it's currently holding true.

The sad thing is that the 4090 Strix was pretty **** in terms of coil whine. There are lots of complaints online about that. Even the poor man's 4090, Zotac, did much better in that regard.

 

[Nitefly]

Ay, no more strix ?

The ‘Astral’ is effectively the new Strix.

There is a Strix but it’s on the lower tier cards and is no longer ‘fully metal’.

 

[silvagti]

DLSS 4.0 will work for previous gen but the multi FFF (fake frame feature) is only for the 50 series only

Thanks for the info. Found this grid for anyone wondering which technologies work on which cards.

NVIDIA DLSS 4 Introduces Multi Frame Generation & Enhancements For All DLSS Technologies

75 DLSS Multi Frame Generation games and apps available on day 0; graphics industry’s first, real-time transformer model enhances image quality for DLSS Ray Reconstruction, Super Resolution, and DLAA.

www.nvidia.com

 

[Nitefly]

It's awful. No matter my base FPS, engaging DLSS3 frame gen on my 4090 is like using a mouse routed connected to my pc via a satalitte. It is the worst of all Nvidia's DLSS/RTX/software features.

Have you watched the vids posted on this? It now works differently and there is not the same sort of latency increases.

 

[Gerard]

Lol, if that astral price is legit the asus really has lost the plot. It seems every new series of gpus they're just pushing and pushing to win the pisstake price for pricing. They won it last time with the 3 grand 4090 with gimmicky water cooler.

 

[BS Dave]

I’m not sure I’ve seen this clarified: will the 40 series cards also be getting the new version of frame gen albeit locked to x2? Or will they be getting DLSS4 upscaling only with the older version of frame gen?

New tensor one. The NVIDIA dude in the DF video confirmed it. They're just chucking out the old one.

 

[Nitefly]

New tensor one. The NVIDIA dude in the DF video confirmed it. They're just chucking out the old one.

Ah, good to know.

FYI @Absolutely Clueless - could be worth giving the new version a whirl if you have a high refresh monitor!

 

[parisv]

frame gen looking good on the 5090. It’s a shame you need the higher base to start as this feature would be great for 5070 buyers.

 

[CAT-THE-FIFTH]

The most important question in the thread is - has @mrk bought one? Do you know @TNA ?

 

[LeBeast]

I'm currently using a 240HZ monitor. So for 3X, it'd be limited to 80 fps base framerate and at 4x, it'd be 60 fps base.

I'm wondering how that'd compare to the 4090 doing 2x but with a higher base framerate like 90-100.

 

[HRL]

I wonder if the A.I can get to a point to auto adjust volts.
If the machine learning can work out frames so quickly, maybe it's possible that it can drop the volts when needed.
Maybe even if they machine learn together and create an online data base for what games require what power at certain stages in the game, so that it avoids any stutter from quick changes.

For instance I game 4k and 120fps is my max.
Maybe on those games if it can generate more, it can just lower the volts to save on energy, heat and the lifetime of the components.
Also Auto frame generation whereby if you max a set power requirement the card can dynamically set if Grame generation is required and by how much in order for best quality and energy consumption based on the users fps requirements.

Surely Vsync would do just that? Can’t imagine the additional latency would be any worse than relying on AI to do what you described instead.

 

[Nitefly]

Ah yes I should have gone to spec-savers, it is pretty clear the 40 series cards are getting the new improved version of frame gen: