Monday, December 16th 2024
32 GB NVIDIA RTX 5090 To Lead the Charge As 5060 Ti Gets 16 GB Upgrade and 5060 Still Stuck With Last-Gen VRAM Spec
Zotac has apparently prematurely published webpages for the entire NVIDIA GeForce RTX 5000 series GPU line-up that will launch in January 2025. According to the leak, spotted by Videocardz, NVIDIA will launch a total of five RTX 5000 series GPUs next month, including the RTX 5090, 5080, 5070 Ti, 5070, and the China-only 5090D. The premature listing has seemingly been removed by Zotac, but screenshots taken by Videocardz confirm previously leaked details, including what appears to be a 32 GB Blackwell GPU.
It's unclear which GPU will feature 32 GB of VRAM, but it stands to reason that it will be either the 5090 or 5090D. Last time we checked in with the RTX 5070 Ti, leaks suggested it would have but 16 GB of GDDR7 VRAM, and there were murmurings of a 32 GB RTX 5090 back in September. Other leaks from Wccftech suggest that the likes of the RTX 5060 and 5060 Ti will pack 8 GB and 16 GB of GDDR7, respectively. While the 5090's alleged 32 GB frame buffer will likely make it more adept at machine learning and other non-gaming tasks, the VRAM bumps given to other, particularly Ti-spec, RTX 5000 GPUs should make them better suited for the ever-increasing demands from modern PC games.
Sources:
VideoCardz, Wccftech
It's unclear which GPU will feature 32 GB of VRAM, but it stands to reason that it will be either the 5090 or 5090D. Last time we checked in with the RTX 5070 Ti, leaks suggested it would have but 16 GB of GDDR7 VRAM, and there were murmurings of a 32 GB RTX 5090 back in September. Other leaks from Wccftech suggest that the likes of the RTX 5060 and 5060 Ti will pack 8 GB and 16 GB of GDDR7, respectively. While the 5090's alleged 32 GB frame buffer will likely make it more adept at machine learning and other non-gaming tasks, the VRAM bumps given to other, particularly Ti-spec, RTX 5000 GPUs should make them better suited for the ever-increasing demands from modern PC games.
173 Comments on 32 GB NVIDIA RTX 5090 To Lead the Charge As 5060 Ti Gets 16 GB Upgrade and 5060 Still Stuck With Last-Gen VRAM Spec
Good comparison here of 8GB vs 16Gb:
Pretty crazy that we have decade long 8GB low-mid range.
Jensen Huang, September 2022:
"Moore's Law is dead … It's completely over, and so the idea that a chip is going to go down in cost over time, unfortunately, is a story of the past."
Yeah I would prefer my daughter to stay inside rather than outside
If a possible rtx 5050 comes with only 6 gb. That will be DOA at launch. As there are now games out there that litterly refuse to start with less than 8 gb vram. As i have seen if i remember correct the new Indiana Jones games would not start on a 6 gb gpu. Just gave a error message with vram. Saw that in a youtube video a few days ago.
Rtx 5060 should at least have 10 gb and preferly 12 gb.
On the low end, 8 GB is tight, but serviceable. The x60 cards are for 1080p and below gamers. Sure, they might be "good enough" for some basic 1440p gaming, but that's where they get you.
kinda feel like nvidia is expecting the AI gravy train is slowing down being willing to bring this level of complexity to the masses.
Maybe 60 series of cards will have hbm on the consumer level
This AI focus is going to be worse than cryptomadness, and people are still pretending it doesn't affect them.
Shit starts to hit the fan when display drivers can't satisfy caching requirements, then stuttering and fps drops are inevitable.
It has been proven multiple times, especially RTX 4060 Ti 8 GB vs Ti 16 GB, that system with less VRAM delivers less 0.1% and 1% lows fps and also uses more system RAM (about 2-3 GB more).
8GB is definitely not okay even for lower mainstream these days, it was already present in mainstream SKUs in 2016.
Game devs can't move forward with quality of graphics assets when they are still contrained by low VRAM amount.
RTX 3060 is most popular GPU according to Steam hardware survey and I'd say it's not the 12 GB variant, rather 6 GB.
Now with Intel introducing 12 GB in low-end segment (B580), Nvidia should really reconsider going past 8 GB. GDDR6(X) chips are not expensive today, you can get 8 GB for $22.
In lowend and lower mainstream it does not actually matter whether you have GDDR7 or GDDR6. Card will most probably don't have enough performance to fully utilize memory
bandwidth whether the chips have 21 Gbps or 28 Gbps. More important than bandwidth is capacity.
The bigger the capacity, the more things you can fit into cache. The lower the capacity, the more you depend on primary storage device's speed and memory bandwidth.
12 GB + 192-bit or 16 GB + 128-bit should become standard even in low-end or lower-mainstream.
Btw, if you recall, GTX x60 series were never low-end cards, they were lower mainstream cards. Nvidia used to make x50 cards and absolute lowend x30.
Since RTX 2000 that has changed, RTX xx70 eries are that what x60 series were meant to be. RTX 4080 12 GB was luckily renamed to 4070 after rich criticism.
But it goes the same way on AMD side. With HD 6000 series, things changed. HD6850/6870 was not a real successor to HD 5850/5870, the HD 6950/6970 was.
As for RTX 5090, forget about <$2k price tag, expect $2299 or more. We saw that there was demand for RTX 4090 even when it cost above $2k.
If the price is worth it as long as there is 16GB of Vram most likely I might end up with a 5080.. I'm hitting the ceiling of the Vram in some 3D stuff I do, RT then the worst offender has been VR where it crashes SteamVR causing the worst sickness I've ever experienced lol
Also Jensen saying that, its an admission of price can only go up, leather jacket man wouldn't even be slightly generous with VRAM or bandwidth unless you buy a flagship card.
Otherwise read my other posts, nothing you said wasn't already answered multiple times.