Speculation and Rumors: Nvidia Blackwell ...

  • Thread starter Deleted member 2197
  • Start date
I'll definitely wait at least until an 18GB 5070 Super or something. Not getting another 12GB card. It's already been a bit of a problem in a couple games I've played.

Agreed. Frame Gen + RT can easily take you over the limit in a few games. I personally wouldn't even be happy with 16GB on my next GPU which is why the rumoured 16GB 5080 probably won't be of much interest. The rumoured 24GB version would be the sweet spot for me depending on price. Anything over £1k is a no go.
 
Jensen Huang will deliver the CES keynote: https://www.ces.tech/press-releases/nvidia-ceo-jensen-huang-to-deliver-ces-2025-keynote/

Cant remember the last time he has done a keynote at CES. So maybe we will see the consumer Blackwell products.
If they’re going to launch so late I hope it at least means it won’t be a paper launch.
Agreed. Frame Gen + RT can easily take you over the limit in a few games. I personally wouldn't even be happy with 16GB on my next GPU which is why the rumoured 16GB 5080 probably won't be of much interest. The rumoured 24GB version would be the sweet spot for me depending on price. Anything over £1k is a no go.
I think the odds a 24GB 5080 is not over £1k are almost zero. Maybe the 16GB model. Hope I’m wrong though.
 
We have just received word from our sources that the NVIDIA GeForce RTX 5080 GPU which will be utilizing the GB203 GPU core will come equipped with 16 GB of GDDR7 memory which will operate at 32 Gbps speeds. This is 4 Gbps faster than the 28 Gbps dies featured on the GeForce RTX 5090. The graphics card will feature a 256-bit wide bus interface and as such, it should feature 1 TB/s of total bandwidth from the memory alone.

This is going to be a major leap over the RTX 4080 series which peaks out at 736 GB/s of bandwidth through their 23 Gbps GDDR6X solution. The G6X solution is absolutely bonkers but GDDR7 will be the game changer and provide GPUs with higher bandwidth, especially those with a narrow bus. It should be noted that 32 Gbps speeds won't be featured across the entire lineup and only the RTX 5080 would get it at first. The possibility that NVIDIA might upgrade the lineup to faster speeds (G7) and higher capacities exists but for now, let's stick to what we can expect from the initial wave of the cards.
 
If it’s 2 slots at <= $1000 and matches the 4090 will people lose their minds? Nvidia has no incentive to give us any deals though. Clearly all the ngreedia anger isn’t leading to lost sales.
 
What the heck are the other slots for anyway? I think the days of customising your PC with additional expansion cards are long gone as the mobos include so much as standard. Is anyone struggling to fit in their GPU, SCSI controller, Soundblaster card, dual USB ports, and modem?

Make a four slot GPU - what different does it really make?
 
What the heck are the other slots for anyway? I think the days of customising your PC with additional expansion cards are long gone as the mobos include so much as standard. Is anyone struggling to fit in their GPU, SCSI controller, Soundblaster card, dual USB ports, and modem?

Make a four slot GPU - what different does it really make?
ITX and mATX are very much a thing and at least I still need a slot for my Sound Blaster
 
Well, there are "mini" cases for which more than 3 or even 2 slots would be a no go.
But ultra high end monster GPUs aren't really intended for such cases. If you're going with small form factor, you should be looking at something less than "peak workstation" hardware.
 
1728483617889.png

There’s also a new update on RTX 5070, the third Blackwell RTX 50 model. It is now said to be arriving alongside 5090 and 5080 at CES 2025, but as usual, it would launch later. Furthermore, the exact specs remain unknown, but the media is now claiming that it will feature GB205 GPU, not GB203 as the 5080 model. In other words, we have a similar situation to the Ada series, where there will be a visible gap between xx70 and xx80 class tiers.


But ultra high end monster GPUs aren't really intended for such cases.
Well there were several generations with top end cards which were able to fit into such cases so some people feel like this should still be the case ¯\_(ツ)_/¯
 
What the heck are the other slots for anyway? I think the days of customising your PC with additional expansion cards are long gone as the mobos include so much as standard. Is anyone struggling to fit in their GPU, SCSI controller, Soundblaster card, dual USB ports, and modem?

Make a four slot GPU - what different does it really make?

I assumed all this heatsink R&D at Nvidia is driven by engineering ambition and not any real consumer need. It’s been a long time since I saw anyone begging for 2 slot graphics cards. If anything we need better blower options for SFF use cases. But instead SFF cases have evolved to support longer and wider cards with standard fan setups.
 
Just for reference, we do know what each max configuration of the Blackwell "RTX 50" GPUs will have to offer and the GB205 maxes out with 5 GPCs or 50 SMs and a total of 6400 CUDA cores.

6400 SPs would be between 4070 and 4070S while 672Gb/s b/w puts on 4070Ti Super level.
 
I recall reading that GB205 would be 8x PCIe5.0. Is that correct? This is the reason I decided to get a PCIe5.0 mobo. I'm sure 8x5.0 is fine, not sure about 8x4.0.
 
If it’s 2 slots at <= $1000 and matches the 4090 will people lose their minds? Nvidia has no incentive to give us any deals though. Clearly all the ngreedia anger isn’t leading to lost sales.
It did, 40 series did not sell all that well compared to 30 series, hence why they released the Super refresh.
 
Back
Top