• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Nvidia Blackwell in Q1-2025

Page 144 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Chopped GB205 for a RTX 5060 Ti (Super) could be pretty good if the price is right ($400?).
But I think that might not appear until next year.
 
but MFG is powerful tech!
in upcoming Doom Dark Ages also will be fun
2f43b1b39fe6cc3fb00d0eb484216ba626a07ac4b91e57ff9428ae0a69e757b4.png
 
Chopped GB205 for a RTX 5060 Ti (Super) could be pretty good if the price is right ($400?).
But I think that might not appear until next year.

The chopped (well more chopped since the 5070 isn't full die) GB205 dies are going to the 5070M and unlike the consumer market Nvidia has contracts with laptop makers that they're going to need to fulfill.

The 5060 Ti is supposed just the 16 GB of the 5060. Maybe this changes, but I wouldn't expect cheaper GB205 until next year.
 
Just installed my TUF 5090, no ROPs missing! It appears to be running PCIE5 x8 according to HWInfo and GPU-Z though, so I'm trying to troubleshoot that. Also, CUDA is failing in ComfyUI/SwarmUI so will have to look into that as well.

Tried a few games though, and there's no doubt its delivering there, but the CPU limitations are ridiculous with my 7950X and from what I'm read its not that much better with the 9800X3D. I will probably get the 9950X3D but it will probably be a bit underwhelming still.

Also, for the record, the OG power connector on the old 4090 (bought on release day) was perfectly fine, not damaged at all. Seems everyone expects the worst these days 😛
 
I wouldn't worry about it only using 8x PCIE. It won't make a difference between 8x and 16x at PCIE5 speeds. I don't even think it matters for PCIE4 based on past testing.

If you have an older CPU that's bottlenecking, just consider it an excuse to crank up the resolution and graphics settings to help balance the load.
 
Practically it's essentially identical. It's technically a bit faster, but not enough to matter at anything. No one could really tell the difference between these cards.

I really wonder what is going on at NVidia. Such negligible upgrades and no stock. It's kind of bizarre.

I think you could sort of argue that the -$50 in MSRP is the upgrade.


Chopped GB205 for a RTX 5060 Ti (Super) could be pretty good if the price is right ($400?).
But I think that might not appear until next year.

I'd bet money that the Super will be just a few hundred MHz of clocks and 18GB VRAM.
 
Took a break and decided to check out Tom's Review. 4 Stars!


LMAO, jumped to the graphs and no 4070S? Really?

1741122669681.png

Open the comments and yup, dumpster fire of nvidia roasting and Jared writing huge "sorry not sorry" paragraphs. "You can't buy a 4070S any more so...." LOL. Best remove all the GPUs from the graphs, I guess.

Peak entertainment at least 😀

At least even he managed to include "probably no stock at MSRP" in the top blurb.

And... a listed con is that the 9070 series drops tomorrow. Bold words, those, right in the top of the article. Might seem like a little, but there.... seems like a lot.
 
If AI bubble pops, Nvidia will care about our dollars again, but not until then.
it's never gonna pop until someone can match nvidia in their shipment volumes, otherwise, the demand is still infinite and probably will be for the next 5-6 years unless some random company decides to break the mold and release a 256gb AI accelerator card for under $1500.

That or someone comes with a real breakthrough that can do full model performance but at 1/4 the file size, but even then maybe something like would merely increase demand of high end AI by 4x. Since the more we can compress a model without compromising intelligence, the smarter huge models will be. And thus demand goes right back up again.
 
Maybe people start buying AMD and just stop caring about the ray tracing scam as much? It's just another Nvidia gimmick anyway.
 
Maybe people start buying AMD and just stop caring about the ray tracing scam as much? It's just another Nvidia gimmick anyway.
Nah. Now even AMD is touting pathtracing demos with RDNA4. I think Nvidia either has to release RTX 6000 series a year earlier or let AMD catch up or even surpass them in pathtracing performance with RDNA5. My feeling is that RTX 5000 is so rushed that the drivers are half baked and we may see at least another 10% additional performance in raster or RT or both in the next year or so.
 
It's just another Nvidia gimmick anyway.
it's going mass when PS6 released. Current problem with RT, you need High Ends cards. When gonna finally playable on xx6x series, it's will be good for everyone.
Currently i like ID tech RT style working fine on most cards, just need good VRAM amount.


think Nvidia either has to release RTX 6000 series a year earlier or let AMD catch up or even surpass them in pathtracing performance with RDNA5
I think it's gonna happens if Rubin just fattier RTX 5xxx, and same generation like 4xxx > 5xxx
 
Nah. Now even AMD is touting pathtracing demos with RDNA4. I think Nvidia either has to release RTX 6000 series a year earlier or let AMD catch up or even surpass them in pathtracing performance with RDNA5. My feeling is that RTX 5000 is so rushed that the drivers are half baked and we may see at least another 10% additional performance in raster or RT or both in the next year or so.

Yeah, for Blackwell I'm guessing it will be right around that traditional 18-36 month mark that the architecture will start to really shine and come into it's own.
 
Yeah, for Blackwell I'm guessing it will be right around that traditional 18-36 month mark that the architecture will start to really shine and come into it's own.
Considering the state of that silicon and the cards' power conectors, that probably won't happen until a refresh is launched. Same with Fermi and NV30 / Geforce FX.
 
Back
Top