RTX 5060s Announced.

Eorzea Time
 
 
 
Language: JP EN FR DE
Version 3.1
New Items
users online
Forum » Everything Else » Tech Support » RTX 5060s announced.
RTX 5060s announced.
Offline
By Afania 2025-04-20 02:48:34
Link | Quote | Reply
 
Asura.Saevel said: »
The only reason nVidia is still selling any consumer GPU's is that Jensen likes to wear leather jackets and stand on stage acting like he's Steve Jobs.


Nah, his social influence in the world has elevated several times higher because of AI hardware. No one, except game nerds, cared about him when he only sell GPU to gamers back then.

These days he can wear leather jackets and stand on stage acting like he's Steve Jobs talking about AI stuff. And every big manufacturing company CEO and politicians listen to him talk below like he is Jesus alive, with media talking about him 3 days straight.

And all those social influence came from AI server, not 5090. Even if he wants social influence, consumer GPU doesn't offer them nearly as much.
Offline
By K123 2025-04-20 06:21:59
Link | Quote | Reply
 
Ok meta might be using 5nm but it definitely wouldn't be out of choice, they just don't have the might to get capacity on 3nm. Noone right now will be designing and making chips for 3nm that aren't already far down the design pipeline. I struggle to believe Nvidia got the demand wrong. Everyone was trashing the 5070 and 5080 since specs were released long ago. Could have been stuck into quantities they ordered before this, maybe, but they are far under market demand for 5090. I'd have bought a 5090 at £1800, maybe at £2000, but I'm not paying any more out of principle.
Offline
By Afania 2025-04-20 06:59:15
Link | Quote | Reply
 
K123 said: »
I'd have bought a 5090 at £1800, maybe at £2000, but I'm not paying any more out of principle.


Don't then. 5090 is luxury item for rich people that can pay anything for the best, it's not a requirement to play games. You are not their target audience. You can play majority of games with lower settings with weaker GPU.
[+]
Offline
Posts: 1212
By Seun 2025-04-20 09:14:01
Link | Quote | Reply
 
Afania said: »
K123 said: »
I'd have bought a 5090 at £1800, maybe at £2000, but I'm not paying any more out of principle.


Don't then. 5090 is luxury item for rich people that can pay anything for the best, it's not a requirement to play games. You are not their target audience. You can play majority of games with lower settings with weaker GPU.

Seconded.

It's discretionary spending for a gamer, but it's a business expense and a write off to a professional. There really isn't any value to be found in a market where some of the consumers don't have to care about the cost. We can keep fooling ourselves with the pretty charts and graphs, but it's fallacy.
Offline
By K123 2025-04-20 10:09:10
Link | Quote | Reply
 
Afania said: »
K123 said: »
I'd have bought a 5090 at £1800, maybe at £2000, but I'm not paying any more out of principle.


Don't then. 5090 is luxury item for rich people that can pay anything for the best, it's not a requirement to play games. You are not their target audience. You can play majority of games with lower settings with weaker GPU.
I only have the 3090 for AI. Same reason the demand is so high for the 5090. Prior to this I was happy with a 4GB AMD card for FFXI and emulators.
[+]
 Asura.Saevel
Offline
Server: Asura
Game: FFXI
Posts: 10086
By Asura.Saevel 2025-04-20 14:39:46
Link | Quote | Reply
 
Seun said: »
Asura.Saevel said: »
Clamshell is rarely used because it reduces the chips effective bandwidth in half and bandwidth is usually far more important to performance then raw capacity.

This is part of the reason why I wanted to see the head to head, specifically in instances where they're not spilling. I'm curious why 28G/s and not 32 tho? It makes my brain cell itch.

I posted it earlier .....

4060 Ti 8/16GB was already done. It's the exact same card with the only difference being memory and you can see when and where more memory starts to matter. And WTH you talking about 28G/s? 4060 TI is 288GB/s memory bandwidth. 5060 Ti is 448GB/s.
Offline
Posts: 1212
By Seun 2025-04-20 15:20:00
Link | Quote | Reply
 
Asura.Saevel said: »
Seun said: »
Asura.Saevel said: »
Clamshell is rarely used because it reduces the chips effective bandwidth in half and bandwidth is usually far more important to performance then raw capacity.

This is part of the reason why I wanted to see the head to head, specifically in instances where they're not spilling. I'm curious why 28G/s and not 32 tho? It makes my brain cell itch.

I posted it earlier .....

4060 Ti 8/16GB was already done. It's the exact same card with the only difference being memory and you can see when and where more memory starts to matter.

MB I thought it was normal vs OC. I didn't think 8GB cards were released to reviewers so I wasn't expecting to see anything before next week.



Asura.Saevel said: »
And WTH you talking about 28G/s? 4060 TI is 288GB/s memory bandwidth. 5060 Ti is 448GB/s.

28 x 16 = 448? That not how it works?
 Asura.Saevel
Offline
Server: Asura
Game: FFXI
Posts: 10086
By Asura.Saevel 2025-04-20 18:21:33
Link | Quote | Reply
 
Seun said: »
28 x 16 = 448? That not how it works?

You need to specify what your referencing to. In this case it's the memory bus speed measured in Gb/s clock.

Convert bits into Bytes we get 3.5GB/s per pin.
28/8 = 3.5

We have 128-bit width meaning 128 pins so 3.5 * 128 = 448 GB/s.

The other way is to do a per-package conversion, (28 * 32)/8 = 112GB/s per chip. Four chips gives us 4 * 112 = 448 GB/s.

28Gbps is the modules that nVidia purchased from their suppliers. The 5080 use's 32Gbps modules instead.
[+]
Offline
Posts: 1212
By Seun 2025-04-20 22:30:32
Link | Quote | Reply
 
I arbitrarily picked the bus speed because 32 sounds better than 28, but I meant more bandwidth in general. Thanks for the breakdown though. I wasn't visualizing the configuration correctly.
Offline
By K123 2025-04-28 05:14:58
Link | Quote | Reply
 
https://videocardz.com/newz/geforce-rtx-5080-super-rumored-to-feature-24gb-memory-rtx-5070-super-with-18gb-config

Says everything I was saying, e.g. 5060 should have used 3GB modules for 12GB, etc.

Nvidia have completely butchered this gen. If there are 2-3 years before 6000 series then I don't even expect desktop PCs to be anything like they are today by then. CAMM modules, etc.
Offline
Posts: 1212
By Seun 2025-04-28 18:45:10
Link | Quote | Reply
 
K123 said: »
https://videocardz.com/newz/geforce-rtx-5080-super-rumored-to-feature-24gb-memory-rtx-5070-super-with-18gb-config

Says everything I was saying, e.g. 5060 should have used 3GB modules for 12GB, etc.

Nvidia have completely butchered this gen. If there are 2-3 years before 6000 series then I don't even expect desktop PCs to be anything like they are today by then. CAMM modules, etc.

Would it really be worth giving up speed and bandwidth for capacity here?
Offline
By K123 2025-04-28 20:10:34
Link | Quote | Reply
 
It was assuming they didn't use such low bus also I suppose.
Offline
Posts: 1212
By Seun 2025-04-29 02:50:05
Link | Quote | Reply
 
I suppose they wouldn't have to, but the 5070 would look even more out of place if a 5060 had the same memory configuration. Plus the 5070 still runs out of memory so I assume a 12GB version of the 5060 ti would have the same issue.
Offline
Posts: 1921
By Felgarr 2025-04-29 04:05:33
Link | Quote | Reply
 
Wow, I hadn't heard of CAMM modules until I read through this thread. However, I expect CAMM modules to be in thin laptops and less-likely to see them in ITX/mATX/ATX motherboards for desktop PCs. Still Interesting though

What other changes do you forsee to desk PCs in 2-3 years?
 Asura.Saevel
Offline
Server: Asura
Game: FFXI
Posts: 10086
By Asura.Saevel 2025-04-29 19:22:00
Link | Quote | Reply
 
Felgarr said: »
Wow, I hadn't heard of CAMM modules until I read through this thread. However, I expect CAMM modules to be in thin laptops and less-likely to see them in ITX/mATX/ATX motherboards for desktop PCs. Still Interesting though

What other changes do you forsee to desk PCs in 2-3 years?

CAMM are nonexistent in the consumer desktop space. Some companies made them but ultimately there was very little demand, it's a solution looking for a problem. The two to four slot DIMM formfactor is just too useful and flexible, it'd be like trying to replace USB peripherals with a proprietary fruit branded... ohh wait. It's like MXM, useful for notebooks or embedded devices, scant everywhere else.