Carzy Zarx’s PC Gaming Emporium - Catch Up on All the Latest PC Gaming Related News

[JEMC]

Captain_Yuri said:

Samsung develops 512GB DDR5 memory based on High-K Metal Gate HKMG process technology

https://videocardz.com/press-release/samsung-develops-512gb-ddr5-memory-based-on-high-k-metal-gate-hkmg-process-technology


Intel’s New IDM 2.0 Strategy: $20b for Two Fabs, Meteor Lake 7nm Tiles, New Foundry Services, IBM Collaboration, Return of IDFb

https://www.anandtech.com/show/16573/intels-new-strategy-20b-for-two-fabs-meteor-lake-7nm-tiles-new-foundry-services-ibm-collaboration-return-of-idf

"As part of Intel’s Foundry Services, the company is announcing that it will work with customers to build SoCs with x86, Arm, and RISC-V cores, as well as leveraging Intel’s IP portfolio of core design and packaging technologies."


Samsung Demos 256Mb 3nm MBCFET Chip: Performance, Density Up, Power Down

https://www.tomshardware.com/news/samsung-3nm-isscc-2021-gaafet-3gae

*pic*


Some interesting things be happening in the Foundry sector. First is Intel opening up their fabs to third party customers which means we will now have 3 main foundries to choose from. And second is Samsung will be using GAA/MBC for their 3nm node instead of FinFET that TSMC is using and GAA is supposed to be significantly better. If they actually manage to do it, Nvidia's early investment into Samsung will pay off significantly similarly to how AMD invested into TSMC and reaped the benefits. And not to mention it will change the monopoly status from TSMC to Samsung.

Imagine one day, maybe in 20 years, getting that kind of capacity for regular PCs...

Intel opening its fabs to others is good news for them, although we'll have to wait and see if their prices are competitive enough to be a good option. Now we need GloGo to get their shit together to make it a 4-way horse race that could scale down a lot of the supply problems we suffer every now and then.

And I wouldn't count TSMC down compared to Samsung's new strategy. They've been in this game long enough to keep a trick or two up their sleeve.

Bofferbrauer2 said:

JEMC said:

There's not enough salt in the world for me to believe that. Nintendo going with the latest hardware? One that hasn't even been announced yet? That would be a first, and also a waste of silicion for "only" a Super or Pro device.

Yeah, that's more like if Nintendo was already shopping for the successor for the Switch, which is not impossible, but even that would take at least a London Eye-sized grain of salt.

Also, if the GPC would stay the same as with Turing (don't know the size for Ampere), then there would be 640 Shader processors compared to 256 right now, so 2.5 times as many. That's the same amount as in the old 1050Ti and MX350 GPUs, but even the MX has a TDP of 20W, way above what something like the Switch could handle in handheld mode without sucking the battery dry in an instant.

So while having 640 SP is not out of question, especially not with a future node, they would need to be seriously tuned down to handle the constrains of a handheld form factor. But between the bigger chip, the higher IPC, the  higher clock rates (Switch clocks so low that it's successor will certainly clock a couple hundred Mhz higher even for handheld mode) and possible other improvements, that could very well result in something between 6-12 times what the current Switch can push in terms of graphics, so about PS4 level in terms of graphics and way enough for a generational uplift.

Since this would be a new SoC, we don't know how much focused on gaming is or what its specs could be. Ampere doubles the shader count compared to Turing for each SM, but would Nintendo prefer to go with 4 SMs and get twice the shaders or would prefer to get half the SMs and the same amount of shaders, relying on its architecture improvements to bring performance forward?

Oh, well, I doubt we'll get any kind of confirmation or proper info for the foreseable future so, is there even a point in thinking about its possible specs?