assuming its real, for the last decade or so TSMC has been better about low power (serving mostly mobile devices) while intel has been better at high power. Last few years TSMC has been better all around but I’d assume 20/18 are targeting high power first, low power down the road if everything works out like GAAFET+Powervia.
TSMC has had better low power performance but the difference isn’t massive. I suspect the main reason is a) 18A is not ready for mass production until too late and b) limited 20A capacity coming online next year is needed for other products (remember intel has to ship absolute shitload of chips). So why not do it on TSMC if that is possible and the alternative is to delay the products to wait for capacity?
Edit: i’m not sure if 20A was again a more limited early version designed primarily for intel desktop products and 18A was supposed to be the long term version of the node with wider array of capabilities?
20A is essentially an early, incomplete release of 18A that can only be used on for the compute tiles. LNL has iGPU, NPU, and x86 cores all on the same tile, so 20A can’t be used.
Intel 3 and 4 weren’t designed with GPUs in mind either.
LNL goes into manufacturing in H1 2024, but 20A will be ready only by H2 2024. Arrow lake will be on both N3B and 20A because 20A is now ahead of schedule.
20A should be manufacturing ready early next year if they are going to launch arrow lake on it in 2024. I suspect it’s already basically done and they are expanding capacity now.
I’m not sure about arrow lake refresh. Right now, we have 2 versions of Arrow Lake. One on 20A and the other on N3B. Both should launch about the same time but will target different core configurations.
If you listen to the investor calls and roadmaps they state when 20A/18A will be ready and on what manufacturing processes.
My guess is IFS will be coming online and Intel is saving that for their customers. And the other is better margin. Intel has better margins on their own manufacturing reserved for their higher priced products.
Lunar Lake being a consumer product will mean lower margins. And saving 20A/18A for server and server gpu products could mean better margins.
Performance is likely secondary. All new silicon perform pretty close that you really can’t tell the difference.
Only NVIDIA has the exception with the software scaling. Like DLSS and Frame Generation.
I’m curious as to why it’s on TSMC N3. Is it because 20A/18A isn’t ready yet or is there another reason?
assuming its real, for the last decade or so TSMC has been better about low power (serving mostly mobile devices) while intel has been better at high power. Last few years TSMC has been better all around but I’d assume 20/18 are targeting high power first, low power down the road if everything works out like GAAFET+Powervia.
TSMC has had better low power performance but the difference isn’t massive. I suspect the main reason is a) 18A is not ready for mass production until too late and b) limited 20A capacity coming online next year is needed for other products (remember intel has to ship absolute shitload of chips). So why not do it on TSMC if that is possible and the alternative is to delay the products to wait for capacity?
Edit: i’m not sure if 20A was again a more limited early version designed primarily for intel desktop products and 18A was supposed to be the long term version of the node with wider array of capabilities?
20A is essentially an early, incomplete release of 18A that can only be used on for the compute tiles. LNL has iGPU, NPU, and x86 cores all on the same tile, so 20A can’t be used.
Intel 3 and 4 weren’t designed with GPUs in mind either.
LNL goes into manufacturing in H1 2024, but 20A will be ready only by H2 2024. Arrow lake will be on both N3B and 20A because 20A is now ahead of schedule.
20A should be manufacturing ready early next year if they are going to launch arrow lake on it in 2024. I suspect it’s already basically done and they are expanding capacity now.
Will Arrow lake refresh be totally on 20A?
I’m not sure about arrow lake refresh. Right now, we have 2 versions of Arrow Lake. One on 20A and the other on N3B. Both should launch about the same time but will target different core configurations.
Because this came out of someone’s ass.
If you listen to the investor calls and roadmaps they state when 20A/18A will be ready and on what manufacturing processes.
My guess is IFS will be coming online and Intel is saving that for their customers. And the other is better margin. Intel has better margins on their own manufacturing reserved for their higher priced products.
Lunar Lake being a consumer product will mean lower margins. And saving 20A/18A for server and server gpu products could mean better margins.
Performance is likely secondary. All new silicon perform pretty close that you really can’t tell the difference.
Only NVIDIA has the exception with the software scaling. Like DLSS and Frame Generation.