You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
I see what you're saying, but OneAPI doesn't really fit with CUDA and ROCM, because it is not GPU specific like the other two. Intel is addressing the targeted application problem with OneAPI libraries (e.g. DDN, VINO, etc), but I agree that it doesn't offer a framework tool like PyTorch...
OneAPI is based on SYCL, which is a multi-vendor industry specification.
https://www.khronos.org/sycl/
OneAPI and CUDA are based on some similar concepts, such as C++ and manual memory management. CUDA supports C, C++, and Fortran, while OneAPI only appears to support Intel's open source Data...
The UXL Foundation is the second I-Hate-Nvidia-Club to get started lately. The first was the Ultra Ethernet Consortium, which aims to displace Nvidia's InfiniBand implementation (the only viable one available). It's sort of amusing. The Ultra Ethernet Consortium is led by AMD, while UXL seems...
I still use Hotmail. Anyone who doesn't like it can... [use your imagination about what pejorative acronym I might insert here]. :p
I agree with you about EV drivetrains, at least on Teslas. I think the legacy carmakers are putting more component-level complexity in their EVs. For Rivian...
I don't have issues with any of that. No one is going to outfit a fab to have it sit 50% idle or worse. (Not to mention the mystery of where all of the EUV machines will come from.) There have been discussions about a fab glut on this site for a long time, and I don't remember anyone...
We certainly have the some of the most convoluted ways of dispensing subsidies in the US. I think the Investment Tax Credit Intel mentioned relates to renewable energy infrastructure, but a search comes back with links about solar energy credits. Whatever.
I agree with you philosophically. I tend towards the we're-understating-the-risk end of the paranoia spectrum regarding Xi and his Taiwan agenda. My point, however, that it is illogical to worry about TSMC's share price due to China's potential actions, but keep a ginormous Apple investment...
Warren Buffett is a smart guy, but I've never understood his geopolitical risk concerns about TSMC, sufficient to cause him to divest TSMC stock, when he holds about $175B of Apple stock. It seems silly, and I can't remember WB ever looking silly other than this situation. And it's not just...
Yeah, this is the dumbest, most poorly conceived part of the *implementation* of the CHIPS Act. It is not in the bill that Congress passed. This is pure Biden Administration. And I wonder... will the USG credit the fab builders for the actual losses they surely will incur while the fabs ramp...
Poland is a much lower cost country than Italy, by like 60% for labor. The unionization rate is also low. I suspect these factors figured prominently into Intel's decision.
Your statement is correct, but in reality there is more to this. It is correct that Congress wrote the bill and Biden just signed it, but the Biden Administration's Dept of Commerce, part of the Executive branch of the USG, did add their own requirements and criteria for the allocation of...
More information and some high-level analysis from TP Morgan, and a bit of silly talk about 450mm wafers. But Morgan is good at pumping execs for comments.
https://www.nextplatform.com/2024/03/14/cerebras-goes-hyperscale-with-third-gen-waferscale-supercomputers/
I don't know for sure, because they don't go into a lot of hardware implementation detail on their web site or in white papers, but they are using optical 100GbE, so I suspect they're using off-chip optical transceivers which are electrically close to the CS-3. I haven't seen one of their...