Their explanation of the issue is sorely lacking. From the description the article provided it sounds like an issue with the instruction decoder blocks but they left out a lot of details. Not sure if this is an issue with the RISC-V instruction set in general or with someone’s specific implementation of it.
Chinese semiconductor thread II
- Thread starter vincent
- Start date
Their explanation of the issue is sorely lacking. From the description the article provided it sounds like an issue with the instruction decoder blocks but they left out a lot of details. Not sure if this is an issue with the RISC-V instruction set in general or with someone’s specific implementation of it.
Maybe this will give LoongArch a boost.
Not sure if it applies to this article in particular, but my understanding was the US lawmakers were attempting to restrict commercial activity between US and/or Chinese implementations of the architecture. That is to say, import/export controls on the hardware, designs, and so on.
Which is still dumb, but not impossible.
antiterror13
Brigadier
Really brutal and who in the right mind want to use US tools and machine, especially for Chinese companies ... great bitter lesson for China to not rely on foreign technologies.
I think American technologies is done in China and many other countries
theres active discussion of kirin 9010 is not using ASML because its alignment mark totally different. tonyget is involved in that discusssion too.
if its not ASML , could it be SMEE SSA800? i dont think SSA800 capable of 7nm at this point.
maybe its huawei's own scanner as it has established a semiconductor equipment division focusing on scanner and full Arf-i industry supply chain is there, Huawei doesn't need to depend on SMEE. maybe huawei is quicker than SMEE to get to 7nm.
if its not ASML , could it be SMEE SSA800? i dont think SSA800 capable of 7nm at this point.
maybe its huawei's own scanner as it has established a semiconductor equipment division focusing on scanner and full Arf-i industry supply chain is there, Huawei doesn't need to depend on SMEE. maybe huawei is quicker than SMEE to get to 7nm.
Actually, from the comment area on that original bilibili video(which has been deleted now). Someone mentioned that the single expose area of ASML is about 6 times of that kirin9010 die area，and one expose only requires one alignment mark，so six dies only require one alignment mark. So it is possible that the die he is got，just happen to be the one without alignment mark，and the video creator agree with this speculation.
I wouldn't take anything huawei says publicly about stuff like this too seriously. After all, they said 5G was impossible last year and then came out with Mate 60 a couple of months later. Everything needs to be put into context. China's industries are mostly underplaying their capabilities to stay off radar.
SAQP does effect yield though.