IT LOOKS LIKE TSMC’s 32nm processes are not doing nearly as well as people think. We hear that things are delayed a bit, and performance is far less than optimal.
If you recall, its 40nm process is about a year late so far and still underperforming in laughable ways. When we first said there were some problems on the 40nm process, people were skeptical. Months later, we found the culprit, but there was no immediate solution. In July, TSMC gave the thumbs up, but it seems to have snatched defeat from the jaws of victory once again and botched things badly in recent weeks. How can it keep stepping backwards?
Now, word has reached our tender ears that TSMC’s 32nm processes are not doing well. We had heard that the 32nm mainstream ‘high performance’ process was worse off than 40nm at a comparable state of development, but then some blue fairy dust was applied. The blue fairy denies dusting TSMC, and given that we are now hearing the 32nm mainstream process just slipped by another 2 months, that denial is seeming more plausible.
Luckily, that is only for the mainstream high performance 32nm process. The 32 low power one is different, but unfortunately has other problems. All of the voices in our heads are telling us the same thing – no, not to get the guns, but that the 32nm low power process is also underperforming. Underperforming by a lot.
So, if you are desperate to buy an unannounced 32nm part made on TSMC’s 32nm process, you might not want to get in line soon. It might be a longer wait than you had anticipated.S|A
Latest posts by Charlie Demerjian (see all)
- SemiAccurate has Skylake-SP die shots, sizes, and more - Jul 21, 2017
- Skylake-SP has a diverged core - Jul 19, 2017
- Intel’s Purley platform architecture is a step forward - Jul 13, 2017
- Intel launches Purley aka metal Xeons - Jul 11, 2017
- AMD’s Epyc has lots of connections - Jun 26, 2017