News for hackers. Stuff that matters.
Re:Intel’s port 7 AGU blunder (2019) by jjoonathan on Monday June 15, 2020 @04:31PM (#23532605)
Simple + plausible explanation: approximately nobody uses fancy instructions with fancy addressing, so they butcher-chopped some verilog (does intel use verilog?) in one generation to meet a target and the chopped code outlived its original purpose.

Does OP really think legacy code dynamics don't apply to intel?

Wasn't there a bug not long ago where you could thermally kill a CPU by pulling clever tricks to get vector unit utilization higher than was supposed to be possible (presumably because nobody cared enough to precisely figure out "possible")?

Re:Intel’s port 7 AGU blunder (2019) by jeffbee on Monday June 15, 2020 @09:21PM (#23534995)
I don't know about "kill" but power viruses have been a research topic for a long time. A decade ago researchers were using automatic techniques for maximizing power consumption on real hardware (https://lca.ece.utexas.edu/pubs/ganesan_pact10.pdf)
Re:Intel’s port 7 AGU blunder (2019) by userbinator on Monday June 15, 2020 @11:20PM (#23535645)
Odd to not see a mention of LINPACK in that paper, when it has reliably produced temperatures exceeding all other stress-testing programs I've tried, as well as uncovered instabilities that others couldn't find. It also turns out to be a very realistic type of benchmark (solving dense systems of linear equations). The paper doesn't predate its use, so I guess the authors just haven't spent much time on overclocking forums...