4 ms·

So this is surely self serving, but... Intel used BitKeeper for their chip dev process for a long time, and during that time the chips got faster with each rel

So this is surely self serving, but... Intel used BitKeeper for their chip dev process for a long time, and during that time the chips got faster with each release. More than a decade of chips were developed under BitKeeper and we saw perf improvements with every chip. They moved to Git and ever since then performance hasn't really moved. The Atom people were the first to move and the mainstream processors followed. I said it was self serving, so rain all over me, but I do wonder if there is anything there. I know we helped them move pretty fast a while back.

grzm 10y ago
If it's anything other than coincidence, I think it's more likely that the cause would be related to why they migrated from BitKeeper to git, rather than the change in tools itself.
- luckydude 10y ago
  I really don't know. As a butthurt BitKeeper guy I would love for it to be SCM reasons. I mostly think that had nothing to do with it but when I look at the data, I wonder. We spent a lot of time helping them with their problems, all stuff that nobody in the Git world steps up to do (like submodules that actually work). All that help may not have made a difference. On the other hand, it went way beyond just SCM stuff, I debugged their various filer problems (that they blamed on BK and I'm happy to say not a single one had anything to do with BK, BK just exposed that the filer was buggy). I like to think our help did make a difference but I really don't know. What I do know is we did whatever Intel wanted for over a decade and they used to have BitKeeper in ever powerpoint they wrote about new processor development. But they have a culture of tools have to be free so I guess they get what they pay for.
  - B1FF_PSUVM 10y ago
    The "toxic culture" is much wider than you realize ... until it bites you. Then it's too late. But good luck to you, you did good and played fair, AFAIK.

grzm 10y ago

If it's anything other than coincidence, I think it's more likely that the cause would be related to why they migrated from BitKeeper to git, rather than the change in tools itself.

luckydude 10y ago
I really don't know. As a butthurt BitKeeper guy I would love for it to be SCM reasons. I mostly think that had nothing to do with it but when I look at the data, I wonder. We spent a lot of time helping them with their problems, all stuff that nobody in the Git world steps up to do (like submodules that actually work). All that help may not have made a difference. On the other hand, it went way beyond just SCM stuff, I debugged their various filer problems (that they blamed on BK and I'm happy to say not a single one had anything to do with BK, BK just exposed that the filer was buggy). I like to think our help did make a difference but I really don't know. What I do know is we did whatever Intel wanted for over a decade and they used to have BitKeeper in ever powerpoint they wrote about new processor development. But they have a culture of tools have to be free so I guess they get what they pay for.
- B1FF_PSUVM 10y ago
  The "toxic culture" is much wider than you realize ... until it bites you. Then it's too late. But good luck to you, you did good and played fair, AFAIK.