 |
Post AqNp90gY1PHU2actt2 by [email protected] |
 |
More posts by [email protected] |
|
Post #AqM262BEuuk2teyQxE by [email protected] |
|
0 likes, 0 repeats |
|
|
|
Blogged: https://brandewinder.com/2025/01/22/fsharp-imperative-performance-opti… |
|
|
|
Post #AqM262MEG3X3Rk7D5E by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder Possible this is due to cache strides?i.e. in version 2 maybe the… |
|
|
|
Post #AqM262TJphCfnjQs8O by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@Tedspence I am not an expert on low level optimization, but yes, I think that&… |
|
|
|
Post #AqM262aPPKsI9ikXBY by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder In addition to this, original and version 1 have a multiple evalu… |
|
|
|
Post #AqM262fN6sqQP74UvA by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@Tedspence yes! I think also that v1 creates more arrays than v2 (1 per column,… |
|
|
|
Post #AqM262nWcZMmoOt0d6 by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder @Tedspence You can accumulate a min and a max pretty easily with … |
|
|
|
Post #AqM8NIf9UKfYfEa6DY by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@greggyb @brandewinder Yep, that would do it! |
|
|
|
Post #AqM8NImx1KuL3QEKNE by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@Tedspence @brandewinder And if we take out the mapping function to minMax, we … |
|
|
|
Post #AqMBoVMQgXBvjAUInA by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@greggyb @Tedspence nice work! I was going to let this rest, but now I might ha… |
|
|
|
Post #AqMBoVVe8GZ2Bknf9s by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder @Tedspence Well, I have bad news if your common case involves an … |
|
|
|
Post #AqMD2KrUcII7AqAzNA by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@greggyb I was actually wondering about that early exit question, and yes I wou… |
|
|
|
Post #AqMD2L006f63bE9mdM by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder If you like this, you should check out the FastFSharp channel. Ma… |
|
|
|
Post #AqMFKXV0ZODFUMT1vs by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder Also, I think in this case it will be tough to match the performa… |
|
|
|
Post #AqNp90Y2X2TXcCe6cq by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@greggyb I agree with the general statement. And, while I generally avoid it, I… |
|
|
|
Post #AqNp90gY1PHU2actt2 by [email protected] |
|
0 likes, 0 repeats |
|
|
|
@brandewinder I think some of this deserves some more test cases, too. Based on… |
