<beach>
At run time, as fast as you can expect. Like other implementations of the LOOP macro, it creates a TAGBODY.
<Josh_2>
Did you decide to go with a dsl or are you writing static html?
<beach>
The latter.
<beach>
I'll put it in a GIT repository soon.
<younder>
Ok I have created a jenkins-gitea CI environment and have started running the preliminary tests of parts of the compiler code
<beach>
What parts?
<younder>
loop-detection and peephole optimization. The SSA-transform and strength-reduction
<younder>
It hasn't been pushed upstream to codeberg yet. It is running on gitea locally.
Guest47 has joined #commonlisp
<beach>
I see.
<younder>
Once I have worked out a few snags in going from gitea to forgejo I will push it and you can have a look if you want.
<beach>
OK.
rkazak has quit [Ping timeout: 276 seconds]
stanrifkin has joined #commonlisp
shka has quit [Quit: Konversation terminated!]
<younder>
Fascinating, AMD uses a DNN to predict branches in new Zen architecures. Not a Markov chain like I thought.
mgl has joined #commonlisp
<Josh_2>
a DNN?
<younder>
Anyhow it is good at CLOS optimations with generic methods. It might not be the great problem that I thought.
<younder>
Josh_2: Apparently. AMD holds it's cards pretty close to it's chest when it comes to card prediction. But I have it from a good source Jim Kelly (the chief architect of Zen)
<younder>
card/branch
* younder
refers to Godbolt ho said virtual functions have the worrst branch prediction patterns andt that 'Clean Code' has the worst performance.
<younder>
who
<younder>
Anyhow the virtual dispatch pattern may not be worth optimizing on modern architectures.
<younder>
I have found a bunch of other optimations which modern architectures don't benefit from.
<younder>
One remaning obstacle is division. It is still expensive. You need to minimize mod and div
<younder>
Although Intel managed to reduce it from 200 cloock cycles to 40..
mala has quit [Ping timeout: 260 seconds]
<younder>
In modern CPU architectures you are profoundly confused by the fact that all though instructions go in and out of the processor in sequential order the are executed asynchronously. In fact a sequence of instructions will never travel the same way twice. Normally that doesn't matter unless you are writing a compler optimizer..
mala has joined #commonlisp
rkazak has joined #commonlisp
mgl has quit [Ping timeout: 260 seconds]
<younder>
The next big one is branch less instructions
<younder>
I am going for the BIG wins
cage has joined #commonlisp
<younder>
Incidentally Beach. The memory layout is also a big part of this. Computers like arrays and trees and slow. They need to be layed out ot sequentially in memory or the performance goes to pots. You can't seperate optimation from memory management.
<younder>
Just my two bits
yazz has quit [Ping timeout: 276 seconds]
yazz has joined #commonlisp
JuanDaugherty has quit [Quit: praxis.meansofproduction.biz (juan@acm.org)]
Guest3730 has joined #commonlisp
cage has quit [Remote host closed the connection]
rkazak has quit [Ping timeout: 248 seconds]
treflip has joined #commonlisp
Guest47 has quit [Remote host closed the connection]
rkazak has joined #commonlisp
Guest47 has joined #commonlisp
rkazak has quit [Ping timeout: 276 seconds]
rkazak has joined #commonlisp
rkazak has quit [Ping timeout: 276 seconds]
treflip` has joined #commonlisp
treflip has quit [Ping timeout: 276 seconds]
treflip`` has joined #commonlisp
rgherdt has quit [Quit: Leaving]
treflip` has quit [Ping timeout: 244 seconds]
rgherdt has joined #commonlisp
rkazak has joined #commonlisp
ejohnson has quit [Ping timeout: 252 seconds]
treflip`` has quit [Ping timeout: 272 seconds]
<beach>
I see.
random-nick has joined #commonlisp
reb has quit [Remote host closed the connection]
rkazak has quit [Ping timeout: 245 seconds]
rkazak has joined #commonlisp
decweb has joined #commonlisp
rkazak has quit [Ping timeout: 260 seconds]
jon_atack has joined #commonlisp
apac has joined #commonlisp
jonatack has quit [Ping timeout: 252 seconds]
surabax has joined #commonlisp
surabax has quit [Changing host]
surabax has joined #commonlisp
rkazak has joined #commonlisp
Guest3730 has quit [Ping timeout: 265 seconds]
<beach>
phoe: So, one option for you would be to just use the main page for the talk. It contains basically the same information as the text file I showed the other day. Then again, you seemed to have other ideas about the presentation, and that's fine too.
rkazak has quit [Ping timeout: 245 seconds]
bitspook has joined #commonlisp
jonatack has joined #commonlisp
jon_atack has quit [Ping timeout: 244 seconds]
pranav has joined #commonlisp
Guest47 has quit [Remote host closed the connection]
rkazak has joined #commonlisp
mwnaylor has quit [Read error: Connection reset by peer]
zxcvz has joined #commonlisp
zxcvz has quit [Client Quit]
rkazak has quit [Ping timeout: 252 seconds]
rkazak has joined #commonlisp
rgherdt_ has joined #commonlisp
rgherdt has quit [Ping timeout: 245 seconds]
rkazak has quit [Ping timeout: 276 seconds]
danza has joined #commonlisp
rkazak has joined #commonlisp
apac has quit [Ping timeout: 268 seconds]
rkazak has quit [Ping timeout: 268 seconds]
Oddity has quit [Ping timeout: 276 seconds]
rkazak has joined #commonlisp
wbooze has quit [Quit: Leaving]
bjorkintosh has quit [Remote host closed the connection]
bjorkintosh has joined #commonlisp
bjorkintosh has joined #commonlisp
kevingal has joined #commonlisp
bitspook has quit [Remote host closed the connection]
emaczen has joined #commonlisp
<skeemer>
i have a couple of questions... 1) how can i view the source code of a function in common lisp? 2) given the sheer amount of libraries without any documentation, what is your workflow to understand how a library works ?
rkazak has quit [Ping timeout: 245 seconds]
<beach>
skeemer: See #clschool. Don't post the same thing in more than one place.