You should also get in before the people who - without testing it - will say "use cmder!" (which is _ten times_ slower on this benchmark) or "use alakritty" (which crashes on this benchmark after two frames), etc.
-
-
Replying to @cmuratori @Jonathan_Blow
Do you have a baseline to compare it to something? Either a fast terminal on a different platform but similar hardware or doing the same render with the deprecated API?
1 reply 0 retweets 0 likes -
Replying to @stephc_int13 @cmuratori
I mean I can just open our Sokoban game console to fullscreen and pick a moderately smaller font and measure how many fps it gets (answer: lots).
1 reply 0 retweets 9 likes -
Replying to @Jonathan_Blow @cmuratori
I don't think this is entirely realistic, even if it can serves to get an idea of the rendering load if correctly implemented with OpenGL.
1 reply 0 retweets 0 likes -
-
Replying to @Jonathan_Blow @cmuratori
Because a terminal emulator is not only a text renderer. I also think that the logic and I/O should be almost free on modern hardware, but it would not be an apple/apple comparison if we don't take the full context into account.
1 reply 0 retweets 0 likes -
Do you honestly think emulating a terminal is more expensive than simulating a 3D game?
1 reply 0 retweets 5 likes -
Absolutely not, I don't think the result of this benchmark is normal. But a benchmark is not very useful if we can't compare to a baseline, and I don't think rendering text is a totally fair baseline. My guess is that what we're seeing is not only a rendering issue.
1 reply 0 retweets 0 likes -
I would assume it is not a rendering issue, actually. It is almost certainly a parsing issue first, and a rendering issue second - ie., the rendering is probably slow, but the parsing of VT codes is almost certainly the bottleneck here.
2 replies 0 retweets 9 likes -
Replying to @cmuratori @stephc_int13 and
But as for your original question, I'm not sure what you mean by "baseline". What do you want to see as a "baseline" for this? It has ~2500 cycles per input byte, ~53000 per output glyph at that speed, assume no threading and no GPU assist.
2 replies 0 retweets 4 likes
So they're like 100x off of what you might expect a rudimentary thing to do, at a minimum. Is that what you mean by baseline?
-
-
Replying to @cmuratori @stephc_int13 and
It really depends, however, on whether they use GPU acceleration or not, as to whether memory bandwidth on the output side is a factor, so I couldn't say about that part. Not using GPU acceleration would be kind of insane these days, but so is 2fps, so, I don't know.
1 reply 0 retweets 1 like -
When using cmd.exe it is not using the GPU. When launched from the most recent Microsoft Terminal, it uses the GPU the result is almost the same.
0 replies 0 retweets 0 likes
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.