Intel isn't going to roll out new desktop chips until 2021/22

[Bires]

[0091/2]

So will intel be selling until then?

[Greasy]

They have released a statement saying they will have 10nm desktop chips, and calling these rumors FUD. I will believe it when I see them, but they are saying this is just FUD.

[CurrentlyPissed]

Originally Posted by Greasy

They have released a statement saying they will have 10nm desktop chips, and calling these rumors FUD. I will believe it when I see them, but they are saying this is just FUD.

It was a pretty solid rumor. But still a rumor, I think it had some play in the field tentatively.

I think stockholders had a play in this remark to go forward with 10nm.

https://www.extremetech.com/computin...sktop-cpus-7nm

[Destroy]

Enough of this 5GHz crap, I want my 10GHz chip NOW!

[CurrentlyPissed]

Originally Posted by Destroy

Enough of this 5GHz crap, I want my 10GHz chip NOW!

I'd take proper coded games to take advantage of correct instructions and core/thread counts over more Ghz. It's times devs figure this out.

[bill dennison]

Originally Posted by CurrentlyPissed

I'd take proper coded games to take advantage of correct instructions and core/thread counts over more Ghz. It's times devs figure this out.

this

they are not going to get much faster soon without sub room temp cooling


but proper thread coding will make the need for faster chips unnecessary

[kboye]

Little story about tech advances vs code optimizations from someone who just went through this days ago.

Been messing around last year making a fan controller using an Arduino Uno. My EEPROM storage/persistent budget was about 1000 bytes, my RAM (SRAM) limit for variables/RAM structure the same as structure itself is cached to avoid spamming the EEPROM wasting limited writes. I have 6 fans with labels (12 bytes per), min/max temp&PWM duty per fan, 8 temp sensor (labels, connections per fan), serial buffering and coms with a C# panel to control the thing, the code itself takes up about 25KB of Flash. I think I can use 10 fans and 16 sensors, but at that point I run out of onboard PWM or RPM IO. Thing boots in 1 second and after setting it up runs by itself.

It felt real good as a programmer to flex my brain a little and think bits and compact data again, hacking around to squeeze 16 bytes down to 2, experiencing overflow when I was running out of SRAM or EEPROM. Controller does more than anything I can buy right now, without software, relying on physical sensors. Using interrupts to avoid halting execution, queue system for polling. I'm happy Still got alot more to learn and optimize.

Yesterday I got a Teensy 3.6. It has around 250 000 bytes more SRAM instead of the total 2000 I was used to, ~1000KB of Flash vs 32. It's quick.

It is completely overkill, except the 60+ IO which I needed.

Point being: I got no idea if I want to use it, too quick a performance leap to make elegant use of it. I know I would be able to make the code more easily understandable and be easier to modify later on by not having to rely on hacks, bit packing etc. I also know it works just the same on this puny thing I got now running at.. What, 16MHz vs 180MHz on the Teensy? I know if I make that transition, I'll probably start getting lazy again and not need to optimize.

Not sure I'm ready for this just yet

Same thing when developing something on desktop, then tweaking it for phones. Instead of 4k textures, suddenly I find myself doing a single 2x2 pixel thing and limiting FPS to avoid overheating the over-agile iPhone that jystcwants to push pixels faster than is good for it