Computers in 97 didn't need much in the way of cooling. A large passive heatsink was plenty for those CPUs. They're not the 300+ watt behemoths we have today.

[–] Pacmanlives@lemmy.world 10 points 5 months ago* (last edited 5 months ago) (1 children)

I really remember heatsinks being a thing on overclocked systems around that time frame and then once we got to P4 cpus the chilling towers appeared those things were massive

[–] fuckwit_mcbumcrumble@lemmy.dbzer0.com 11 points 5 months ago (2 children)

The lower power 486s didn't even need a heatsink. The P3 was the first to take a heasink resembling what we have today, but damn did the P4s need some serious cooling.

It's kinda funny how we think the 100 watts of a desktop P4 was insane when now the TDP of a high end laptop CPU is more than that.

[–] vaionko@sopuli.xyz 4 points 5 months ago

My Pentium 100 even says "Heatsink req'd"

[–] Illecors@lemmy.cafe 2 points 5 months ago (3 children)

It's kinda funny how we think the 100 watts of a desktop P4 was insane when now the TDP of a high end laptop CPU is more than that.

It really isn't. Modern mobile cpus barely sip power.

[–] mbfalzar@lemmy.dbzer0.com 3 points 4 months ago* (last edited 4 months ago)

PL2 on a 14900T is 106W

Edit: I'm an idiot, T series is low power socketed, not mobile. 14900HX has a TDP of 55W but boosts short term to 157W, which is still pretty ridiculous

[–] ICastFist@programming.dev 1 points 4 months ago (1 children)

If you meant cell phones and tablets, that's mostly due to the different architecture. RISC processors are super energy efficient, which also makes them much cooler to run.

x86-64 is a CISC architecture, which tends to be much more power hungry. There are only a couple of very low power Celeron CPUs that work under 10W of TDP, while that's very common among phones' CPUs.

[–] Trainguyrom@reddthat.com 3 points 4 months ago* (last edited 4 months ago)

x86-64 is a CISC architecture

In many cases it's actually RISC under the hood and uses an interpreter to translate the CISC commands and run them in the most optimal manner on the silicon

ARM and RISC-V absolutely scale up to multi-hundred watt server CPUs quite easily. Just look at the Ampere systems you can rent from various VPSes for example

The big benefit that ARM and RISC-V have is they have no established backwards compatibility to keep carrying technical debt forwards. ARM versions their instruction sets and software has to be released for given versions of ARM cores, and RISC-V is simply too new to have any significant technical debt on the instruction set side.

Atom cores were notable for focusing the architecture on some instructions then other instructions would be a slog to execute, so they were really good at certain things and for desktop use (especially in the extremely budget machines they got shoved into) they were painful. Much like how eCores are now. They're very carefully architected for power efficiency, and do their jobs extremely well, but an all eCore CPU is a slog for desktop use in many cases

[–] fuckwit_mcbumcrumble@lemmy.dbzer0.com 1 points 4 months ago

My 11950H (and all other “full power” Intel mobile CPUs) have a PL1 of >100 watts (109 for mine), and mine a PL2 of 139 watts. This laptop is about an inch thick.

Nothing about this laptop sips power, I’ve gotten as bad as 30 minutes of battery life out of a 90 watt hour battery not playing games.

[–] psud@aussie.zone 8 points 5 months ago

I helped set up a friend's "586" (about equivalent to a Pentium 1) and he had neglected to buy a heat sink or fan

A hammer was a sufficient heat sink for the time it took to set up windows

[–] Dave@lemmy.nz 4 points 5 months ago

Super impressive since we used to play Quake 2 all day on it!