autotldr@lemmings.world
on 07 Jun 2024 02:40
nextcollapse
This is the best summary I could come up with:
Back in February Intel changed the Meteor Lake EPP from a value of 128 to 115 when operating in the default “balance_performance” mode but that turns out to not be entirely ideal.
The one-line patch put out today bumps it from 115 to 64 and is yielding nice gains in performance and power efficiency.
With the revised EPP value, the Intel Meteor Lake performance improved by 19% for the Speedometer web browser benchmark.
Meanwhile the Microsoft Teams video conference power consumption dropped by 35% with that EPP adjustment.
This patch provides that updated tuning magic for the Intel Meteor Lake SoCs with the P-State driver.
Hopefully the patches will be merged for Linux v6.11 if not being submitted early as a “fix” for the current 6.10 cycle…
The original article contains 375 words, the summary contains 127 words. Saved 66%. I’m a bot and I’m open source!
narc0tic_bird@lemm.ee
on 07 Jun 2024 06:27
nextcollapse
It’ll hopefully arrive before 6.11.
gravitas_deficiency@sh.itjust.works
on 07 Jun 2024 13:11
collapse
I suppose it’s possible they’ll back port it to 6.10 if they miss the window… but as it’s not security-oriented, that’s probably pretty iffy.
darkfiremp3@beehaw.org
on 08 Jun 2024 00:34
collapse
I’m so bad at remembering all these different Intel code names
threaded - newest
This is the best summary I could come up with:
Back in February Intel changed the Meteor Lake EPP from a value of 128 to 115 when operating in the default “balance_performance” mode but that turns out to not be entirely ideal.
The one-line patch put out today bumps it from 115 to 64 and is yielding nice gains in performance and power efficiency.
With the revised EPP value, the Intel Meteor Lake performance improved by 19% for the Speedometer web browser benchmark.
Meanwhile the Microsoft Teams video conference power consumption dropped by 35% with that EPP adjustment.
This patch provides that updated tuning magic for the Intel Meteor Lake SoCs with the P-State driver.
Hopefully the patches will be merged for Linux v6.11 if not being submitted early as a “fix” for the current 6.10 cycle…
The original article contains 375 words, the summary contains 127 words. Saved 66%. I’m a bot and I’m open source!
It’ll hopefully arrive before 6.11.
I suppose it’s possible they’ll back port it to 6.10 if they miss the window… but as it’s not security-oriented, that’s probably pretty iffy.
I’m so bad at remembering all these different Intel code names