-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Raptor Lake Leaks + Intel 4 developments
- Thread starter g67575
- Start date
The thread scheduler in Windows 11 supports and interfaces with the on-chip thread director.
Windows 10 doesn't do this. It's thread scheduler can still differentiate between the core types but without the assistance of the director, workloads can be misassigned.
Windows 11 will also have a better implementation of DirectStorage due to the new storage subsystem. Windows 11 will be better the gaming platform.
Windows 10 doesn't do this. It's thread scheduler can still differentiate between the core types but without the assistance of the director, workloads can be misassigned.
Windows 11 will also have a better implementation of DirectStorage due to the new storage subsystem. Windows 11 will be better the gaming platform.
Last edited:
Soldato
Soldato
Ι don't know how 10 behaves, but turning ecores off with windows 11 is just....well let's call it weirdl The interaction between e and p cores is insanely nice, games offload non latency sensitive tasks to them (like spiderman and hitman 3), and it makes it seemless to game while using a heavy application on the background, it doesn't impact your gaming experience at all. Pretty crazy stuff without having to resot to process lasso
Last edited:
Soldato
- Joined
- 14 Aug 2018
- Posts
- 3,519
No wonder you have such a poor experience with Alderlake. Unless you are using a program like ProcessLasso you are likely to have a substantially subpar experience using Win10 with AL. It really does need to be used with Win11.
Also you are in error conflating IPC with latency. If I had time I would explain it but I'm sure a bit of googling should enlighten you.
Last edited:
Soldato
Latency is the time it takes me respond to a thread, IPC is how thorough my answers/posts are in that thread.
Last edited:
Associate
Alder Lake is a great chip the P cores that is. Hybrid arch I hate and cannot stand!! Best 8 core chip there is. Its a pity Intel has no 8 core only option without the e-crap cores.
OcUK Staff
Disable them.... Or use switcher....
OcUK Staff
Temperatures are off if your getting 95c with this result or cooling is not good or efficiency is poor..... What cooler no way should you get this temp with this score.95c benches coming in hot!
Associate
- Joined
- 28 Sep 2018
- Posts
- 2,422
A random guy in our group. Just an arctic ii 360 but his ambients are like 26c. Also not voltage tuned. It’s just meme’ing the 95c convo mainly.
Last edited:
OcUK Staff
Well 360 should defo be better....
Associate
- Joined
- 28 Sep 2018
- Posts
- 2,422
It’s always gonna be relative to ambients. Look at his mins. Mid 30’s. If he ran adaptive with offset he can shave a lot more but he likes his fixed voltage. Don’t ask why….
Associate
Still not a fan of hybrid arch at all regardless. But love the P cores Intel has.
Yes I disable them at BIOS level and treat it as an awesome 8 P core chip!! My 12700K that is!! The 8 cores in it match the 7700X 8 cores in IPC at same clock speed with lower latency and it costs less too!!
Associate
Should Gamers Disable E-Cores? Core i9-12900K, 41 Game Benchmark
Support us on Patreon: https://www.patreon.com/hardwareunboxedJoin us on Floatplane: https://www.floatplane.com/channel/HardwareUnboxedBuy relevant products ...
www.youtube.com
Per that benchmark e-cores ironically hurt performance in CPU intensive games like Far Cry 6 and CyberPunk 2077 and even Hitman 3 (Which actually has a thread that is optimized for e-cores).
But yet somehow performance is hurt with e-cores disabled in single or 2 threads at most games like CS GO and Starcraft 2 which defies all logic.
If there something else being affected on Alder Lake CPUs with e-cores on or off besides just the cores not being there. I mean single or 2 threads a most games should be way better with them off. Someone mentioned maybe L2 cache being less, but the L2 cache is only per core and if e-cores are off, it should have 0 to do with the P cores or any P core disabled L2 cache should not affect the other P cores. And L3 cache remains the same.
Associate
- Joined
- 28 Sep 2018
- Posts
- 2,422
Example on how much more efficient the new die is at heat transfer. This is on a loop. Starting temps are 28c and even pulling 373watts, cpu reaches only 74c
Because the heat tranfer from the die is so good, the strength of your cooling will continue to scale for a change.
Before anyone makes things up, no, it’s not running at stock by any means.
Soldato
thats insane, My KS would have been melting at this point, can't wait to play!!
Example on how much more efficient the new die is at heat transfer. This is on a loop. Starting temps are 28c and even pulling 373watts, cpu reaches only 74c
Because the heat tranfer from the die is so good, the strength of your cooling will continue to scale for a change.
Before anyone makes things up, no, it’s not running at stock by any means.
Now it's his room that's melting. High wattage + low temps = melting room
Last edited:
Associate
- Joined
- 28 Sep 2018
- Posts
- 2,422
I think I wrote a mini guide on the adl thread anwhile back but for adl and rpl, using a fixed core and llc is not the way to get the most out of the platform for daily.
You’ll want to use adaptive with CEP. Then apply small offsets until you see throttling from cep. That’ll give you a baseline from there you can play a bit more with using avx guardbands to reduce vcore for SSE. The guardbands act as voltage buffers for avx/avx2 workloads. You kinda get the best of both worlds this way. Finally you can try to find the sweet spot for your boards switching frequency.
Doing it right, you‘ll pick up 200-300mhz additional headroom over doing the traditional fixed vcore and llc method.
You’ll want to use adaptive with CEP. Then apply small offsets until you see throttling from cep. That’ll give you a baseline from there you can play a bit more with using avx guardbands to reduce vcore for SSE. The guardbands act as voltage buffers for avx/avx2 workloads. You kinda get the best of both worlds this way. Finally you can try to find the sweet spot for your boards switching frequency.
Doing it right, you‘ll pick up 200-300mhz additional headroom over doing the traditional fixed vcore and llc method.