Arrow Lake testing woes

[@an_easy_name_to_remember4749]

Appreciate your content. I’ve always OC’d intel over the years, direct die, etc… but never worked with AMD. Used your 9800x3d guide to learn the ropes on it and cannot express my gratitude. Thank you for sharing your passion here as it’s helped me. 👍☕️

[@fatihak7298]

Can you please tell me how to make these settings on biostar b650mp-e pro motherboard? processor 8600g. this is the most i could do! gpu:3000mhz mem:3200…

[@rasporilac]

One question I have x870e Gigabyte Aorus Pro, ryzen 7 9800x3d, 2x32gb ram 6000mhz, 1350w psu. The problem is that cpu clock is fluctuating in the games, it's not holding the max boost and it should, the cpu temp is 49-52c in the games and clock goes down to 4900mhz, 5000mhz etc. I tried bios update, high,ultimate power plan, installing chipset drivers, undervolting, using pbo, scalar, increasing temp limit even if cpu is not hot. It's windows 11, I tried core isolation on/off, in cinebench 24 it's holding the max boost at all cores without fluctuating. I tried to oc manual and then there is no drops but I don't want to use manual oc. Anyone know what this could be?

[@kolegabatric]

why don't you benchmark it? if the results are fine why bother? You could be clock stretching if you tinkered with voltages/clocks

[@Vitis-n2v]

Games don't load the cpu the same way a heavily multithreaded synthetic workload like cinebench does which makes the cpu not hold the full boost. Look at your gpu usage in the game you're testing. If it's maxed out at 99-100% your cpu has nothing more to do so it uses the extra time to clock down or sleep while waiting for gpu. Also high power profile keeps all cores at high clock but that actually hurts single core boosts very often so i personally don't even use high performance anymore.

[@WSS_the_OG]

Under even heavy gaming loads, particularly when GPU limited, it is expected that CPU clocks will fluctuate. Gaming loads are not flat or static, and I don't know of many games that fully load all cores of a CPU at 100%. CPU clocks vary based on loads vs core temps and Voltage.

[@docbrody]

For CPU testing using games, ALWAYS unbind the GPU first. set all graphics settings to the lowest settings with a resolution of 1080p (or less). Then run your CPU benchmarks and see where the clocks hold. Also, keep in mind that games don’t put the CPU under the same load as cinebench. At most they will only use 8 cores, but even that is very rare.

[@paperandpavement]

Im about to go to sleep to this, with all due respect 😊

[@nitrouspeed3583]

I am struggling with the exact same thing on my 245K. On the timespy leaderboard, norefills and myself nitrous have been trying to get anything dynamic to do what we tell it to do but it just doesn’t listen or puts 1.44V into the cores

[@Zurvv]

Looking forward to more Arrow Lake stuff. Maybe the fast ez OC for 285k? :)

[@SkatterBencher]

I'm getting the 245K guide ready now. It taught me a lot about ARL OC, so next would be the 285K.

[@yerachmielb1]

My very first overclocking experience of any kind was (looks at watch) 15 hours ago.

[@sgredsch]

i have a similar issue and i am pulling my hair out. i run a 265k on an ASRock z890 Livemixer and for the Love of god, i cannot get a uefi based dynamic undervolt/OC to work properly. using Intel xtu i know what voltage and frequency offsets the cores can do and it works properly, but whenever i change anything in the uefi about the CPU cores (core voltage or frequency) the DLVR goes crazy and blasts a 1,5v input voltage, even thought it should be around 1,3v due to the undervolt, totally ruining the efficiency gains and the temperature. that even happens when i overwrite other dlvr sourced voltages with a fixed value. right now, according to my experience, tuning for a daily setup with dynamic clock/voltage behavior only works with xtu and not through uefi. also, anyone had any luck overclocking 2x32gb ddr5 6000 cl30 hynix memory? im stuck on 6600MTs, anything higher barely posts. i didnt try blasting SA and memory Controller voltage, because its a Workstation and needs to be Long Term stable

[@docbrody]

Do you trust the voltages that HW Info is showing you?

[@SkatterBencher]

Sure. Trusting the readings is fine when you know what's being measured.

[@docbrody]

@SkatterBencher What I meant is: we all know HWinfo is not always reliable. Especially on newer hardware and especially on power… So perhaps that is worth considering when the shown voltages do not match what you know was set in bios. Just a thought.

[@FireCestina1200]

​@@docbrodyOne sure thing. Asus Mobo suck at reading voltage and is well known for that. I think that's even said on Hwininfo64 that if you have something from Asus some data could be false/or wrongfully named. Asus being Asus. I'm ngl, Asus is good at one thing: Making high-end board. The other stuff either suck. Have a higher failure rate (Like the Asus ROG Strix 4090 without fuse). Or worse price/performance

[@docbrody]

@@FireCestina1200 exactly.

[@Wasmachineman]

could be worse lol, I popped my 9590 by accident earlier today 💀

[@arturb5152]

This 120mm fan should be on the back plate . To lover CPU temp ...... Is it 1.4 Volts make a faster degradation Arrow lake CPUS ??