F5F Stay Refreshed Hardware Desktop Discussing 11980hk overclocking tips and seeking advice is valuable.

Discussing 11980hk overclocking tips and seeking advice is valuable.

Discussing 11980hk overclocking tips and seeking advice is valuable.

Pages (2): Previous 1 2
A
AlmightyEag
Posting Freak
785
12-20-2023, 03:46 PM
#11
Here are the latest improvements you've achieved so far from this chip, and some suggestions could be very useful while you upgrade your 10850k setup.
A
AlmightyEag
12-20-2023, 03:46 PM #11

Here are the latest improvements you've achieved so far from this chip, and some suggestions could be very useful while you upgrade your 10850k setup.

P
PERKSIE
Junior Member
48
12-26-2023, 12:10 PM
#12
It seems memory adjustments typically aren't possible through Windows unless the hardware supports it from the start. Training for memory performance usually comes with updates or changes in technology, not direct settings in the OS.
P
PERKSIE
12-26-2023, 12:10 PM #12

It seems memory adjustments typically aren't possible through Windows unless the hardware supports it from the start. Training for memory performance usually comes with updates or changes in technology, not direct settings in the OS.

J
JVMASTER11
Member
127
12-26-2023, 01:31 PM
#13
I've been getting more comfortable with Intel XTU and noticed a great capability for newer boards like this—voltages stay intact after a reboot. I'm using VCCSA to boost performance, but it only pushes me up to 3333. It seems the board isn't designed well for high-speed memory. After tweaking the settings as much as possible, I'm still seeing some latency and bandwidth issues, though it's not ideal. I'm mainly counting on increased IPC and extra cache per core to give me an edge over my 10850k, especially in power efficiency.
J
JVMASTER11
12-26-2023, 01:31 PM #13

I've been getting more comfortable with Intel XTU and noticed a great capability for newer boards like this—voltages stay intact after a reboot. I'm using VCCSA to boost performance, but it only pushes me up to 3333. It seems the board isn't designed well for high-speed memory. After tweaking the settings as much as possible, I'm still seeing some latency and bandwidth issues, though it's not ideal. I'm mainly counting on increased IPC and extra cache per core to give me an edge over my 10850k, especially in power efficiency.

R
Rafa
Member
58
12-28-2023, 08:39 AM
#14
The clock isn't fixed to specific timing values, so you can relocate it within the operating system. But because the timing parameters are critical and cannot be changed, you'll soon exceed acceptable limits. It's probable that 1333 isn't the true cutoff for the controller, but without direct control over the timings, it won't function beyond that point. Even with a training clock target set at boot, you might still push past it.
R
Rafa
12-28-2023, 08:39 AM #14

The clock isn't fixed to specific timing values, so you can relocate it within the operating system. But because the timing parameters are critical and cannot be changed, you'll soon exceed acceptable limits. It's probable that 1333 isn't the true cutoff for the controller, but without direct control over the timings, it won't function beyond that point. Even with a training clock target set at boot, you might still push past it.

H
holystefan
Junior Member
20
12-28-2023, 08:11 PM
#15
It's been a while since the last update, but there are some new developments and improvements. I've moved up to a BCLK of 101.75, achieving an effective mem clock of 3391. There are minor gains in bandwidth and latency, though I had to adjust multipliers which slightly impacted core and cache clocks. Performance remains within acceptable limits. I'm still working on increasing core speeds beyond 1.36v without throttling, and I believe optimizing a few dozen millivolts could significantly boost single-core performance on the top two cores.
H
holystefan
12-28-2023, 08:11 PM #15

It's been a while since the last update, but there are some new developments and improvements. I've moved up to a BCLK of 101.75, achieving an effective mem clock of 3391. There are minor gains in bandwidth and latency, though I had to adjust multipliers which slightly impacted core and cache clocks. Performance remains within acceptable limits. I'm still working on increasing core speeds beyond 1.36v without throttling, and I believe optimizing a few dozen millivolts could significantly boost single-core performance on the top two cores.

Pages (2): Previous 1 2