The bumps appear to be coming from the little metal tensioning tabs pressing against the back of the screen through the copper heat transfer tape a little too hard.
https://youtu.be/xEBYVJ2_r_Q?t=103
https://youtu.be/0poyvpB7Fkw?t=193
iPhone 15 has these tabs but only on the reinforced edges of the frame and Samsung doesn’t use them at all for Galaxy S series
Bizarre design choice from Google to have the tabs press directly into the back of the screen…
Use these settings on top of XMP
SVID Behavior = Typical
Digi+ LLC = 4
TX VDDQ = 1.35v
IMC VDD = 1.35v
Strix Z790-H should be cheaper than Strix Z790-A
You should check if you actually need the A/E features over H.
If you carried over the Windows install from another chipset motherboard, you should reinstall the Chipset INF utility and also reset your Windows power plans.
The board was more insulated than usual from the actual company because of the weird corporate structure of OpenAI and some members were uniquely unqualified.
Follow these instructions for ASUS
Skip the TVB part since you’re trying to reduce power, but everything else applies.
I recommend using Process Lasso proper. It’s not hard to learn and much more powerful. CoreDirector looks like Process Lasso’s Efficiency mode/CPU set/CPU affinity rules repackaged into one on/off toggle for every listed app.
You can do more interesting rules with Process Lasso like keeping non-interactive applications off of the highest-turbo P-cores or restricted to E-cores to improve idle power. My 60/60/60/60/57/57/57/57 13900K idles at 8-9W with MSI AB, Discord*, Steam, and other apps banished to E-cores and never allowed to turbo the P-cores.
*Discord gets 57x P-cores 6 and 7 as a treat.
Disabling HT usually gives another 100-200MHz at the same load line just because the core pulls less power and runs cooler.
Try 5.8/5.6/5.5 with OCTVB break points at 60C and 70C.
Sadly, no other details were shared at this time on these upcoming CopprLink PCIe cables. The name is arguably a bit odd and separately there is already CopperLink Ethernet cables from a company Patton Electronics. CopprLink hasn’t appeared yet on the USPTO trademark search for reavealing its styling or other details.
…
As for Metro Exodus, it’s more like:
When game devs get too ambitious and spawns SystemInfo.processorCount() workers for a 32-thread battleroyale in the L3$
APO still slightly outperforms E-cores off and HT off in this game, so I’m assuming APO does something useful with the 16MB of L2$ to take pressure of the L3$
I recently got one of these as a curiosity when their Amazon storefront had a coupon. Mildly impressed at how ‘quiet’ they are for the 2400 RPM rating. Still wouldn’t ever run them at 2400 RPM but I’d rather these at 2400 RPM than the NF-A14 iPPC at 2000 RPM if I had to run a 140mm fan that fast.
More watts = more amps = more voltage drop from LLC AND higher voltage requirement from increased temperature.
You need to increase the AC load line configuration to counter the increased voltage drop, but first consider whether an additional 2-3% all-core performance is worth the 20-30% higher power required.
Thread Director doesn’t do any directing, it’s a a set of new registers the OS scheduler is supposed to read for feedback on how well a thread is running on a core. If APO can do it right, it means the scheduler is wrong.
15.6 HARDWARE FEEDBACK INTERFACE AND INTEL® THREAD DIRECTOR
Intel processors that enumerate CPUID.06H.0H:EAX.HW_FEEDBACK[bit 19] as 1 support Hardware Feedback Interface (HFI). Hardware provides guidance to the Operating System (OS) scheduler to perform optimal workload scheduling through a hardware feedback interface structure in memory.
Components literally popping off of the M2 PCB is absurd. A lot of the M2 sticks do flex a little-to-alot when installed under a heatsink but WD should’ve tested for this.
Reset all of your motherboard settings, all of them. Use the clear CMOS if possible.
Set the following
XMP enabled
Load Line Calibration 7
MSI Lite Load 6
and then try again.
I’ve had so many issues today in regards to my bios and having constant bsods from the settings
Yeah that’s what happens when you set a static voltage unless it’s real high. Don’t do that for these CPUs.
HWInfo is saying the same as CPU-Z. You’re looking at the XMP profile box which is what’s baked on the SPD, not what’s running.
Your XMP boot is likely failing and ASRock bios booted it in the slowest possible configuration.
Set these voltages manually up to the max in parenthesis
VCCSA 1.3V (max 1.35V)
VDD_CPU 1.35V (max 1.4V) - this should be CPU VDD2, verify in HWInfo
VDD_IMC 1.35V (max 1.4V) - this should be VDDQ_TX, verify in HWInfo
and then try booting 5200/5000/4800. You then want to test the IMC using y-cruncher stress test on the “VST” test after you successfully boot. You only need to run it for 10 minutes. If it fails, drop the clocks by 200.
Alternatively if you can live with 96GB, 2x48GB is much more likely to work out of the box.
If you don’t want to experiment with timings, the shortest way is to just drop the speed by 200MHz increments until Y-cruncher VST or HNT no longer errors out within 10 minutes. Even minor instability should blow up VST or HNT in a couple of 2-min cycles and a 200MHz drop is enough to fix it.
Ryzen division’s issues has to do with the gigantic inventory pileup that started in mid-2022 and forced them to undership the market for almost a year. It’s bound to make shipment data reported by these research firms look funky.
P3, P5, P7 are all on one side of the CPU, so it’s likely the cooler pressure on the CPU IHS isn’t even for some reason.
Some ideas:
Use more paste and make sure the paste imprint is even.
Double check to make sure you didn’t mix the LGA1200/1700 standoffs which are different lengths.
Make sure the excessively large universal Arctic mounting frame isn’t catching on a VRM component.
Make sure the contact frame tension is even across all four screws