It’s up to AMD to explain why their CPUs do what they do; what we’re attempting to do in this piece is replicate the behavior and see if we can find a way to undervolt effectively.
The average core VID numbers stack up differently than the reported vcore. These next charts are all frequency plots. The most interesting part of these results, however, are the scores with a negative voltage offset: the Godlike board was stable with an offset as low as -0.1V using the 3900X, and gained 2% over the stock multithreaded score.
Once it get close too or surpasses 95c it starts reducing the speed of the processor to maintain the Max temperature at or under 95C.If you are getting such high temperature with a liquid CPU Cooler, then your CPU Cooler is not installed properly or is defective or the Processor Temperature sensor is defective.
I was only able to get “-50mv” and “-100mv” to stick, any other values would cause it to be reset to “Auto” or display an error if -100 was exceeded.A BIOS error message when going sub -100mv offsetThus I ran the -50mv offset three times in cinebench, and then moved onto a -100mv offset, which proved stable as well.Results from applying a vcore offsetThis may seem a little underwhelming, but I’m happy to accept this. At stock, the Godlike performed slightly better than the Master, and the best undervolt offset we could manage didn’t significantly change either board’s score.One-point-zero is a ridiculously low voltage to run for Ryzen 3000 CPUs, but both boards and both CPUs allowed us to set and boot at this voltage. If the CPU isn’t performing any actions at 3GHz, it won’t consume significant power. Undervolting your CPU can make your system run cooler, quieter, and more efficiently. Single-threaded scores were largely unaffected and actually decreased slightly on the Godlike board with the voltage offset, so the -.05 offset for the 3900X on the Master is the best combination we’ve tried.There’s an unholy number of charts that could be generated with the data we’ve gathered, so we’ll be focusing on the MSI Godlike board for most of this piece, just because the graphs for it are a little easier to read. Would highly recommend to first coat everything around CPU and GPU with nail polish or acrylic conformal coating , to not ruin your laptop , even tho it is highly unlikely , it is really cheap and more than worth to be safe , than sorry.
I saw that wattmeter was also reporting idle power usage (this is why we measure at idle!
AMD Ryzen™ 7 3750H Processors are equipped with Zen Core and the intelligent SenseMI Technology & Radeon™ RX Vega 10 Graphics.
It sounds like either the thermal paste hasn't been applied properly or the CPU is faulty.Water cooling should bring you lower temperatures than air.Also these chips usually shutdown at 100c so the fact it is getting to 102c indicates a problem.Can you do an RMA to try another one?Thank you for your reply so fastMy thoughts are it is to do with the CPU as with the stock fan and the watercooler came with their own thermal paste pre-applied. Zen 2 processors don't have the processor chiplets nor the I/O die in the middle.
Set the cTDP to 65 (or 65000 depending on the BIOS).Updating the cTDP value to a lower wattageBut make sure you have your sensors and benchmarks readied.
Just to be clear, it stays in the 60's and 50's for only the first minute and once the cooling has properly got going it then sits within the 40's whilst idling.get a new cooler that is suitable for AM4 which has much different requirements than older solutionsbest bet is my favorite, liquid nitrogen gets real results but humidity is a problem In multithreaded and single-threaded results for the R5 3600, it’s immediately obvious that applying 1V vcore tanked performance.