Steam telepítése
belépés
|
nyelv
简体中文 (egyszerűsített kínai)
繁體中文 (hagyományos kínai)
日本語 (japán)
한국어 (koreai)
ไทย (thai)
Български (bolgár)
Čeština (cseh)
Dansk (dán)
Deutsch (német)
English (angol)
Español - España (spanyolországi spanyol)
Español - Latinoamérica (latin-amerikai spanyol)
Ελληνικά (görög)
Français (francia)
Italiano (olasz)
Bahasa Indonesia (indonéz)
Nederlands (holland)
Norsk (norvég)
Polski (lengyel)
Português (portugáliai portugál)
Português - Brasil (brazíliai portugál)
Română (román)
Русский (orosz)
Suomi (finn)
Svenska (svéd)
Türkçe (török)
Tiếng Việt (vietnámi)
Українська (ukrán)
Fordítási probléma jelentése
it can also be throttling due to cpu cooling (cpu cooler and thermal paste)
and may also be the way that specific cpu wants to run itself
they have multiple ccx and will run each differently depending on load, but it will do the best it can
forcing a higher overclock can force it to throttle sooner and lower than it would by default
I understand each CPU will perform differently and to its own limits, but my CPU under higher core count loads seem way, way below par. If I was at par and not getting above par results, yeah, I'd get it, but why is it so far below par? That signifies some awful limit somewhere, right?
if you are using the stock cooler do not try to overclock
for most ryzen cpus jsut let them do what they do
if you are comparing stock to oc performance, they will be different
if the case is getting warm, check the fans and airflow
If using the stock AMD Wraith cooler try using a negative offset voltage on vcore of 0.05 - 0.15 volts. Lower vcore means less heat.
Are you using PBO or auto settings for the CPU?
The FIT, otherwise known as the Silicon Fitness, usually has a limit based on lottery, which can also scale based on the amount of cores the chip has; a Ryzen 9 CPU usually has a safe limit of around 1.325v while Ryzen 5 may have a bit more leeway and be closer to 1.35v, but it's unique to the chip, part of the lottery.
There's also an issue with current, voltage overshooting, motherboards tricking software into thinking its using less or more voltage than is actually happening, AGESA versions with different generations, etc.
There are quite a few factors, but I myself have noticed that my 3900X can't get the same clocks at 1.35+ that it once could, but it seemingly stays stable for longer at ~1.3.
This is why my recommendation is to never exceed 1.325v when overclocking any CPU, and to be safe if you want your CPU to last closer to 10 years, finding the best clocks while locking voltage to ~1.325v is a good idea. Potential performance loss in the single core could be worth it, we don't know what will happen in 5+ years.
Or simply try Google your Motherboard model (such as MSI B450 Tomahawk Max) followed by words/terms such as: in-depth review -or- overclocking guide
Yeah, searching the internet is showing me that most people are getting ~4750-ish at stock (not overclock) on this CPU. There some in the 4600, one 4500, but I only found one lower than that, and it was someone who had even lower than I did at around 3900, but resetting the BIOS to default got them back up to normal scores. This includes people with B350 and B450 boards running stock, so I can't imagine my B550 or RAM should be dropping me nearly 20% of my score?
So, I tried that. There was an improvement, but something tells me I'm still way off base here.
Here's with everything set to default. RAM is running at 2,133 MHz (not sure about timings) and Infinity Fabric was like 1,200 MHz.
https://i.imgur.com/Ncd4kyD.png
So, I reset, and set D.O.C.P. with Infinity Fabric to 1,800 MHz and nothing else. Ram is running at 3,600 MHz (16-19-19). As in, back to where it was when I got my result posted in the OP.
https://i.imgur.com/DOHtXLA.png
So, uh, I guess my single core score stayed up but that's not my concern. From 2,133 MHz RAM/1200 FCLCK to 3,600 MHz RAM/1,800 FCLCK and I'm losing performance atop my already well below par results. Keep in mind it's not this one benchmark itself I'm really concerned with (another mimics this symptom), but the fact that it may be a symptom of less than normal all core load performance. My scores are basically between a Ryzen 5 3600X and Ryzen 3700X, so I guess I have a Ryzen 6 3650X here.
I'm starting to worry my board or more so RAM choice was bad. Considering I spent the most on RAM (because of quantity), that's disappointing. I knew I wasn't getting top grade stuff, but surely 3600 MHz CL 16 stuff shouldn't be getting me atrociously below par results?
Might need to do more manual exploring. Maybe there's a setting on this board/BIOS I need to set.
If the first # is say 16, but the rest are higher... such as say, 16-18-18
Then this could be why.
Review this in CPUZ
You can always try something like 3200mhz and see if you can get the timings lower and tighter.
https://imgur.com/a/64i3ZhK
4400 ish in CB R20 is not that far off I think. Could be some bloatware running in the background.
Here's a CPU-Z validation link though, in case I'm really overlooking something obvious.
https://valid.x86.fr/8gnpwi
It's always been hit or miss for Ryzen desktop CPUs to work with 4 RAM sticks. There is definitely something funny happening there. For example, FCLK clock halved on its own.
Lower your RAM speed to 3000~3200, or just take out two of the sticks, see what happens.