Haven't overclocked anything for about 10 years and just upgraded to a Ryzen 9 5950X 16C/32T CPU on an X570 Gigabyte Aorus Master motherboard with 32GB G.Skill 3600MHz CL16 RAM. Am wondering whether it's worth the time to re-learn how to overclock for any gains or whether anyone have any quick pointers on how to overclock AMD CPUs (if anyone here bothers to overclock their systems )?!

 

I refuse to overclock anything. If something isn't running fast enough I prefer to upgrade. I hate having things burn out before their time. (That use to happen all the time when I overclocked)

 

My 10700k Is running fine at stock levels using about 20 % all cores for my projects no need to over clock at all.

 

I wouldn't bother since is a beast of a CPU. This also means you'd have to use a powerful cooler. Maybe in 5-7 years if is worth it...
One thing that you can tweak and gain quite something is the RAM settings. Ryzens have always had it harder to make the most of the RAM than Intels. AMD with the Ryzen 7 5K offers a pretty good tool to tweak it.
Also, without going pure OCing (within the clock, voltage limits), if you can tweak to manage that the CPU at 100% doesn't lower much from their max turbo is almost like an OCing.

 

nope.

 

Yep, ASUS motherboards utilize one-click overclocking which makes it super simple. My bios reports that my i9-10900k system runs overclocked at 43%. No crashes. No issues.

 

Yes I do overclock for audio. Why not? It's a supersimple process nowadays, you don't have to mess with jumpers on the mobo like in the old days and the safety thresholds are supertight on AMD and intel as well, it is really hard to burn things even on purpose, you just can't, the mobo shuts down itself. I got my 3950x at 4.3 every core all day long on liquid for 5 years in row now.

 

i have done quite a few times , mainly a cross between not being able to upgrade, squeezing every last drop from whats essentially cheap in the pc world, so if by any freek accident i ruin anything in the process its not a major thing

 

What is the easiest way to overclock nowadays? Does it still have to be BIOS fiddling or are there in-Windows tools that make it easy?

I have a:

• AMD Ryzen 9 5950X
• Gigabyte X570 Aorus Master Motherboard
• 32GB (4 x 8GB) G.Skill TridenZ CL14 3600MHz RAM
• MSI GTX1070 8GB Armor Edition GPU

....just in case there's any tools specific to the hardware, e.g. Gigabyte tools, that make it easy and fool-proof

 

@buns, in my experience the OC software is in the mobo bios. The fancier mobos have one click oc like another person mentioned above.

I do oc with unlocked cpus that are advertised as oc'able. I figure that's why I paid extra for the custom clock ability. Never had issues but I'm sure some people do. If you start crashing it's easy to undo or reset to default. Not fun to crash infront a client though

 

Last edited: May 16, 2023

Be sensible:

Bottom line: overclocking your CPU or graphics card by hand has more risks than benefits. You will lose the warranty, make your system more unstable and a stronger heat generation often entails a higher noise level due to noisy fans. The benefit, on the other hand, is a modest performance increase in the single-digit or low double-digit percentage range, which rarely has a noticeable impact in practice.

 

I decided to run at a constant clock speed of 4.8 on CCD1 (8 cores) and 4.65 (stable) on CCD2 (other 8cores) on my 5950x.
Tried all the tools provided by AMD/ASUS in the bios but a fixed speed has fixed all my problems.
Was a heavy Acustica user, no so much today, but everything runs smoothly with great temps and huge sessions are a breeze.
Depends entirely on the quality of silicon of your chip ! Some can go higher, some much lower.

Next build : No overclocking but a stable over 5Ghz clock speed. But I'm good for many years with the 5950x.

 

Nah, audio doesn't have FPS bragging rights anyway.

Personally: why spend the time for marginal gain for my computer usage? why risk even slight instability? If anything I'd rather underclock if it means less temps (and therefore less noise).

We've pretty much hit diminishing returns with OC already.

 

Also, why can't my 24core 5ghz CPU with 64gb 5hz ram not produce low latency like a dual core UAD SHARC dsp chip?

 

Because DSP chips are purpose built to provide fast calculations and fast calculations only.

 

And folks, don't read too much into the DPC numbers. It can basically tell you just one thing: if there's an offending driver on your system.

It can't be used to qualify hardware quality in isolation. It's just a number, but if your Wifi driver is causing higher DPC latency than your GPU then it tells you that your Wifi (or rather, the Wifi drivers) is most probably at fault. That the GPU driver is causing highest latency is normal operation as expected.

In short: it's a debugging tool, not a benchmarking tool. For instance, on my system, running on an NVidia discrete card, the GPU driver not unexpectedly causes most DPC events, but no dropouts on audio at 128 samples buffer with decent load.

 

Understood, thanks. Still I wish chip architects could dedicate some of my 24 physical cores for optional specific processes only like audio/video processing. God knows I don't need all 24 for phub

Some SOC(sytems on chips) used for phones and edge ai devices use a dedicated second chip solely for encryption. Would love a second chip for dsp only in windows. Certainly a business opportunity here for anyone with a microprocessor factory