AMD Ryzen 5 4600 overclocked to 4.2 ghz @ 1.312v - does it matter?

seward

n00b
Joined
Mar 6, 2014
Messages
32
I just spent a couple of weeks taking a $99 cpu from 3.7ghz to 4.2ghz on a $129 motherboard (Asus Prime B550-PLUS) with some decent performance RAM (2x8GB Corsair Vengeance DDR4 3600). I haven't overclocked a new chip for over a decade, and I didn't put this build together to overclock. The priority was having a working home office computer for less than $400 (I already had some parts, including a very over-powered Antec 1200 psu). But I educated myself about the cpu and board and was happily surprised to find that they could do stuff, so I went into BIOS and did stuff, and ran Prime95, made adjustments, and did it again. I ended up at 4.2ghz with 1.312vcore (BIOS), with RAM running at DOCP settings and fclk at 1800 (temps are good, peaked at mid-70s; I have a DeepCool AG400 air cooler on the cpu). I started with setting cpu clock ratio to 40 with Auto volts, then went to 4.1ghz with 1.28v (manual), and eventually 4.2ghz at 1.312v, with 12+ hours of error-free Prime95, which is at least a good start in terms of stabilization and tweaking, the general goal of which will be to lower voltage. I've actually backed off from 4.2ghz to 4.1 ghz 24/7, primarily to keep volts low.

That sort of brings me to the other part of this post. I've been away from the OC scene for a while, and I was pretty uninformed with regard to what chips and boards do now. In particular, and specifically with regard to AMD, it sounds like cpus now have built-in performance-boosting functions like Performance Boost Overdrive which make overclocking sort of pointless. The general sentiment seems to be that these built-in or BIOS-driven boosters, when considered along with the relative fragility of 7 nm chips, makes turning up the volts and heat for "just a few more frames per second" not worth the cost to the chip's long-term performance and lifespan.

This is my first AMD build, and it looks like there's some stuff in BIOS, like SOC voltage which is perhaps similar? to old northbridge/southbridge settings as complements to cpu clocks and volts, which could be used to lower vcore. Those might be helpful if I want to go higher, and also to get cpu volts down at present clocks. I enjoy doing these things, and I tell myself that I am getting a few more quality frames per second..,but I haven't run any performance tests yet, like Cinebench, got tired of them a decade ago, but they're legit ways to measure performance and show it to others, so I'm gonna do them, at least to compare stock settings to this overclock. But I'm wondering if I should be prepared to see very little change or improvement? Is overclocking as a way of unlocking otherwise-inaccessible levels of performance in cpus no longer really needed, because of things like PBO?
 
It's actually worse than that, you can easily go backwards.

Amd chips in particular are designed to boost individual cores as required while maximizing performance. They can pick which core needs it and feed only that core more volts and clock and end up boosting higher than your overclock. When you all core oc it's possible to get less performance due to being thermally limited.

Gross simplification but unless you're doing some extreme cooling I think PBO won the overclocking wars
 
Overclocking still makes sense if you do the kind of work that uses all cores /threads but it usually comes at the cost of lowere single core boost
 
Thanks very much for replies! The things that I plan to do that will challenge this cpu are mostly gaming-related. I play old games (like StarCraft 2 and SupCom 1/2) and Factorio at 1080p (might go to 1440), and I love eye candy, so I like to play them maxed out. RTS-type games do benefit from cpu power, I think, but it may be at the cost of single-core performance here. It's funny, I just couldn't say No to that BIOS...but then I educated myself a little bit, and began to wonder if I was making a mistake. I got so tired of benchmarking years ago, but I think I have to go back there - at least use Cinebench - to get an objective sense of whether this overclock is worth it.
 
Does Ryzen Master work on the chip? I would do my work their and auto overclock it, it will know what is safe on voltage. I do not use PBO scaler

This is the chip doing work load



and the game clock is higher than work load, Ryzen Master is running the chip.

 
I just spent a couple of weeks taking a $99 cpu from 3.7ghz to 4.2ghz on a $129 motherboard (Asus Prime B550-PLUS) with some decent performance RAM (2x8GB Corsair Vengeance DDR4 3600). I haven't overclocked a new chip for over a decade, and I didn't put this build together to overclock. The priority was having a working home office computer for less than $400 (I already had some parts, including a very over-powered Antec 1200 psu). But I educated myself about the cpu and board and was happily surprised to find that they could do stuff, so I went into BIOS and did stuff, and ran Prime95, made adjustments, and did it again. I ended up at 4.2ghz with 1.312vcore (BIOS), with RAM running at DOCP settings and fclk at 1800 (temps are good, peaked at mid-70s; I have a DeepCool AG400 air cooler on the cpu). I started with setting cpu clock ratio to 40 with Auto volts, then went to 4.1ghz with 1.28v (manual), and eventually 4.2ghz at 1.312v, with 12+ hours of error-free Prime95, which is at least a good start in terms of stabilization and tweaking, the general goal of which will be to lower voltage. I've actually backed off from 4.2ghz to 4.1 ghz 24/7, primarily to keep volts low.

That sort of brings me to the other part of this post. I've been away from the OC scene for a while, and I was pretty uninformed with regard to what chips and boards do now. In particular, and specifically with regard to AMD, it sounds like cpus now have built-in performance-boosting functions like Performance Boost Overdrive which make overclocking sort of pointless. The general sentiment seems to be that these built-in or BIOS-driven boosters, when considered along with the relative fragility of 7 nm chips, makes turning up the volts and heat for "just a few more frames per second" not worth the cost to the chip's long-term performance and lifespan.

This is my first AMD build, and it looks like there's some stuff in BIOS, like SOC voltage which is perhaps similar? to old northbridge/southbridge settings as complements to cpu clocks and volts, which could be used to lower vcore. Those might be helpful if I want to go higher, and also to get cpu volts down at present clocks. I enjoy doing these things, and I tell myself that I am getting a few more quality frames per second..,but I haven't run any performance tests yet, like Cinebench, got tired of them a decade ago, but they're legit ways to measure performance and show it to others, so I'm gonna do them, at least to compare stock settings to this overclock. But I'm wondering if I should be prepared to see very little change or improvement? Is overclocking as a way of unlocking otherwise-inaccessible levels of performance in cpus no longer really needed, because of things like PBO?
Okay? :rolleyes:
 
I did a bunch of runs of Cinebench at stock and overclock settings...stock multi/single with PBO averaged higher than overclock. I've been reading a lot about PBO, so it wasn't traumatic. Back to stock, with Auto volts, unless they're too high.
 
I did a bunch of runs of Cinebench at stock and overclock settings...stock multi/single with PBO averaged higher than overclock. I've been reading a lot about PBO, so it wasn't traumatic. Back to stock, with Auto volts, unless they're too high.
Something to read up on is under volting.

Chips have a little extra volts by default to ensure stability, sometimes you can open up some thermal head room and get higher boost numbers by reducing the voltage. Ryzen master is the best way to do this. I'm not the expert, I just run stock because it's way easier, but I'm curious and read stuff.
 
Something to read up on is under volting.

Chips have a little extra volts by default to ensure stability, sometimes you can open up some thermal head room and get higher boost numbers by reducing the voltage. Ryzen master is the best way to do this. I'm not the expert, I just run stock because it's way easier, but I'm curious and read stuff.

This is what I'm trying to do now. Auto cpu volts at stock 3.7ghz were pretty high before I started changing settings (1.4+ at times, ridiculous). They've gone down a bit since I undid my overclock, though I have no idea why, but I'm glad that it's now generally under 1.4v. PBO seems to work, for my purposes, at least. I'd like it if someone would explain to me what the benefits of a static overclock versus PBO would be. Temps weren't an issue for me during tests, mid-70s C was the worst of it, but if there's literally no performance gain, just a vcore that doesn't go down...why bother (aside from the fun of it, I really did enjoy the process, surprised that there wasn't a pushback of some sort against PBO, maybe there was, been away for a bit). Anyway yeah, would be nice to lower volts with Offset, run some tests then...that's an excuse to use Prime95, that'll be fun...
 
If Ryzen Master allows +100 auto overclock on the chip, going ahead and apply and test, after it passed the test, you need to run Curve Optimizer test, it will decide what the cpu can do with your cooler.

It used a CO All Core Value -12 -offset on my cores and it decides the voltage needed.

Also it un- coupled my memory, but it is still running memory at 1800 and Fabric clock at 1800 at 16 -19 - 19 -19
 
Back
Top