First Overclock- AMD FX 4100

[RGone]

I doubt your current power supply is 'too' weak just yet.

RGone...

 

[bassnut]

I think I would go for the power supply first. Build a good backbone for your system.

 

[KookieB]

I installed the new SSD one day last week. I've been wanting to start overclocking my system for a while now but am a little overwhelmed by all the information I've been reading recently. Today I read this guide but it may be outdated, I also read this guide, and combed through this and watched this video
The ultimate guide I read says to find the Max FSB, then max memory FSB, then Max clock speed.
Inside the BIOS I was able to disable all the useless features easily, but am unsure about what to do next. which settings other than those things to change and which ones to leave alone and in what order. Below is a bunch of the options i've found inside the BIOS that I know I need to understand to tweak and find a good overclock. I'm looking to hopefully gain some insight on what some of these settings should be set to to start and in what order I should be increasing them, which ones to avoid changing, and if any of the stuff i've read is outdated and no longer helpful please let me know.

processor max frequency:6300Mhz
NB Max Frequency:6200Mhz
Processor Max Voltage:1.5500V

Options available and corresponding defaults:
CPU Frequency Multipier x18.0 3600Mhz
CPU Voltage 1.3250V
NB Frequency Multiplier x10.0 2000Mhz
CPU NB Voltage 1.2250 V
HT Bus Speed Auto (Min. 200Mhz Max. 2400Mhz)

DRAM Timing Control

Voltage Control (these can be set to auto or manual but manual shows no default voltages only min and max)
DRAM Voltage (Min. 1.250V Max. 2.065V)
NB Voltage (Min. 1.110 V Max. 1.400V)
HT Voltage(Min. 1.210V Max. 1.400V)
CPU VDDA Voltage (Min. 2.56V Max. 2.70V)
PCIE VDDA Voltage (Min. 1.81V Max 1.92V)

 

[xrror]

I just looked up his power supply for jollies...

and...

Antec wasn't messing around with the 550BP apparently:

BP550 Plus 550 Delta Electronics Yes 1 22 22 25 - 384 W 3 80 Plus certif. pending

You can't really get much better than Delta for an OEM.

edit add: And now I realize that my tab for this thread was stale, and this question was answered 14 days ago. d'oh

 

[KookieB]

This post will be to clear up my last post so it has more direct questions. My BIOS gives me options for increasing the Multi of both CPU Frequency and NB Frequency. I also have an option for setting HT Bus Speed. My first bit of confusion is about the multipliers. The CPU multi is set at x18 default, the NB multi is at x10 default.
Should I be increasing both of these multipliers? Should I be increasing them at the same rate? or one at a time until I run into instability?

The second bit of confusion was with the HT Bus speed. Should I be increasing the Multiplier until I run into instability, back off, and then raise the bus speed?

Am I correct that I should raise voltages last and again only if instability occurs after raising the other two to their stable limits?

 

[Johan45]

Start with only the CPU multiplier, raise it up and test until you find instability or the temps get out of hand. 62c for the package and 72c for the socket. If it gets unstable and the temps are still in line then add a bit of VCore voltage and test again. The HT and NB busses should be left at stock for now. Have you added any cooling to the VRM?? That board has no heatsinks.

 

[xrror]

Yea Johan beat me to it. Be sure to take notes on what voltages you needed to stabilize the various speeds you hit (via cpu multiplier)... that'll come in useful later for the next phase =)

I think the "challenge/fun" will be when you get to finding out what your memory will do. But that's after you kick the tires on your cpu speed first =)

 

[KookieB]

Thanks guys I appreciate the help. I have not added cooling to the VRM's yet. I have been reading up about the different techniques though. I'll definitely be adding a 60mm* fan to blow directly on the vrms, do you think it would be a good idea to try and install heat sinks something like this on the vrm's or stick with just the fan to start.

 

[Johan45]

Star with just the fan, I wouldn't push that board too far any way. It's just not built for speed.

 

[KookieB]

Thanks. Yeah I wasn't really planning on getting any kind of a great overclock on this machine but I would like to be familiar with the mechanics of it before I build my next machine.

 

[Johan45]

Doesn't hurt to start somewhere just be mindful of the temps and if the board starts throttling your CPU or something like that then back off on your overclock.

 

[KookieB]

Be sure to take notes on what voltages you needed to stabilize the various speeds you hit (via cpu multiplier)... that'll come in useful later for the next phase =)

I think the "challenge/fun" will be when you get to finding out what your memory will do. But that's after you kick the tires on your cpu speed first =)

I already have a paper journal started for this PC and oc.

 

[KookieB]

I'll be making a stop at microcenter someday after work this week to pick up two 40mm fans to place over the vrm's on the Mobo.

I've had two stable overclocks so far and am testing a third now.
The first overclock was tested with prime 95 blend for 12 hours. the temperatures reported are from hwinfo64 and were started at the same time as the test to avoid throwing off avg temps.
CPU Voltage: 1.3250v(stock)
Multiplier: X20.5 at 4100 MHz
Avg package temp: 30.1 degrees
Max package temp: 35.1 degrees

The second overclock was tested for 2 hours with prime 95 blend, I will be testing it again this week for 12 hours to ensure stability. the temperatures reported are from hwinfo64 started at the same time as the test to avoid throwing off avg temps.
CPU Voltage: 1.3375v
Multiplier: x21.5 at 4300 MHz
Avg package temp: 32.1 degrees
Max package temp: 33.9 degrees

 

[Johan45]

Looks good so far if those temps are accurate you have quite a bit of headroom still. I wouldn't bother testing for 12 hours until you've come to your final OC

 

[KookieB]

I was unable to find a stable clock at 1.35V. I tried X23.0 at 4600MHz which failed p95 in the first few minutes, I lowered the multi to X22.5 at 4500MHz and had similar results, at X22.0 4400MHz Prime 95 ran fine with no temp spikes for 20 minutes. I ran it again overnight for 8 hours and when I returned one of the cores had error-ed out during the test. Should I just forget about 1.35v and try the next voltage setting, or try and adjust the bus speed a small amount to find a good stable clock at 1.35v?

 

[Johan45]

You do have room to increases the voltage. Most likely another bump and the 4.4 OC would pass. Then it's just rinse and repeat. If you need 1.4 for 4.5 that's just what it is. Keep an eye on temps and monitor the CPU usage as well to see if it's throttling.

 

[KookieB]

Thanks Johan, I'll keep working at it then over the next week or two and see what i'm able to find. I will update this as I continue to find more stable speeds.

 

[Johan45]

You're on the right track and small steps are better than big ones when it comes to overclocking.

 

[KookieB]

You're on the right track and small steps are better than big ones when it comes to overclocking.

Because science is exact, reminds me of a chemistry teacher I had my freshman year of college that encouraged students to double round numbers. Meaning if you have 4.447, it would be rounded to 4.45 then rounded again to 4.5. I dropped the class after the first day because as any scientist or mathematician will tell you you should never round more than once so as to preserve accuracy. 4.447 rounded to one decimal place is 4.4 not 4.5.

 

[Johan45]

Good analogy. It just helps keep the voltages lower than higher in most cases