I forgot how much of a difference it can make

GhengisKhan

Supreme [H]ardness
Joined
May 16, 2005
Messages
7,776
I have a system that I've had since 2012. AMD FX6300 @ 3.5. I never OCd it, so I just left the stock HSF on it. Well, yesterday I decided to see what I could get out of it. I started OCing it, and ramped it up to 3.8Ghz @ 1.2v. Idle was at about 36°C. When I put it through a stress test last night the temps climbed up to and leveled out at 59°C after about 15 min. This morning I went to Microcenter and picked up a CM 212 Evo that they're running a good sale on, and some AS ceramique 2. After installing it the idle temp was about 22°C, and the load temp after 20min only raised up to 29°C max. An extremely good improvement. Color me impressed. Over the next few days, I can't wait to see how far I can push this thing.
I know it's not as good of a setup as a lot of you guys have, but I am happy with it now.
 
Nice. But take a 3770k and put 1.4v through it and let me know how that holds up.
 
that 212evo on your 6300 should get you to 4.5 at 1.45-1.48ish volts and should run around 50c at full load. that chip WILL start to wig-out when it hits 65c so anything under 60 is safe. I was able to do 4.5 with my 212evo on two different 4100s and my current 8120. also, adding a second fan for push/pull will knock off another 3-5C. anything more than 4.5/4.6 and 1.48-1.5 and it will heat-soak pretty quick though so that's about the limit with that cooler/chip combo.

Nice. But take a 3770k and put 1.4v through it and let me know how that holds up.
yeah that's a whole different ball game but not really relevant to his post...
 
that 212evo on your 6300 should get you to 4.5 at 1.45-1.48ish volts and should run around 50c at full load. that chip WILL start to wig-out when it hits 65c so anything under 60 is safe. I was able to do 4.5 with my 212evo on two different 4100s and my current 8120. also, adding a second fan for push/pull will knock off another 3-5C. anything more than 4.5/4.6 and 1.48-1.5 and it will heat-soak pretty quick though so that's about the limit with that cooler/chip combo.

yeah that's a whole different ball game but not really relevant to his post...
Yea, the first thing I did when installing it was to put another 120mm fan on it, with both of them blowing up into the exhaust fans on the top of my case. I'll be very happy if I can reach 4.5
I work evening shift though, so it'll have to wait until I can get home tonight to start pushing it.
 
I would be VERY surprised if you cant do 4.5 with that setup. bump you voltage a bit, make sure your vrms are cooled well and set LLC to med/high if you have it and you should be there. even if your mobo is not an "OC" mobo its doable. the second 4100 I mentioned was a system I built for a friend back in '12 based on a GB 970a-d3, not an OC board in any way. the psu died and while I was replacing that I tossed in the 212evo and had no prob getting it to 4.5 @ 1.488v. it was rock solid and after several hours of realbench stress test it topped out at ~53c.
 
Alright, I figured out that the CPU temp reporting within Linux wasn't working right. I figured out how to get it straightened out. Now that I've gotten a bit of an OC out of it (4.1Ghz @ 1.025v), it's running at about 39-40° idle and 65-67° with all 6 cores running at 100% for 15 minutes now (will run it for at least 20 min, just to be sure it doesn't go any higher). I am slowly increasing the multiplier to see how far I can push it @ 1.025v before it starts giving me problems.

*EDIT*
Gotta say, I'm pretty damn impressed by this 212 HSF. I shut off the stress test, and within about 45 seconds the temp dropped from 67° down to 44°
 
Last edited:
67/44 F or C? if F that temp is still not correct. amd fx chips have an issue with the temp not showing correctly until above ~25C and is typically 10-12C higher than it shows. so that 67(19c) is probably about 86(~30C). so it is most likely higher than it appears BUT you still have tonnes of temp room, up to 140(60C). not to mention 44 is ~7C...

if those temps are in C then something is wrong and the temps are way too high. that chip should be able to do 4.1GHz(that's its boost speed) @ stock voltage with a stock cooler.
 
67/44 F or C? if F that temp is still not correct. amd fx chips have an issue with the temp not showing correctly until above ~25C and is typically 10-12C higher than it shows. so that 67(19c) is probably about 86(~30C). so it is most likely higher than it appears BUT you still have tonnes of temp room, up to 140(60C). not to mention 44 is ~7C...

if those temps are in C then something is wrong and the temps are way too high. that chip should be able to do 4.1GHz(that's its boost speed) @ stock voltage with a stock cooler.
The temps are in °C. They're higher than what I want them to be when I'm done with the OC.
I researched, and I found that with this 6300 chip, that the max long-term safe high temps are around 70°C. I raised the voltage up first until I reached a high temp I am comfortable with running at as a top stable.

While during normal usage at stock speeds/volts I never got near 100% usage out of all 6 cores simultaneously. At most one will be at around 100%, while the others are running somewhere between 40% and 50%.

I'm doing a full load testing using a program called "Stress" which can run all 6 cores at 100% usage with the workloads of 12 cores.

I will now be increasing the multiplier by .5 each time for a stress test of at least 20 minutes.

I am taking my sweet time doing this OC to squeeze every last little bit I can get out of this thing before upgrading my video card from the current GTX 460 that's in it now to something that will play games more modern that CS:GO @ 90fps with all options at max.

I hope this explains why the temps are so high while I'm going through this process.

*EDIT*
BTW, it's currently at 4.1 @ 1.025v. My OP was wrong, it's 1.025v not 1.2v.
 
Last edited:
ok C then. you said youre at 4.1 with 1.025v with 67C max, is that correct? 'cause that is really high then but you are right aboot 70 being max so like I said shoot for ~60c max(you want the breathing room, trust me!). so if your temps are really that high at that speed/voltage, I would double check your mounting/paste/tightness. you should be able to just go straight to 4.4/4.5 @ 1.4-1.45ish volts(llc med/high) and get high 50s(22C ambient). then you can tweak from there. that is a pretty normal for all the FX chips(I have personally don't it on several of each 4xxx, 6xxx and 8xxx chip). as far as testing goes, I use AMD overdrive's stability test for a couple hours(the cpu status page shows C to TJMax which is handy) then real bench for a couple hours(tests cpu and gpu) then game. I personally don't use prime95 but if you do decide to make sure you get version 26.6 or older to avoid the problem with the avx overworking/heating the cpu on newer versions.

ps: in the bios, make sure you have turned off boost mode so it stays at the speed you set, ALL power savings and HEPT. good luck!
 
ok C then. you said youre at 4.1 with 1.025v with 67C max, is that correct? 'cause that is really high then but you are right aboot 70 being max so like I said shoot for ~60c max(you want the breathing room, trust me!). so if your temps are really that high at that speed/voltage, I would double check your mounting/paste/tightness. you should be able to just go straight to 4.4/4.5 @ 1.4-1.45ish volts(llc med/high) and get high 50s(22C ambient). then you can tweak from there. that is a pretty normal for all the FX chips(I have personally don't it on several of each 4xxx, 6xxx and 8xxx chip). as far as testing goes, I use AMD overdrive's stability test for a couple hours(the cpu status page shows C to TJMax which is handy) then real bench for a couple hours(tests cpu and gpu) then game. I personally don't use prime95 but if you do decide to make sure you get version 26.6 or older to avoid the problem with the avx overworking/heating the cpu on newer versions.

ps: in the bios, make sure you have turned off boost mode so it stays at the speed you set, ALL power savings and HEPT. good luck!
Once I get a chance I'll double check the mounting. I overslept this morning, and have a few things I have to get done first, lol.
Strange thing is, I set the BIOS to 1.025v, and I just noticed that it's showing 1.6v in the reporting software. If this is true, it could explain the high temps I'm seeing.
Also, I can't use the programs you suggested. I'm running Mint Linux. I use I-Nex / CPU, instead of CPUID. Psensor to chart the temps over time. System Monitor to display the stress levels (for CPU, RAM, or HDD testing). And instead of Prome95 I use a Terminal based program called Stress that puts 100% load on all 6 CPUs.

Here's a screenshot of it all running at idle.
Testing.jpg
 
oh right I forgot the Linux part....
and yes the 1.6v is the high temp issue. make sure that all the voltages are manually set and your llc is set to med/high. you need to get the voltage down, even idling that is way too much for an air cooler. you are doing it all via bios not a software tool right? what mobo is it?
 
Well. I switched from I-Nex over to a program, appropriately called HardInfo. And it's giving me quite a bit more info.
sensors.jpg

I reseated the CPU cooler, and the idle temps dropped to 37C. I started stress testing it again, and previously within about 2 min the temp would go up to at least 60C. I've been running the stress test now for about 15 min and the temp is now sitting at 46C and it hasn't gone up in the last 3-4 minutes. You were definitely right about that.

As for the voltages, I'm not sure which sensor to look at on this reporting program. In BIOS it's set at 1.175V, but the closest one to that is in5 @ 1.272V. And I know the +3.3V = 1.664V isn't the real reading of my PSU output. BIOS shows that at about 3.4V.

I've got to go to work now, so I'll get back to plugging away at it once I get home tonight.
 
it looks like those are, from top to bottom, nb temp, cpu package temp and cpu core temp. typically the easiest way to distinguish the temps is start a stress test and the one that is the highest will be cpu pack and the one that is ~10C lower is the cpu core. then the other one that usually stays the most constant is the nb as it doesn't change temp much. that newest pic doesn't look like any of the voltages are correct. it says both your 3.3v and batt are at ~1.6v which would cause all sorts of issues. the 1.59v looks closest to what I-nex showed above. I would stick to I-nex as it was showing the correct stock and oc voltages.
 
it looks like those are, from top to bottom, nb temp, cpu package temp and cpu core temp. typically the easiest way to distinguish the temps is start a stress test and the one that is the highest will be cpu pack and the one that is ~10C lower is the cpu core. then the other one that usually stays the most constant is the nb as it doesn't change temp much. that newest pic doesn't look like any of the voltages are correct. it says both your 3.3v and batt are at ~1.6v which would cause all sorts of issues. the 1.59v looks closest to what I-nex showed above. I would stick to I-nex as it was showing the correct stock and oc voltages.
I-Nex wasn't showing the correct voltage though. It was showing the same thing as what the 3.3V line says in HardInfo (at one point I had them both running at the same time), but wasn't showing any of the other voltages, and there is a whole lot more types of info, and logging of everything.
 
ok then just use I-nex to check the cpu voltage(it does look correct, but still high unless you lowered it) before you start hardinfo then you should be able to match the closest value.
 
While messing with the OC a little bit this morning, I raised and lowered the voltage a few times. One of those times required a BIOS reset. Just out of curiosity I checked I-Nex (before switching to HardInfo), and the voltage still reported at 1.6V
 
try just setting the cpu voltage to default, 'cause like I said the chips normal boost is to 4.1 so it should be able to do that that at stock voltage. see what the bios reports under health menu and see what the program says. I'm hoping its mis-reading. just don't do any heavy stressing until you get it below 1.5v or you risk burning up that chip.

edit: in the bios set the voltage to manual but leave it at "normal"
 
I did have everything set back to "normal", because I did a BIOS reset. That put everything back to defaults, which is .920V, 200Mhz FSB, and multiplier at 17.5. Both programs showed the voltage at 1.66V still, but the idle temp was down to about 26-28C, so I'm certain that the reported voltage is wrong. I'm still trying to figure out how to get it to read the correct voltage, but I know what it's set at in BIOS so I'm not really worried about it anymore.
With the HSF working so much better now, I'm going to try and push it a bit further.
 
yeah those voltages seems really odd. the default should be 1.25ish. BUT if you can get it stable and cool, great! also some ppl have found, myself included, that using a higher fsb will get you lower voltgaes/temps. 240 fsb is common and keeps it evenly devisable to get the ht, nb and ram at the correct speeds. that's why I use 20x240 instead of 24x200. good luck and let me know how it goes!
 
OK, I think I figured out something I had wrong. I had changed the name in Psensor, from temp2 to CPU, as this was the one with the highest temperature. I'm pretty sure I had the wrong temp labeled as being the CPU. I did some searching online and found that if you click on each sensor and go to Properties, that there was some info about that sensor. I clicked on the one I had labeled as CPU and it says "Chip: f71889ed". I checked the other sensors, and #1 through 3 all say the same thing. On the other hand, temp4 says "AMD CPU", I am trying to figure out what the ones with "Chip: f71889ed" are. I think it might be my NB, but I'm still looking.

I tried an OC up to 235 FSB, 18.5 multiplier (which means I was running at 4.347GHz, CPU voltage @ 1.175v. With it cranked up that high, the temp4 (one I now think is the CPU) went from 15C at idle, up to only 34C at 100% load after 15 minutes.

It's the one labeled temp2 that is running 41C at idle, and going up to 71C at full load (again 15 min).

Should I be worried about the other component that is getting up to 71C?
 
that 1.174v still seems strangely low it should be at least ~1.22v
the 71C is most likely the cpu socket temp and 80C is the absolute max. 70ish is a little warm but still safe and would put your cpu core at aboot 60C. remember that you want a 10C safety zone, absolute max is 70 core and 80 socket and it sounds like your hittin 60/70. my chips max is 54/75 so right now at idle my core is ~30 and package is at ~40 with my room temp at 21.5C. full load gets me ~55/65. I hope that was clear enough and helps. basically I think youre on the right track just need to dial it in. seems like your gonna top out around 4.4/4.5 which is aboot right. try 18.5x240 that was a sweet spot for my FX4100.

ps: this is my sensor info it looks similar to yours but comes from speed fan

IT8720F
temp1 = general mobo temp
temp2 = NB
temp3 = cpu socket

then my cpu core is listed as AMD K15
 
that 1.174v still seems strangely low it should be at least ~1.22v
the 71C is most likely the cpu socket temp and 80C is the absolute max. 70ish is a little warm but still safe and would put your cpu core at aboot 60C. remember that you want a 10C safety zone, absolute max is 70 core and 80 socket and it sounds like your hittin 60/70. my chips max is 54/75 so right now at idle my core is ~30 and package is at ~40 with my room temp at 21.5C. full load gets me ~55/65. I hope that was clear enough and helps. basically I think youre on the right track just need to dial it in. seems like your gonna top out around 4.4/4.5 which is aboot right. try 18.5x240 that was a sweet spot for my FX4100.

ps: this is my sensor info it looks similar to yours but comes from speed fan

IT8720F
temp1 = general mobo temp
temp2 = NB
temp3 = cpu socket

then my cpu core is listed as AMD K15
I will probably bump the voltage up a bit before I OC it any more, I dropped it back down to 4.0GHz because it wasn't 100% stable.

But it's going to be a little while before I do any more. I found that 1 of my case fans wasn't spinning and one was turning but not pushing any air. I'm going to replace them, and reapply new thermal paste to the NB to see if it helps any. But to do the NB, I have to pull the mobo out since the screws are on the back. These 3 pictures are what my computer case looks like right now.

20170207_151423.jpg

20170207_151429.jpg

20170207_151436.jpg
Crappy picks from the front-facing camera on my GS6 (main cam is broken)
 
your nb isnt really a concern it doesnt get that hot. so you might want to skip it. the vrms need cooling though. you should rotate your 212 so its going front to back instead of sucking air off the gpu. it does make a difference and will also help cool the vrm.
 
Unfortunately, I think the 212 can only go up or down. There is a post sticking up in the middle of the block that holds the bracket in place. I'll have to check more though. If I could get it sideways, it would line up directly with the rear case fan exhausting out the back of the case and that would be awesome.

Problem right now is that my daughter showed up with my grandson, so it will be a little while before I get back to it. He's just watching TV right now, so I can browse the forums every couple of minutes. (took me 20 min to write this post)
 
you can adjust the bracket so it will work, its designed that way. have fun with the fam!
 
Well, the bracket adjustment worked. Just took me a sec to figure it out. I had to spread the hold-down in the opposite direction, put it on the block, and then spread the arms back in the other direction to get them to go around the heat pipes.

Also, I partially figured out which temp was which. I booted into Windows 10, used Open Hardware Monitor and Prime95. The high temp is something on my motherboard, and the CPU is the lowest temperature that I was seeing previously. OHM actually shows which temp is the CPU (blue line on chart), and which ones are from the mobo.

WinTemps.jpg

I did notice something that has made me starting to look for a new mobo. If you look at the CPU VCore you'll see that the Max is listed as 1.176V, which is what I have it set at in BIOS, and what it was showing up until the time I started Prime95. But, the value while running at 100% dropped down to 1.128V. I don't think my current mobo can handle the power draw increase by the CPU. Either that, or there's a setting that I'm missing somewhere in the BIOS that is allowing it to throttle the VCore. I'm not exactly sure why it isn't causing instabilities, but that screenshot was a little while ago, and P95 has been running for the past 40 minutes without any further increase in the temps or anything crashing. While it's been running, I've browsed through files on all my HDDs with Explorer, ran Gimp for a little while (to crop the screenshot and export it to a .jpg).

Also, I noticed that my BIOS, Windows, MemTest, and Mint are all reporting that I only have 16Gb of RAM, while in reality there is 32Gb installed. I'm going to reboot a few times while moving and swapping out the sticks to see if there might be a bad stick or slot.

*edit*
When I just shut down P95 to shut down, the VCore jumped back up to 1.176V right away.
 
Nevermind about the VCore drop during load. Apparently it's called Vdroop, and is built in to prevent voltage spikes from damaging the CPU when dropping out of a high usage state. It's built into all mobos and there is no way to turn it off. Just strange that I've never even heard of this until I started looking into a reason for the drop.
 
I couldn't find anything about LLC. Also, no matter how I swap the RAM around (tried every stick in every slot) it will only register 1/2 of the actual size. I have 4x8Gb sticks, and it shows them all as 4Gb. This mobo is supposed to be able to recognize up to 32Gb of RAM, so that's what I put in it. So, I'm using the computer now with only 1/2 of the actual available RAM. Guess I'll be making a trip to microcenter tomorrow morning. I'm going to bed now so I can get up early enough to go before work.
 
Got back from Microcenter a little bit ago. Picked up a Gigabyte 990FX Gaming motherboard. Actually has cooling for the VRMs, unlike the one I had. Also has several other advantages over the last one. More USB3 ports and SATA3 ports. 64Gb max DDR3 @ 2400 instead of just 32Gb max. Hell, everything's better about it.
I'll be installing it after work tonight. Then I'll update. Thx for all of your help so far. I really appreciate it.
Also, I don't think I mentioned it before, but the old mobo was an MSI 970a-G43 which was a budget board back when I bought it in 2013.
 
that's a good board I love GB! this board will give you much better control and hopefully better results too. I know it will have LLC to take care of any vdroop too!
 
Damnit. I made a mistake. This board only supports 32Gb memory, not 64. It was the ASRock Fatal1ty board I was looking at that was 64... I've got some more comparisons to look at. I might be going back to Microcenter tomorrow morning. :(
 
unless you will actually use the ram(rendering/editing) don't worry aboot it, 32GB is plenty for 99% of uses.
 
Damnit. I made a mistake. This board only supports 32Gb memory, not 64. It was the ASRock Fatal1ty board I was looking at that was 64... I've got some more comparisons to look at. I might be going back to Microcenter tomorrow morning. :(

To get 64GB on a 990FX motherboard you will need mythical 16GB DDR3 DIMMs that may or may not exist in non-ECC form. That motherboard probably supports unbuffered ECC if you want to go that route, but you'll end up spending a lot more on RAM than you did on the entire build.
 
OK. But I'm still going to look at the Fatal1ty board again before installing this one. It was on sale for $10 cheaper than the one I did buy, even though the regular price is $30 more. I won't worry about the amount of RAM though. Thanks guys. Of course being the Fatal1ty name may be the only reason for the higher price.
 
Yea, it's like I thought. After going through all of the options on both boards, it looks like the Gigabyte one I picked up is going to work best for me. Most testing reviews that I could find had the GB one edging ahead with connections and ability to OC.
 
Well... I'm beyond pissed right now. All the time and energy I've put into this so far. Cable management, using a grounding strap, being slow and careful. And the damned thing is DOA.

I went through and disconnected everything one by one, to the point of having one stick of RAM, one CPU fan, and one SSD plugged in. Nothing else. No video card. No USB devices. No extra case fans. Not a fucking thing. I hit the power button and absolutely nothing happens. No fans spin up (not even the PSU), none of the lights on the Mobo that are supposed to light up do anything.

I know it has power because it has a power switch built into the board that lights up. I've tried resetting the BIOS. I've pulled the battery. I've tried everything I can think of. Looks like I'm going to have to make another trip to Microcenter in the morning. If I can wake up in time before I have to go to work. Damn place is 1/2 hour drive each way too.

I'm trying to go to sleep, but I'm just so pissed. This is the first DOA part I've ever gotten in 37 years of messing with computers...

:mad::mad::mad::mad::mad:
 
WTF! I exchanged the mobo. I brought the new one home. Hooked up the bare essentials. (Main power cable, +12v plugs, CPU, CPU fan, one stick of RAM). I hit the power button... And nothing. No post. No power to the fans... Exact same thing as the other one.

I jumpered the green and black pins on the PSU, hit its power switch and it turns on and the fans spin up just fine.

I'm at a loss. Could the CPU have gone bad in between mobos? I used an ESD wrist strap. I left the board in the bag it comes in untill I had touched it to the side of the case I was strapped to.

I guess I'll have to put the old mobo back together and see if it boots.
 
do everything outside the case until you get it booting. then add everything back even load windows/Linux, make sure it all works outside the case. then put it back together. 2 doa in a row isn't very likely. double check your power/led connections, even try just shorting the power pins with a screw driver to see if it powers up
 
Well, I'm just going to go back to Microcenter again and exchange it for the ASRock Fatal1ty board.

I tried booting it outside of the case, no change. So I swapped everything back to the old MSI board. It boots right up without a pause.

I suspect that the BIOS of this particular Mobo doesn't recognize my FX6300 CPU. I checked the compatibility list, and it's there, but I don't trust it now.

Other than getting 2 DOA boards in a row, there's nothing else even connected. I don't have any other CPUs to test the theory out with the 990FX board, or to even flash the BIOS to a newer version. And I don't have the money to buy a new one.

Sigh...
 
Back
Top