|
The 965P chipset could only realistically get up to around 360-380mhz if you were lucky. 333mhz isn't officially supported on it but both of the 965P-DS3 boards I had ran well above that since Gigabyte "supported" the higher FSB via overclocking. 400mhz+ FSB wasn't really common until later P35 boards. I have no idea what specific settings I used on my old e6600 since that was years ago, although I remember running it around 3.2ghz or so on that board without too much effort. future ghost fucked around with this message at 17:43 on Oct 19, 2013 |
#
?
Oct 19, 2013 17:39
|
|
|
|
| # ? May 26, 2024 22:04 |
|
|
I have a couple questions about overclocking my Phenom II X4 955 Black Edition CPU lightly with AMD overdrive. Below are the specs as it stands now, I haven't changed anything.  As you can see it is currently at 3.2GHz with a Vcore of 1.35V. I have aftermarket air cooling and my CPU does not exceed 45C under full load. From what I've read and seen online the process is to increase the core multiplier and run a stability test. If temperatures are ok and the stability test does not fail, increase the clock again. If the system becomes unstable, slightly raise the Vcore. Is that correct? I've seen similar builds get to 3.8GHz but I would be happy to see even 3.6GHz. I'm really just trying to not have to buy an intel chipset for a couple more months. Is there anything you guys see in my hardware or plan that would be inadvisable for this kind of light OC? I know that the motherboard isn't built for OC like some are but I figured since I'm doing a software OC it shouldn't matter. Another question - if I'm doing a stability test and it fails, and I restart, do I restart with the unstable CPU clock/voltage? Or does AMD overdrive reset to previous settings in the event of a crash? Thanks in advance. E: If I'm already at 1.35V stock it seems I don't have much room to move? I'm guessing I still want to stay below 1.4V and really as low as possible. VelociBacon fucked around with this message at 17:57 on Oct 20, 2013 |
|
#
?
Oct 20, 2013 17:49
|
|
|
VelociBacon posted:I have a couple questions about overclocking my Phenom II X4 955 Black Edition CPU lightly with AMD overdrive. Below are the specs as it stands now, I haven't changed anything. I'm only speaking strictly from experience and what I've taught myself, but you should really get in the swing of doing your modifications from the BIOS instead. Software can be convenient but it may not always work, your mileage may vary and whatnot. With aircooling I would not exceed more than 1.45V, perhaps even 1.5V. One of the big factors here is VRM quality and power phases. From what I'm finding about your particular motherboard it has been recommended to purchase an aftermarket cooling solution for the VRMs (yes, such a market exists) but you're in luck because Gigabyte makes quality VRMs compared to a few trashy MSI boards, one of which I had experience with and of course drove me crazy. Heat + AMD = not a very good result ever. See how high you can get at only 1.4V and keep your temperatures below like 57C period. If the temperatures are too high, back off of the voltage and see if you can retain stability while dropping heat. If that doesn't work, you'll have to back off the clock and stop or go through the process again and see what you can do differently. If the stability test fails I think it falls back to either a stock or last known good configuration. It's been a while since I overclocked a Phenom system, sorry. I would just skip that business and do stuff in the BIOS, then test with something like Prime95 instead. Good luck! p.s. as with any type of overclocking you run the risk of burning up a chip (not very probable unless you cross over into 1.6V territory on air...) and shortening its life of course. Take it easy, don't drive your head into the wall if it just doesn't go up anymore. Not all chips are cut equal. Also save up for an Intel system because when you make the jump goddamn you will notice the difference. An Unoriginal Name fucked around with this message at 18:09 on Oct 20, 2013 |
|
#
?
Oct 20, 2013 18:04
|
|
|
An Unoriginal Name posted:I'm only speaking strictly from experience and what I've taught myself, but you should really get in the swing of doing your modifications from the BIOS instead. Software can be convenient but it may not always work, your mileage may vary and whatnot. Thank you for the reply! I'm not super keen on heading into BIOS stuff at the moment because I have a little bit of a difficult time retaining the information about FSB, interaction with RAM, 9-9-9-3, timings, etc. I'd rather just slide the unlocked CPU a little higher with the software  I realise it's not considered as good a solution.So far I'm at 3.6GHz with no increase in voltage and it hasn't crashed yet on the short stability tests I've ran. One thing though is after running a 3dMark benchmark I noticed this 6.0GHz reading in my HWiNFO64:  Is this just an erroneous reading or is my CPU getting some kind of spiking issue that I should be concerned about? Thanks in advance. Going to see where the limit is on stock voltage.
|
|
#
?
Oct 20, 2013 19:10
|
|
|
VelociBacon posted:So far I'm at 3.6GHz with no increase in voltage and it hasn't crashed yet on the short stability tests I've ran. One thing though is after running a 3dMark benchmark I noticed this 6.0GHz reading in my HWiNFO64: That could very well just be an error in the monitor/sensor due to overclock fuckery. I would give Core Temp a try and maybe even CPU-Z to see if you get more consistent results. Note that CPU-Z does not give a temperature readout, but it should accurately display (current) clock rate and voltage. An Unoriginal Name fucked around with this message at 19:22 on Oct 20, 2013 |
|
#
?
Oct 20, 2013 19:19
|
|
|
An Unoriginal Name posted:That could very well just be an error in the monitor/sensor due to overclock fuckery. I would give Core Temp a try and maybe even CPU-Z to see if you get more consistent results. Note that CPU-Z does not give a temperature readout, but it should accurately display (current) clock rate and voltage. I just ran another stability test + 3dMark at 200MHz higher than when I got that 6GHz reading and it didn't show up again so I'm going to assume it was a one-off misread. I'm now about to test it at 4.0GHz, still on 1.35V and I'm sorta expecting it to be unstable at this point (stock is 3.2GHz 1.35V).
|
|
#
?
Oct 20, 2013 19:32
|
|
|
VelociBacon posted:I just ran another stability test + 3dMark at 200MHz higher than when I got that 6GHz reading and it didn't show up again so I'm going to assume it was a one-off misread. I'm now about to test it at 4.0GHz, still on 1.35V and I'm sorta expecting it to be unstable at this point (stock is 3.2GHz 1.35V). Well at this rate if you can keep going while staying down on stock voltage I'd say you've got quite the chip going for you! Just make sure that when you reach a point where the stability tests fail, you knock it back to the last stable clock rate and see how long that one lasts instead, before increasing voltage.
|
|
#
?
Oct 20, 2013 19:36
|
|
|
An Unoriginal Name posted:Well at this rate if you can keep going while staying down on stock voltage I'd say you've got quite the chip going for you! Just make sure that when you reach a point where the stability tests fail, you knock it back to the last stable clock rate and see how long that one lasts instead, before increasing voltage. It did fine on the stability test + 3dMark, going to boot up some BF4 beta and thrash it for awhile. Tonight I'm going to be at a thanksgiving dinner (Canada) and then work so I was going to let it run stability tests through that time. Is prime95 the way to go? I'm going to be amazed if it can run 4.02GHz on stock voltage with no issues. My CPU temps are still just under 50C at full load. E: Nevermind, BF4 beta is down! alas.
|
|
#
?
Oct 20, 2013 19:48
|
|
|
VelociBacon posted:It did fine on the stability test + 3dMark, going to boot up some BF4 beta and thrash it for awhile. Tonight I'm going to be at a thanksgiving dinner (Canada) and then work so I was going to let it run stability tests through that time. Is prime95 the way to go? Prime95 is a pretty basic and almost standard kind of stress tester to use, at least in my eyes, so I'd give it a shot unless you prefer a different type of test. If the computer goes unstable then it will more than likely BSOD and reboot. I would be concerned of the temperatures with it running so long, but if more than a few hours of straight gaming is accurate to you then it would be a pretty good tell of how the overclock is.
|
|
#
?
Oct 20, 2013 19:55
|
|
|
Am I missing something? The OC is well on it's way to finishing an 8 hour stability test with no issues. It's an AOD OC of a phenom II x4 955 from 3.2GHz to 4GHz. I didn't have to adjust the voltage which strikes me as very strange because I've seen online other people having to bump their vcore up a bit to acheive 3.8GHz on the same chip. I am seeing significant increases in benchmarking utilities so I know that it's actually changing the CPU clock. Is there this much variation chip-to-chip even within the same model?
|
|
#
?
Oct 21, 2013 05:58
|
|
|
VelociBacon posted:Is there this much variation chip-to-chip even within the same model? Yes, there can even be as large as greater than 1GHz differences between good and bad CPUs.
|
|
#
?
Oct 21, 2013 07:02
|
|
|
Is there a database somewhere of reported overclocks for a particular CPU? Basically looking to find out whether what I've got is within a particular range of being OK or whether there's room for improvement (either by adding cooling or trying lower voltage). I've got a i3570K at 4.4ghz, peak core temp (using Real Temp) reached around 72C running Prime95 blend.
|
|
#
?
Oct 21, 2013 14:44
|
|
|
If there are, you're doing yourself a disservice by consulting them. They can only tell you what that model of CPU can do, not what your CPU can do, and juggling CPUs is time-consuming, exhausting, and is likely to become a money pit, even before you admit it's for chasing the OC dragon. Now if you're just looking for what the bone-simple overclock is then sure, but you're already clocked past that point.
|
|
#
?
Oct 21, 2013 15:15
|
|
|
When I set mine to a high FSB occasionally it will fail to start (like, it will turn on but nothing happens) and reverts to the factory settings on the next boot. It seems that when I let the thing "warm up" a little, then reboot and raise the FSB it just runs perfectly well at the higher FSB, the problem only seems to appear after I power down overnight. Could this be a power supply issue?
|
|
#
?
Oct 21, 2013 15:57
|
|
|
Rockker posted:Is there a database somewhere of reported overclocks for a particular CPU? Basically looking to find out whether what I've got is within a particular range of being OK or whether there's room for improvement (either by adding cooling or trying lower voltage). I've got a i3570K at 4.4ghz, peak core temp (using Real Temp) reached around 72C running Prime95 blend. Yes, please take a look at this overclock.net thread: http://www.overclock.net/t/1247869/official-the-ivy-bridge-stable-suicide-club-guides-voltages-temps-bios-templates-inc-spreadsheet
|
|
#
?
Oct 21, 2013 17:32
|
|
|
Stuff like that's going to be self-selected for people showing off. It's an overclocking-specific enthusiast site, the forums thereof, and requires a number of hoops to be jumped through to get your name up.
|
|
#
?
Oct 21, 2013 21:17
|
|
|
that overclock.net thread posted:**This thread is not for discussing safe Temps / Voltages for Ivy Bridge** Well there you go.
|
|
#
?
Oct 21, 2013 21:26
|
|
|
So I decided to man up and use BIOS instead of AMD overdrive since I wasn't getting it to save my clock on next-boot for some reason. I notice that when using the same multiplier to get my 4GHz it no longer throttles the CPU according to load, it just runs at 4GHz all the time. Is there a way around this in my BIOS? I tried disabling cool&quiet, made no difference as you'd expect. Board is a Gigabyte GA-870A-USB3. Processor is an AMD Phenom II X4 955. VelociBacon fucked around with this message at 22:39 on Oct 21, 2013 |
|
#
?
Oct 21, 2013 22:34
|
|
|
You actually want to *enable* Cool'n'Quiet to get idle throttling.
|
|
#
?
Oct 21, 2013 22:36
|
|
|
Factory Factory posted:You actually want to *enable* Cool'n'Quiet to get idle throttling. It was enabled and it was still sticking to 4GHz.
|
|
#
?
Oct 21, 2013 22:39
|
|
|
Shibawanko posted:When I set mine to a high FSB occasionally it will fail to start (like, it will turn on but nothing happens) and reverts to the factory settings on the next boot. An Unoriginal Name posted:From what I'm finding about your particular motherboard it has been recommended to purchase an aftermarket cooling solution for the VRMs (yes, such a market exists) future ghost fucked around with this message at 23:29 on Oct 21, 2013 |
|
#
?
Oct 21, 2013 23:25
|
|
|
What heatsinks you guys recommend for AMD Fx8350's? I got some $15 cooler master 212 and can hit 4.5GHz but it looks lame in my new case. I have a darker mobo and black interior with asus 280x. Should i get one of those Phantek heatsinks for $50 so I can be cool? http://www.newegg.com/Product/Product.aspx?Item=N82E16835709020
|
|
#
?
Oct 22, 2013 00:11
|
|
|
You should really look at a 140mm, 240mm, or 280mm closed-loop liquid cooler. That Phanteks cooler isn't really significantly better than a CM Hyper 212 EVO, based on the reviews I could find. Also you are a tiny bit ridiculous, what with the AMD FX CPU and the color-coordinating of system internals.
|
|
#
?
Oct 22, 2013 00:21
|
|
|
Had prime95 running, went out to run some errands, came back and the computer was in sleep mode (3 hours later) and prime 95 had crashed. The computer came out of sleep mode normally and is running normally. Is this indicative of the OC being unstable or did Prime95 just crash because the computer went to sleep? E: So I found out that C&Q only works at 4GHz if you change the multiplier from overdrive. Doing it from BIOS gives you a limit of 3.8GHz for C&Q, anything higher and you're at max clock the whole time. VelociBacon fucked around with this message at 01:44 on Oct 22, 2013 |
|
#
?
Oct 22, 2013 01:04
|
|
|
Factory Factory posted:You should really look at a 140mm, 240mm, or 280mm closed-loop liquid cooler. That Phanteks cooler isn't really significantly better than a CM Hyper 212 EVO, based on the reviews I could find. Also you are a tiny bit ridiculous, what with the AMD FX CPU and the color-coordinating of system internals. Yeah yeah, I actually have a much lower cpu cooler, the tx3 without heatsink grease right now :/. The 140 and above seem too big for not much increase in performance and if I go water I might be better off buying the FX series that includes one no? Thanks http://www.newegg.com/Product/Product.aspx?Item=N82E16835103064&Tpk=cooler%20master%20tx3
|
|
#
?
Oct 22, 2013 09:03
|
|
|
What, you haven't even bought the FX 8350 yet? If so, drop the idea completely and build an Intel box. It's really, really worth it. Best case, an FX 8350 is slightly more buff than an i5-4670K on fully-threaded applications while using twice the electricity. Worst case, an FX 8350 is half the single-threaded performance of an i5-4670K while still using twice the electricity.
|
|
#
?
Oct 22, 2013 09:32
|
|
|
I have the 8350 and mobo already
|
|
#
?
Oct 22, 2013 10:17
|
|
|
VelociBacon posted:Had prime95 running, went out to run some errands, came back and the computer was in sleep mode (3 hours later) and prime 95 had crashed. The computer came out of sleep mode normally and is running normally.
|
|
#
?
Oct 22, 2013 19:31
|
|
|
http://www.techpowerup.com/192997/cooler-master-announces-the-glacer-240l-cpu-liquid-cooler.html Looks like Cooler Master has licensed the H220?
|
|
#
?
Oct 22, 2013 20:01
|
|
|
BITCOIN MINING RIG posted:If you want to do a quick test use IntelBurnTest set to maximum. If it can't survive 10 runs of IBT (or it gets scorching hot) your voltage, power supply, and/or cooling is insufficient for the CPU clock. 10 runs of "times to run: 10" or just one "times to run: 10"?
|
|
#
?
Oct 23, 2013 12:35
|
|
|
Guni posted:10 runs of "times to run: 10" or just one "times to run: 10"?
|
|
#
?
Oct 23, 2013 17:07
|
|
|
Odd temperature spikes during Prime95 I'm trying to figure out... Just put together an Asrock Extreme4 and I5-4670k. Using the Hyper 212 EVO. Started off with speedfan which reported absurd temperature and voltages. 14C idle, which is less than freaking ambient temperature, and ~40C under load with .9 showing for Vcore. Obviously not accurate. So I load up HWinfo, which for some reason isn't reporting Vcore, so I also grab CPU-Z which does report what I've set in the bios. Running prime95 tests, starting off at 4ghz everything seems fine. Then I go to 4.5ghz. At first, temperature seems great, holding steady at 62-67C under load for the full 15 minutes of the first tests that Prime95 runs. Immediately after the "Self-test 448k passed!" message, Prime95 switches to a length of 8k. (whatever the gently caress that means) Immediately, temperatures spike to the high 80s and creep into the mid-90s. First couple times, I abort the test. This time, I ride out the high temps and let it finish the 15 minutes at this type of test, where it switches to a 512k length whatevermabob. Temperatures promptly drop right back to the 62-67 range. Attempting to eliminate variables, I got vcore to stay fixed at 1.25v, same results. One thing I did notice, in the lower-temperature tests, HWinfo reports the clock rates as actually bouncing back and forth between 4.3ghz and 4.5ghz. When the 8k hot tests are running, they stay pegged at 4.5ghz. edit: Disabling speedstep keeps things pegged at 4.5 during the "cold" P95 tests, but temperatures remain in the 62-69 range. I'm not too familiar with Prime95 testing... is that a normal occurrence? Is there something about that second test segment that is just cruel to the CPU? I've only done this sort of thing once before on my previous i5-2500k build, and don't recall this much variance in a long run of Prime95. edit2: Poking around a bit, yes, it seems that small FFTs place a greater stress on the CPU. They consistently put my temps into the 90s at 4.5ghz and 1.25vcore. Deuce fucked around with this message at 02:28 on Oct 25, 2013 |
|
#
?
Oct 25, 2013 01:46
|
|
|
Haswell overclocking question: I built a new PC with a 4670k and a NH-D14 a few days ago. Overclocking the 4670k, I've only managed to hit 4,2 GHz @ 1,25V while staying stable, at least so far, and within somewhat reasonable temperatures. 4,2 GHz @ 1,24V was unstable (did IBT-very high 40 fine, but froze after an hour or so of Prime95-blend), as well as 4,3 GHz @ 1,27V. Did I just get unlucky with the great Haswell chip lottery, or is there something else I could try to get it stable at a lower voltage while maintaining the clock speed?
|
|
#
?
Nov 4, 2013 07:31
|
|
|
If you were heat limited, you could try delidding the chip and re-mounting the heatspreader, but while that might help your temperatures some, I'd say it's a chance worse than a coin flip that it would get you stable at 4.3 GHz. Probably just unlucky. 
|
|
#
?
Nov 4, 2013 07:50
|
|
|
Factory Factory posted:If you were heat limited, you could try delidding the chip and re-mounting the heatspreader, but while that might help your temperatures some, I'd say it's a chance worse than a coin flip that it would get you stable at 4.3 GHz. Probably just unlucky. Thought as much. I actually considered delidding, but decided against that as my temperatures shouldn't be the limiting factor, since I'm getting maximum temps of a bit over 80C in IBT at 1,25V. Guess I'll just keep this OC or even go lower to drop the voltage. Thanks for the confirmation!
|
|
#
?
Nov 4, 2013 08:15
|
|
|
This might be an elementary question, but it wasn't answered in the very thorough OP. I have a 7850 with a decent overclock (both sliders in the AMD overclock utility nearly all the way to the right -- I think 1050 core, 1425 mem). This is at default voltage, which leads to my next few questions: 1) How do I know if I need more voltage? Would the card lock up if it wasn't getting enough voltage, or frame rates take a sudden dive? 2) Temps seem pretty good -- even high stress benches like Unigine - Valley and Furmark only take it to about 65 max. I'm guessing this is because I have the TwinFrozr model. Good idea to push it further? 3) If everything is working "just fine", is there any reason to bump voltage up in the AMD utility just to be safe? I assume I should if I try to get another 100 mhz or so out of the core and mem? regulargonzalez fucked around with this message at 10:53 on Nov 14, 2013 |
|
#
?
Nov 14, 2013 10:48
|
|
|
regulargonzalez posted:This might be an elementary question, but it wasn't answered in the very thorough OP. 1) If it's unstable. Video cards can tolerate more instability than a CPU. If your overclock breaks down after three hours of stress but you only play for two hours at a time, that's stable enough, you know? Especially when the most common fault is just a driver reboot. Instability in a graphics card looks like:
2) Go ahead, push it further. 3) Bumping up the voltage more than strictly needed might help with stability in the case that your current environment is better for the card than it may be at another time. For example, if your card is currently well-cooled by winter's chill, your overclock may not be stable once summer rolls around; higher voltage would head that off. Other than that, higher voltage than needed for stability does not help and can even hurt minorly.
|
|
#
?
Nov 14, 2013 15:05
|
|
|
Factory Factory posted:Great stuff Thanks! I don't suppose you (or anyone else, for that matter) would happen to know of a site that shows something like a bell curve for overclocking results with various cards? Just curious what target I should be aiming for -- I have no idea if this card maxes out around 1050 or 1400 core, so a general reference would be nice 
|
|
#
?
Nov 14, 2013 22:32
|
|
|
regulargonzalez posted:Thanks! I don't suppose you (or anyone else, for that matter) would happen to know of a site that shows something like a bell curve for overclocking results with various cards? Just curious what target I should be aiming for -- I have no idea if this card maxes out around 1050 or 1400 core, so a general reference would be nice If they're anything like CPU's you apparantly see a great deal of variation from chip to chip, even within the same model.
|
|
#
?
Nov 14, 2013 22:41
|
|
|
|
| # ? May 26, 2024 22:04 |
|
|
Overclocking a 4770K on a Gigabyte Z87 board. I'm using the following settings: CPU VRIN Override LLC: Extreme CPU VRIN Override Voltage: 1.7V VCore: 1.24V Ring Voltage: 1.15V CPU Multiplier: 46X BCLK: Auto Turbo: Auto C1E, C3, C6/C7, and EIST: All Enabled Chip seems pretty drat good so far, with a stable quick and dirty overclock of 4.6 GHz out of the gate without any fine tuning. Temperatures are all over the place during the AIDA64 stress test, but they're normally 57-70 C, based on the instructions being loaded/run. How should I go about optimizing this overclock. I might push to see if it's stable at 4.7, but I'm happy with 4.6 right now so I'm also considering dialing back the VCore and upping the uncore to try to get more performance. I'm not too thrilled about using Extreme Vdroop, so I might try dialing that back to Turbo. Thoughts on further optimization before I start tweaking little things for small gains? Edit: Couldn't resist...stress testing with the same settings at 47X. Still stable. Is this a golden chip? OK, it needs 1.26V to stay up for >20 min at 4.7. Temps still floating around 70C. Protokoll fucked around with this message at 01:49 on Nov 15, 2013 |
|
#
?
Nov 15, 2013 01:24
|
|
