• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AM3 720 Black Edition OC + Chipset Temps???

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Mjolnir

Member
Joined
Jun 5, 2009
Location
Sydney, Australia
Firstly, being my first post, I'd like to say hi! =D. Anyway moving straight to business.
This is my hardware, specs, temps, etc.

MONITORINGSTUFFS.jpg


Now. Firstly, OC'ing. You can see that my HT link is limited thanks to the budget AM2+ board I'm using. My 'lovely' board also only does increments of 50mv 100mv and 150mv in terms of vCore. So right now its on 50mv+ . Running stable at 3.4 ghz. 1.41 volts bugs me though, should be able to reach this at stock.. But I found that to be unstable. Hmm. (for the moment.. Only 'testing' I've done properly is play games for several hours, + render out a 3D Scene I created for tafe class, which utilises all cores at 100% load).

I was wondering if I'm very much limited in this OC due to my chipset/Mobo as it's fairly budget and isn't a proper AM3 board. I.e. is 3.4 probably the best I'm going to get at stable? (reference to above, concerning voltage, etc)

Anyway 3.4 is fine for me if I can't up it, but higher would be nicer.

What concerns me most, however, is the temp in Hardware Monitor: Nvidia nforce3 pro. Temps are 85 degrees (Idle). Everything else is fine however, my cpu temps are great, as are my gfx card. But this 85 degrees bugs me, I'm not 100% sure about its actual position on my board.. Can anyone help me with this? Regardless, is this 85 stable/safe? I only recently used hardware monitor and noticed this. But I've never had any problems thus far so I guess it's ok? Just want a bit of input. Ta guys.

EDIT: By the way, temps given under the AM3 X3 720 are around 20 per core. Is this counted as the 'CPU' Temp, or are the temps under TMPIN0 and TMPIN1 (i.e 30 odd degrees) the actual CPU temp? Also: I noticed something. The stock voltage for the 720 is supposed to be 1.325, correct me if I'm wrong, but on 'Auto' Setting in my Bios, (there's not standard option), it runs at 1.4, Why? ><. Even at 2.8 ghz stock. Stupid board >.>

Addition: Played with some more settings.. It's at 1.5v, sporting 210 fsb x 17, around 3.563 ghz. Seems stable, played games for a while, ran Cinebench several times.. *shrugs*. I'm a noob overclocker, so ya XD. I've started reading some of the guides here though.. But yeah, Cheers for any responses..
 
Last edited:
:welcome: to OCF!


If your board can use ProbeII that's a good program for ASUS boards. IDK what HWMonitor is showing there. :shrug:


CoreTemp is a good program for core temps if you're unsure about HWMonitor, though I'd guess those CPU temps are correct (18°C). HWMonitor is a little confusing because the AMD K10's only have one CPU temp sensor so, by all rights, there should only be one core temp showing. http://www.alcpu.com/CoreTemp/

For OC'ing you really should be measuring load temps, that is, with the CPU working 100%. Prime95 and OCCT are both good programs for loading the CPU.


Your HT Link speeds aren't a big deal (IMO) but your northbridge speed is a real problem and may be holding back your OC. Check Dolk's Guide for correct NB speeds based on CPU speed. Post back after you've experimented a little. There's a lot of stuff there and questions are normal. ;) http://www.ocforums.com/showthread.php?p=5988622#post5988622


You've gotten past the hardest part of OC'ing - the willingness to jump in and try new things ... :)
 
Last edited:
coretemp.jpg


??? Lol 2.933 ghz, 172 x 17.0? Wha? I'm running what it says in my nick.. Lol. Whats going on there xD. I've looked at the guide, problem is, my motherboard is very limited. I can't modify the NB myself, only by increasing the FSB my ram/NB automatically clocks higher to compensate. I can add voltage to the NB but thats about it. I think maybe I should be happy with my OC for now, and wait til I get a new mobo.. Lol. Oh and yeah I've watched load temps using Cinema 4D rendering out a file for tafe, and my max load temp with a 1.5v @ 3.56 ghz is 39 degrees C in a fairly coolish room (it's winter here in Aus :p). And the temp I mean is the one as TMPIN0 in the above pic on my first post. My actual CORE temps have yet to go above 28 Degrees, at 100% load. Cheers for the reply mate. I'll re-read the thread and see If I can play around a touch more..
 
Last edited:
That 4850 is really cool. What are the temps at load (look at sensors in GPU-Z) and what heatsink/fan are you using for it?

I haven't run any full 100% load GPU tests, but playing Left 4 Dead (max settings, etc) usually runs the GPU at 90-99% (depending on areas), and the max load temp so far is 52 degrees for the core, and 55 for Memory. Using the Thermalright T-Rad 2 with a SCYTHE 120mm Gentle Typhoon Fan @ 1150rpm. Dead silent, and excellent cooling. The stock fan on the HD4850 was crappy and very noisy. It'd be on max, sound like a jet engine at 9000 rpm being only a (i think) 92mm fan, and still run at 60-70 degrees +. I'm very happy with it =).

EDIT: One thing to note about the T-Rad 2, it's fairly limited on Cards it works on, I think the highest ATI card it can run on is the HD4870, the 4890 doesn't support it. I got it cos I knew I wouldn't be upgrading my GPU for a while so yeah.
 
coretemp.jpg


??? Lol 2.933 ghz, 172 x 17.0? Wha? I'm running what it says in my nick.. Lol. Whats going on there xD. I've looked at the guide, problem is, my motherboard is very limited. I can't modify the NB myself, only by increasing the FSB my ram/NB automatically clocks higher to compensate. I can add voltage to the NB but thats about it. I think maybe I should be happy with my OC for now, and wait til I get a new mobo.. Lol. Oh and yeah I've watched load temps using Cinema 4D rendering out a file for tafe, and my max load temp with a 1.5v @ 3.56 ghz is 39 degrees C in a fairly coolish room (it's winter here in Aus :p). And the temp I mean is the one as TMPIN0 in the above pic on my first post. My actual CORE temps have yet to go above 28 Degrees, at 100% load. Cheers for the reply mate. I'll re-read the thread and see If I can play around a touch more..
That's the older version of CoreTemp so the speed will be off. IIRC the new version corrects for that. Regardless, the core temp will be correct - it reads it straight off the CPU.

Have you tried something like 222x16? That would correct the NB speeds. The only issue then would be your RAM, which would be running at 444 MHz (DDR2-888). Most RAM will run that fast without problems but not all ...
 
There's usually a 10 -12 C offset for the 720 in Coretemp, in order for it to be correct. What is the ambient temperature? You're on water, so add 3-4 C (QuietIce is the H2O pro, so he'll correct me if I'm wrong here) to your ambient, and that should be closer to your true temp. I use a 12C offset for my 720. This was all proven with a temp probe, and readings from the Heatsink base, back around the release of the 720.
 
Run prime95 while your have HWmonitor up and watch the numbers for a bit. The ones that start to climb quickly are the CPU and the core(s). Prime will not use the gpu so you can isolate the temps in and under the processor that way. The cpu tems and the core temp ought to be within 2-3 degrees of one another. Any temps that show below ambient in those monitoring programs are not accurate if you are on air cooling. They need to be offset. Some of the monitoring utilities make provision for that, like Fanspeed and Coretemp. Unfortunately it is the case that motherboard and processor temp sensors are not always calibrated well so there are a number of variables one has to deal with. CPU and core temperatures ought to be around 10 degrees C. greater than ambient temp in a well-ventilated case. That might give you a starting point for offsets.
 
Last edited:
Ok ty for the input/info guys. Water? Says who? (nzaneb) Lol. Ambient temp would be roughly 20 degrees, ish. Just an estimate, now that I think of it I don't have even a thermometer for room temp haha. I'll run some benchmarks in the next few days when I have time and let you's know.

Quietice, k I'll try those numbers and see how I go. Cheers guys! Glad I joined these forums! Others left me hanging for a long time, and when they did reply it was useless -_-. I'll post results soon.
 
CPUZ.jpg


I popped in 222x16.5 for the sake of it. I ran wPrime 1024m test, 455.172 seconds. No issues (lockdowns) And my temps as shown by Hardware Monitor were 36 Max under TMPIN0, and the actual cores themselves maxed out at 26 degrees. So theres a 10 degrees difference between the cpu temp and the core temps. Strange?

I'll run the other benchmarks for longer times another day, but for the moment that's where I stand. I suppose just under 3.7 ghz is a fairly happy OC. I'll try upping it again later to see how it runs wPrime again. Cheers guys

UPDATE: I ran Prime95 on the lowest mode testing FPU only, lasted for a little while, thennnnn CRASH BANG ...K :p. I dropped the multiplier back down to 16, giving me 3.546 ghz. *Headdesk* :p
 
Last edited:
Not strange to get a 10 degree temp difference in core temps and cpu temps. Like I said, sensor calibrations are sometimes way off. When they are not the core and cpu temps are usually a within a couple of degrees of each other.
 
Last edited:
CPUZ.jpg


I popped in 222x16.5 for the sake of it. I ran wPrime 1024m test, 455.172 seconds. No issues (lockdowns) And my temps as shown by Hardware Monitor were 36 Max under TMPIN0, and the actual cores themselves maxed out at 26 degrees. So theres a 10 degrees difference between the cpu temp and the core temps. Strange?

I'll run the other benchmarks for longer times another day, but for the moment that's where I stand. I suppose just under 3.7 ghz is a fairly happy OC. I'll try upping it again later to see how it runs wPrime again. Cheers guys

UPDATE: I ran Prime95 on the lowest mode testing FPU only, lasted for a little while, thennnnn CRASH BANG ...K :p. I dropped the multiplier back down to 16, giving me 3.546 ghz. *Headdesk* :p
The idea with 222x16 was to increase the NB/CPU ratio. If you don't have a setting in BIOS to raise the NB multiplier then dropping the CPU multiplier and raising the clock speed is the only way to do that. I suggested those numbers to keep the CPU OC you had while, hopefully increasing stability. Now that 222x16 is working you can try OC'ing farther by increasing the clock speed, which will keep the NB speed more in-line with the CPU speed.

You might also notice better benchmark scores with these settings since the internal NB acts very much like the old external ones. Having a faster NB also increases CPU <-> system data rates, including the L3 cache ...
 
Kayyy, will do (nzaneb). I'll play with the clock speed and let you know how i go running prime95. (If it crashes on me or not). Cheers guys! =)
 
Update guys.
cpuzhardware.jpg


That's pretty much the maximum STABLE speed I can manage for the moment. I've tried running a bit higher, but it dies doing FTT test in prime95. This includes both the multiplier and the bus speed separately, multiplier any higher it instant fails on 100% load in prime95.

I ran Prime95 for roughly an hour mind you.. I know it should be longer, but I'm impatient sue me :p. Also I had to add .05 volts to my ram (from 1.85 to 1.9) as to stabilize it at its higher speed. 3.5 ghz, in the real world I doubt I'll notice the difference between that and perhaps 100-200 mhz higher.. Though I'm sure my 3D works will render minutes faster.. That time could be valuable who knows! Lol.

Thanks for all the help, when I get a nicer board and perhaps some ddr3 ram in the future, I'll give it a better go ay. Oh and the Max load temp was 43 degrees Celsius in a warmed room (it's winter here, so I had the heater on). After updating my Bios from 0405 to 0602, the bios corrected the voltage problem it was having (stock for 720 is 1.325, but on 'auto' it was pumping 1.41), now its fixed, so I'm running 1.46 (i.e. +150mv). I tried at 1.41, but no cigar, so yeah. Cheers again! =D
 
Let me be the first to say.. I'm a retard -_-. I think I'm trying to jump the gun too much here. As I was writing an edit to the post above saying how funny I thought it was that Prime95 was still running (cos I forgot), the system crashed. I was like 'sigh'. No prob right? Lower the clock right? WRONG. Computer wouldnt turn back on, turned on for 2 seconds, and cut out straight away.

Opened it up, Clear CMOS.. Luckily that worked, I was worried something fried (I.e. the ram volt increase, which (I'd think) was the reason for the crash).

I think I'll give up to be honest (for now.. :p). I'll stick it at 3.2 ghz with a +50mv (1.36v) and leave it at that. I highly doubt this'd be unstable, but I will still test it shortly. You win some you lose some.. Lol. Taking out my gfx card is a pain though.. With the custom cooler, gah. It blocks my SATA ports, bends the cable almost 90degrees, but they still work.. Lol mATX board = fail. XD. Oh well, was fun while it lasted :p.
 
Check to see what your RAM is rated for. Most DDR2 is rated between 2.0 and 2.1v unless you happen to have low voltage dimms, those tend to be in the range of 1.8-1.9v. If you were getting instability, your ram could have been an issue because it wasn't getting enough juice or your timings were too tight. You may have a bit more success stress testing if you give it that extra voltage or loosen up the timings a bit.
 
Check to see what your RAM is rated for. Most DDR2 is rated between 2.0 and 2.1v unless you happen to have low voltage dimms, those tend to be in the range of 1.8-1.9v. If you were getting instability, your ram could have been an issue because it wasn't getting enough juice or your timings were too tight. You may have a bit more success stress testing if you give it that extra voltage or loosen up the timings a bit.

Second that, I did a quick drivethru on corsairs website and it looks like both the xms2 in 800mhz and 1066mhz both are rated at 2.1 volts stock....try raising the voltage to that setting and see what happens, and no worries, if your ram gets fried from the voltage and it's rated at that voltage, it's under a lifetime warranty per corsair.
 
http://www.corsair.com/_datasheets/TWIN2X4096-6400C5.pdf

Thats teh RAM I have, rated at 1.9v, which is what I had it at. So with those times, right now they're at 5-5-5-18. By 'loosening' them am I dropping them to higher timings? I.e 7-7-7-20 or something? I don't wanna push it past 1.9 if it's rated at 1.9 exactly just in case something does happen.. That said, I've already pushed the 720 past its recommended max of 1.425.. So wth am I saying??? Lol. :p I'll give it another go just for you guys haha. Cheers for the info.
 
Back