M&Ps Ivy Bridge Over-Clocking Experience

That's correct - MSI updated the BIOS to % rather than levels. You can quite happily set 100% to counter Vdroop because MSIs vdroop control is not as aggressive as other manufacturers'.


Yes C1E is a pathway through which the OS can tell the motherboard to back off the pedal. You should have it enabled for a turbo overclock. It can also be enabled for a fixed overclock (this will allow the CPU to rest @ 1.6GHz whilst drawing fewer Watts even if the volts remain the same although it may adversely affect stability).


I can't think of anything that would specifically affect shockwave, however:
a. General stability is always a factor. What stress testing have you been doing? I would advise using the latest Prime95 and running it for 8-24 hours.
b. Make sure that you have the latest shockwave installed along with the latest graphics drivers too (I assume you have not overclocked your GPU).
c. If it persists then try using a different internet browser and see if the issue persists.

Hope that helps :D
Well, I ran AiDA64 for at least 8 hrs. Fine there. I checked shockwave, up to date there. I did have the card overclkd. I backed off a bit, from 845 to 800 and mem down to1250. I also switched to IE. so far, so good. Thanks!
 
Hey, a few questions from a noob OC'er.

I had been running at a 4.2 GHz fixed OC using OCGenie (laziness ftl), but I figured it was time to give real OC'ing a try. After fooling around with a fixed OC, I decided on using Turbo. I set the Turbo multipliers to 45, and everything seems perfectly kosher. However, I did have a few questions.

1. One core runs consistently hotter than the others under load and all 4 are basically all over the place with one lower and the last 2 somewhere in the middle (OCCT screenie below). Is this par for the course with Ivy Bridge, or is it something that's normal when OC'ing most chips?

1a. I've heard that removing the heat spreader and re-applying the tim on Ivy cpu's may even out the temps and lower them in general. Worth the bother, or no?

2. When applying a fixed OC, I've noticed that I'm unable to adjust the CPU ratio and the CPU Core Voltage in the UEFI BIOS (Z77 GD55), but I can change them in the ClickBIOS through MSI Control Center in the OS. Am I missing an option that would enable these? I see in your screenshots that yours seem to be adjustable, but I'm running the newest BIOS revision which is apparently a bit different.

3. Is there a huge difference in strain on the cpu when comparing a fixed OC to a Turbo one? Say for arguments sake that I got an amazing chip that would run 4.5GHz fixed on relatively low volts and temps. Would I still be better off (for chip longevity) with a Turbo OC with the volts on Auto, or...? (Light-ish to medium gaming usage, no rendering or other heavy loads - if that matters).

Anyway, thank you for the guide M&P. It's been very helpful, especially since our hardware is so similar.

.
 

Attachments

  • OCCT_3570k.jpg
    OCCT_3570k.jpg
    98.7 KB · Views: 212
1. One core runs consistently hotter than the others under load and all 4 are basically all over the place with one lower and the last 2 somewhere in the middle (OCCT screenie below). Is this par for the course with Ivy Bridge, or is it something that's normal when OC'ing most chips?

1a. I've heard that removing the heat spreader and re-applying the tim on Ivy cpu's may even out the temps and lower them in general. Worth the bother, or no?

2. When applying a fixed OC, I've noticed that I'm unable to adjust the CPU ratio and the CPU Core Voltage in the UEFI BIOS (Z77 GD55), but I can change them in the ClickBIOS through MSI Control Center in the OS. Am I missing an option that would enable these? I see in your screenshots that yours seem to be adjustable, but I'm running the newest BIOS revision which is apparently a bit different.

3. Is there a huge difference in strain on the cpu when comparing a fixed OC to a Turbo one? Say for arguments sake that I got an amazing chip that would run 4.5GHz fixed on relatively low volts and temps. Would I still be better off (for chip longevity) with a Turbo OC with the volts on Auto, or...? (Light-ish to medium gaming usage, no rendering or other heavy loads - if that matters).
Grats on your first proper OC :D

1. It's very common for the cores to have a spread of temperatures. 10-15C between the highest and lowest is quite normal.

1a. Delidding does indeed make a huge difference. My temps dropped by about ~20C. It is also highly risky and involves a good pair of steady hands. My advice would be to only do it if:
a. You don't mind killing your chip (on top of losing the warranty if you are successful).
b. You want to push further than your CPU will let you because of temps.

If so then have a look at the IB delidding club:
http://forum.overclock3d.net/showthread.php?t=51850

2. This guide is a little outdated as I too have a recent BIOS but no issues with changing the multiplier or any other settings. I have heard however that having ClickBIOS installed can conflict with the BIOS. I'd suggest uninstalling ClickBIOS and just working from the BIOS. Also try setting your BIOS to defaults (that works in most cases too I find!).

3. A Turbo/Offset voltage will always last longer because it will, on average, use fewer volts and have lower temps. Are you likely to see the difference? Well probably not to be honest but it depends on the overclock and how hard you use it. A decently controlled OC should last 2-4 years with minor adjustments and the CPU will probably work for years after that too although you will see some degradation (degradation is not avoidable, even at stock - that's why auto volts are so high).

In most cases the true lifespan of an overclocked CPU will outlast it's useful life for performance users. In your scenario I'd expect you to still be setting overclocks at over 4GHz in a few years time. People don't really know because Ivy hasn't been out that long yet but it will react in a similar way to Sandy. I've seen chips which were folded on 24/7 for month after month at 80C, 100% load and after a year they lost only 100MHz. I'd say that as long as you keep the temps under 80C and the volts in the low 1.3x or below you probably won't need to drop more than 1-300Mhz over several years of owning it.

M&P
 
Thanks for the advice M&P. Think I'll be leaving my lid on. :p

Turns out that the weirdness in the BIOS is apparently by design. I found a thread on the MSI forums started by a dude with the same problem - no brackets around any of the voltage settings or the multiplier, and clicking or hitting enter does nothing. However, the values are adjustable using the +/- keys on the numpad. MSI apparently expected people to come by this info through osmosis.
 
Thanks for the advice M&P. Think I'll be leaving my lid on. :p

Turns out that the weirdness in the BIOS is apparently by design. I found a thread on the MSI forums started by a dude with the same problem - no brackets around any of the voltage settings or the multiplier, and clicking or hitting enter does nothing. However, the values are adjustable using the +/- keys on the numpad. MSI apparently expected people to come by this info through osmosis.
Ah - yea now you mention it I remember several people on the forum coming across the same problem and solution. I'm so used to adjusting it this way now I forgot that it was an issue!

Can you not use the numpad to directly type in the figures too?
 
i don't know if im doing anything wrong but my 3770k needs 1.32v in bios just for 4.5ghz i have tried tiny toms method but i ended up with the same result

extreme llc
vcore 1.33 bios
cpu pll 1.6v
everything else auto

gigabyte z77ud5
3770k Malay (L223C097) h60 cooler (corsair sp performance fans)
corsair vengeance 1600mhz memory cas9
gtx 670 windforse x3 gpu
corsair ax 850 watt psu
corsair 120gb force 3 ssd
 
i don't know if im doing anything wrong but my 3770k needs 1.32v in bios just for 4.5ghz i have tried tiny toms method but i ended up with the same result

extreme llc
vcore 1.33 bios
cpu pll 1.6v
everything else auto

gigabyte z77ud5
3770k Malay (L223C097) h60 cooler (corsair sp performance fans)
corsair vengeance 1600mhz memory cas9
gtx 670 windforse x3 gpu
corsair ax 850 watt psu
corsair 120gb force 3 ssd
Have you heard of the silicone lottery? Some chips need more volts than others - it's just luck of the draw. All you can do is run the volts and temps up and see where it gets you :D
 
Have you heard of the silicone lottery? Some chips need more volts than others - it's just luck of the draw. All you can do is run the volts and temps up and see where it gets you :D

yeah i have herd of the silicone lottery just making sure that i was ok with the volts im using but looking into it i shold be ok with the vcore im using it seems like i always get the cpus that needs loads of vcore for high oc's
 
Excellent post :) Earlier today i was just seeing how low my cpu would run voltage wise and something weird happened i basically lost internet connection and my windows aero theme aswell as programs not responding . Granted i didn't put the voltage that low , could running the voltage low or even possibly general overclocking casue something like this ?

Thanks .
 
Excellent post :) Earlier today i was just seeing how low my cpu would run voltage wise and something weird happened i basically lost internet connection and my windows aero theme aswell as programs not responding . Granted i didn't put the voltage that low , could running the voltage low or even possibly general overclocking casue something like this ?

Thanks .

it might be possible i know sometimes when im running stability tests my pc can act strange
but also i have seen that i can be stable with a certain voltage and get x amount of gflops and im stable
but if i add a little more v-core i can get more performance

ie i have better performance in Intel burn test and cine bench at 4.5ghz than someone with the same cpu at 5.0ghz
strange
 
couldn't work out the vdroop offset on my board i have an msi board as well but there displayed in percentages .

( +12.5 - 100 % ) so im presuming 12.5 is level 0 maybe someone can clarify .
 
Last edited:
yeah i have herd of the silicone lottery just making sure that i was ok with the volts im using but looking into it i shold be ok with the vcore im using it seems like i always get the cpus that needs loads of vcore for high oc's
Yea that's the sad story for most of us, I always get middle of the road chips for my personal rigs. It's usually more than enough though.

Excellent post :) Earlier today i was just seeing how low my cpu would run voltage wise and something weird happened i basically lost internet connection and my windows aero theme aswell as programs not responding . Granted i didn't put the voltage that low , could running the voltage low or even possibly general overclocking casue something like this ?

Thanks .
Yea that sounds like a classic case of undervolting the clock. Add more volts and it will go away.

it might be possible i know sometimes when im running stability tests my pc can act strange
but also i have seen that i can be stable with a certain voltage and get x amount of gflops and im stable
but if i add a little more v-core i can get more performance

ie i have better performance in Intel burn test and cine bench at 4.5ghz than someone with the same cpu at 5.0ghz
strange
Yea, again, if the CPU is running with fractionally too few volts then the processor might hang on cycles, even if it recovers to pass the test. That would affect performance. More volts!

couldn't work out the vdroop offset on my board i have an msi board as well but there displayed in percentages .

( +12.5 - 100 % ) so im presuming 12.5 is level 0 maybe someone can clarify .
100% = level 0 in the old BIOS.

Anyone know about ASRock mobos? My bios is quite a bit different
All motherboards will be based on a similar set of functions so you should just be able to Google the names and figure it out :D I still haven't done an ASrock yet but I hear decent things about them.

M&P
 
7. CPU features is where we adjust the performance of our machine. If I remember correctly with out Enhanced Turbo enabled the thing we need to alter doesn't appear. Another screenshot:
QMk73.jpg

God dammit I looked bloody everywhere for that !!

I managed to clock up my FSB to 106 and get it stable but bugger me they do a good job of hiding the turbo cycle overclock !!!!!

My Xeon is basically a I5 2400 with added features in disguise. I found this article yesterday...

http://www.bit-tech.net/hardware/cpus/2011/01/03/intel-sandy-bridge-review/14

Which shows them clocking it to 3.99ghz from the stock 3.3 by using the FSB and the 4 cycle turbo multiplier.

Argh ! I can't wait to get my board back now. Thanks for showing me where that is dude I looked bloody everywhere !!!

Quick note to those who didn't know..

All of the non K series CPUs with Turbo (note you MUST have Turbo) can still be overclocked. In fact, Intel even tell you this ! So many people thought that the non K were completely locked down but they are not.



Basically Intel left some room for overclocking. Basically they give you 4 cycles (or 4x multi) over your regular Turbo overclock to use for overclocking. That, combined with a small BCLCK increase allows you to get quite a substantial overclock out of a 'locked' CPU.

I will be doing a guide on this using my G45 once it's been RMAd.

BTW.. M&P do you mind if I use some of your pics in a guide for overclocking non K cpus?
 
Last edited:
Thanks again for the post, Mr. M & P. I am having pretty good success with the both methods. I did notice that on my board MSI GD65. I have to set the core ratio on the first overclocking screen in bios. If I leave that to auto, with the xmp profile it will only go up to 38. So I have set the ratio limit under cpu features to 43 also, but I 'm not sure what affect this will have as I've already set the core ratio.
 
Greetings, Mr. M&P. Thanks for the great info.I have been fooling around with your recommendations and have had success. I have noticed a few differences in settings. On my board MSI GD65 the droop control is set in %. So at 100%, with your settings set forth in the turbo section, the volts stays at about 1.15-1.16 under stress test. Also not that I have to set the multiplier under group 1.(adjust cpu ratio). If I leave this on auto even after setting the core limit at group 8. to 44, under stress test, it will only go up to 38. So I am now testing v-droop at 50%, where my volts are at about 1.13-1.12. Note that I 'm using Aida 64. After just an hour , I can see the temps are lower compared to the previous settings by about 8-10 deg. I'll let it go for about 8 hrs. to see whats goiing on. Thanks!
 
Back
Top