Core i7 Overclocking by Batch #

[Avan]

Yeah a real stability test is really needed otherwise I want this as my entry

Awesome bclk man! So close to beating Runmc on his, I think it was 241. How much voltage did you have to feed it for that?

[FabricioGS]

Ok I am updating the spreadsheet. May I make a suggestion?

I say we need to have at least one small standard. How about 1 hour of your favorite stability app? OCCT/Prime/linx/IBT whatever floats your boat. That way someone does not see 4.5ghz batch runs out to get it only to find out it only runs 3.8 with HT stable? Just a thought, let me know what you think.

I vote for at least 1 hour Prime95 stable or at least 5 Linx runs.

[MisterMagoo]

I vote for at least 1 hour Prime95 stable or at least 5 Linx runs.

How about some temps for your latest entry? It must run a bit cooler?

[MisterMagoo]

3841A371 water cooling.

Any temps you would like added?

Edit: Also what about running OCCT on all 8 threads? I noticed HT was not checked.

[MisterMagoo]

Updated through this post.

Note: You may have to clear cache for it to refresh.

This is how it goes. If you posted a stable shot IE 1+ hours you went to the top. Duplicate speeds were then sorted lowest voltage to highest voltage. If you have a "?" for hours stable and need it corrected please let me know so I can do so.

[GenTarkin]

alright, this is somewhat frustrating...Ive added my 3836B043 to that spreadsheet on google docs a couple times and its still not listed in the table your making....

[The Nemesis]

Okay I give up. I've mentioned a couple times that my particular dfi board reports all my voltages .05v over actual measured voltage. So the 1.408v in cpu-z is 1.356v measured and the 1.328v for hte 3838a was 1.27v measured. Someone will be unecessarily overvolting there cpu if trying to acheive my clock with similar batch if the go by those misreported voltages. I'm gonna grab an eVGA board so I guess I'll post some results then with more accurate readings.

[MisterMagoo]

alright, this is somewhat frustrating...Ive added my 3836B043 to that spreadsheet on google docs a couple times and its still not listed in the table your making....

I am not going to google docs so you need to post in here with a screenshot.

Okay I give up. I've mentioned a couple times that my particular dfi board reports all my voltages .05v over actual measured voltage. So the 1.408v in cpu-z is 1.356v measured and the 1.328v for hte 3838a was 1.27v measured. Someone will be unecessarily overvolting there cpu if trying to acheive my clock with similar batch if the go by those misreported voltages. I'm gonna grab an eVGA board so I guess I'll post some results then with more accurate readings.

Sorry dude, will get that fixed up.

[MisterMagoo]

Updated through this post.

Note: You may have to clear cache for it to refresh.

[headala]

Updated through this post.

Note: You may have to clear cache for it to refresh.

This is how it goes. If you posted a stable shot IE 1+ hours you went to the top. Duplicate speeds were then sorted lowest voltage to highest voltage. If you have a "?" for hours stable and need it corrected please let me know so I can do so.

Thank you VERY much for doing this. I know it's a pain but it's a big help to everyone.

[MisterMagoo]

Thank you VERY much for doing this. I know it's a pain but it's a big help to everyone.

NP, glad I could help.

[FabricioGS]

How about some temps for your latest entry? It must run a bit cooler?

Oh, it's in the print screen, but core temp was minimized :P

Anyway, here it is: 69/68/66/65

[MisterMagoo]

Oh, it's in the print screen, but core temp was minimized :P

Anyway, here it is: 69/68/66/65

Cool, I forget to look at the tray.

[road-runner]

Any temps you would like added?

Edit: Also what about running OCCT on all 8 threads? I noticed HT was not checked.

Aww I was wondering what that done now I know. I will rerun it, I cant remember temps exactly but it seems like it was around 70c...

[jcool]

Nice work Magoo

Even though you didn't include any of the CPUs I tested except for the 940... some are here for example. Or do I need to "prove" the numbers every time for you to believe it

Anyhow I am keeping the 3837A707 920 for now, backed it down do 4,3Ghz for 24/7 operation.

I got the same board as Nemesis (DFI x58 UT) and get the same voltage overreporting by Smart Guardian and CPU-Z, so I can only assume my true volts are substantually lower. Unfortunately the measure points on the mobo are blocked by my rad, so I can't measure for now.

Bios settings:

205x21, HT: on Vcore: 1,4435V VTT: 1,42V Vdimm: 1,68V IOH: 1,23V
Cooling: Single Stage with extremely crappy mount, evap temp at around -28C

[Pisklink]

I have the feeling we get different reports of vcore. I mean it looks some people are reporting voltages when idle while others are reporting when stressed. I think it should be clear for everybody what voltage is need. I would suggest both bios voltage as well as stressed voltage.

[M@ster]

Here are my results 24/7, WC
2 folding consoles working
3838B079 Just like Jvidia

[GenTarkin]

well that may be hard because for me I have 3 different voltages for vcore .. there is BIOS, cpuz idle and then cpuz load.
bios shows like 1.28v cpuz 1.26v and for load 1.23v

So, still bios and cpuz load?

Also, I was thinking would it be important to put QPI/VTT voltage on that spreadsheet? because that is a major overclocking determining factor too in some cases...

[MisterMagoo]

well that may be hard because for me I have 3 different voltages for vcore .. there is BIOS, cpuz idle and then cpuz load.
bios shows like 1.28v cpuz 1.26v and for load 1.23v

So, still bios and cpuz load?

Also, I was thinking would it be important to put QPI/VTT voltage on that spreadsheet? because that is a major overclocking determining factor too in some cases...

Could add VTT easily. If everyone wants that let me know.

[jcool]

It should be clear that the voltage that counts is load Vcore. Just sucks if even that one gets misreported like on the DFI x58's... causes a lot of confusion

[GenTarkin]

the one that counts is load vcore? hrm...so, if we just have one vcore listed on the sheet that should be the one displayed? In that case all I have to go off of is what cpuz tells me for load vcore. cuz Ive been quoting idle vcore in all my oc posts =P

[MisterMagoo]

This is the MAIN problem. There were never any standards initially set for this thread. So we have "conflicting" information across the board. IMO BIOS voltage and vtt is all that is needed. The purpose of this list is to see what batches overclock well. If we use bios voltage the user who is inquiring about a batch knows what it will take from a bios stand point to get his rig running at those speeds.

Load and idle voltage can vary from different PSU's, CPU-z/eleet/everest versions etc etc.

This is not a benchmark test and we dont need to be at eachothers throats to get a higher spot, if you know what I mean. This is for the greater of the i7 users, not our individual selves.

[jcool]

Exactly. Which is why we shouldn't mislead people with confusing voltages ^^
On the GA-EX58 series you have a vdroop of around 0,03V, plus they usually undervolt. Setting 1,3000V in bios results in 1,248V load on my Extreme.
Now on the DFI (if the CPU-Z reading is even accurate), I get 1,31V loaded at 1,3000V bios setting when Vdroop is disabled. Quite the difference...

[loonym]

For me the list is nice but I like to see as much info in the post as possible. The reason behind this for me is simple. If I contact a seller and he tells me he has 2 batches, for instance XYZA1234 and ZYXA4321. I can log on XS, search the batch number and maybe get an idea of the potential before I buy. Keep the posts coming folks.

[anzial]

This is the MAIN problem. There were never any standards initially set for this thread. So we have "conflicting" information across the board. IMO BIOS voltage and vtt is all that is needed. The purpose of this list is to see what batches overclock well. If we use bios voltage the user who is inquiring about a batch knows what it will take from a bios stand point to get his rig running at those speeds.

Load and idle voltage can vary from different PSU's, CPU-z/eleet/everest versions etc etc.

This is not a benchmark test and we dont need to be at eachothers throats to get a higher spot, if you know what I mean. This is for the greater of the i7 users, not our individual selves.

Magoo, I did post a screenshot with prime95 12hours+, the one you got my data off a few pages back BTW, I also have a shot of 7hours+ linx with a tad lower vcore but I can't find the pic right now