TPU's Core i7 Overclocking and Feedback

[Fitseries3]

i'll double check but i think you need it the other way.

i'll explain more if im right. not too sure as of now.

looks good though.

 

[SystemViper]

Now those are some sweet Pics' Especially that board...

Here she is. Sorry for the mess. Haven't secured the clamps yet, etc.

http://img.techpowerup.org/090321/S7300508.jpg

http://img.techpowerup.org/090321/S7300509.jpg

 

[Fitseries3]

yep.

turn your block 90 degrees either way..

 

[Fitseries3]

also...

you could water the NB....

 

[PaulieG]

also...

you could water the NB....


http://forums.techpowerup.com/attachment.php?attachmentid=23982&stc=1&d=1237688922

Honda over at XS says that the orientation that I have the block at now gave the best temps. I really don't plan to WC the NB since this board is heading to SV next week, but I do plan to cool the NB on my Blood Rage, whenever it gets here.

http://67.90.82.13/forums/showthread.php?t=219081

 

[Binge]

So my thermal grease (Shin-Etsu x23-7783D) has finally set in.

Ambient temps:

Unrealistically high load:

The only reason I do core damage for 1 min is that after 72 hours of running BIONC and folding my cpu doesn't even get that hot.

 

[PaulieG]

So my thermal grease has finally set in.

Ambient temps:
http://img.techpowerup.org/090322/RealTemp5am.jpg

Unrealistically high load:
http://img.techpowerup.org/090322/RealTemp5am2.jpg

The only reason I do core damage for 1 min is that after 72 hours of running BIONC and folding my cpu doesn't even get that hot.

I'll run core damage in a bit to compare. What's your actual vcore here?

 

[Chicken Patty]

great temps binge, what rad a PA120.3???

 

[SystemViper]

So my thermal grease has finally set in.

Ambient temps:
http://img.techpowerup.org/090322/RealTemp5am.jpg

Unrealistically high load:
http://img.techpowerup.org/090322/RealTemp5am2.jpg

The only reason I do core damage for 1 min is that after 72 hours of running BIONC and folding my cpu doesn't even get that hot.

WhaT GREEEEZE DO YOU USE?

 

[Chicken Patty]

WhaT GREEEEZE DO YOU USE?

that and also what voltage was used during that run of core damage?

 

[Binge]

Shine Etsu x23-7783D, the Vcore is 1.408V in bios, the multimeter reads 1.408V

 

[mudkip]

Hey guys,

I'm making a list of Core i7 batches so we can have a good view on which batches are good and which are bad.
Therefore I'd like your batch numbers , your highest STABLE overclock ,and your Vcore and QPI/VTT (QPI/DRAM) voltage.

E.G.

Core i7 920 #3838B153 Stable @ 1.35 Vcore , 1.32v QPI/VTT

This will also be posted on others forums. I'll keep the list up-to-date myself.

Thank you guys and don't forget to post your Core i7 batches!

 

[Binge]

Mudkip, check this out: i7 Batch List

We later determined that there is no good or bad batch. Every single chip has the chance to be a winner or a bad apple.

 

[Thrackan]

I'm going to make one last attempt to get to 4Ghz Prime stable using a lot of volties, but I have one tiny project I want to fix first.
To give you a small hint: Stock and aftermarket coolers do mix.

 

[Chicken Patty]

Mudkip, check this out: i7 Batch List

We later determined that there is no good or bad batch. Every single chip has the chance to be a winner or a bad apple.

the same way i've seen B batches hitting some high clocks with minimal voltages, I see A batches doing the same. Its just luck with the i7. Although usually B batches are good, but I have seen some bad B batch CPU's.

 

[mudkip]

Mudkip, check this out: i7 Batch List

We later determined that there is no good or bad batch. Every single chip has the chance to be a winner or a bad apple.

Really intresting. I've heard that batch 3835xxx was good but I'm not sure about it.

I did a little overclocking today:

As you guys know I can't reach 4Ghz normally.. but today I started to play with the 21 multiplier.

Normally I run 3,8Ghz 100% stable:

20x190 HT on
QPPx36
Mem Multiplier 8 , 7-7-7-20
Uncore x18
1.31vcore
1.31QPI/VTT

When I started with the 21 multiplier I set the QPI/VTT votlage on 1.5v and the memory voltage on 1.66v and 1.2 for the Vcore.

21x182 @ 1.22v it was stable but not stress stable...
Also memory refused to work... I raised the memory till 1.68 , 1.70 , 1,72 and finally 1,74v . Memory was finally stable! Then I raised the vcore and ended with 1.275v. Then I had to ran OCCT if it was stable. OCCT passed succesfully with large data set and the temps didn't go higher then 64 degrees. Tomorrow I'm going to lower the QPI/VTT voltage.

Settings for today:

21x182
QPI x36
Uncore x20
Memory multiplier 10 , 8-8-8-24 @ 1.74v
1.275 Vcore
QPI/VTT 1.5v


My batch number: 3841A403

Is 1.74v for the memory dangerous? I'm afraid my sticks will die. Normally it's rated for 1.65v @ 7-7-7-24

 

[n-ster]

Nah... 1.74v is fine... never knew OCZs could OC to 1820... especially not those! making the opiniom i have on ocz change... especially after hesitating ever since Binge and some others told me they would stay away from OCZ

 

[Binge]

Is 1.74v for the memory dangerous? I'm afraid my sticks will die. Normally it's rated for 1.65v @ 7-7-7-24

No. As long as your memory is within .5v of your QPI then there is no risk to your CPU. The DRAM is like any other piece of hardware if you go above spec it will have lower expected life, but here's the catch. They've already done it to you because current DDR3 is really rated for 1.5V. They sticker it at 1.65V because they want to sell faster memory. yadda yadda... lots of useless info, but something you should know is that putting your QPI up that high could be dangerous to the CPU. Most boards I play with... when you set the QPI that high the number field turns RED.

 

[Fitseries3]

ive run ddr3 up to 2.3v and its still kickin today.

 

[mudkip]

No. As long as your memory is within .5v of your QPI then there is no risk to your CPU. The DRAM is like any other piece of hardware if you go above spec it will have lower expected life, but here's the catch. They've already done it to you because current DDR3 is really rated for 1.5V. They sticker it at 1.65V because they want to sell faster memory. yadda yadda... lots of useless info, but something you should know is that putting your QPI up that high could be dangerous to the CPU. Most boards I play with... when you set the QPI that high the number field turns RED.

n my motherboard this voltage is ''purple'' and it starts with the red zone at 1.6v or something like that. Anyway I lowered the QPI/VTT voltage to 1.375v and I've ran 1 hour OCCT large data set stable.

Do you think this voltage is safe?

 

[Binge]

Yeah I think that voltage is safe and your temps are a lot better. Impressive results on air, really impressive.

 

[mudkip]

Can somebody please tell me why my memory @ 1820Mhz is slower than memory running on 1600Mhz?

Bench 1:

Core i7@ 4Ghz
QPIx36 = 3600Mhz
Uncore x18
Memory multiplier 8 @ 7-7-7-20

Bench 2
Core i7 @ 3,822Ghz
QPI x36 = 3276Mhz
Uncore x20
Memory multiplier x10 @ 8-8-8-24

Is it because the QPI link is higher?

 

[Binge]

It's not slower, it's just not that much faster. And your Cache is not your ram d00d.

 

[mudkip]

It's not slower, it's just not that much faster. And your Cache is not your ram d00d.

Why is the cache slower?

 

[Binge]

It doesn't really matter. The latency is less on the cache with the overclock. I prefer latency being lower over transfer rates sometimes. That's a quirk of your CPU. Nothing I can give you an answer on.