Core i7/X58 Overclocking Thread

Quote:

---

Originally Posted by rge

SteveRo can you run your 2000 rated mem easier using higher multis and lower bclk's? Reason I am asking, trying to see if just my cpu, my ram or what. My OCZ 3x2gb 7,7,7 1600 I can run at 2100mhz 8,8,8 (12mem multi x175bclk) with 1.6qpi/vtt and 1.96DRAM for benching spi, etc without issue. However running mem at 10x202, I cant even boot (tried uncore 20 and 21). I can run 21x200, mem 10x200 2000 mhz but barely stable. Going to 201 and no boot, no matter what even trying much higher vtt, dram qpi pLL, even tried raising everything...different timings, taking outside in cold...no go. I dont know if it is because I have to crank my volts way up to run at 2000 to begin with and just cant handle anymore...curious how low voltage ie mem rated 2000 behaves.

---

rge - I am using bclock 143 because the gskill xmp turned me onto it.
I already knew my max prime stable oc for ht off was 4.43 (21x211) previously - it just worked out that I could get 4.43 exactly at 31x143.
What where the chances of that!
I am able to run 2k, 9-9-9 at qpi 1.55v and mem v at 1.84.
I think 1.55, 1.66 is the xmp v's for 2k, 999, I will need to check that.
So, I haven't tried any low bclock - high multi's, at least not lower than 143 anyway.
Give me a particular test case you would like to see - I will run it and let you know.

Quote:

---

Originally Posted by SteveRo

rge - I am using bclock 143 because the gskill xmp turned me onto it.
I already knew my max prime stable oc for ht off was 4.43 (21x211) previously - it just worked out that I could get 4.43 exactly at 31x143.
What where the chances of that!
I am able to run 2k, 9-9-9 at qpi 1.55v and mem v at 1.84 (maybe down to as low as 1.66).
I think 1.55, 1.66 is the xmp v's for 2k, 999, I will need to check that.
So, I haven't tried any low bclock - high multi's, at least not lower than 143 anyway.
Give me a particular test case you would like to see - I will run it and let you know.

---

Thanks, try to run/boot 19 or 21 cpu multi x 205 bclk with 10x mem multi, 20x uncore, 2050 mhz mem. If that fails, then try 2050 with higher multi and lower bclk.

First Post so apologies if I get this wrong or if I am posting in the wron area. I want to thank everyone who contributes to this Forum as I have learned a ton from all the great information and in the last 2 weeks successfully built and OC'ed my first system over 4Ghz. I have tried to include all the relevant information about my system below (in signature) and I have a question I have not seen an answer to:

When I flashed from the standard F4 BIOS to the F5g I noticed a nice 3-4 degree drop in my Real Temp 3.0 core temps but I am also having the posted issue recovering from S3 state. Is there a F5 series BIOS that will allow recovery from C3 and still give the temp drop?

Thanks

Edit: Noticed that you guys were using F5e so I thought I would give that a try and it did the trick. Did have to wait 10 secs before waking it back up for it not to hang though.

I think VCORE is to high, this chip seems to do it at 1.20v. We should see ;)

Strange thing happening at 200x20 (QPI:3600/NB:3200) SuperPI score its a little better than 211x19 (QPI:3794/NB:3372), any sugestions?
Why memory voltage is 1.80v AUTO? It should be at 1.50v-1.65v? Bios issue? I've got to select it manually ;)

I could try 4.2 or higher but on AIR i think it's not possible...

http://img14.imageshack.us/img14/1940/i9202wc5.png

Ok guys. This thread is 105 pages so obviously I can't be expected to read it to look for a solution.

I have an Asus P6T Deluxe, with the 1102 BIOS. I just got my i7 under water, and the friggen 20x multiplier is driving me nuts. It WILL NOT stabilize at any bclock above 200 basically. Trying to load the CPU will result in an instant BSOD.

But, 222 * 19 works fine, and so does 200 * 20 with turbo enabled.

So what the hell is the issue here? Also, I need 1.5v for the CPU set in the BIOS in order to be fully stable (pass prime95) at 222x19. It seems like my CPU is either a dog, or I have no idea what I'm doing. Also, I can't seem to get my memory stable at stock speed, voltage and timings. DDR3 1600 will crash while running prime95... and 1mhz above that won't even load windows. I mean wtf it can't be that bad??! Please I am running in circles here, what can I do/try to get some better results here?

Quote:

---

Originally Posted by rge

Thanks, try to run/boot 19 or 21 cpu multi x 205 bclk with 10x mem multi, 20x uncore, 2050 mhz mem. If that fails, then try 2050 with higher multi and lower bclk.

---

I can pass linpackx5 at 21x205, mem at 10x but I need high v's to do it.
Temps are through the roof with cpu and qpi v so high.

Quote:

---

Originally Posted by 003

Ok guys. This thread is 105 pages so obviously I can't be expected to read it to look for a solution.

I have an Asus P6T Deluxe, with the 1102 BIOS. I just got my i7 under water, and the friggen 20x multiplier is driving me nuts. It WILL NOT stabilize at any bclock above 200 basically. Trying to load the CPU will result in an instant BSOD.

But, 222 * 19 works fine, and so does 200 * 20 with turbo enabled.

So what the hell is the issue here? Also, I need 1.5v for the CPU set in the BIOS in order to be fully stable (pass prime95) at 222x19. It seems like my CPU is either a dog, or I have no idea what I'm doing. Also, I can't seem to get my memory stable at stock speed, voltage and timings. DDR3 1600 will crash while running prime95... and 1mhz above that won't even load windows. I mean wtf it can't be that bad??! Please I am running in circles here, what can I do/try to get some better results here?

---

Welcome 003 - yes many of us have discovered there is something unusual with the 200 bclock - particularly 20x200.
Even when we can get 19x211 or 21x191, 20x200 will not work.
I am impressed you can get a 222 bclock stable, best I can do is 211.

Regarding your memory - when at high bclocks (190+) you need to keep the mem multi at 6, 8 probably no greater than 10.
The uncore multi needs to be at least 2x you mem multi, preferably 2x plus one.
So for example mem at 6x, uncore at 13x and qpi at 36x.
QPI must stay below 4k (or it will not boot I think).
Uncore wants to be below 4k, sometimes you can run a little over 4k.
As you increase your mem multi, the uncore multi very quickly begins to push you into the 4k barrier.
For example if you are at a 205 bclock, if you set mem multi to 8 for 1640 memory, your uncore needs to be at 17 or 3484
With qpi set to 36 you are at 3690 qpi
I think most folks find better stability keeping uncore lower than qpi freq at about 9/10 ratio.
Using the above example - uncore/qpi: 3484/3690=0.944
Another thought - does your memory have xmp settings?
If yes, turn xmp on and see what your memory maker has the xmp speed and v's set to.
You may need to check these values in windows - my mobo just shows them as auto (when xmp is turned).
Hope this helps.

Chance


I could get 4.1 (ht on) prime stable on air but max temp was 97.
This was with nochua and 2x120 fans flat out in push/pull.
When trying to go much over 4 on air - heat gets to be a big problem..

Quote:

---

Originally Posted by SteveRo

I can pass linpackx5 at 21x205, mem at 10x but I need high v's to do it.
Temps are through the roof with cpu and qpi v so high.

---

Thanks SteveRo for doing that:up:. If you can run linpack at all, it would definitely be bench stable which is what I was looking for.

Making me wish I had bought some 2000 rated ram just so I could try on mine. I have to use high qpi/vtt and dram v to start with (just to get 2000) which may be the problem, tried even up to 1.8vtt but no go, or maybe just my mem controller cant handle >2000mem at 200 bclk.

Quote:

---

Originally Posted by 003

Ok guys. This thread is 105 pages so obviously I can't be expected to read it to look for a solution.

I have an Asus P6T Deluxe, with the 1102 BIOS. I just got my i7 under water, and the friggen 20x multiplier is driving me nuts. It WILL NOT stabilize at any bclock above 200 basically. Trying to load the CPU will result in an instant BSOD.

But, 222 * 19 works fine, and so does 200 * 20 with turbo enabled.

So what the hell is the issue here? Also, I need 1.5v for the CPU set in the BIOS in order to be fully stable (pass prime95) at 222x19. It seems like my CPU is either a dog, or I have no idea what I'm doing. Also, I can't seem to get my memory stable at stock speed, voltage and timings. DDR3 1600 will crash while running prime95... and 1mhz above that won't even load windows. I mean wtf it can't be that bad??! Please I am running in circles here, what can I do/try to get some better results here?

---

Like SteveRo said, the 20x multi sucks on nearly everyones i920, you just have to stay away from it. Just like the stock 22 multi sucks on my i940. I have to use 7 notches higher vcore to get anything stable on 22 multi, comparing same mhz to 21 or 23(turbo) multi, and getting into higher mhz 22 just wont work at all.

Regarding 1.5 vcore, is that with LLC enabled or not? If not, then it is not so bad for 4.2, but sounds like you neither have worst or best.

Regarding mem I had one bad stick of ram of 3 after 2 weeks prior to ever ocing my ram, that was screwing mine up, I tried one stick at time, found the problem, then rma solved my mem issue. But, I would try SteveRo suggestions on mem, 8x mem, 17x UC, qpi 36, DRAM 1.66v (stock v for most) also see what auto does for dram v, qpi/vtt 1.4 (just to make sure when testing). then raise bclk a little, keeping cpu vcore higher than need so cpu does not interfere. If that does not work try with one stick at time, and see if one is bad.

Quote:

---

Originally Posted by rge

Thanks SteveRo for doing that:up:. If you can run linpack at all, it would definitely be bench stable which is what I was looking for.

Making me wish I had bought some 2000 rated ram just so I could try on mine. I have to use high qpi/vtt and dram v to start with (just to get 2000) which may be the problem, tried even up to 1.8vtt but no go, or maybe just my mem controller cant handle >2000mem at 200 bclk.

---

All that I would come up with - I am sure you already tried -
no xmp for your memory?
qpi at 1.55 to 1.6
mem at 1.84
timings at 10-10-10

my 2k memory was just 200 bucks - not so much.

Quote:

---

Originally Posted by SteveRo

Another thought - does your memory have xmp settings?
If yes, turn xmp on and see what your memory maker has the xmp speed and v's set to.
You may need to check these values in windows - my mobo just shows them as auto (when xmp is turned).
Hope this helps.

---

How do I turn on XMP for my memory?

Quote:

---

Originally Posted by 003

How do I turn on XMP for my memory?

---

good morning,

I don't know where it is in the p6t bios but near or with all the memory adjustments there should be something called xmp (or something like that).
it should be an on or off switch - if you don't have it in the bios odds are your memory doesn't have it.

Quote:

---

Originally Posted by SteveRo

Chance


I could get 4.1 (ht on) prime stable on air but max temp was 97.
This was with nochua and 2x120 fans flat out in push/pull.
When trying to go much over 4 on air - heat gets to be a big problem..

---

My CPU max temperature was 71, i don´t have that problem, and my cooler is the same has yours ;)

Quote:

---

Originally Posted by chance

My CPU max temperature was 71, i don´t have that problem, and my cooler is the same has yours ;)

---

say again, you are prime stable at 4 on air with 71c?

Quote:

---

Originally Posted by SteveRo

say again, you are prime stable at 4 on air with 71c?

---

LOL, yeah thats right 71º was the maximum reached, the normal temperature is 67º but some times its scales to 70-71º but maybe thats because its cold here in Portugal, i think in the summer it will be at least more 10º => 81º ;)
Artic Silver 5 or Noctua Thermal Compound? I mixed them both ;) I've put the coolers in the opposite directions, have you? That way the air will go to one unique direction, going out through the power supply. Maybe a better batch, some are hotter ones. Whats your opinion?

Quote:

---

Originally Posted by chance

LOL, yeah thats right 71º was the maximum reached, the normal temperature is 67º but some times its scales to 70-71º but maybe thats because its cold here in Portugal, i think in the summer it will be at least more 10º => 81º ;)
Artic Silver 5 or Noctua? I've put the coolers in the opposite directions, have you? That way the air will go to one unique direction, going out through the power supply.

---

Looking good! - what is your vcore set at in the bios?
How long can you run prime95 (all cores) at those settings?

Quote:

---

Originally Posted by SteveRo

Looking good! - what is your vcore set at in the bios?
How long can you run prime95 (all cores) at those settings?

---

Vcore is 1.27v (1.3000v in BIOS), i've prime yesterday but at 6h30m it has showed a blue screen/reboot, maybe i need more voltage, or not. I think it's a multiplier problem cause 20x200 only primes at > 1.32v, i think a new bios will come soon, mb is P6TDLXv2.

PS: Priming in this moment at 1.28v, hope it pass the 6h30m, cause i want a error free machine ;)

Quote:

---

Originally Posted by chance

Vcore is 1.27v (1.3000v in BIOS), i've prime yesterday but at 6h30m it has showed a blue screen/reboot, maybe i need more voltage, or not. I think it's a multiplier problem cause 20x200 only primes at > 1.32v, i think a new bios will come soon, mb is P6TDLXv2.

PS: Priming in this moment at 1.28v, hope it pass the 6h30m, cause i want a error free machine ;)

---

yep - your on your jouney - hang in there!

:shocked:

Quote:

---

Originally Posted by SteveRo

yep - your on your jouney - hang in there!

---

Thanks ;)

I'm testing my 2nd i7 920 3836A756 here and it's noticeably different from my 1st i7 920 of same batch in various ways, i.e. handles lower cpu vtt and higher bclk, but when it comes to high mem clocks >DDR3-2000Mhz and high QPI frequencies, it seems it has a more strict relationship between cpu vtt and vcore once above 1.48v cpu vtt and likes vcore to be 0.02-0.08v close to it. My first i7 920 didn't have so a strict relationship.

So guys anyone have a i7 920 with such a similar relationship to my 2nd i7 920 ?

Quote:

---

Originally Posted by eva2000

it seems it has a more strict relationship between cpu vtt and vcore once above 1.48v cpu vtt and likes vcore to be 0.02-0.08v close to it. My first i7 920 didn't have so a strict relationship.

So guys anyone have a i7 920 with such a similar relationship to my 2nd i7 920 ?

---

Wow, great discovery. :clap: When I was trying to make mine i7 920 stable (with vCore 1.3~1.35v) I was wondering why it was booting into Windows with less vQPI/DRAM(1.40~1.44v), while it couldn't POST with higher(1.5v+).

Quote:

---

Originally Posted by eva2000

I'm testing my 2nd i7 920 3836A756 here and it's noticeably different from my 1st i7 920 of same batch in various ways, i.e. handles lower cpu vtt and higher bclk, but when it comes to high mem clocks >DDR3-2000Mhz and high QPI frequencies, it seems it has a more strict relationship between cpu vtt and vcore once above 1.48v cpu vtt and likes vcore to be 0.02-0.08v close to it. My first i7 920 didn't have so a strict relationship.

So guys anyone have a i7 920 with such a similar relationship to my 2nd i7 920 ?

---

that is interesting...I will have to give that a go with my i940. My mem will run spi 32m with 2100mhz no problem, my cpu will do 4.7+ghz no problem with spi32m, but so far have not been able to run two together. Have to give that a try...but will have to wait a day or two until cold enough to run 4.7 again.

my best super pi till date

4.4 GHz validation
http://valid.canardpc.com/show_oc.php?id=503214

4.4 GHz super pi run 9.359 sec.
http://img.techpowerup.org/090208/Capture044.jpg

[QUOTE=Chicken Patty;3637366]my best super pi till date

4.4 GHz validation
http://valid.canardpc.com/show_oc.php?id=503214

4.4 GHz super pi run 9.359 sec.

Good run! and did I see that right - with only dual channel memory?