dave851 Woot!

just put my water cooling loop together and look what happens!

Core i7 920(C0) BATCH# 3836B230
CPU clocked at: 4043.5 MHz   

Congrats -- it's not leaking! To give it a really good workout, turn HT on.

DrNip

Did ya notice how my 4.420 ghz non Hyper Transport run smoked both of your guys 4.7 something ghz Hyper Transport runs?   bogus.

 
You mean this kind of smoke?
 
56m 26s @ almost 70 GFlops
 

overclocker333
HT Off
Xeon W3570   Batch #: 
CPU clocked at: 4820.9 mhz  (HT off)
Voltage: 1.59v
Idle temps: 7/0/7/-1 
Load temps:  64/58/61/54

Yeppers I seen that as well as you missed my point.  I can kinda see the logic behind that as that test was almost a .1 ghz higher than your 4.74 ghz test but I really can't see the logic behind the 4.42ghz beating out the 4.74ghz test as that is .32 ghz lower.  That was my point as I was comparing my non hyper to your hyper, not my non hyper to your non hyper.  It was purely a observation not a my test is better than your test.  Either way it is loop legged and that is what had me perplexed.  I wonder if it has to do with higher BCLK?  Kinda like the 775 whereas higher FSB with lower mult trumped lower FSB with higher mult in just about every bench.  I didn't think it was like that though with the i7's.  Maybe if you could find a setting where you could run your mult high with lower BCLK and then at the same speed run it with a lower mult but higher BCLK just to see which one finishes faster or if they finish the same?  If you have time.

DrNip 

 That was my point as I was comparing my non hyper to your hyper, not my non hyper to your non hyper.  It was purely a observation not a my test is better than your test.    

 
No worries... I was just having a little fun. 
 

DrNip 

 I wonder if it has to do with higher BCLK? 

 
I still have problems at high clocks with my w3570 which has an unlocked multi whether running a low or high bclk.

From vaperstylz's sig  
 
"Corsair H50 w/Delta 5500 rpm 252cfm Fan attached "
 
That's what I'm talking about... "Get It Colder!"
 
I am a big "fan" of Delta fans!

DrNip

OK, I scraped the Hyper Transport as I was getting too many inconsistencies as well as it was just too squirmy with LinX.  I am also starting to think Hyper Transport is a gimmick and you won't see much real world difference with it enabled.  Without it my temps were waaaay cooler, less volts required as well as I was consistent time after time.  Here is my latest run.

Core i7 920 BATCH# 3845B027   
CPU clocked at: 4420.0 MHz     
Voltage: 1.270 idle/1.282 load  
Idle temp: 37,38,36,35   
Load temp: 71,71,67,64 <- Priceless 
Air 
Hyper Transport Off

i see you 160 / 65.1

and lower / raise you a... 153 / 67.8

  [img=http://img9.imageshack.us/img9/2682/testinglinx.th.jpg]




didn't have time to let that one play out...just threw it out there on the fly...:)

DrNip

I wish my lil o 920 was unlocked as I too run into high BCLK probs.  No way I would get it stable at 230 BCLK to match your best.  

hell, i can't get stable 4 linx past 224, but can boot into windows at 231...damn shame...i really need to seriously work at this.

DrNip

I wish my lil o 920 was unlocked as I too run into high BCLK probs.  No way I would get it stable at 230 BCLK to match your best.  

If getting to the higher bclk is the problem... "Johnksss" and "Shock-The-Monkey" have figured out the that problem.  I booted my i7 920 at over 5ghz last night thanks to them!  More IOH VCORE and PCIE... I will try and post the voltages I used later to get there or maybe they can jump in.

johnksss
 i see you 160 / 65.1

and lower / raise you a... 153 / 67.8
didn't have time to let that one play out...just threw it out there on the fly...:)

Well now this just beats all that i have seen.   Starting to think LinX is a gimmick now too!

yeah, was not going to say anything about his validated high bclk...lol

i seen it the other day.

back when i got *slapped in the face* with using the correct os for the job..haha

DrNip

Haha.  It wasn't like that man or at least I didn't mean it like that.

lol

nah, im already knowing. you we're totally right....i was just sulking a bit because i totally forgot about it.
we are only as good as our tools! and if they ain't good......we ain't good

edit: see you have the lucky nf200 board...those seem to do far better at booting high bclks...(speculation of course)

I made a comment a few pages back about including RAM speeds in the screenshot but nobody took onto it.  I am interested in seeing what people are using.

@johnksss - I always boot at 220 BCLK and use Eleet to get higher.  No need booting any higher than that really.  The 760 can hang if not beat the 759 in BCLK and vise versa.  Pretty much equal in my opinion.

Oh no.  I set my volts and other various settings in BIOS to what I need to get me where I want to go, boot at 220 BCLK and then use Eleet to get me the rest of the way.  Plain and simple.  I do this with every bench.

Here are the rest of my settings used:





The above picture should have QPI Frequency Selection at 4.8 not auto.


As far as RAM settings just depends on the bench.  Usually fast and tight except for CPU-z I run it loose just to take it out of the equation.  Also on the PCIE frequency I only run it that high when going for CPU-z validation.  All other benches are around 5ghz and PCIE around 105-110.

I think it is a problem with ht at higher BCLK not LinX.  Can you guys pin point at what BCLK this starts to happen.

If there is no bug with LinX then why is it when you run LinX and you notice the timer freezes does LinX continue to run with full 100% load? I have observered this a number of times and when it does happen, the time duration for the frozen clock is not compensated for and LinX shows a shorter time yet the GFLOPS and CPU speed do not corilate. Also even tho it indicates 100% load, actually only threads 0,1,2,3 are fully loaded while 4,5,6,7 drop in load levels indicated by lower heat produced by cores 2 and 3. Task manager shows 100% on all threads but the level of stress within 100% is not as significant. Much like that or Prime95 is in Blend. Much lower heat produced but still indicates 100% on all threads.

johnksss

DrNip

OK, I scraped the Hyper Transport as I was getting too many inconsistencies as well as it was just too squirmy with LinX.  I am also starting to think Hyper Transport is a gimmick and you won't see much real world difference with it enabled.  Without it my temps were waaaay cooler, less volts required as well as I was consistent time after time.  Here is my latest run.

Core i7 920 BATCH# 3845B027   
CPU clocked at: 4420.0 MHz     
Voltage: 1.270 idle/1.282 load  
Idle temp: 37,38,36,35   
Load temp: 71,71,67,64 <- Priceless 
Air 
Hyper Transport Off

i see you 160 / 65.1

and lower / raise you a... 153 / 67.8


didn't have time to let that one play out...just threw it out there on the fly...:)

DrNip

I wish my lil o 920 was unlocked as I too run into high BCLK probs.  No way I would get it stable at 230 BCLK to match your best.  

hell, i can't get stable 4 linx past 224, but can boot into windows at 231...damn shame...i really need to seriously work at this.

 


I don't mean to quote pictures and usually don't but I am this time so excuse me.  Upon looking at this further I see you are running Windows 7 or maybe Vista but without all the visuals.  I will disable all my visuals and see if I get the same.

The inconsistencies I had weren't of it pausing or anything like that but different gflops and times.  Now mind you I would start a run and let it run a few times and then stop it.  Then restart it with a totally different gflop and run time.  Nothing changed, nothing running, just start, stop and then start, stop again.  Then I would start it again for it to be running the same as it was two runs before.  I'm talking a difference of 9 gflops and pretty big time difference.