Intel Core i7 920 Overclocking and Heat @ [H]

DrNash said:
Well it turns out turning the pins is what UNLOCKS the HSF. All you have to do is push them down once to get it locked.
Click to expand...

Full respect for admitting your mistake, and giving the full debrief, your words may just save some other poor soul from making the same mistake, or at least knowing how to fix it.
 
Excellent result Rossi! I must do testing with 19x multi! I have done some initial testing with 20x multi for 4ghz:

Vista Ultimate SP1 64-bit
Gigabyte GA-EX58-UD5 @ F5d
G.Skill F3-12800C8-2GBPI-B x 2 @ 1600mhz 2:8, 8-8-8-21-1T
Ci7-920 (3841A383 from Newegg)
TRUE 120 Extreme with push-pull Panaflo 120x38mm M1-BX fans @ 12v @ ~2100 rpm
4000mhz Cores, 3200mhz Uncore, 3600mhz QPI

Temps are from Real Temp 2.90
vCore is from CPU-Z 1.49
Stability tested with LinX x 20 and Prime Blend for 6 hours

Relevant BIOS settings:
Turbo: off
QPI link: x36
Uncore: x16
Load-Line Calibration: enabled
Everything else not mentioned: Auto/Default

4.0ghz (20x200) HT enabled @ 1.360v load, 1.376v idle
vCore: 1.40625v
vDIMM: 1.660v (1.65v in HW Monitor)
vQPI: 1.375v
LinX max temps: 83, 83, 81, 82

4.0ghz (20x200) Hyperthreading DISABLED @ 1.264v load, 1.280 idle
Timings: 8-8-8-21-1T
vCore: 1.30000v
vDIMM: 1.660v (1.65v in HW Monitor)
vQPI: 1.335v
LinX max temps: 67 66 66 66 (edit: my temps increased with higher ambient)

Edit, added recently:
3.8ghz (19x200 turbo off) @ 1.21875 bios, 1.184v load (and idle!) with HT on, 1.275v QPI, 8-8-8-21-1T, 1.60v DRAM (1.62 BIOS), max LinX temp 63c.

Idle temps with HT on @ 1.376v: 37 35 38 37
Ambient temp: ~20C
I haven't done testing with the 3rd stick of RAM yet because my second fan gets in the way of the heatsink on the PI Black stick.

I have the motherboard mounted to a removable motherboard tray laying flat on a table. My TRUE 120 extreme is not lapped and has a very uneven surface. I used a big blob of the paste that came with the TRUE 120 extreme in the middle of the IHS and screwed her down evenly.

Is this chip worthy of liquid cooling?

It seems to me that I'm ready to go for gaming at the moment with no cooling upgrades.

I'm sure the cooling will still be adequate for HT disabled even in a hot room with one less fan based on the LinX temps I am getting.
 
PhantomAu said:
Full respect for admitting your mistake, and giving the full debrief, your words may just save some other poor soul from making the same mistake, or at least knowing how to fix it.
Click to expand...
Yeah I made the same mistake the first time I installed the HSF. Then wondered why my temps were so bad until I fixed it.
 
cec1l said:
Excellent result Rossi! I must do testing with 19x multi! I have done some initial testing with 20x multi for 4ghz:

Gigabyte GA-EX58-UD5 @ F5c
G.Skill F3-12800C8-2GBPI-B x 2 @ 1600mhz 2:8, 8-8-8-21-1T
Ci7-920 (3841A383 from Newegg)
TRUE 120 Extreme with push-pull Panaflo 120x38mm M1-BX fans @ 12v @ ~2100 rpm
4000mhz Cores, 3200mhz Uncore, 3600mhz QPI

Temps are from Real Temp 2.90
vCore is from CPU-Z 1.49
Stability tested with LinX x 20 and Prime for 1-6 hours

Relevant BIOS settings:
Turbo: off
QPI link: x36
Uncore: x16
Load-Line Calibration: enabled
Everything else not mentioned: Auto/Default

4.0ghz (20x200) HT enabled @ 1.344v load, 1.360-1.376v idle
vCore: 1.40000v
vDIMM: 1.660v (1.65v in HW Monitor)
vQPI: 1.355v
LinX max temps: 83, 83, 81, 82

4.0ghz (20x200) Hyperthreading DISABLED @ 1.248v load, 1.264 idle
Timings: 9-9-9-24-1T (Haven't tested with 8-8-8-21-1T yet, should work)
vCore: 1.29375
vDIMM: 1.60v
vQPI: 1.335v
LinX max temps: 61 61 61 61

Idle temps with HT on @ 1.376v: 37 35 38 37
Ambient temp: ~20C
I haven't done testing with the 3rd stick of RAM yet because my second fan gets in the way of the heatsink on the PI Black stick.

I have the motherboard mounted to a removable motherboard tray laying flat on a table. My TRUE 120 extreme is not lapped and has a very uneven surface. I used a big blob of the paste that came with the TRUE 120 extreme in the middle of the IHS and screwed her down evenly.

Is this chip worthy of liquid cooling?

It seems to me that I'm ready to go for gaming at the moment with no cooling upgrades.

I'm sure the cooling will still be adequate for HT disabled even in a hot room with one less fan based on the LinX temps I am getting.
Click to expand...

Thanks!
I managed to get it to 1.29v now which was nice.

Nice reults from you too, very interresting results with the HT off, i may have to investigate into that as im not sure i really need an extra 4 threads so the power save and less heat would be nice :)

As for liquid cooling, yes, if you want to overclock over 4Ghz and keep it there cool but as for a 4Ghz overclock i say if its under 85C load and at a safe voltage you're pretty much going to be ok on air.
 
DanNeely said:
I successfully completed 5 passes a memtest86 at 20x200 DDR3-1600 last night.

210x19 was stable for about 5 min before crashing under prime95, approx the same level of stability I got with 195x20. I tried 220x18 which failed to post. 205x19 has been stable for 5m and counting.
Click to expand...

i7 920 @ 3.2 w/ EVGA x58 and 280 GTX in TT Mozart TX and ran through COD WAW on all difficulties through a 12 hour period w/ temps in the mid to high 80s on air and no crashes. I like my i7, thanks Kyle for your video work.
 
I am pretty impressed with my 920. There were a few problems with my very poorly documented Biostar TPower X58's BIOS, however once I found my way around it was the easiest thing ever to reach my 3.57GHz. I am kind of thinking about going higher, but this is a computer I plan on running 24/7 or so, and I don't really want to push it. The ONLY settings I had to change in the BIOS were to turn off SpeedStep, and ramp the clock up.

cpu.png


I'd say I did pretty well for the very small amount of effort I put in to it. The high temps were at 100% load after running wprime to 1024M.

Cooling it I am using a Cooler Master V8 and once wPrime was finished it went right back down to the idle temp you see there in no time at all.

Pretty damn impressive if you ask me.

Since the BIOS handled the voltage automatically I'm not sure if I can get away with less or not, I may try that soon. Any ideas?
 
Don't rely on auto voltages, your mobo might be overvolting everything, you run a potential risk of a burn-out or something
 
Here are my voltages as reported by the BIOS (just restarted to check em out:

CPU Vcore - 1.312
CPU VTT - 1.344 (but changes to 1.360 briefly then back to 1.344 at a regular schedule. not sure what that's about)
North Bridge VCC - 1.104
DRAM Voltage - 1.536

Does any of this seem out of whack to you guys? Since my RAM's frequency isn't shown in that picture, it is clocked at 1360
 
I have read several sites about overclocking i7 now. But I am still fuzzy on the voltages. Can someone clarify what "CPU Vtt" is? I know I need to keep QPI/uncore within 0.5v of Vmem [I guess QPI must default to 1.15v since there's that 1.65v memory warning]. And not to exceed 1.375v on Vcore. I've also seen "IOH voltage" and "QPI voltage" on one BIOS screenshot. Another site mentioned "CPU PLL voltage" and that +0.2v worked for Penryn.
 
According to most people WC seem to be better that air cool. But I could never get my WC to cool more than my air cool setup. How come?
 
kemathon said:
According to most people WC seem to be better that air cool. But I could never get my WC to cool more than my air cool setup. How come?
Click to expand...

Well, a good WC setup will be better than high-end air. If you're using lousy parts or don't have a big enough rad or strong enough pump, then that would hurt performance a lot.
 
This is the farthest I'm willing to push the chip on air, and even then just to see what it can do in case I want to improve my cooling setup:

i7-920, 3841A383 from Newegg:

21 x 200 4.2ghz, turbo on, HT on, 1.408v load, 1.424v idle, 1.395v VTT, 3200mhz Uncore, 1600mhz memory @ 9-9-9-24-2T with 6GB.

TRUE-120 Extreme with push-pull Panaflo M1-BX fans with motherboard laying on table.

89c max temp @ 18c ambient. It was hitting 97c in normal ambients but I had the voltages higher (too high lol).

http://img222.imageshack.us/my.php?image=nehalem4pt2ht6gblinxts1.jpg
 
For those in the DFW area, Microcenter has the 3841A batch I7 920's in stock. Pretty good overclockers.
 
Just ordered a P6T + Intel i7 920 + 6GB Kit 2x3GB 1333MHz kingston memory.

Will the stock fan be good enough for an overclock to 166MHz BCLK from 133MHz? That is 3320MHz on default voltage.
 
magnusr said:
Just ordered a P6T + Intel i7 920 + 6GB Kit 2x3GB 1333MHz kingston memory.

Will the stock fan be good enough for an overclock to 166MHz BCLK from 133MHz? That is 3320MHz on default voltage.
Click to expand...

yea should be fine. the voltage is what increases temps i think and if the voltage is not touched, your temps shouldnt be effected.
 
Thx for your answer.

Im not familiar with the P6T yet. But since I ordered 1333MHz DDR3 I dont need to adjust anything else than the BCLK to 166 right? That will give memory speed 8*166 = 1328MHz which should be fine right?
 
Yea that overclock should be fine, temps will increase because power usage and therefore, heat are increased.
 
i7 920 batch # 3837B317 purchased at microcenter

Asus P6T (non-deluxe) purchased at microcenter
bios revision 0202

OCZ Platinum 6GB DDR3 1600 OCZ3P1600LV6GK purchase at newegg
http://www.newegg.com/Product/Product.aspx?Item=N82E16820227381

Cooling: TRUE + MX2 thermalpaste + LGA1366 BoltThru kit + Thermalright fanclip
http://www.jab-tech.com/Intel-Core-i7-1366-nehalem-c-338.html

PSU: Corsair 1000HX 1000W
http://www.newegg.com/Product/Product.aspx?Item=N82E16817139007&Tpk=corsair 1000w



ai: manual
cpu ratio 20
blck 175
pcie freq 100
dram freq 1403
uclk 2807
qpi link data rate 6316

dram timings: 8-8-8-24

cpu voltage 1.15
cpu pll auto
qpi/dram (vtt) 1.3 (the difference between DRAM bus voltage and QPI should not be more than 0.5V)
ioh voltage auto
ioh pcie voltage auto
ich voltage auto
ich pcie voltage auto
dram bus voltage 1.64 (max is 1.64)
loadline calibration: enabled (must be enabled else crash prime95 instantly).

cpu settings:

cpu ratio: 20
c1e: enabled
hardware prefetcher: enabled
adjacent cache line prefetch: enabled
intel virtualization tech: disabled
cpu tm: disabled
execute disable bit: enabled
intel HT tech: enabled
active processor cores: all
a20m: disabled
Intel c-state tech: disabled

final CPU frequency: 3.5Ghz

Untitled.jpg

Update: I was able to bring the Vcore down to 1.15!. Very impressive.
1.jpg
 
3836B043



Runs hot, although it's under water - GTZ+PA120.3+DDC3.2&XSPC top. Linx bumps up the temps to mid 70C (77/75/75/72), so it's 5C hotter than prime95. I fault my fans, and I'm running them in pull config w/o any gasket or shroud, should try them with a shroud and in push I think. GTZ orientation might be at fault, as well. room temp around 75F
 
When people sit down with their i7 and increase the BCLK up to 160 or 180 or so, and increase the voltage accordingly to 1.3, and it runs perfectly fine(with aftermarket HSF of course)...are they the exception or the norm?

I see some people going to 1.3v and 180 BCLK and it running like a rock. I am also seeing other i7 guides that recommend a max vcore of 1.25. By increasing the BCLK, won't that cause RAM instability?

Is it really that simple, to just bump up the BCLK and voltage as long as you aren't interested in crazy high speeds, say 3.6 or so?
 
I have seen at least one thread where it took 1.35v for 3.6GHz, but I don't know the specifics. It's never guaranteed, but 1.3v @ 3.6 is an aggressive start. Personally I found a huge temp diff between 1.3v and 1.35v, so I set 1.3v as the max during testing.
 
arkk said:
When people sit down with their i7 and increase the BCLK up to 160 or 180 or so, and increase the voltage accordingly to 1.3, and it runs perfectly fine(with aftermarket HSF of course)...are they the exception or the norm?

I see some people going to 1.3v and 180 BCLK and it running like a rock. I am also seeing other i7 guides that recommend a max vcore of 1.25. By increasing the BCLK, won't that cause RAM instability?

Is it really that simple, to just bump up the BCLK and voltage as long as you aren't interested in crazy high speeds, say 3.6 or so?
Click to expand...

yup i got my blck at 165 and thats it. 63 max on load. the thing is, when it gets hot in the summer. those temps may rise up so 3.4ghz is the safe side of things
 
I just increased the overclock on my computer to 3.8GHz and the temps are perfectly fine. It idles at 38C and at full load it gets up to like 80C or so. I'm not 100% positive on the voltages (I am not at my computer right now) but I am very impressed with the results.

Basically my bclk is at 200, muliplier is at 19, and my RAM is therefore running at the rated DDR3-1600. I was kinda wondering about getting Mushkin RAM since it was one of the cheapest options on Newegg, but also had the lowest latency ratings (at 1600) but I am pretty impressed.

The coolest thing is my motherboard's voltages were all auto-set by the BIOS, and they're about in line with what I have seen by other people overclocking manually. It's a Biostar TPower X58.

499460.png
 
I am running at 200x19 now also... gigabyte ex58-ud5 (f5e) with a 920 (3837a741) and 3x 2gb gskill 1600 9-9-9-24

200x19 (3800)
qpi 36x
mem 8x (1600)
uncore 3200

vcore = 1.300 in bios (w LLC on)
vdimm = 1.60
qpi = 1.360
cpu pll = 1.80

I hard set the mem timings and left everything else alone.... seems pretty happy here... with HT off my load temps hit about 64c max with 24ish ambient.. with HT on it goes to like 75c. Right now I am leaving HT off because I only game on this machine and I love the lower temps... I think I can get away with less vcore also.. I'm going to experiment with 1.275 or so...
 
First post...here it goes.

Vcore 1.14
Dram V: 1.64
QPI 1.22

Multiplier x20 + Turbo on
Bclck 180
Ram: Defualts to somthing like 1450ish (Corsair Dominator 1600 6x1G)

Runs stable at 3.88 idle temps at 45-50 degrees but maxes to 75 with prime95. Using a Mugen Scythe 2 in a small heavily modded Raidmax Scorpio 868 case. I pushed to 4.0 (200 bclck) but hit temps in the 80's in prime. So backed down. Ambient temp is typically 25 degrees, but likely higher come summer so I wanted some wiggle room. Also got a GTX260 core 216 in this case tossin some heat in there. I put a full case mode here:

http://www.eggxpert.com/forums/thread/476270.aspx

I moved the fans around, so I have an 80mm fan pushing air in above the memory sticks and the 120 pulling out towards the rear case fans, bringing all my temps down about 5 degrees (in the review above I hadn't notice that I had the 120 in reverse pulling air towards the gpu!).

I may look at water cooling in a bit, but I blew my budget building this thing.

Ian
 
Yeah, I've had no problems at 4.0 (actually just over 4) other then heat. I think with decent water cooling getting around 4-4.2 should be no problem. What I found was that belo 3.8 you can actually run quite low core voltage. At 3.6 I was at 1.125V then a little bump for stability at 3.8. However anything over 3.8 needed a bigger bump over 1.25 and exponentially increasing the voltage as the clock speed increases. Likely as most of it is lost to heat, as the temp exponentially increases as well. I'll try to post some screenshots this weekend so people don't think I'm telling tales.

Ian
 
Thislilfishy said:
First post...here it goes.

Vcore 1.14
Dram V: 1.64
QPI 1.22

Multiplier x20 + Turbo on
Bclck 180
Ram: Defualts to somthing like 1450ish (Corsair Dominator 1600 6x1G)

Runs stable at 3.88 idle temps at 45-50 degrees but maxes to 75 with prime95. Using a Mugen Scythe 2 in a small heavily modded Raidmax Scorpio 868 case. I pushed to 4.0 (200 bclck) but hit temps in the 80's in prime. So backed down. Ambient temp is typically 25 degrees, but likely higher come summer so I wanted some wiggle room. Also got a GTX260 core 216 in this case tossin some heat in there. I put a full case mode here:

http://www.eggxpert.com/forums/thread/476270.aspx

I moved the fans around, so I have an 80mm fan pushing air in above the memory sticks and the 120 pulling out towards the rear case fans, bringing all my temps down about 5 degrees (in the review above I hadn't notice that I had the 120 in reverse pulling air towards the gpu!).

I may look at water cooling in a bit, but I blew my budget building this thing.

Ian
Click to expand...

what's the batch number? and how long did you prime? "stable is considered at least 8 hours of large or blend fft. Even then overclocks that pass 12 hours of large fft can fail with just 5 cycles of Linx. Also run that program. to make sure.
 
I forget the batch number, I'll check when I get home. I ran prime for 5 hours, while browsing the internet. So not quite the 8 hours you require. It has been running stable in games and stuff, so although the other tests may show something not quite right, seems pointless if everything I use is running properly without crashing...

Ian
 
Just got my 920 today with the plain P6T (got the plain because of the pci slot located at the bottom where I wanted it). I went for a conservative overclock of 3.5 gig. CPU and QPI/DRAM voltages at 1.25 volts, memory at 1.64 volts, turbo boost on, hyperthreading on, everything else on auto. Memory running at 700mhz. Very stable so far with good temps. I hit the above first try right out of the gate with no glitches at all. I'm sure this will go higher but I'm pretty happy with what I got.
 
I have the same setup. Intel I7 920, Corsair DDR3-1600 12gig on an Asus P6T (non deluxe). I am at 4ghz , 800 bus.

I have the Thermalright IFX cooler- big and cools ok. I would like something smaller and silent- but it works.

http://www.thermalright.com/new_a_p...ifx-14-775/product_cpu_cooler_ifx-14-775.html

It runs at 50c Idle and 80-90 at full load with one fan on silent mode.

Wish I could run cooler, but it is faster than my old Q6600 at 3.0 ghz
 
I'm running about 32c at idle. 50c at idle is a bit much for me. It's why I stuck with the speed I did, wanted everyday temps I could live with and a decent speed increase over stock. With turbo boost on (CPU TM Function disabled), I get around 3.675 gig so I'm cool with that.
 
I turned Turbo to off. If I left turbo on, my OC was 3.8/ The thermal specs of the I7's are higher than I am used to,, but it seems in the range of everyone else. Liquid cooling is coming closer and closer to being a better option.
 
I could never hang with liquid cooling. I'm not that much of an extreme tweaker and the thought of mixing water with electrical components is a concept my brain doesn't know what to do with, heh.
 
Here some results with my chip (3838A batch). Stable = P95 large FFT > 12 hours
Also, HT is enabled, vdroop is on and using air cooling (TRUE).
4.2GHZ @ 1.38125v. Temps in mid 80s. Too much voltage for my taste!
4.0GHZ @ 1.3000v. Temps in low 70s. This is my 24/7 setting. :D
3.8GHZ @ 1.2125v. Temps in low 60s.
 
So just want confirmation. We are seeing 4.0 as the current ceiling at this point regardless of batch?

*not referring to anything more then air/water cooling
 
RadXge said:
Here some results with my chip (3838A batch). Stable = P95 large FFT > 12 hours
Also, HT is enabled, vdroop is on and using air cooling (TRUE).
4.2GHZ @ 1.38125v. Temps in mid 80s. Too much voltage for my taste!
4.0GHZ @ 1.3000v. Temps in low 70s. This is my 24/7 setting. :D
3.8GHZ @ 1.2125v. Temps in low 60s.
Click to expand...

I pulled out a P95 crashed after 16 hours once ... 4GHz on 1.3v? Try P95 blend 4 threads (1.6GB) + IBT (4 threads, 2GB) and fire up a 3D game (with your remaining 2GB).
 
You must log in or register to reply here.
Back
Top