How to overclock your Nehalem (Core i7)

Tempered81

Diamond Member
Jan 29, 2007
6,374
1
81

Prohibited links edited
Markfw900
Anandtech Moderator



Courtesy of Saaya at Xtremesystems.org

Core i7 920 at 4.2ghz
Cpu-Z validated:
http://valid.x86-secret.com/show_oc.php?id=440058
(people at XS think this might be a fake screenshot because it uses a 22x multi on an i7 920 model - which is supposed to max at 21x.)

Core i7 920 Priming at 3.8ghz:

http://www.matbe.com/images/bi...ivers/000000078330.png

Core i7 Retail Box pictured:
http://i373.photobucket.com/al...75/gooddog2008/004.jpg

Retail Box Stepping & CPU-z pic
:
http://valid.x86-secret.com/show_oc.php?id=440677

Core i7 965 XE running 200mhz base clock on MSI X58:
http://img2.abload.de/img/200cpuzql1z.jpg

Core i7 965 XE at 5.0ghz Cpu-Z validated:
http://valid.x86-secret.com/cache/screenshot/440470.png
http://uneit.com/xpres/wp-cont...08/11/5005-1-thumb.gif
http://uneit.com/xpres/wp-cont...2008/11/5037-thumb.gif

5.4ghz:

http://fugger.netfirms.com/nhm/cpu.JPG

Update:
Nehalem 101

"Hey guys,

thought id post some details so your already prepared for nehalem and how to clock it.
Ill try to post more details here bit by bit as i have the time

Clocks:
Instead of the FSB Nehalem uses a reference clock like AMD, but its at 133Mhz and not 200Mhz.
This reference clock is multiplied to create all clocks inside the CPU:

Bclock x CPU Multiplier = CPU Clock (12-256)
Bclock x QPI Multiplier = QPI Clock (18, 24)
Bclock x Uncore Multiplier = Uncore Clock (10-30)
Uncore clock / Memory Divider = Memory clock

Great Clockspeed Expectations:

222Mhz Bclock screenshot on ln2
200Mhz Bclock stable on air
8GT/s or 4Ghz or below QPI speed, cooling doesnt matter
6Ghz or below screenshot with 1 core on ln2
5.5Ghz or below 3d stable with all 4 cores on ln2
4Ghz or below 3d stable with all 4 cores on air
2100Mhz or below tri channel stable on air

Overclocking:

If you have an XE cpu its all easy, you just change the CPU Multiplier to increase the CPU Clockspeed, thats it...
If you dont have cpu multiplier access, then things get a bit tricky
You need to push up the Bclock, which is related to all other clocks though... so when you increase the Bclock you overclock every part of the processor.
QPI, Uncore and the memory controller/memory.

Crime Scene Investiagtion errr i mean QPI:

You dont really want to overclock QPI though...
Just like HyperTransport QPI is way overpowered for a Desktop system and only really useful for Servers, and you wont see any notable improvements from clocking the QPI bus higher. So while overclocking it wont really give you any benefits, thats what you have to do if you want to overclock your cpu via the Bclock.

It usually maxes out at around 8GT/s, which is around 4Ghz, and as you have seen from the details above, there are only 2 multipliers you can select unfortunately, which will limit your overclocks on Bclock for the 920 and 940 parts. The lowest multiplier you can select is 18. If you do the maths, that means you will be limited to a Bclock of 222. That is IF you processor and board can run 4Ghz QPI... some cant run that high...

Can I get an Uncore do you want more?

Now about Uncore, Uncore is the L3 cache and memory controller, and its powered not by Vcore, but by VTT. I guess Intel couldnt decide what to call this part of the cpu and then went for Uncore which pretty much means its not the core... not very creative, but it gets the job done
Im sure youve seen many if not all mainboard manufacturers using one or two or even 4 phases of a CPU PWM to power the VTT. While on Core2 VTT consumed around 10-15W and you could use a basic PWM to feed it, it now needs double that power, around 30W, and when your overclocking, even more, naturally...

3-some Memories: mhhh good times, good times

The memory controller inside the "Uncore" needs to run at two times the memory BUS speed or more. So for DDR3 1066 (533Mhz memory clock) you need the Uncore to run at 2133Mhz or higher. There are different memory dividers though to clock the memory lower than half the speed of the memory controller/Uncore. According to Intel all CPUs will have all memory dividers unlocked, and will have all the Uncore ratios unlocked, so you guys will be able to run even the highest memory speeds, even with the slowest Nehalem cpus.

Well the highest means an uncore speed of 30x133=4000Mhz /2=2000Mhz
If you want memory speeds beyond DDR3 2000 (1ghz memory clock) you will have to overclock the reference clock. The highest theoretical memory clock with 222Mhz as the highest Bclock is 30x200=6000Mhz /2=3000Mhz.
So DDR3 3000 (1500Mhz memory clock) so this shouldnt be a limit

Memory kits above 2ghz, like the corsair 2133 Kit or the rumored DDR3 2400 kits we all heard of wont be possible without overclocking the cpu though! keep that in mind... then again so far it seems memory can clock higher on x48 than nehalem, but nehalems 3some goodness kills those huge clocks of x48 in every benchmark...

Apprently it seems to be quite useful as well to overclock the Uncore even if you dont want to run huge memory clocks. Remember it contains the L3 cache, so overclocking it will reduce L3 cache latencies and we all know that the higher the memory controller is clocked, the more efficient it works and the higher the bandwidth, even if the memory is still running at stock.


Here are some recommended settings for you i got a from a friend at intel.
if your reading this, cheers man

How to find max Bclock:

Bclock 133Mhz
CPU Multiplier 12x/14x
Uncore Multiplier 2x memory speed or higher
QPI Multiplier 18x
High memory Divider for low memory clock
Vcore 1.25v
VTT 1.45v
Vdimm 1.65v

push up the Bclock and increase VTT and QPI volts to get higher.
Max 100% safe VTT seems to be 1.65v


How to find max Mem clock:

Bclock 133Mhz
CPU Multiplier 12x/14x
Uncore Multiplier 2x memory speed or higher
QPI Multiplier 18x
Chose Memory Divider and Timings
Vcore 1.25v
VTT 1.45v
Vdimm 1.65v

push up the Bclock and increase VTT and Vdimm to get higher.
more vdimm and vtt than 1.65v at your own risk
1.8vdimm should be 100% safe though



How to find max CPU Clocks on i7-965:

Bclock 133Mhz or lower
CPU Multiplier 24x
Uncore Multiplier 2x memory speed or higher
QPI Multiplier 18x
High Memory Divider for low memory clock
Vcore 1.25v
VTT 1.45v
Vdimm 1.65v

reduce Bclock, then increase the cpu multipliers step by step.
push up vcore to get higher clocks stable


How to find max CPU Clocks on i7-940/i7-920:

max Bclock
CPU Multiplier 15x
Uncore Multiplier 2x memory speed or higher
QPI Multiplier 18x
High Memory Divider for low memory clock
Vcore 1.25v
VTT 1.45v
Vdimm 1.65v

increase CPU multipliers step by step, push up vcore to get higher clocks stable. if you cant get a multiplier stable, reduce Bclock until your stable.
Max bclock for high multis is slightly lower than for low multis.


The return of the Turbo-Button?

The max turbo multipliers are stored in seperate registers from the "normal" max allowed cpu multipliers, and they can not be changed (unless you hack the cpu ) The turbo multipliers can not be forced to stay on all the time... at least not directly... you can fool the cpu into believing all the conditions that are necessary to run turbo mode are there, even if they are not
the highest boot multiplier is the highest "normal" multiplier...

for the i7-965 it works like this:

IF turbo is disabled
OR vcore is unstable
OR current is too high
OR Tdp is too high:

4 cores will run at 24x133=3200Mhz
3 cores will run at 24x133=3200Mhz
2 cores will run at 24x133=3200Mhz
1 cores will run at 24x133=3200Mhz

IF Turbo is enabled
AND vcore is stable
AND current is ok
AND the TDP is below the limit

4 cores will run at 26x133=3450Mhz
3 cores will run at 26x133=3450Mhz
2 cores will run at 26x133=3450Mhz
1 core will run at 27x133=3600Mhz


for the i7-940 it works like this:

IF turbo is disabled
OR vcore is unstable
OR current is too high
OR Tdp is too high:

4 cores will run at 22x133=2933Mhz
3 cores will run at 22x133=2933Mhz
2 cores will run at 22x133=2933Mhz
1 cores will run at 22x133=2933Mhz

IF Turbo is enabled
AND vcore is stable
AND current is ok
AND the TDP is below the limit

4 cores will run at 22x133=2933Mhz
3 cores will run at 22x133=2933Mhz
2 cores will run at 22x133=2933Mhz
1 core will run at 23x133=3050Mhz

so you see, turbo is really not that useful for most of the cpus, its only useful for the 965, where it basically auto overclocks the cpu... so basically the Core i7 lineup looks like this:

920 2800Mhz BUT if vcore is unstable, current is too high, or temps are too high, it will run slower.
940 3050Mhz BUT if vcore is unstable, current is too high, or temps are too high, it will run slower.
965 3600Mhz BUT if vcore is unstable, current is too high, or temps are too high, it will run slower.

What turbo really is, is a more advanced form of throttling the cpus from those clockspeeds down to a SAFE speed that will work even with high temps, fluctuating vcore and high current. instead of advertising a 3600mhz cpu that throtles down to 3400 or less depending on the situation, which would cause a lot of complains, the cpus are rated at 3200mhz and everything else above it is a BONUS, so nobody can complain


This is it or now, ill add more details later on. if you have any questions, just ask "

------------------
Original post:

"Instead of the FSB Nehalem uses a reference clock like AMD, but its at 133Mhz and not 200Mhz. This reference clock is multiplied to create all clocks inside the CPU:

Bclock x CPU Multiplier = CPU Clock (12-256)
Bclock x QPI Multiplier = QPI Clock (18, 24)
Bclock x Uncore Multiplier = Uncore Clock (10-30)

If you have an XE cpu its all easy, you just change the CPU Multiplier to increase the CPU Clockspeed, thats it...

If you dont have cpu multiplier access, then things get a bit tricky
You need to push up the Bclock, which is related to all other clocks though... so when you increase the Bclock you overclock every part of the processor.
QPI, Uncore and the memory controller/memory.

You dont really want to overclock QPI though...
Just like HyperTransport QPI is way overpowered for a Desktop system and only really useful for Servers, and you wont see any notable improvements from clocking the QPI bus higher. So while overclocking it wont really give you any benefits, thats what you have to do if you want to overclock your cpu via the Bclock.

It usually maxes out at around 8GT/s, which is around 4Ghz, and as you have seen from the details above, there are only 2 multipliers you can select unfortunately, which will limit your overclocks on Bclock for the 920 and 940 parts. The lowest multiplier you can select is 18. If you do the maths, that means you will be limited to a Bclock of 222. That is IF you processor can run 4Ghz QPI... some cant run that high...

Now about Uncore, Uncore is the L3 cache and memory controller, and its powered not by Vcore, but by VTT. I guess Intel couldnt decide what to call this part of the cpu and then went for Uncore which pretty much means its not the core... not very creative, but it gets the job done
Im sure why youve seen many if not all mainboard manufacturers using one or two or even 4 phases of a CPU PWM to power the VTT. While on Core2 VTT consumed around 10-15W and you could use a basic PWM to feed it, it now needs double that power, around 30W, and when your overclocking, even more, naturally...

The memory controller inside the "Uncore" needs to run at two times the memory BUS speed or more. So for DDR3 1066 you need the Uncore to run at 2133Mhz or higher. There are different memory dividers though to clock the memory lower than half the speed of the memory controller/Uncore. According to Intel all CPUs will have all memory dividers unlocked, and will have all the Uncore ratios unlocked, so you guys will be able to run even the highest memory speeds, even with the slowest Nehalem cpus.

Apprently this seems to be quite useful as well to overclock the Uncore... remember it contains the L3 cache, so overclocking it is supposed to reduce L3 cache latencies and we all know that the higher the memory controller is clocked, the more efficient it works and the higher the bandwidth, even if the memory is still running at stock."

source:
http://www.xtremesystems.org/f...?p=3379790&postcount=1
 
Last edited by a moderator:

faxon

Platinum Member
May 23, 2008
2,109
1
81
so i take it these are going to be a bit harder to overclock than 775 chips at first, but once you learn how you still can? and what does it mean for overclocking on a multiplier locked chip if you can only select 2 QPI dividers? sure, you could OC the QPI to fine tune the OC, but if you do this will it have any adverse effects on your system as a whole? also, is there any word yet on the 1166 CPU ocability? i read on here somewhere that there were plans to bus lock those CPUs, making them effectively impossible to overclock
 

Tempered81

Diamond Member
Jan 29, 2007
6,374
1
81
i also read that the next gen will be locked and that this socket nehalem was the one to get for ocing
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
hehe cool. at least this socket is expected to have some affordable chips (920 for starters). the other socket will probably be mainly the dual cores and cheap quads that have less cache and what not. hopefully the deneb release will keep everything competitive.
 

Concillian

Diamond Member
May 26, 2004
3,751
8
81
Originally posted by: jaredpace
i also read that the next gen will be locked and that this socket nehalem was the one to get for ocing

well that's the nice thing about competition. If I can't get a cheap & fast CPU from Intel, there's always Deneb. As long as nobody is irrationally loyal, it's fine.

However They said 775 was going to be "locked" and unable to be overclocked when it came out too. But the Mobo MFRs found ways around it and it worked.

If raising Bclock is the only way to OC and the QPI is the limiting factor on that, then the Mobo MFRs will find a way to make high QPI clocks stable. If I'm reading him right, he's saying that Bclock starts at 133 and the main ceiling is going to be QPI when Bclock reaches ~200-222. That's still ~50-60% overclocks, which pretty much what we see in the realm of possibility from Wolfdale on air (e7200 from 2.53 --> 4 GHZ is 1.58).

The only major effect a ceiling like that would have is on phase change and maybe water. The "normal" overclocker would not see a really major impact from such a limitation.
 

Tempered81

Diamond Member
Jan 29, 2007
6,374
1
81
yah i heard max oc's for 920 & 940 models is ~4400mhz with current conditions. still thats crazy fast speed.
 

brencat

Platinum Member
Feb 26, 2007
2,170
3
76
Thx Jared for the sneak peak. But my head hurts already...getting a bit more complicated it seems. (Actually I'm just tired and don't want to think anymore after a long week...)
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Interesting. This whole "uncore asynchronous to the core" is rather phenom'esque and I'm curious to see what happens here if folks with 965 run the corner condition where they overclock the L3$ to the point its clockspeed is greater than the cores.

Do latencies continue to improve or does it just level off?

If the plan is to have L3$ run synchronous clockspeed with the cores, unless the user changes the mutlipliers and makes them run asynchronous, I wonder why they went to the trouble of creating the chip to do this. Future SKU's intended to have less than 1:1 clockspeed with the L3$? (celeron versions)
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
this could mean wonderful things for cache performance, and could potentially mitigate the latency issues that were hampering nehalem's clock for clock performance vs penryn in single threaded apps. the only question is how big of a difference. i guess we will just have to wait until the 3rd for NDA to lift eh?
 

Tempered81

Diamond Member
Jan 29, 2007
6,374
1
81
Expected Air Overclocks by model number:

core i7 920: ~3600mhz
core i7 940: ~3700 - 3900mhz
core i7 965: ~3900mhz +

source:
http://www.xtremesystems.org/f...p=3390745&postcount=23

"my i965 run great at 1.30V with AIR COOLING at 3,9GHz Prime95 stable ...
second i920 is pretty stable at 3,6GHz with AIR COOLING and base clock at 170MHz ...
best choice for money/value is i940, is overclockable to 3.7 - 3,8GHz on AIR ... with best mobo"
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: jaredpace
Expected Air Overclocks by model number:

core i7 920: ~3600mhz
core i7 940: ~3700 - 3900mhz
core i7 965: ~3900mhz +

source:
http://www.xtremesystems.org/f...p=3390745&postcount=23

"my i965 run great at 1.30V with AIR COOLING at 3,9GHz Prime95 stable ...
second i920 is pretty stable at 3,6GHz with AIR COOLING and base clock at 170MHz ...
best choice for money/value is i940, is overclockable to 3.7 - 3,8GHz on AIR ... with best mobo"

I like that, expected overclocks...based on a sample size of exactly one CPU per model. :roll:

Despite the lacking statistics, I don't get how the 940 could be ranked above 920 for money/value if these data are to be believed.

I'd rather spend $300 less and have a system that OC's to within 100-300MHz of the 940.

Is anyone really going to notice the difference between a 3.6GHz and 3.9GHz i7?
 

Martimus

Diamond Member
Apr 24, 2007
4,490
157
106
It looks very similar to overclocking an A64, which is pretty much what I think most of us expected.
 

harpoon84

Golden Member
Jul 16, 2006
1,084
0
0
So now instead of raising the FSB we raise the QPI. Just like the ~500MHz FSB walls on C2Qs, we'll naturally have a 'QPI wall' that could potentially limit overclocking. Though if the QPI is capable of at least 200MHz as the post claims then it won't be much of a problem for most overclockers, as even an i920 could potentially hit 4GHz before hitting the QPI wall.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
chances are im not going to be getting one of these until after they release a revised stepping (a little $ issue, but im upgrading from a 939 A64 4000+, its so worth it), so given intel's stepping history, can i expect the OCs i get to be maybe 100-200MHz higher than the numbers at launch?
 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: harpoon84
So now instead of raising the FSB we raise the QPI. Just like the ~500MHz FSB walls on C2Qs, we'll naturally have a 'QPI wall' that could potentially limit overclocking. Though if the QPI is capable of at least 200MHz as the post claims then it won't be much of a problem for most overclockers, as even an i920 could potentially hit 4GHz before hitting the QPI wall.

I think you might have just failed at reading the OP there Harpoon ;) :)

We overclock the Bclock (base clock). Overclocking QPI is possible, but does nothing to overclock the Core or Uncore parts of the system. Overclocking the Bclock is what overclocks all of the above.
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
Here is my take on it.

I think every motherboard will be capable of 178 Bclock, since with an 18x QPI that equals to 3204 mhz > 6.4 G/s QPI. Since every board needs to support the 965, they will probably run 178 Bclock at stock settings.


Bclock = 178

920 = x20 = 3560 mhz (at probably very reasonable vcore perhaps even stock vcore)
940 = x22 = 3916 mhz (within safe vcore limits, air coolable)

178 Bclock x18 QPI = 3204 mhz > 6.4 G/s QPI
178 Bclock x18 Uncore = 3204 mhz (good for DDR3-1600, raise VTT at own risk for more)




Thats your 24/7 no problem overclock. Anything between 178 and 200 Bclock is luck. Anything above 200 is only for benching. It will take a skilled overclocker, one who knows how to tweak many parameters and how to isolate and pinpoint instability culprits. I can already see tons of posts saying I cant pass prime/occt/linpack/whatever you masturbate to, while people are trying to overclock just by raising vcore.

It's going to be harder then that folks.




Here is another sweet overclock, if your motherboard will be capable:

Bclock = 200

920 = x18-20 = 3600-4000 mhz (within safe vcore limits, air coolable)
940 = x18-22 = 3600-4400 mhz (vcore at limits for >4000, only water coolable)

200 Bclock x18 QPI = 3600 mhz > 7.2 G/s (motherboard better have balls for this)
200 Bclock x16 Uncore = 3200 mhz (good for DDR3-1600, raise VTT at own risk for more)
 

Mr Fox

Senior member
Sep 24, 2006
876
0
76
Originally posted by: JAG87
Here is my take on it.

I think every motherboard will be capable of 178 Bclock, since with an 18x QPI that equals to 3204 mhz > 6.4 G/s QPI. Since every board needs to support the 965, they will probably run 178 Bclock at stock settings.


Bclock = 178

920 = x20 = 3560 mhz (at probably very reasonable vcore perhaps even stock vcore)
940 = x22 = 3916 mhz (within safe vcore limits, air coolable)

178 Bclock x18 QPI = 3204 mhz > 6.4 G/s QPI
178 Bclock x18 Uncore = 3204 mhz (good for DDR3-1600, raise VTT at own risk for more)




Thats your 24/7 no problem overclock. Anything between 178 and 200 Bclock is luck. Anything above 200 is only for benching. It will take a skilled overclocker, one who knows how to tweak many parameters and how to isolate and pinpoint instability culprits. I can already see tons of posts saying I cant pass prime/occt/linpack/whatever you masturbate to, while people are trying to overclock just by raising vcore.

It's going to be harder then that folks.




Here is another sweet overclock, if your motherboard will be capable:

Bclock = 200

920 = x18-20 = 3600-4000 mhz (within safe vcore limits, air coolable)
940 = x18-22 = 3600-4400 mhz (vcore at limits for >4000, only water coolable)

200 Bclock x18 QPI = 3600 mhz > 7.2 G/s (motherboard better have balls for this)
200 Bclock x16 Uncore = 3200 mhz (good for DDR3-1600, raise VTT at own risk for more)



the biggest issues will be staying within the VDIMM (1.5v - 1.65v) voltages, and not getting too wild at VTT also.. I know that the early BIOS will have a few Quirks...

It will be interesting to see what the partner MOBO's will all offer, , and How the ASUS ROG pieces will allow perhaps more with less ? They have had the multi-Unlock for awhile, and it will be interesting if it will be able to take a $300.00 bottom end chip, and deliver top end performance.

Micron once again is the preferred IC vendor, and staying low voltage will assure POST, and not having SIDS with the new baby.. Corsair, Mushkin, and GSkill all have 6 GB (3 x 2GB) Kits..I'm sure more will be out soon..

The current line-up of Air Cooling units is not there yet...I'm thinking that there will be some monolithic copper based upon some of the stuff that i have seen in the pipeline.

the larger die is more prone to hot spotting, and The best way to prevent is quick dispersal in the thermal mass of a large chunk of Copper.

 

Idontcare

Elite Member
Oct 10, 1999
21,110
64
91
Originally posted by: JAG87
Here is my take on it.

I think every motherboard will be capable of 178 Bclock, since with an 18x QPI that equals to 3204 mhz > 6.4 G/s QPI. Since every board needs to support the 965, they will probably run 178 Bclock at stock settings.

I don't follow how you rationalize the Bclock requirement for supporting the higher-end QPI clockspeed.

QPI supports 18x and 24x multipliers for the necessary 6.4GT/s QPI.

Supporting 6.4GT/s for 965 support does not require a Bclock of 178, just standard 133MHz Bclock and standard 24x QPI multiplier.

I wonder whether QPI multiplier will be locked or unlocked though for the non-EX models. If the QPI multiplier is unlocked on the CPU then 920's and 940's can run at 6.4GT/s QPI by simply changing the multiplier in the BIOS (presuming the option is available in the BIOS).
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
21,067
3,574
126
Originally posted by: jaredpace
10-4

Thread: Core i7/x58 Overclocking Thread View Single Post
Today, 05:08 AM #67
OBR
Xtreme Addict

Join Date: Dec 2005
Location: Czechia
Posts: 1,370 On all of my 5 boards is multi with Turbo mode on i920 only 21, 22 multi screen must be FAKE





that looks like today to me.