Download the Tom's Hardware App from the App Store
The reference for current tech news
Yes No
Ads
Tom's Hardware > Forum > Overclocking > CPUs > i7 920/930/9xx Sweet Spots

i7 920/930/9xx Sweet Spots

Forum Overclocking : CPUs i7 920/930/9xx Sweet Spots

Page:    Previous 1 2 Next Bottom Search this thread
Word :    Username :           
 

Hello everyone! :hello:

 

The past few days I have been wandering around the internet trying to find people's best 920/930 overclocks based on voltages more so than maximum frequency, and it's been a difficult hunt. There are tons of people with the chips at 4GHz and above, but I'm curious as to what your "sweet spot" is for the chip... or most "efficient" OC, rather than max stable.
For example: I'm seeing people report chips hit up to 3.8 on stock voltages, then a decent voltage jump to 4.0, and even a larger one to 4.2.

 

-What's the lowest voltage your chip runs with @ 3.6? Lowest voltage @ 3.8? Lowest voltage @ 4.0? 4.2? etc..
(What are your other voltages at? not just CPUv)

 

-What cooler do you have? What temps are you getting at each setting?

 

-What's your batch number?

 

Finding the max stable is really fun, but how about an efficient 24/7 setting? with the choice of loading your max setting profile when you want the extra juice.
I just think this would be helpful for newbies and experienced OC'ers alike, so if you'd care to add a reply and share your experiences, that would be awesome.

 

(If anyone knows of a similar thread anywhere, or if this is a double post, let me know please!)

 


SUBMITTED CHIPS:

 

blckhaze ----------------------------- 920 D0 --- 3990MHz --- 1.231v --- 38C/71C --- HT, Turbo
SpinachEater ----------------------- 930 D0 --- 3900MHz --- 1.250v --- 38C/61C ---
redechelon -------------------------- 930 D0 --- 4011MHz --- 1.1875v - 41C/60C ---
Hard Line ---------------------------- 930 D0 --- 4300MHz --- 1.296v --- 45C/75C --- HT
mdbrotha03 ------------------------- 920 D0 --- 4011MHz --- 1.200v --- 45C/75C --- HT, Turbo


Message edited by redechelon on 07-27-2010 at 08:17:25 PM
Reply to redechelon
Register or log in to remove.

Maybe a template would help organize... if this thread gets posts, I'd be happy to update the original post with everyone's settings.

 

920/930/other:
Revision:
Stepping:
Batch No.:

 

Frequency:
Voltage:

 

Motherboard:
Memory:
Case:

 

Cooler:
Idle Temp:
Load Temp:
Ambient Temp:

 

Hyper-Threading:
LLC/vDroop:
SpeedStep:
Turbo:

 

Other Voltages:


Message edited by redechelon on 07-15-2010 at 11:59:43 PM
Reply to redechelon

Well, I'll try to get this thing started I guess.

 

920/930/other: 930
Revision: D0
Stepping: 5
Batch No.: 3002A574

 

Frequency: 3717MHz
Voltage: 1.106v

 

Frequency: 3801MHz
Voltage: 1.150v

 

Frequency: 4011MHz
Voltage with HT: 1.206v
Voltage w/o HT: 1.1875v

 

Motherboard: EVGA 132-BL-E758-A1
Memory: 6GB OCZ Obsidian @ 1450Mhz 8-8-8-21
Case: Antec Twelve-Hundred

 

Cooler: Sunbeamtech Core Contact Freezer 120mm
Idle Temp: 40C @ 3.8GHz, 41C @ 4.0GHz
Load Temp: 63C @ 3.8GHz, 71C @ 4.0GHz
Ambient Temp:

 

Hyper-Threading: Enabled
LLC/vDroop: Disabled
SpeedStep: Disabled
Turbo: Disabled

 

Other Voltages:

 

CPU PLL - 1.875v
QPI PLL - 1.35v
VTT - [+25mV] 1.21v
IOH - 1.125v
ICH - 1.125v


Message edited by redechelon on 07-21-2010 at 07:24:41 AM
Reply to redechelon

Hey I am about to embark on my 930 OC as well. I need to research all of the different settings but I will contribute to this post in a few days when I start. I am with you though, information seems to be scattered about so it is helpful to see what voltages people are using.

Reply to SpinachEater

SpinachEater wrote :

Hey I am about to embark on my 930 OC as well. I need to research all of the different settings but I will contribute to this post in a few days when I start. I am with you though, information seems to be scattered about so it is helpful to see what voltages people are using.



Right on, yeah let me know how it goes. It'd just be nice to compile some settings other than 4.0 only. There's a significant temp/voltage drop to 3.8 for me, and a very small performance drop.. so it seems to be more practical. Just wondering what other batches/chips are like, just for people's reference and so I know what to expect for OC'ing friends' chips and things.

Reply to redechelon

It is nice to see a few examples, but realizr that there are several variables, some unknowable. :
1) Chips are all slightly different, even from the same batch and stepping.
2) How well the cpu cooler was mounted may be more important than the cooler type.
3) What is the ambient temperature.
4) How good is the case cooling.
5) Are there some hot vga card/s involved?

Is the overclock a maximum effort, a best stable effort, or just a minor tweaking.

For my part, I just use a 920 c0 with a slight OC to 3.2.

Reply to geofelt

geofelt wrote :

It is nice to see a few examples, but realizr that there are several variables, some unknowable. :
1) Chips are all slightly different, even from the same batch and stepping.
2) How well the cpu cooler was mounted may be more important than the cooler type.
3) What is the ambient temperature.
4) How good is the case cooling.
5) Are there some hot vga card/s involved?

Is the overclock a maximum effort, a best stable effort, or just a minor tweaking.

For my part, I just use a 920 c0 with a slight OC to 3.2.



1) Of course they are, that's the point of this thread... to get an IDEA of what to expect, based on other setups.
2) I'm assuming that if people can overclock to 3.6-4.0, they know how to seat their cooler.
3) We can add this to the template.
4) We can add this to the template.
5) We can add this to the template, although this is correlated to the ambient.

And whether the OC is maximum or minor also is the point of this thread... the sweet spot... in other words, your best voltage to frequency ratio.

Reply to redechelon

redechelon wrote :

I'm assuming that if people can overclock to 3.6-4.0, they know how to seat their cooler.


That's a BIG assumption :D

Dammm I need a 930, my 860 sucks up the juice like a 10 year old at a pool party.

Reply to RJR

geofelt wrote :

It is nice to see a few examples, but realizr that there are several variables, some unknowable. :
1) Chips are all slightly different, even from the same batch and stepping.
2) How well the cpu cooler was mounted may be more important than the cooler type.
3) What is the ambient temperature.
4) How good is the case cooling.
5) Are there some hot vga card/s involved?

Is the overclock a maximum effort, a best stable effort, or just a minor tweaking.

For my part, I just use a 920 c0 with a slight OC to 3.2.



1. For sure but it will be nice to compile a range. Also, mobos are different and given the manufacturer, chipset, and, cooling, it will impact OCs for certain.

2. Mega for sure. I was just installing mine and I put it on several different ways just to play around with various thermal paste adjustments. Depending on fan orientation and how I applied thermal paste, I was getting a Prime95 run temps (at stock settings of 2.8Ghz) from 56-70C. It is crucial to make sure your heatsink is installed to be at it's top notch cooling potential. If I took my 70C install, I would have been screwed for OC range.

3. Definitely. I will include that with my settings. winter temps make a huge difference. Lucky we are in the summer so this should be roughly the worst case scenario.

4. Case style would be good info as well.

5. Meyeah. Not really important with a well vented case but will definitely play a factor if someone has a stuffy case.

Reply to SpinachEater

There's no way to get a thread like this perfectly accurate. In forums, we're just going to have to expect human error. :)
The point of the thread was to gather information from numerous people to see a sort of... average. It would help give you an idea of what to expect before overclocking and would be used as a guide, not a bible. Also, you could compare your chip to others to see where you specific CPU's performance is at and how you can improve it in certain areas.

Reply to redechelon

Yeah all of the little deviations are just part of posting up OC settings. I was finding that turning of HT drops temps and stabilizes higher clocks quite a bit. Have you tried messing around with that?

Reply to SpinachEater

SpinachEater wrote :

Yeah all of the little deviations are just part of posting up OC settings. I was finding that turning of HT drops temps and stabilizes higher clocks quite a bit. Have you tried messing around with that?

 

Yeah, it does decrease temps a good chunk (often 10C?), but I'm already satisfied with my temps and I'd prefer to not sacrifice performance if it's avoidable. I haven't tried benchmarking a difference though... would be interesting. In articles I've read, HT seems to make a 0-20% gain, usually about 10% or less. If you could reach a much higher clock without hyper-threading enabled, it might be worth it, I dunno though.


Message edited by redechelon on 07-17-2010 at 06:45:51 AM
Reply to redechelon

Yeah that is what I am curious about. Can you get higher clocks with HT off to compensate performance at lower clocks with HT on? I wanted to play with that a bit to see what was going on. Most games don't use more than 4 cores. I was watching my core usage and benchmarks are always just using 4 threads. The only thing I see using all 8 is Prime95 atm.

Reply to SpinachEater

Yeah... I think it would depend on the application being used, but the more I read about it, the more I'm tempted to turn off HT. I don't think it helps in games at all, so obviously a higher clock would do something there (Even though it would be minor, it's mostly in the graphics cards). In general productivity though, I'm not so sure... I guess it would depend on how much higher of a clock you could achieve. If you do rendering and converting really often, I'm sure it's worth it... but I don't use programs like that often enough.

I'm stuck on my laptop for the next few days, otherwise I'd see how high I could go. 10C would provide some nice headroom =]

Reply to redechelon

When I watch core usage, the load only alternates between 4 threads and the other 4 always look dead. Bit-tech just had a good article about core usage with games and it looks like most of them only need 3 right now. (Edit - Most of the games they tested that is.... I wish they looked at a lot more games but this type of test should be easy for anyone to do)
http://www.bit-tech.net/hardware/c [...] mes-need/1

Yeah I'm with you on that one. If I was doing heavy processing with video conversion I might not consider turning it off but since I am primarily doing this OC for max FPS in games, I want to really look into it because I am thinking higher clocks across 4 cores will trump lower clocks on 8 threads.

Same here. I haven't been home in a while so I haven't gotten a chance to test this out. I got a stable 4Ghz with HT off and was getting excellent temps but I didn't get to benchmark thoroughly yet.


Message edited by SpinachEater on 07-18-2010 at 04:58:20 AM
Reply to SpinachEater

Jeez, sorry to highjack your thread on this HT tangent but this is pretty interesting. It looks like you can indeed benefit from turning HT off to get higher clocks. I am finding a lot of evidence that HT mostly causes a small drop in gaming performance.

http://ixbtlabs.com/articles3/cpu/ci7-turbo-ht-p2.html

Message quoted 1 times
Message edited by SpinachEater on 07-18-2010 at 06:13:02 AM
Reply to SpinachEater

SpinachEater wrote :

Jeez, sorry to highjack your thread on this HT tangent but this is pretty interesting. It looks like you can indeed benefit from turning HT off to get higher clocks. I am finding a lot of evidence that HT mostly causes a small drop in gaming performance.

http://ixbtlabs.com/articles3/cpu/ci7-turbo-ht-p2.html



Nah, I'm just waiting for someone to post their results if it will ever happen =P. When I get home, I'll change HT settings around and benchmark... I'll post differences.

Yeah, interesting articles... they all seem to get different percentages, but the basic idea of them is the same. HT aids in certain programs, but for games, it even often hurts performance... which is pathetic. I wish more games supported more than 3 cores/threads currently.

Reply to redechelon

Alright, I tried disabling Hyper-Threading at 4.0 and it lowered my temps at load to 63C with Prime95. With HT, it loads at about 72C. It allowed me to lower my voltage further also: from 1.206 to 1.1875. So, now it loads at 60C... that's pretty damn good, but later tonight, I'll bench it to see if there's a significant performance shift. If it's a very similar or identical score in programs like 3DMarkVantage and Heaven, then I'll keep it disabled for sure. I don't record and edit often enough to care about performance in those areas... mainly gaming and simple programs.


Message edited by redechelon on 07-19-2010 at 08:44:27 PM
Reply to redechelon

Wow, interesting. I am playing around with mine tonight to see what is up. My RAM seems to be giving me some sass though.

Reply to SpinachEater

Hey red, I am having some weird issues can you help? I am hitting 3.8 and 4.0 GhZ just fine and with great temps (64C max at 4Ghz with HT off) I am running Prime 95, memtest, Stone Giant, AVP, Heaven, and Final Fantasy benchmarks with no issues what so ever at 4.0 Ghz but when I start up a real game like Crysis or BF BC2, I go to BSOD.

Do you have an idea as to what would cause a game to crash but not any of the benchmarks? I am starting with high voltages and working my way down but nothing at 3.8 and up seems to allow games to run.

I first thought it was my RAM but I downclocked the crap out of it and I still BSOD. Would you have any ideas regarding why I am crashing in games but not with benchmarks?

Reply to SpinachEater

SpinachEater wrote :

Hey red, I am having some weird issues can you help? I am hitting 3.8 and 4.0 GhZ just fine and with great temps (64C max at 4Ghz with HT off) I am running Prime 95, memtest, Stone Giant, AVP, Heaven, and Final Fantasy benchmarks with no issues what so ever at 4.0 Ghz but when I start up a real game like Crysis or BF BC2, I go to BSOD.

Do you have an idea as to what would cause a game to crash but not any of the benchmarks? I am starting with high voltages and working my way down but nothing at 3.8 and up seems to allow games to run.

I first thought it was my RAM but I downclocked the crap out of it and I still BSOD. Would you have any ideas regarding why I am crashing in games but not with benchmarks?



Weird... what tests are you running in Prime95, Blend?

What RAM do you have and what settings/timings/voltage?

Maybe if you post your CPU overclock settings too, that would help. Maybe something will stick out.

Reply to redechelon

920
Revision: D0
Stepping: 5


Frequency: 3.990
Voltage: 1.231


Motherboard: P6T V2
Memory: 12gb 1600 Dominator 8-8-8-24 @ 1520mhz
Case:


Cooler: H50
Idle Temp: 38~
Load Temp: 71~
Ambient Temp: 24c


Hyper-Threading: 8
LLC/vDroop:
SpeedStep: ON
Turbo: ON

Other Voltages: Dram 1.35 & 1.66

1.225vcore ran for 3 hours before prime95 stopped working
1.231vcore ran prime95 for 9+ hours in 2 different instances, with memtest x4.
Overclocking to 4.2ghz required the vcore to be 1.263, so it was quite a big jump for that (the 200 bclk) so I think 1.231 with bclk 190 is the sweet spot on this 920.

Reply to Blckhaze

Ok, I am currently stuck at 3.5GHz and I am trying to minimize Vcore to drop below 1.29375V because it sounds like I should be able to go a lot lower ATM. I will update this post as I find the lowest voltages.

920/930/other: 930
Revision: D0
Stepping: 5
Batch No.: ?

Frequency: 3.90
Voltage: 1.250V
Multiplier: 21x
Blck: 186
Uncore: 2x RAM clocks

Motherboard: Gigabyte GA-X58A-UD3R (v2.0)
Memory: 6 GB Mushkin (3x2GB) 9-9-9-24
Case: Antec 900

Cooler: Xigmateck S1283

Hyper Threading OFF
Idle Temp: 38-35-36-33C (core 1-2-3-4)
Load Temp: 61-58-57-55C (core 1-2-3-4) with Prime 95

Hyper Threading On
Idle Temp 38-34-35-32C (core 1-2-3-4)
Load Temp 68-66-64-63C (core 1-2-3-4) with Prime 95


Ambient Temp: 26C

Hyper-Threading:Off (on for some tests)
LLC/vDroop: Off
SpeedStep: Off
Turbo: Off

Other Voltages:
QPI/VTT: 1.255v
CPU/PLL: 1.80v
PCIe: 1.50v
QPI/PLL: 1.10v
IOH Core: 1.240v
ICH I/O: 1.50v
ICH Core: 1.240v
DRAM: 1.540v (1333 Mhz)

Message quoted 1 times
Message edited by SpinachEater on 07-22-2010 at 06:47:06 AM
Reply to SpinachEater

Blckhaze wrote :

920
Revision: D0
Stepping: 5

 

Frequency: 3.990
Voltage: 1.231

 

Motherboard: P6T V2
Memory: 12gb 1600 Dominator 8-8-8-24 @ 1520mhz
Case:

 

Cooler: H50
Idle Temp: 38~
Load Temp: 71~
Ambient Temp: 24c

 

Hyper-Threading: 8
LLC/vDroop:
SpeedStep: ON
Turbo: ON

Other Voltages: Dram 1.35 & 1.66

 

1.225vcore ran for 3 hours before prime95 stopped working
1.231vcore ran prime95 for 9+ hours in 2 different instances, with memtest x4.
Overclocking to 4.2ghz required the vcore to be 1.263, so it was quite a big jump for that (the 200 bclk) so I think 1.231 with bclk 190 is the sweet spot on this 920.

 

Nice, thx blckhaze. :wahoo:

 

Yeah, the jump to 200 BCLK is much larger for me too, so it's not really worth it.


Message edited by redechelon on 07-20-2010 at 08:57:48 AM
Reply to redechelon

SpinachEater wrote :

Ok, I am currently stuck at 3.5GHz and I am trying to minimize Vcore to drop below 1.29375V because it sounds like I should be able to go a lot lower ATM. I will update this post as I find the lowest voltages.

920/930/other: 930
Revision: D0
Stepping: 5
Batch No.: ?

Frequency: 3.50
Voltage: 1.29375V
Multiplier: 21x
Blck: 166
Uncore: x20 ( = 2x RAM clocks)

Motherboard: Gigabyte GA-X58A-UD3R
Memory: 6 GB Mushkin (3x2GB) 8-8-8-24
Case: Antec 900

Cooler: Xigmateck S1283
Idle Temp: 40-36-38-34C Max (core 1-2-3-4)
Load Temp: 61-59-58-55C Max (core 1-2-3-4)
Ambient Temp: 26C

Hyper-Threading: Off
LLC/vDroop: Off
SpeedStep: Off
Turbo: Off

Other Voltages:
QPI/VTT: 1.255v
CPU/PLL: 1.80v
PCIe: 1.50v
QPI/PLL: 1.10v
IOH Core: 1.240v
ICH I/O: 1.50v
ICH Core: 1.240v
DRAM: 1.540v (1333 Mhz)



Just a thought, but try increasing your QPI PLL to 1.3-1.35 or something to see if it helps.

Reply to redechelon

Spot on sir. I am up to 3.90 now and running pretty stable. I updated my post for the current settings. Now I can't seem to break 4.0 but I am happy with what I am getting at 3.9

I can prob go through and drop some more voltages to fine tune it. I am pretty shocked at how high you can get the CPU clocks with such a low Vcore.

Message quoted 1 times
Message edited by SpinachEater on 07-21-2010 at 06:11:45 AM
Reply to SpinachEater

SpinachEater wrote :

Spot on sir. I am up to 3.90 now and running pretty stable. I updated my post for the current settings. Now I can't seem to break 4.0 but I am happy with what I am getting at 3.9

 

I can prob go through and drop some more voltages to fine tune it. I am pretty shocked at how high you can get the CPU clocks with such a low Vcore.

 

Awesome, that looks pretty good. :sol: 1.25v & 61C. I'll update the main post if it means anything lol...

 

And I dunno how it does it, it just does. I raised a few of the voltages a bit actually, they crashed in long Prime95 tests. I updated the posts a bit back...


Message edited by redechelon on 07-21-2010 at 07:31:34 AM
Reply to redechelon

Yeah people poo on the Xigmatek S1283 cooler for the new i7s but it is rocking out for me.

I was down by your vcore (1.206v) but I was getting a single core crashing out in Prime95. I am prob a bit over the necessary vcore but I gave up tweaking last night because I wanted to get some gaming in!

What is great is that when I game, nothing uses 100% of all 4 cores so I am even below 61C in real life temps.

Reply to SpinachEater

The cooler rates pretty well on frostytech.

 

Yeah, Prime95 and other stress tests always produce much higher temps than full load in games. Some overclock their CPU to temps further than recommended because they can count on the CPU never reaching that temp in real-world applications... I wouldn't though. =P


Message edited by redechelon on 07-22-2010 at 08:37:20 AM
Reply to redechelon

Hmm, I did some HT ON vs OFF tests. I ran 3D Mark 06 at my 3.9Ghz

HT ON
Overall: 25316
CPU: 6747

HT OFF
Overall: 24876 (-440 points, -1.7%)
CPU: 6280 (-467 points, -6.9%)

There is an ever so slight difference but it is probably all due to the CPU score from the CPU rendering tests.

Reply to SpinachEater

I think i got me a golden chip I got a 930 running stable at 4.3 205x21 with hyperthreading on with 1.296v never tops 75C under prime/Linx ( I'm using a Noctua D-14 and AS5 btw)

------------------------------ i7 930 @ 4.3ghz | GA-X58A-UD3R v2| 6GB Mushkin Redline Enhanced DDR3 1600 6-8-6-24 1T |XFX 5870 1GB | 2 Samsung F3 500GB Raid 0 Noctua D-14 | HAF 922 http://valid.canardpc.com/cache/banner/1304158.png
Reply to Hard Line

Can you post your settings with red's format way above? It will be helpful to see what all of your voltages are set to other than Vcore

Reply to SpinachEater

Interesting results... thanks Spinach, I still have to test mine too.

 

Thanks Hard Line, I'll post yours template or not, but it would be nice to see. That is a nice chip.


Message edited by redechelon on 07-23-2010 at 03:28:14 AM
Reply to redechelon

My i7 920 runs at 21x191 1.2v

Ran Intel Burn in with 20 runs at max memory and 8 threads.

Test was successful. 75c was the highest core temp using a D14 in a warm room.

Idle Temps around 45c

Gaming Temps in the Mid 50s

HT and Turbo on

All other voltage on Auto

All other settings at default values.


Message edited by mdbrotha03 on 07-23-2010 at 07:50:44 AM
------------------------------ http://valid.canardpc.com/cache/banner/1688524.png
Reply to mdbrotha03

I appreciate all the info you gave, but if I'm going to post it, I need all of it.. Idle temps, other voltages, HT?, etc

Reply to redechelon

redechelon wrote :

I appreciate all the info you gave, but if I'm going to post it, I need all of it.. Idle temps, other voltages, HT?, etc



updated

------------------------------ http://valid.canardpc.com/cache/banner/1688524.png
Reply to mdbrotha03

 

Thank you. Just to make sure, it's 1.200v? and I'm assuming that it's a D0?


Message edited by redechelon on 07-23-2010 at 07:45:39 AM
Reply to redechelon

Yes, its a DO. I also have my CPU z val in my sig now. When I put 1.2v in the bios it comes out as 1.192v. Also its 191x21. If you at CPU z its 190.84

Message quoted 1 times
Message edited by mdbrotha03 on 07-23-2010 at 07:51:35 AM
------------------------------ http://valid.canardpc.com/cache/banner/1688524.png
Reply to mdbrotha03

mdbrotha03 wrote :

Yes, its a DO. I also have my CPU z val in my sig now. When I put 1.2v in the bios it comes out as 1.192v. Also its 191x21. If you at CPU z its 190.84



Alright, updated =]

Reply to redechelon

Hey guys, I am having some trouble getting my rig stable at 4ghz. Although, I am not exactly sure what to set my voltages at to get it stable. As it stands now, i can run prime95 for about 5min before it will freeze or bsod. I had it running stable at 3.52 ghz (22/160) at 1.2125 vCore with temps getting into the upper 60s... so i know i can push it further. Any guidance would be greatly appreciated!

920/930/other: 930
Revision: D0
Stepping: 5
Batch No.: ? (not sure, at work)

Frequency: 4.01ghz (21/191)
Voltage: 1.225

Motherboard: Asus P6X58D-E
Memory: 8gb G.SKILL Ripjaws Series DDR3 1600 (1.6v 7/7/7/24)
Case: Silverstone Fortress FT02B

Cooler: Cooler Master Hyper 212+
Idle Temp: 42-46 across all for cores
Load Temp: 72-75 across all 4 cores
Ambient Temp: ~28

Hyper-Threading: Yes
LLC/vDroop: Yes
SpeedStep: No
Turbo: No

Other Voltages: Everything is Auto except for vCore and DRAM voltage

Reply to awilson

I wouldn't run with auto voltage settings because you have no clue where the instability is coming from. It could be your RAM, could be CPU, could be the chipset...

Reply to SpinachEater

Hey awilson. Yeah try setting some of our voltages to match ours. The first thing I always ask is what your memory multiplier is at... if you're increasing your BCLK, it's raising what your memory is running at. So at 191, you need it at 8x for a total of 1528MHz.

Reply to redechelon

I just tested Heaven with HT on and off max settings, I saw a 1.8% loss also... which is tiny. I'm sure in a more CPU taxing program or in the Physics test of 3DMarkVantage, it would score lower.. but for gaming, it practically makes no difference in performance with a 10C temperature drop.

 

(Graphics card side note... I also tried turning on ambient occlusion for one of them which looks awesome, but made some textures act strangely. In a game like CS:S, or tf2, it helps add much more detail and doesn't hurt performance too much. Have you checked this out?)


Message edited by redechelon on 07-23-2010 at 08:42:16 PM
Reply to redechelon

Alright, so i read through a few more threads and tutorials and messed around with my voltages a little more and got it stable running prime95 for about 30min. I know this isnt really a long enough time to tell if it will be stable or not, but i just wanted to check and make sure my voltages looked good before i run a more extensive stress test.

according to cpu-z, i have:

4009.1 MHz (21x190.9)
vCore: 1.2875 (CPU-Z is reading 1.264??)
QPI: 1.275
PLL: 1.88
IOH: 1.14
DRAM: 1.6 (7/7/7/24)

and i got temps of: (I figure these are about right since the ambient temp is 34 degrees right now)
idle: 48 46 48 45
load: 81 80 77 78

Reply to awilson

awilson wrote :

Alright, so i read through a few more threads and tutorials and messed around with my voltages a little more and got it stable running prime95 for about 30min. I know this isnt really a long enough time to tell if it will be stable or not, but i just wanted to check and make sure my voltages looked good before i run a more extensive stress test.

according to cpu-z, i have:

4009.1 MHz (21x190.9)
vCore: 1.2875 (CPU-Z is reading 1.264??)
QPI: 1.275
PLL: 1.88
IOH: 1.14
DRAM: 1.6 (7/7/7/24)

and i got temps of: (I figure these are about right since the ambient temp is 34 degrees right now)
idle: 48 46 48 45
load: 81 80 77 78



Looks pretty good. 30 min isn't a long time like you said, but it's long enough to suggest you're close.

80C is about right considering your ambient and your cooler. I always recommend the 212+ from a value standpoint, but at 4GHz 75-80C sounds right.
I think your voltages are reading lower because you have LLC enabled.

Reply to redechelon

Got a chance to do more hyperthreading experiments. I play a lot of bad company 2 so since it doesn't have a benchmark I thought I would use FRAPS and try something like walk through an empty map on the same path. I used Panama Canal and walked down the middle of the road with HT on and HT off.

I noticed when I loaded the game up it maxed out all 8 threads and also when I play, it seems to use a lot of hyperthreading (using all 8 threads as well) vs some games that only use 4 threads and the other 4 have no activity. I thought HT being on would be a great benefit in this case but it turns out that it doesn't make much of a difference with my settings.

CPU - 3.90 Ghz
GPU - 780Mhz
Resolution - 1680x1050


HT ON
MIN-55,
MAX-89,
AVG-70.345

HT OFF -
MIN-57,
MAX-91,
AVG-70.172


Message edited by SpinachEater on 07-24-2010 at 11:01:32 PM
Reply to SpinachEater

Huh... Half way through after you said it was using 8 threads, I was expecting you to say it made a huge difference.

 

Well, from looking at multiple reviews, benchmarking, and seeing your results too... HT doesn't benefit games. I mean... yes maybe a very very small percentage, but not only is it invisible to the naked eye... numbers confirm that it's VERY small, if any. A lot of game benchmarks I was looking at even saw a 1-2% loss on average, which is similar to your findings.

 

If all you do is game... and don't render/edit/model/etc, it would seem foolish to have Hyper-Threading on. Even if it yields a 5C drop, that's nice.


Message edited by redechelon on 07-25-2010 at 06:23:18 AM
Reply to redechelon

Me too. I saw how it was using all of the threads pretty heavily so I was thinking that it would make a difference. My only other thought is that perhaps the high CPU clocks are making me GPU bound and HT doesn't matter at this point.

I can either downclock the CPU and do HT ON/OFF or go down to a lower resolution and maybe I will become CPU bound again.

Reply to SpinachEater

That's a good point... Lowering resolution would work I bet, I'm not sure if you could lower an i7-930 enough though. Well, at least we know that at 4.0GHz, it doesn't make a difference =P

Reply to redechelon

ok here goes

920/930/other: i7 930
Revision: D0
Stepping: 5
Batch No.: 3002A652

Frequency: 4300
Voltage: 1.296v

Motherboard: Gigabyte X58A-UD3R rev 2
Memory: 6GB Mushkin Redline Enhanced 6-8-6-24-1T @1640mhz
Case: Haf 922

Cooler: Noctua D-14
Idle Temp: 45C
Load Temp: 75C
Ambient Temp: 33C

Hyper-Threading: ON
LLC/vDroop: LVL 2
SpeedStep: ON
Turbo: OFF

Other Voltages:

CPU PLL: 1.8
QPI/Vtt: 1.355
DRAM 1.66
DRAM Termination: 0.75
QPI-PLL: 1.1
IOH Core: 1.18
ICH I/O: 1.5
ICH Core: 1.1

------------------------------ i7 930 @ 4.3ghz | GA-X58A-UD3R v2| 6GB Mushkin Redline Enhanced DDR3 1600 6-8-6-24 1T |XFX 5870 1GB | 2 Samsung F3 500GB Raid 0 Noctua D-14 | HAF 922 http://valid.canardpc.com/cache/banner/1304158.png
Reply to Hard Line
Register or log in to remove.
Previous
1 2
Tom's Hardware > Forum > Overclocking > CPUs > i7 920/930/9xx Sweet Spots
Go to:

There are 1454 identified and unidentified users. To see the list of identified users, Click here.

Please mind

You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.

Add a reply Cancel
  • Ask the community now
  • Publish
Ad
Ads
Latest best answer
Will this be a good build to overclock 2500k to 5.0GHz?
By DJDeCiBeL, 17 hours ago:

I'm the one that told him that a Silver Arrow will get to 5Ghz, so I'll take the blame...

Best offers
They won a badge
Join us in greeting them
Top experts