Sign in with
Sign up | Sign in
Your question

How's this overclock of i7 860? Anything wrong?

Last response: in Overclocking
Share
December 7, 2009 4:47:24 PM

I've tested this and it is completely stable (even if only one core's working). The temperatures stay in range. But I'm worried that my lack of OC knowledge has left my system inefficient. In particular, I'm wondering if I should loosen the RAM timings and increase the RAM speed. How does the 1:1 ratio work with QPI? Also, will Turbo Mode or hyperthreading hurt my speed or is that only when you get to higher overclocks? Does anything stick out as bad?

BCLK: 145Mhz
Multiplier 22-26x
CPU Speed: 3190-3770Mhz
QPI Link: 2610Mhz

Turbo Mode: On
Hyperthreading: On

RAM ratio: 10x
RAM Speed: 1450Mhz
RAM Timings: 7-7-7-20

Voltages: All on stock

System: i7 860, 4GB DDR3, GA-P55M-UD4, HIS 4890, Vraptor

More about : overclock 860 wrong

December 7, 2009 5:24:05 PM

3770 on stock is pretty nice and safe ! one good cpu i guess . my i5 goes to ~3.6 (180x20 , turbo disabled) on stock
m
0
l
December 7, 2009 7:38:50 PM

Yeah, I've got a foxconn socket, so I'm nervous about raising the voltage. Being able to get almost an entire 1Ghz over the stock 2.8Ghz without a vcore change is great. I haven't tried above 3.6Ghz on all four cores with turbo disabled.

I'm really just confused about whether I should try for a higher QPI or RAM speeds. How does the FSB:RAM ratio work when there is no FSB to speak of?
m
0
l
Related resources
December 8, 2009 4:30:40 AM

no , with an integrated memory controller dont worry about fsb:D ram and stuff , you want low latency , and then higher speed . 7-7-7-20 on 1450 is pretty neat , post a link to your memory . more speed will produce more heat especially if its 1.65 volts .
m
0
l
December 8, 2009 4:41:01 AM

Here's the RAM I have. I had heard that low latency was better than higher speeds, so I immidiatly went to 1600Mhz 7-7-7-20. My overclock gave me the options of 1450Mz or 1740Mhz, but 1740Mhz wasn't stable. I haven't changed the voltage, so it's still at 1.6V.

http://www.newegg.com/Product/Product.aspx?Item=N82E168...

So, I shouldn't worry about getting QPI or Uncore speeds up?
m
0
l
December 8, 2009 5:43:05 AM

QPI does not exist in lynnfield .

for uncore , check this out :-

http://rampantspeculation.com/forums/showthread.php?p=3...

that guy is getting gains in superPI and stuff with higher uncore multiplier .

you can try the 1740 mhz by bumping ram voltage to 1.65 volts and try different CAS settings , but dont go above that (max for core i5/i7) .
m
0
l
December 8, 2009 7:03:50 AM

Thanks for the link. It's exactly what I was looking for.

With the Lynnfields, the Uncore core multiplier is locked at 18x. This means that I can only change it by changing my BCLK. From the testing I did earlier, changing the QPI (DMA?) and Uncore speeds showed a small level of improvement in games, but I'll try some other benchmarks.

Would raising the memory to 1740Mhz be worth it if I had to change the CL to 8? And 1.65V won't be too hard on my RAM will it?
m
0
l
December 8, 2009 8:05:07 AM

ripjaws have an official upper limit of 1.65 volts .

as latency = CAS cycles / cycle time , 1740 mhz will give you slightly better latency and much better bandwidth .

your memory is 1.6v 9-9-9-24 1866 , so :-

1. run 1.65v(not 1.6) 1740 9-9-9-24 . it of course should work , relaxed than base timings and more voltage (max allowed on p55 platform) .

2. try tightening the timings from there . with so many combinations , in 4 variables , it can be a long process indeed .

post your result !
m
0
l
a b à CPUs
a b K Overclocking
December 8, 2009 1:51:23 PM

Quote:
If I had a foxconn socket, I would return it before you burn it up and void your warranty.


Incomplete uninformed advice.
m
0
l
December 8, 2009 4:31:11 PM

notty , what vcore are you using for ur 24x160 ? bumped any other voltages too ? ure one lucky fellow if u say "stock" . thanks in advance .
m
0
l
December 8, 2009 5:33:08 PM

When I went to my memory settings, I realized that the RAM voltage was only set to 1.5V. So, I bumped it to 1.55V and have 1740Mhz 8-8-8-20 running stable. Does the RAM voltage go through the CPU in any way? I'm having more and more people tell me what zipzoomflyhigh said.

My computer is taking 2-3 minutes to get past the post screen, and I have to wait 3-5 minutes at the Windows Login screen before my wireless mouse and keyboard work. This is making overclocking a real pain in the ass.
m
0
l
December 8, 2009 5:42:18 PM

ive just read ur post , will reply for the socket issue as i have to sleep now as its late in india ! i just finished reading all 16 pages from extremesystems regarding more on that :-

http://www.xtremesystems.org/forums/showthread.php?t=23...

for the memory , yes , the cpu drives the ram . the memory voltage setting is for the cpu's integrated memory controller which can get fried above 1.65 volts according to intel .
m
0
l
December 8, 2009 6:32:08 PM

Update: Even with the voltage at 1.65v, CL 7 timings are very unstable.

Zipzoomflyhigh: Although you are right that problems have been found to occur with these sockets, I think that the stories have been exaggerated. Foxconn sockets are still on nearly all 1156 Asus and Gigabyte boards being sold. If the percentage of breaking boards/cpus was that high, I hope that they would stop selling them by now. Foxconn has even reportedly fixed the problem. However, I've decided that I won't up the voltages on anything related to the CPU.
m
0
l
December 8, 2009 11:27:20 PM

Dougx1317 said:
When I went to my memory settings, I realized that the RAM voltage was only set to 1.5V. So, I bumped it to 1.55V and have 1740Mhz 8-8-8-20 running stable. Does the RAM voltage go through the CPU in any way? I'm having more and more people tell me what zipzoomflyhigh said.

My computer is taking 2-3 minutes to get past the post screen, and I have to wait 3-5 minutes at the Windows Login screen before my wireless mouse and keyboard work. This is making overclocking a real pain in the ass.



Interesting that you are having slow posting. I have been running my machine for about one month now at ~ 3.5ghz (160x22)...stable based on prime 95. Running my memory at stock 1600 (160x10). This is all on air with no increase in default voltages. Rock solid stable.

But....I was finding it annoying to wait for the boot, and it 'felt' like applications took a long time to respond to mouse input.

On a whim, I reset my bios to "optimized default", which ran the processor in turbo mode, ran the bclk at 133mhz, and set the multimplier at anywhere between 9 at idle and 26 at peak. It underclocked my memory at 1333mhz. Result: lightning fast boots, and nearly instantaneous response of applications to mouse click.


Hmmm....does this make sense? Bottlenecks somewhere?

While I did lose a few frames/second in certain applications, the pleasure of responsiveness is attractive. I'd like to have the power of both the responsivness and the frames from the overclock.

Bob

i7 860
Cooler Master Hyper TX3 hsf
Gigabyte GA-P55A-UD4P
G.Skill Ripjaws series 8 GB (4x2GB) ddr3 1600
G92 8800 GTS (512 mb) video
520 Watt Crucial modular power supply
m
0
l
December 9, 2009 12:22:51 AM

bob5568 said:
I reset my bios to "optimized default"...lightning fast boots, and nearly instantaneous response of applications to mouse click.

Thanks. I'll try this and let you know if it works. It only started when I began overclocking, so it's a good idea.
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 12:44:20 AM

bob5568 said:
Interesting that you are having slow posting. I have been running my machine for about one month now at ~ 3.5ghz (160x22)...stable based on prime 95. Running my memory at stock 1600 (160x10). This is all on air with no increase in default voltages. Rock solid stable.

But....I was finding it annoying to wait for the boot, and it 'felt' like applications took a long time to respond to mouse input.

On a whim, I reset my bios to "optimized default", which ran the processor in turbo mode, ran the bclk at 133mhz, and set the multimplier at anywhere between 9 at idle and 26 at peak. It underclocked my memory at 1333mhz. Result: lightning fast boots, and nearly instantaneous response of applications to mouse click.


Hmmm....does this make sense? Bottlenecks somewhere?

While I did lose a few frames/second in certain applications, the pleasure of responsiveness is attractive. I'd like to have the power of both the responsivness and the frames from the overclock.

Bob

i7 860
Cooler Master Hyper TX3 hsf
Gigabyte GA-P55A-UD4P
G.Skill Ripjaws series 8 GB (4x2GB) ddr3 1600
G92 8800 GTS (512 mb) video
520 Watt Crucial modular power supply


Did you run 1600MHz at specified voltage(1.65V)? My machine(RAM@1440MHz,CL8) runs significantly faster with 1.65V compared to 1.58V(minimum stable voltage).
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 1:43:44 AM

Dougx1317 said:
Voltages: All on stock

Do you mean that you have all of the voltages running at the default setting of Auto? Or did you set them by hand to 'stock' levels? If you set by hand, which did you set and what values? I'm also curious what version of BIOS you are running.
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 2:04:17 AM

bob5568 said:
On a whim, I reset my bios to "optimized default", which ran the processor in turbo mode, ran the bclk at 133mhz, and set the multimplier at anywhere between 9 at idle and 26 at peak. It underclocked my memory at 1333mhz. Result: lightning fast boots, and nearly instantaneous response of applications to mouse click.

Did you try running your overclock with all speedstep, turbo, etc settings turned on? This would show if it is (a lack of) 'moving' multipliers or turbo functionality that is causing your delays.

I discovered that even some tweaks that I considered minor to BIOS were being considered OCing by Gigabyte and a number of BIOS settings that were set to Auto were being treated as Disabled, where they had previously been treated as Enabled. Certainly in your case where you did OC your bclk the BIOS would have done this. By making a few changes in BIOS I was able to keep turbo on and get my multiplier to idle at x9 and ramp up to x26 with one core maxed out.

The changes I had to make in the BIOS were on the Advanced CPU Core Features page:
Intel Turbo Boost Tech. - changed from Auto to Enabled
CPU Cores Enabled - left as All
CPU Multi-Threading - left as Enabled
CPU Enhanced Halt (C1E) - changed from Auto to Enabled
C3/C6/C7 State Support - changed from Auto to Enabled
CPU Thermal Monitor - changed from Auto to Enabled
CPU EIST Function - changed from Auto to Enabled
Bi-Directional PROCHOT - changed from Auto to Enabled

At ~3.5GHz you may want to try with Turbo disabled and enabled, although I have seen articles stating that was doable with Turbo on. (Hm. On quick glance this may be 3.5 with Turbo running? http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=364... )
m
0
l
December 9, 2009 3:23:35 AM

I had left all of the voltages on auto. I have a Gigabyte P55-UD4P. It has Award Software International, Version F3, Date 08/01/2009 bios.

Unfortunately, returning to optimized defaults didn't speed up the post.
m
0
l
December 9, 2009 4:41:28 AM

ekoostik said:
Do you mean that you have all of the voltages running at the default setting of Auto? Or did you set them by hand to 'stock' levels? If you set by hand, which did you set and what values? I'm also curious what version of BIOS you are running.


I left voltages at Auto, and was running bios f4. Based on your question, I looked...and there was a bios update...so I flashed up to f5. Although I couldn't boot into turbo mode...no settings seem to allow that, I returned my overclock, and now windows 7 is responsive.

It will be interesting to see if this was an important change. I also upped my memory voltage to 1.6.

My preference is to overclock as I have some apps that use multi cores, and those run fastest if all 4 cores can be active and fast at the same time!

Cheers,

Bob
m
0
l
December 9, 2009 4:46:30 AM

ekoostik said:
Did you try running your overclock with all speedstep, turbo, etc settings turned on? This would show if it is (a lack of) 'moving' multipliers or turbo functionality that is causing your delays.

I discovered that even some tweaks that I considered minor to BIOS were being considered OCing by Gigabyte and a number of BIOS settings that were set to Auto were being treated as Disabled, where they had previously been treated as Enabled. Certainly in your case where you did OC your bclk the BIOS would have done this. By making a few changes in BIOS I was able to keep turbo on and get my multiplier to idle at x9 and ramp up to x26 with one core maxed out.

The changes I had to make in the BIOS were on the Advanced CPU Core Features page:
Intel Turbo Boost Tech. - changed from Auto to Enabled
CPU Cores Enabled - left as All
CPU Multi-Threading - left as Enabled
CPU Enhanced Halt (C1E) - changed from Auto to Enabled
C3/C6/C7 State Support - changed from Auto to Enabled
CPU Thermal Monitor - changed from Auto to Enabled
CPU EIST Function - changed from Auto to Enabled
Bi-Directional PROCHOT - changed from Auto to Enabled

At ~3.5GHz you may want to try with Turbo disabled and enabled, although I have seen articles stating that was doable with Turbo on. (Hm. On quick glance this may be 3.5 with Turbo running? http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=364... )



Thanks for your detailed post. I am now running 3.5 with turbo "on" at the slowest setting, "cruise", I beleive its called. Seems to have very little if any effect. I did use your tip to "enable intel turbo boost tech. rather than auto.

I'll experiment more as the days go....I just flashed the new bios, f5, and I haven't decided if its made a difference or not.

Cheers,
Bob
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 12:49:05 PM

bob5568 said:
Thanks for your detailed post. I am now running 3.5 with turbo "on" at the slowest setting, "cruise", I beleive its called. Seems to have very little if any effect. I did use your tip to "enable intel turbo boost tech. rather than auto.

The setting in BIOS that you can update to cruise is C.I.A.2. I assume this is what you are talking about? This doesn't directly impact Turbo, at least not as 'Turbo' is most often referred to these days. Turbo is your CPUs ability (so Intel provided, native to the CPU) to dynamically ramp up under heavy loads based on the number of cores that are active. C.I.A.2 is a feature of the motherboard that Gigabyte has provided to dynamically overclock based on load needs.

Since you have already OCed your board, I would recommend setting C.I.A.2 to Disabled. Leave Intel Turbo Boost Tech on Enabled as long as it is stable to get max use out of the i7 860's increased multipliers for single or dual threaded/core use.

Here's a quote from one article on C.I.A.2, somewhat dated but still helpful in explaning what it does.
Quote:
... [GIGABYTE’s C.I.A 2] is a dynamic overclocking feature and it will overclock the bus speed of the motherboard when extra processing power is needed. This has several downsides; the most noticeable is that your RAM will be overclocked as well. Since you can still set your RAM speed manually even with DES enabled, this could cause your system to crash when C.I.A 2 kicks in, at least if you’ve tweaked it to its limits.

http://www.tweaktown.com/articles/1275/closer_look_at_g...
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 1:02:16 PM

Dougx1317 said:
I have a Gigabyte P55-UD4P. It has Award Software International, Version F3, Date 08/01/2009 bios.


Dougx1317,
You may want to consider updating your BIOS. Even if it doesn't fix the issue at hand, you are running with the BIOS as of release date. Gigabyte has fixed and improved a number of capabilities with their board in BIOS updates since then. Your board currently has an F5 version available since 11/20. You can find it here: http://www.gigabyte.us/Support/Motherboard/BIOS_Model.a...

If you do update your BIOS, do not use the windows based software provided @BIOS. It has been known to brick a number of motherboards if not done exactly right. The process for updating BIOS using QFlash - from within BIOS itself - is easy enough that there is no reason not to do it that way. Basically, you:
1. Download the latest BIOS file
2. 'Run' the file (double-click it, whatever your prefered metaphor is) to unpack the contents onto a USB drive.
3. Reboot your machine with the USB drive plugged into a USB slot
4. Enter BIOS
5. Enter Qflash - I believe using F8 but it will say at the bottom of your screen (it is possible to enter Qflash during POST without entering BIOS, either method is ok)
6. Find your USB drive - it may be labeled HDD
7. Select your BIOS file from the USB and let it run
8. When you get back into BIOS, select Load Optimized Defaults, then make any changes you have made previously

You can find additional information here: http://www.gigabyte.com.tw/FileList/NewTech/old_motherb...
There's a helpful pdf link at the bottom of the page, these instructions use a floppy disk but you can simply use a USB instead as I have described. The usb stick does not have to be bootable.
m
0
l
December 9, 2009 4:26:50 PM

ekoostik said:
The setting in BIOS that you can update to cruise is C.I.A.2. I assume this is what you are talking about? This doesn't directly impact Turbo, at least not as 'Turbo' is most often referred to these days. Turbo is your CPUs ability (so Intel provided, native to the CPU) to dynamically ramp up under heavy loads based on the number of cores that are active. C.I.A.2 is a feature of the motherboard that Gigabyte has provided to dynamically overclock based on load needs.

Since you have already OCed your board, I would recommend setting C.I.A.2 to Disabled. Leave Intel Turbo Boost Tech on Enabled as long as it is stable to get max use out of the i7 860's increased multipliers for single or dual threaded/core use.

Here's a quote from one article on C.I.A.2, somewhat dated but still helpful in explaning what it does.
Quote:
... [GIGABYTE’s C.I.A 2] is a dynamic overclocking feature and it will overclock the bus speed of the motherboard when extra processing power is needed. This has several downsides; the most noticeable is that your RAM will be overclocked as well. Since you can still set your RAM speed manually even with DES enabled, this could cause your system to crash when C.I.A 2 kicks in, at least if you’ve tweaked it to its limits.

http://www.tweaktown.com/articles/1275/closer_look_at_g...


Thank you for this information. You've corrected a misconception of mine, I thought CIA2 was simply the lever to enable the feature of Turbo...I had no idea how it really worked. This explains why using cia2 almost always defeated my effort to boot when running oc on memory and cpu.

I will do more exeriments tonight!

Cheers,
Bob
m
0
l
December 9, 2009 4:31:36 PM

Quote:
That's because Foxconn fixed the problem after Tom's hardware contacted them about the problem. They pulled all the problem sockets off the shelves.

How have the stories been exaggerated? Both ANandtech and Tom's hardware both reported the problem in their own testing, not to mention a ton of other sites.
That's not exaggeration, that's fact.

I simply stated what I would do and what I think would be the smart thing to do. Return the board now while you still can -OR- risk having to buy a new one after it burns.


I sympathize with Doug's conundrum. I too struggled with this issue, having initially ordered the Gigabyte p55-ud2. I eventually rma'd my board and bought the p55a-ud4p, but I almost didn't. One day I was sure that it wouldn't be a problem for me, and the next day I was sure that I didn't want to take the chance.

There is nearly no way to quantify the issue for any one person, so it boils down to how he feels.

Best,
Bob
m
0
l
December 9, 2009 5:48:01 PM

Quote:
That's because Foxconn fixed the problem after Tom's hardware contacted them about the problem. They pulled all the problem sockets off the shelves.

How have the stories been exaggerated? Both ANandtech and Tom's hardware both reported the problem in their own testing, not to mention a ton of other sites.
That's not exaggeration, that's fact.

I simply stated what I would do and what I think would be the smart thing to do. Return the board now while you still can -OR- risk having to buy a new one after it burns.




As you said, they pulled the problem boards. Wheres the problem?
m
0
l
December 9, 2009 5:52:54 PM

bob5568 said:
I sympathize with Doug's conundrum. I too struggled with this issue, having initially ordered the Gigabyte p55-ud2. I eventually rma'd my board and bought the p55a-ud4p, but I almost didn't. One day I was sure that it wouldn't be a problem for me, and the next day I was sure that I didn't want to take the chance.

There is nearly no way to quantify the issue for any one person, so it boils down to how he feels.

What really makes returning the board such a pain is that it was a gift from my parents, so I'd have to go through them to return it. It was also purchased as a combo deal, so I'd have to return the RAM that I've found is surprisingly good. There is also the matter of the 15% restocking fee that will be around $35, since the RAM would go with it.

I know that I'm at risk right now. I'm just not sure how to quantify the risk. Is it really worth all the hassels, time with no computer, and $35?
m
0
l
December 9, 2009 5:57:25 PM

I also wanted to add that despite the socket, I really like this board. It has freindly bios and works great. Overclocking is incredibly simple. And I did flash it to F5 bios; It took about 2 minutes.

I've got the same question as Sonic-Boom.
m
0
l
December 9, 2009 5:58:43 PM

If the board is less than 1 month old and you do not plan to overclock past 3.9ghz then keep it, there is no reason what so ever to RMA it if you are just going to keep it at 3.6ghz.
m
0
l
December 9, 2009 6:03:51 PM

no its not worth the hassle . i would be satisfied with 3770 at stock voltage . my i5 does not reach that , could be because of lower binning of 750 than 860 .

i checked my ud2 foxconn after 10 days since it started , total about 40 hours of usage and hours running, all at around 1.45 Vcore @4.00 ghz . while i was at it , i ran wprime to max out cpu and left in running overnight . to my relief , both the cpu pads and socket pins and even the plastic were all right when i checked it in the morning . so i could suggest that u dont have to be worried at stock voltage at your speed . but i wound not go any higher than that even for benchmarking , and as safety , re-clock it to around 3.8 at lower voltage . i would still be ok at 200-220 mhz less , although i admit 4.0 is a nice figure to reach !
m
0
l
December 9, 2009 6:08:20 PM

and for the ram , u could try 7-8-7-22 or something like that , basically a bit relaxed than 7-7-7-20 only if u wanna try it , not that it would make that much of a difference ! and compared to stock 9-9-9-24 for 1866 , your memory is fine at cas 8 managing 1740 mhz . nice ram . u cannot surely expect such result from a different piece of the same model !
m
0
l
December 9, 2009 6:10:29 PM

cyberkuberiah said:
no its not worth the hassle . i would be satisfied with 3770 at stock voltage . my i5 does reach that , could be because of lower binning of 750 than 860...

...i wound not go any higher than that even for benchmarking , and as safety , re-clock it to around 3.8 at lower voltage . i would still be ok at 200-220 mhz less , although i admit 4.0 is a nice figure to reach !

Thanks for reassuring me. 3.77Ghz is high enough for me, and it only gets that high with one core. Are saying that I should undervolt the processor? Would that make it safer? I thought undervolting was bad for a processor.

And as for the RAM, thought that the first number was the only one that effected performance significantly? Considering that I have my RAM at 1740Mhz CL 8 with only 1.5V, I'd say I've got some great RAM.
m
0
l
December 9, 2009 6:25:44 PM

i was being literal , referring to myself that i will undervolt the i5 1.45V 4.0 to around 1.33V ~3.8 ghz for safety .

yeah , keep that ram !
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 7:27:29 PM

Quote:

How have the stories been exaggerated? Both ANandtech and Tom's hardware both reported the problem in their own testing, not to mention a ton of other sites.
That's not exaggeration, that's fact.


You may want to take a look at XS forums, under UD6 Burn, same exact pictures as Anand article (because that's where they got their info from) and it was from EXTREME OCing and not just normal OCing that caused the socket burn problem.

m
0
l
December 9, 2009 7:41:51 PM

I'm sure that a few people may get burned processors without extreme overclocking, but I think it's a much smaller percentage of people than we've been led to believe. Socket 1156 processors have only been out for two months, nobody really knows what the long term percentage of burnouts will be right now.

If there really is that high of percentage of burnouts, we should all start a class action lawsuit against the companies that sell the destructive motherboards. But I have a feeling that the percentage is not going to be that high. Asus and Gigabyte rely too much on a good reputation to let a large percentage of their customers' computers blow. It would be a bad business move on their part, at least.
m
0
l
December 9, 2009 8:39:10 PM

I'll checked the processor pads and pins this weekend to see how they look. I'm still not thrilled about the idea of returning this, but Zipzoomflyhigh has effectively ruined any peace of mind that I had before. And he seems to be in the majority of people and reports I've found on the internet.

You're getting me paranoid now. Does anyone know if Newegg will wave the restocking fee, because of the Foxconn socket? Is there anyway to return the motherboard but not the RAM if they were sold as a combo? I assume there is no way for me personally to replace the socket, right?

I'm still not completely sure what I'll do, but I'm forming a contingency plan. The idea of $500 of hardware blowing because I was lazy seems bad to me. Plus I've got my netbook to hold me over.
m
0
l
December 9, 2009 9:02:02 PM

Wow really. Just chill out its fine. Anandtech said WORRY ABOUT IF YOU OVERCLOCK PAST 4.0GHZ. They didn't say send them all back. I have an Asus board with a Foxconn socket that I am using with an 860 and I have zero worries. This has been blown way out of proportion. Are you extreme overclocking? Your answer to that is weather you should RMA it or not.
m
0
l
December 9, 2009 9:04:54 PM

Oh and the Capacitor issue...You have it at stock voltage so those are fine.
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 9:24:57 PM

I gotta agree with Sonic-Boom on this one.
Dougx1317 said:
And he seems to be in the majority of people and reports I've found on the internet.

That's the vocal majority, many of which who are just out to stir people up. It's like a couple posters I see who every time they post a 'solution' to a problem it's: "RMA that [insert problem device here] immediately!!" No time taken to address or understand an issue, just shoot from the hip response that spreads misinformation and creates a lot of useless work but of course none for themselves. The rest of the folks out there you don't hear from are quite happy with their hardware and not worried about malfunctions occuring at pushing-the-envelope speeds at supercooled levels. Of course if you do plan on pushing your rig past 4.0GHz then you may want to considered RMA. But that kind of logical information is lost in the hysteria of the RMA IT ALL!! response.
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 10:31:05 PM

sonic-boom said:
Wow really. Just chill out its fine. Anandtech said WORRY ABOUT IF YOU OVERCLOCK PAST 4.0GHZ. They didn't say send them all back. I have an Asus board with a Foxconn socket that I am using with an 860 and I have zero worries. This has been blown way out of proportion. Are you extreme overclocking? Your answer to that is weather you should RMA it or not.


+1 ^

I've got a UD4p and an i7 860 and have had it over 1.5v (just while benching of course) and know a bunch of others who have run theirs over 1.6v WITHOUT a problem.

@4.0 GHz, I run 1.3v and not at all worried about it. Just people like zip who think the sky is falling because of EXTREME OC's that were generated by members at XS.

YES, if you are going to run LN2, Dice or phase by all means pick up a Lotes socket board, besides that don't worry at all.


m
0
l
December 9, 2009 10:36:28 PM

I see it halfway between ekoostik and zipzoom.

On one hand, there is no strong evidence of failure at levels of oc around 3.5...say. I learned long ago, that absence of information is not presence of comfort. Not intended to add fear to an already emotional concept...just recognize that engineering designs are normally presented with a factor of safety. If there is a reason why a poor manufacturing choice has reduced that factor of safety, its not a good thing.

That's about all we can say. Right???? All can agree that foxconns socket producing defects at high voltages exposes that there is a limit to the factor of safety that is unusual.

Fragility of design is not a step function....that is, old foxconn boards are NOT working "perfectly" until 4 ghz, and then magically "burn up" after 4ghz, even thou we hear that testers only found visible evidence after 4 ghz.

And remember that of a production lot of board you have a normal curve of quality, so if you are unlucky, you will need your board's design safety margin more than the lucky guy who got one of the best boards. That's why testimony is weak reasoning.

My sense, and what pushed me over the line to return my early foxconn board was that the engineers design was compromised. Pins were not making complete contact. That's simply a bad situation. Its not all that important to me that the testers feel that in their limited sample size there was still "some" safety margin.

Now...some pragmatic answers for Doug...

I also bought my board as a "combo deal". They simply accepted the board and took back the bundle savings instead of providing a complete refund. So, you don't have to rma the memory also.

In fact, Newegg was amazing...my purchase had been more than 30 days old, yet they understood that I'd waiting until receiving my win 7 before beginning my build, and they knew that in between the evidence of foxconns problems emerged. Based on that, they waived the 30 day limit and provided me the rma.

I do sympathize with the gift complication. You have to decide what to do based on how it feels to you. Anyone that professes to KNOW what you'll experience is predicting the future.

No one can really do that.

Best,
Bob
m
0
l
December 9, 2009 10:41:11 PM

I went through all of this when I first bought my motherboard, and now Zipzoomflyhigh has resurrected all my original concerns. I had originally agreed with Sonic-Boom and ekoostik. I'll consider the RMA option, but I'm enjoying the board I have now. So, I'll let you guys know what I decide.

Thanks for your input Bob. Knowing that I can keep the RAM makes this a little easier. If they'll wave the restocking fee, I'd probably return it. Part of what motivates me is that the P55A-UD3 has a Lotes socket, USB 3.0, SATA-6GB/s, and would actually be $5 cheaper. Although I still think the Foxconn effect is overemphasized, I'll admit that peace of mind is worth something to me.
m
0
l
a b à CPUs
a b K Overclocking
December 9, 2009 10:41:14 PM

OK, if everyone is saying the Foxconn socket boards will burn up at +/- 4.0 GHz explain why at HWBOT the top 5 record OC's (CPU-Z) for i7 860 are all Foxconn socket MB's.

Like to have someone explain that one, all over 5 Ghz's.

m
0
l
December 9, 2009 10:48:54 PM

RJR said:
OK, if everyone is saying the Foxconn socket boards will burn up at +/- 4.0 GHz explain why at HWBOT the top 5 record OC's (CPU-Z) for i7 860 are all Foxconn socket MB's.

Like to have someone explain that one, all over 5 Ghz's.

The processors have only been out for 2 months. Run them like that for a year or so and they could blow. I'm not saying they will, I just thought I'd bring up the long term aspect of it.

And maybe they just got lucky?
m
0
l
December 9, 2009 11:10:29 PM

RJR said:
OK, if everyone is saying the Foxconn socket boards will burn up at +/- 4.0 GHz explain why at HWBOT the top 5 record OC's (CPU-Z) for i7 860 are all Foxconn socket MB's.

Like to have someone explain that one, all over 5 Ghz's.



As I posted earlier, testimony and scant data is a foolish premise to base an argument on.


When I buy a product, I'd like it to conform to it's designers plan. Wouldn't you agree that the designers at Intel and Foxconn intended each pin on the CPU to make good electrical contact to the mother board?

If it does so, then there would not be the demonstrated difference between the lotes and foxconn sockets.

Since such difference HAS been demonstrated, accepting ownership of the flawed product is only logical if the pain of doing anything else is significant.

Afterall, we build our computers once every 2 years, or so. We USE them every day. Shouldn't the build be given priority and be the best we can acheive.

Why argue that a person should build his box with a known defect? Especially if the defect can be eliminated at no significant cost?

Bob
m
0
l
a b à CPUs
a b K Overclocking
December 10, 2009 1:11:36 AM

bob5568 said:
As I posted earlier, testimony and scant data is a foolish premise to base an argument on.


When I buy a product, I'd like it to conform to it's designers plan. Wouldn't you agree that the designers at Intel and Foxconn intended each pin on the CPU to make good electrical contact to the mother board?



Exactly, as you have stated, scant data is all that is available. I can't seem to find any credible source that can provide proof (one way or the other) of your concluded design flaw. If you are coming to a conclusion that a design flaw exists and is reproduceable and is being covered up by both Intel and Foxconn with the 10's of thousands of 1156 motherboards out there already I'm thinking a credible source would have unvailed this conspiracy by now. Well, I guess I'll just enjoy my OCed i7 860 until the conspiracy breaks then.



m
0
l
December 10, 2009 4:24:06 AM

RJR said:
Exactly, as you have stated, scant data is all that is available. I can't seem to find any credible source that can provide proof (one way or the other) of your concluded design flaw. If you are coming to a conclusion that a design flaw exists and is reproduceable and is being covered up by both Intel and Foxconn with the 10's of thousands of 1156 motherboards out there already I'm thinking a credible source would have unvailed this conspiracy by now. Well, I guess I'll just enjoy my OCed i7 860 until the conspiracy breaks then.



Sorry RJR, I didn't realize you did not accept the evidence presented by the photos on Anandtech and other places. I call that evidence of a design flaw. If you believe such photos are evidence of a conspiracy, then we are from such different world views that further conversation would be a mistake.

Cheers,

Bob
m
0
l
December 11, 2009 7:37:38 PM

I just thought I'd let you guys know that I did return the motherboard. I went with the GA-P55-UD3. I did this for the following reasons:
- Newegg waved the restocking fee.
- It was two different orders, so I get the new board before I send the old one back. (Always have a working computer.)
- The Lotes socket does give me some peace of mind and overclocking confidence.
- The new board has features like USB 3.0 and SATA 6Gb/s.
- The new board was cheaper.
- I only lost features that I won't use like more eSATA ports, TPM, and 8x/8x crossfire.

I don't know if this was necessary, but it didn't take much effort and saved me some money. When it comes to high end computer hardware, I decided that it wasn't worth the risk. Even if it was a very small risk. Plus I got to trade features that I'll never use for features that I might end up using.
m
0
l
December 12, 2009 3:58:17 AM

no problemo . no more worrying !
m
0
l
December 12, 2009 4:00:51 AM

although i checked it before as i wrote above , by the way ill keep my cpu at 4.0 on this foxconn ud2 and report if there is anything comes up in the long term ... so far so good at least till now .
m
0
l
!