Intel Core i7 Review Thread

[JumpingJack]

So if i can set memory multiplier to 6, i can easily reach 200 x 20 = 4Ghz ( 200 Bclock x 6 Memory Multiplier = 1200Mhz Mem). In that case my 1333Mhz rams doesnt block my overclocking . Is it right ?

You are correct.

[ap4lifetn]

My system is up and running




P6T has the memory dividers and also list your ram speed at that certain multiplier.

[MGreg]

Seeing stock at an online store now.
Hope its a good sign.

[herderien]

My system is up and running

P6T has the memory dividers and also list your ram speed at that certain multiplier.

what cooling do you use ?

[T_Flight]

You probably won't be seeing him again for hours at least...maybe days. He's gotta be having a blast!

[Jacky]

Even though this thread has turned to more of a general Q&A you can add another review, this time from ars. http://arstechnica.com/reviews/hardw...-review.ars/11

[ap4lifetn]

What are you guys getting as VID on your Core i7's? Realtemp says mine is .9750 but I don't think its right considering CPU-Z reads 1.048V at idle, and about 1.2V in BIOS

[strange|ife]

lol been so out of loop just found out x58 supports SLI

these are looking very nice. keep the reviews coming in people

green with envy

[Drwho?]

if you want to put face on people who did the baby .. here is few of them: http://www.podtech.net/home/5436/intels-core-i7
Glenn, Ronak,Rani, and don't forget all of the others in the pics!
You can see my lab into the video, where Matt, Regina are standing and where I am standing.

You can see, we have A LOT of TOYs

[Donnie27]

if you want to put face on people who did the baby .. here is few of them: http://www.podtech.net/home/5436/intels-core-i7
Glenn, Ronak,Rani, and don't forget all of the others in the pics!
You can see my lab into the video, where Matt, Regina are standing and where I am standing.

You can see, we have A LOT of TOYs

Good stuff, thanks!

[DeanZ]

hey found this on anandtech apparently says max vcore is 1.55 before damage http://download.intel.com/design/pro...hts/320834.pdf page 22...on school wireless so I cannot check myself it takes too long

also apparently The maximum VTT voltage is 1.35V (for the UNCORE portion of the processor)
The maximum VDIMM voltage 1.875V
The maximum VPLL voltage is 1.89V

[Hornet331]

hey found this on anandtech apparently says max vcore is 1.55 before damage http://download.intel.com/design/pro...hts/320834.pdf page 22...on school wireless so I cannot check myself it takes too long

also apparently The maximum VTT voltage is 1.35V (for the UNCORE portion of the processor)
The maximum VDIMM voltage 1.875V
The maximum VPLL voltage is 1.89V

nothing new, intel also spec vcc 1.45max for C2D.

If you go over 1.55V it does not instantly damage your you...

[JumpingJack]

hey found this on anandtech apparently says max vcore is 1.55 before damage http://download.intel.com/design/pro...hts/320834.pdf page 22...on school wireless so I cannot check myself it takes too long

also apparently The maximum VTT voltage is 1.35V (for the UNCORE portion of the processor)
The maximum VDIMM voltage 1.875V
The maximum VPLL voltage is 1.89V

Yeah, what Hornet said above... nothing really to this... every processor has specs so that system builders (motherboard makers) know how to make their boards and what not, and also to establish warrantee boundaries.

Intel and AMD set their spec conditions based on testing the product under various conditions, where they determine the safe allowable operating zone and still enable a 3 year warrantee ... if you keep your voltage at or below 1.55 V, then you have not violated the warrantee agreement ... on this parameter.

This is no different than the total allowed weight over a bridge, or the max tire pressure for your car.

[DeanZ]

Yeah, what Hornet said above... nothing really to this... every processor has specs so that system builders (motherboard makers) know how to make their boards and what not, and also to establish warrantee boundaries.

Intel and AMD set their spec conditions based on testing the product under various conditions, where they determine the safe allowable operating zone and still enable a 3 year warrantee ... if you keep your voltage at or below 1.55 V, then you have not violated the warrantee agreement ... on this parameter.

This is no different than the total allowed weight over a bridge, or the max tire pressure for your car.

yeah I know that...but for us safe overclockers its kinda helpful until we learn what these chips can take...I want to see how these chips handle at or above this voltage before I try to make a 24/7 machine thats high volted

[T_Flight]

Exactly. Many of us do not want to risk these things until we see what they are gonna do long term. Now I would be glad to do testing for Intel to find the limits if they supplied me with chips if I torch them, but at a 1000 bucks a pop Extreme i965's don't grow on trees. hehe 10 CPU's would equal 10 grand and that's not even a representative sampling.

I'm personally staying on the conservative side with mine until we see, then I will increase voltages if it helps performance and doesn't cause slow death. I guess we could call that SiDS...short for Slow i7 Death Syndrome.

I don't wanna experience SiDS, so I will chill, so I don't kill, then go Extreme, and make sure the OC is "clean".

[cool_case]

Well, there's a kill voltage that's much higher than the reduce-the-chip's-life-by-half voltage. Running temps are a good indicator of chip life. For c2d's, running in the 50s reduced chip life by half on an expected 7-year 24-7 life, and in the 60s reduced it half again. I'm guessing, but with a tmax of 105, Nehalem's normal full-life running temp, looks closer to 50s than c2d's 40s.

So, a little above 1.55 (as noted above) isn't going to fry your chip right away, and you could probably get away with 1.6v 24/7 for a while, though, not if it's at load often. You're not likely to fry a chip at a too-high-for-air voltage of 1.8v. It just wont' boot, or it will shut down pretty soon after it does.

[Shintai]

Well, there's a kill voltage that's much higher than the reduce-the-chip's-life-by-half voltage. Running temps are a good indicator of chip life. For c2d's, running in the 50s reduced chip life by half on an expected 7-year 24-7 life, and in the 60s reduced it half again. I'm guessing, but with a tmax of 105, Nehalem's normal full-life running temp, looks closer to 50s than c2d's 40s.

So, a little above 1.55 (as noted above) isn't going to fry your chip right away, and you could probably get away with 1.6v 24/7 for a while, though, not if it's at load often. You're not likely to fry a chip at a too-high-for-air voltage of 1.8v. It just wont' boot, or it will shut down pretty soon after it does.

Fish?

Thats maybe the most made up stuff I heard in along time. Only OC/voltage affects its life, unless you get over the throttle temperature.

The expected life of CPus are 25-250 years with normal operations. And 60C is normal. Overvoltage and OC is the killer. And those parts usually makes small errors first where operation still seem safe and secure. Prime makes no errors etc. But some transistors are broken and from there it continues.

But your numbers are completely pulled out of the air.

[cool_case]

Fish?

Thats maybe the most made up stuff I heard in along time. Only OC/voltage affects its life, unless you get over the throttle temperature.

The expected life of CPus are 25-250 years with normal operations. And 60C is normal. Overvoltage and OC is the killer. And those parts usually makes small errors first where operation still seem safe and secure. Prime makes no errors etc. But some transistors are broken and from there it continues.

But your numbers are completely pulled out of the air.

Not true. Those c2d temperature numbers came from a professor of EE with whom I work who was part of a team that did a series of therrnal testing experiments for processors.

Now, I didn't claim they were still true for Nehalem. But, it's a safe assumption that higher temperature deteriorate chip life for Nehalem too.

60C may be normal for Nehalem, but it was not normal for c2d.

It's also true that over-voltage affects chip life and that it, like you say, can damage transistors.

I said 7 years of normal chip life for a chip running 24/7, under fairly constant load, such as a server chip. A desktop chip would have a much longer lifespan, but 250 years?

If what you are saying is true Shintai "only OC/voltage affects chip life unless you get over the throttle temperature" then why should anyone bother with extreme air cooling, or with water cooling? Keeping temps down, gives better OCs and it reduces chip-related errors.

Here's further evidence that it's not just voltage and OC that affects chip performance and life. On LN2, many of us have run our C2Ds at 1.9v vcore with OCs from the mid 5Ghz to over 6Ghz. If high voltage or a high OC alone damaged transistors, those chips would have often died in these OC runs. And they didn't die often, only rarely. Temperature is a big factor. For extremely low temps, resistance is lower and conductivity is higher. Temperature is a factor on the high end too.

I'm going to stick with: 'keeping chip temps down well below the throttle temp lengthens chip life life' until I see definitive proof that this is not the case.

Nehalem is a different architecture than c2d. Perhaps the c2d temperature re: chip life information is outdated. If you have results from internal Intel experiments that show that constantly running a chip just below, but not over, its throttle temperature will not affect its life, or cause errors, or limit overclocks, then we would be happy to see that. It would change the way people approach chip cooling.

[Jacky]

I said 7 years of normal chip life for a chip running 24/7, under fairly constant load, such as a server chip. A desktop chip would have a much longer lifespan, but 250 years?

7 years is way, way too short. What do you mean anyway? 50% (X%?) of chips should die after 7 years 24/7 loads?

If what you are saying is true Shintai "only OC/voltage affects chip life unless you get over the throttle temperature" then why should anyone bother with extreme air cooling, or with water cooling? Keeping temps down, gives better OCs and it reduces chip-related errors.

I'm pretty sure temperature and OC is important, but voltage is probably most important.
Low end cooling cannot stop chips running high voltage from throttling due to temperature -> no overclock...

[Hornet331]

60C may be normal for Nehalem, but it was not normal for c2d.

i guess you never run a C2Quad on stock cooling?

they can reach 70°C+ in no time on the cooling intel supplies.

[cool_case]

i guess you never run a C2Quad on stock cooling?

they can reach 70°C+ in no time on the cooling intel supplies.

I had meant to say idle temps. C2d temps, even for quads, were below 60c at idle. We're seeing Nehalem idle temps up higher at idle with stock cooling - though there's not really enough data yet to know if they really run hotter than c2ds, 2-core or quad.

[cool_case]

7 years is way, way too short. What do you mean anyway? 50% (X%?) of chips should die after 7 years 24/7 loads?

I'm pretty sure temperature and OC is important, but voltage is probably most important.
Low end cooling cannot stop chips running high voltage from throttling due to temperature -> no overclock...

The reports my EE colleague referred to said that Intel uses 7 years as the life span of server chips on 24/7 load. He is not an Intel employee, but worked on studies that Intel used. An Intel employee working in testing could clear this up quickly - though the information may be proprietary.

Voltage is definitely a big factor, but it's not the only factor. First, we know that with extreme cooling, voltages could be in the 1.9v range for c2d's, without doing damage. You wouldn't have even been able to post if trying that kind of a voltage with air or water. With Nehalem, I haven't seen anyone trying a voltage above 1.6v yet. ( I'll be putting my 965 on LN2 this week, so that will provide a little more data.)

Second, electronic components don't like heat. The throttle temperature is set because if the chip hits that, damage would almost instantly be done. But, below the throttle temperature, especially as you get closer to that temperature, there is gradual depredation of the chip's electronics that increases as the temperature gets closer to the throttle temperature. If you run a chip 24/7 at 5-10 degrees below throttle, vs. 30 degrees below throttle, it just makes intuitive sense that depredation of the components will happen significantly faster in the hotter case.

[Hornet331]

You wouldn't have even been able to post if trying that kind of a voltage with air or water. With Nehalem, I haven't seen anyone trying a voltage above 1.6v yet. ( I'll be putting my 965 on LN2 this week, so that will provide a little more data.)

Oh there are already quite some, even nearing 1.8V

Its not even hard to find.

[cool_case]

Oh there are already quite some, even nearing 1.8V

Its not even hard to find.

Still only seeing a couple above 1.6v for the i7 965 on hw_bot. There are 8 above 1.6v on the overclocking world database and two above 1.7v there, highest 1.78v Shamino.

[T_Flight]

1.78...For people with Unlimited access to trays of CPU's.