Socket 1366 obsolete, SMT a 'gimmick'

[warmon6]

[:jaydeejohn:7]

wounder how long this will last.

 

[Scotteq]

Bold Letters! Caps!!

Babble
Burble
Banter
Bicker bicker bicker
Brouhaha
Balderdash
Ballyhoo

Comments
Cliches
Commentary
Controversy
Chatter
Chit-chat
Chit-chat
Chit-chat
Conversation
Contradiction
Criticism


Debates
Discussions
(These are words with a 'D' this time!)
Dialogue
Dualogue
Diatribe
Dissention
Declamation





Expugnations! Exclamations!! Enfadulations!!



[:jaydeejohn:7]

 

[cjl]

My reply:

Both TOM and Anandtech shows that LGA1366 runs much hotter than LGA1156 and no one has problem with that. ---1)

What is the reason of rise of temp? Heat

Where is heat come from? Lost of power

What is lost of power due to? Leakage current

What is leakage current dependent on? TRANSISTOR EFFICIENCY

Ok, I won't argue with you on the transistor type anymore as it will be endless debate unless I can show you the transistor spec which I can't.

Now, let's assume LGA1156 and LGA1366 has the same transistors.

Yes, leakage current is based on the voltage and frequency as well.

Hence, transistor has different characteristics(leakage current, EFFICENCY...etc) in different states. --- 2)

Conservation of Energy: Less heat(or power-lost) state is the higher efficiency state. --- 3)

The state of higher frequency with less heat is the higher efficiency state. --- 4)

i7-860 has higher frequency, more transistors and lower temps( -> less heat generated) than i7-920 AND 1), 2), 3), 4).

*[ Thus, 860 is in more EFFICIENT state compared to 920 and so it is still more efficient than 920 EVEN THEY HAVE THE SAME TRANSISTOR. ]*

p.s. You kept calling for proof through the whole thread. Where's your proof? Nothing, absolutely nothing since the start of this argument.

I did show proof - I demonstrated that the system-level power gap between an i7-920 and an i7-860 was nearly entirely due to the northbridge and extra memory channel, both at idle and at load. This implies similar power consumption for the CPU itself.

As for my frequency/voltage comment, you really don't understand the idea of multiple clock domains, do you. A QPI link might need a higher clock speed than a PCI-E controller for example (both of which are on separate clock domains from the core). That is why you need to account for what the additional transistors are used for in a CPU, not just how many there are.

 

[SpidersWeb]

I haven't done enough research to properly participate in this debate however the plants to manufacture the i7 9xx chips would have been setup well before the i5/i7 8xx plants. In that time technology would've moved forward slightly, so personally I would expect the i5 and i7 860 to be more efficent.

My i7 930 does put out a lot of heat, and I need to upgrade my heatsink, but 3.6Ghz (166x22) at stock voltage stable, have to be happy with that. I haven't owned an i5 setup, so can't compare the experience with anything unfortunately though.

IF Sandy Bridge is expected on 1366, then (going back to the point of this topic, rather than power consumption), 1366 would definately not be obsolete. 1156 is a mainstream platform that was developed later, so it should/has slightly better technology applied in parts - but I wouldn't expect this to last forever.

Not that I care, because when sandy bridge is out, so will better motherboards! I mostly went with 1366 because I felt it was lower risk (been out longer, less problems previously), rather than the small percentage gains in benchmarks or upgradability, in fact I've actually turned off HT to save on heat output.

PS cjl - what are you using for cooling? My Xigmatech S1283 is my limitation for overclocking at the moment.

 

[yannifb]

Well by the looks of this it seems Sandy Bridge isn't going to be a high end/enthusiast part:
http://images.anandtech.com/reviews/cpu/intel/gulftown/roadmap.jpg

 

[SpidersWeb]

bummer! I missed that link.

Well if that roadmap is accurate, then it does seem 1366 will die off.
Personally before this conversation I'd just assumed that both 1366 and 1156 would disappear with sandy bridge, 1156 doesn't have enough PCI-E lanes built in (at the moment it does for most, but give it a few years), adding more would mean more pins.

 

[yannifb]

bummer! I missed that link.

Well if that roadmap is accurate, then it does seem 1366 will die off.
Personally before this conversation I'd just assumed that both 1366 and 1156 would disappear with sandy bridge, 1156 doesn't have enough PCI-E lanes built in (at the moment it does for most, but give it a few years), adding more would mean more pins.

I think Sandy Bridge was designed to be the high end part at first, but all the Larrabee business screwed that up. SB was closely tied with it in the beginning after all. And also, since the 980x and 970x remain the high end parts, we can assume there will be no 6 core + SB chips.

 

[cjl]

I'm using a TRUE with a dual fan push-pull setup.

 

[cjl]

Well by the looks of this it seems Sandy Bridge isn't going to be a high end/enthusiast part:
http://images.anandtech.com/reviews/cpu/intel/gulftown/roadmap.jpg

Sandy bridge could fit into that ">i7-980x" category in Q1 '11.

 

[EXT64]

Maybe they'll keep both. Seems to me (at least right now) both have a market, so until we all really need triple channel or six cores, I don't see why they can't keep a redone (new chipset, etc.) 1156 around for the mid/low range.

 

[yannifb]

Sandy bridge could fit into that ">i7-980x" category in Q1 '11.

The performance section still says the 970 however, without the greater than or equal to sign. My guess is that they are going to have another higher clocked 980. Like what happened with the i7 960 being replaced with the 975.

 

[cjl]

Entirely possible. Right now, I'd say it's difficult to say. They might also release it a quarter or two later for the upper segment, similar to the release of 32nm.

 

[yannifb]

Entirely possible. Right now, I'd say it's difficult to say. They might also release it a quarter or two later for the upper segment, similar to the release of 32nm.

Well honestly this surprised me. Is SB still going to use the 256 bit FPU?

 

[cjl]

For AVX? I'm pretty sure that's still on the roadmap...

 

[yannifb]

For AVX? I'm pretty sure that's still on the roadmap...

Well i remember them making a big deal about it along with Larrabee, and since LRB was basically scrapped for dt for now, i thought it may have been affected. I wonder how it will compare to BD though. Despite the 256 bit FPU (as opposed to BD 128 bit) it seems BD has several more instruction sets than SB.

 

[andy5174]

@cjl:

Again, again and again. OFF-DIE NB has nothing to do with """CPU""" temp. We are talking about """CPU alone""".

I did show proof - I demonstrated that the system-level power gap between an i7-920 and an i7-860 was nearly entirely due to the northbridge and extra memory channel, both at idle and at load. This implies similar power consumption for the CPU itself.

As for my frequency/voltage comment, you really don't understand the idea of multiple clock domains, do you. A QPI link might need a higher clock speed than a PCI-E controller for example (both of which are on separate clock domains from the core). That is why you need to account for what the additional transistors are used for in a CPU, not just how many there are.

Forget every single theory now. Forget about efficiency too.

Just to let you know:

1. 860 performs better than 920 when both at stock.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3641&p=2

2. 860 """CPU alone"""(Not system)performs cooler than 920 even with higher frequency at stock.

3. OFF-NB has no effect on """CPU"""

In conclusion, 860 performs better with less heat generated.

You judge it yourself what does it mean.

p.s. I showed the proof of temp diff and performance diff which are related to the efficiency. You showed no proof at all on your efficiency arguments.

 

[cjl]

The slight additional performance is due to the higher turbo modes. The cooler running is questionable.

Try again.

 

[Kon37]

@cjl:

Again, again and again. OFF-DIE NB has nothing to do with """CPU""" temp. We are talking about """CPU alone""".


Forget every single theory now. Forget about efficiency too.

Just to let you know:

1. 860 performs better than 920 when both at stock.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3641&p=2

2. 860 """CPU alone"""(Not system)performs cooler than 920 even with higher frequency at stock.

3. OFF-NB has no effect on """CPU"""

In conclusion, 860 performs better with less heat generated.

You judge it yourself what does it mean.

p.s. I showed the proof of temp diff and performance diff which are related to the efficiency. You showed no proof at all on your efficiency arguments.

The i7 860 performs better at stock only because of the turbo mode, at the same clock without turbo 1366 is faster.

 

[Kon37]

The performance section still says the 970 however, without the greater than or equal to sign. My guess is that they are going to have another higher clocked 980. Like what happened with the i7 960 being replaced with the 975.

The 975 replaced the 965, the 960 replaced the 950.

 

[andy5174]

The slight additional performance is due to the higher turbo modes. The cooler running is questionable.

Try again.

Neither TOM and Anandtech has problem with LGA1156 running much cooler.

It is all you can say when you lose arguments in valid sense as you did in debating the efficiency.

cjl wrote:

Nope. Try again.

Except that they don't.

Fair enough. You'll note that I'm not comparing the 980 to Lynnfield above though - I'm comparing Bloomfield (specifically the 920 most of the time) to Lynnfield.

All you said is no no no, "cjl" is right with NO evidence at all. Absolutely no evidence since the start.

You are just wasting my time. I will say "Nope, fail again" to you from now on as you are just being unreasonable.

 

[andy5174]

The i7 860 performs better at stock only because of the turbo mode, at the same clock without turbo 1366 is faster.

We are debating on efficiency.

860 runs cooler than 920 even with higher stock and so 860 will runs much cooler than 920 if they both at 2.8GHz.

Hence, 860 generates less heat than 920. And heat is from lost of power

 

[SpidersWeb]

Andy, does your P55/i5 setup show CPU Power in HWMon etc? Would be interesting to compare figures since we're running the same clock rate, and both without HT.

Mine hits 150.8W in prime95.

(I'm simply just curious)

 

[andy5174]

My HW monitor doesn't work for some reason. It always show 95W.

If you really insist, I will upload a picture of screen shot after I go home.

I am currently doing the darn hard programming which is a compulsory and most hated course.

 

[andy5174]

BTW, what is your OC, case, cooler, ambient temp and its corresponding temps under prime95 large FFT?

 

[jimmysmitty]

Neither TOM and Anandtech has problem with LGA1156 running much cooler.

It is all you can say when you lose arguments in valid sense as you did in debating the efficiency.

Nope. Try again.

Except that they don't.

Fair enough. You'll note that I'm not comparing the 980 to Lynnfield above though - I'm comparing Bloomfield (specifically the 920 most of the time) to Lynnfield.

All you said is no no no, "cjl" is right with NO evidence at all. Absolutely no evidence since the start.

You are just wasting my time. I will say "Nope, fail again" to you from now on as you are just being unreasonable.

I think you are being pretty unreasonable. I thought that the D0s came out around the same time. Either way the D0s run cooler, on less voltage and can OC to normally 4GHz on air for most people.

You went from using how much power a CPU uses to heat. CJL stated they usse the same power (CPU) you agrued it didn't.

I myself am still convinced that the CPUs are the same process and silicon. There could be many reasons as to the differences in heat, such as SMT or the larger IMC that the 920 has or even the fact that QPI is much faster than DMI.

I personally would rather have Intel explain it than argue with you or anyone else as they probably know why.

BTW, I am sorry I mixed up the Q6600 and Q6700s release. Oh well. Sue me. My point was the pricing. They all started high end only then trickled down to affordable. The same will happen with Gulftown even IF they don't show it on the road map.