First Shanghai benchmark results.

[ryboto]

Watch QX9650 @3Ghz VS Core i7 920 .Core2Q wins in power draw test by a large margin while running at 11% higher clock.

sick burn! I was waiting for someone to post this.

[STaRGaZeR]

snip

When the i7-965 significantly outperforms the QX9770, its power consumption is around the same - thus giving us much better performance per watt. When the i7-965 can't really outperform the QX9770, for example in some of the gaming benchmarks, the total system power consumption is much lower.

http://www.anandtech.com/cpuchipsets...spx?i=3453&p=3

[gosh]

http://www.anandtech.com/cpuchipsets...spx?i=3453&p=3

One strange(?) thing about these tests is that big sites in english seems to be more positive compared to sites from other countries.

Here is a picture from a swedish review of i7

http://sweclockers.com/articles_show.php?id=6125&page=5

[Hornet331]

its quite funny....
when it turned out that phenom consumed more then kentsfield (cpu only), it was irelevant and only platform consumption was relevant.
Now suddenly people who said platform only counsumption matters, say cpu only is relevant again?

Seems quite hypocritical to me....

[Rammsteiner]

*cough* performance difference *cough*

And thus?

That makes no sense.

No matter what, I do think Intel messed something up with Nehalem regarding power usage. First K10 is called Prescott because of its power usage, now Nehalem uses the same architecture and it's all good?

I mean, 1K for a Extreme Edition/FX is completely normal because of the performance difference? Your comment ain't really valid there

[informal]

http://www.anandtech.com/cpuchipsets...spx?i=3453&p=3

Keywords : QX9770,current leakage,3+Ghz clocks,extreme suckage/low power efficiency,fail comparison

[informal]

http://www.anandtech.com/cpuchipsets...spx?i=3448&p=9

look at the power consumption at 2.66ghz....

better start pm'ing mods for removing this post again....

btw, its almost logical nehalem would consume more power at same freq/voltage as a C2q, it has more transistors
although intel used static cmos

Don't waste your breath man,he will spin it like a good spintel worker.

[Hornet331]

One strange(?) thing about these tests is that big sites in english seems to be more positive compared to sites from other countries.

Here is a picture from a swedish review of i7

http://sweclockers.com/articles_show.php?id=6125&page=5

Absolute max. power consumption of ci7 is higher then for C2Q, but if you compare it running the same apps (as anand did) they need less power for the same task. The 4 "virtual cores" have a price.

Sometimes i wonder why people dont critize that 4 cores need more power then 2 cores more often.
Now you have the same, people complain that 4+4 cores need more power then 4 cores.

btw does anyone have power consumption comparsions with SMT/Turbo disabled for Ci7 to C2D?

[Calmatory]

its quite funny....
when it turned out that phenom consumed more then kentsfield (cpu only), it was irelevant and only platform consumption was relevant.
Now suddenly people who said platform only counsumption matters, say cpu only is relevant again?

Seems quite hypocritical to me....

Did this come out as a surprise? Same with ATI vs. Nvidia. Same will happen if/when AMD goes past Intel once again.

Intel is offering a CPU, AMD is offering a whole platform, but I know this is about CPUs.

Absolute max. power consumption of ci7 is higher then for C2Q, but if you compare it running the same apps (as anand did) they need less power for the same task. The 4 "virtual cores" have a price.

Your point being? Sounds like a poor excuse for the higher power consumption. Does it matter really what is in if the results speak for themselves?

[Hornet331]

Did this come out as a surprise? Same with ATI vs. Nvidia. Same will happen if/when AMD goes past Intel once again.

Intel is offering a CPU, AMD is offering a whole platform, but I know this is about CPUs.

Platform -> aka CPU and Chipset (compared with the same gfx).

[Calmatory]

Platform -> aka CPU and Chipset (compared with the same gfx).

I'd like to include gfx to that, but fair enough.

To be more precise: AMD is marketting their own platform which is capable of running games with reasonable price. Majority of people here would be the target audience. Intel does not get there anytime soon.

But as I said, fair enough.

[STaRGaZeR]

Keywords : QX9770,current leakage,3+Ghz clocks,extreme suckage/low power efficiency,fail comparison

What leakage? Are you saying that they're not comparable at 3,2GHz because Nehalem is new or something? We are not talking about 65nm AMD process here

Another round of hypocrists please...

[informal]

What leakage? Are you saying that they're not comparable at 3,2GHz because Nehalem is new or something? We are not talking about 65nm AMD process here

Another round of hypocrists please...

Read the post in which i asked for 2.66 vs 2.66Ghz comparison.Other user provided the link to the same anandtech website where 920 consumes a lot more than york @ 2.66Ghz.So picking up the most problematic part from power draw level(QX9770) just to paint a nice picture about i7 power draw is lame.
I provided the Lost Circuits measurement where QX9650 draws 20% less power while working at 11% higher clock.Shintai claimed the complete opposite-a lower power draw for same clocked core i7 and c2q

[gallag]

Some people are sure trying hard not to like i7. It has faster ipc and clocks higher all within the same power consumption, Some tests it uses more and others less, This also includes qpi so system power should be the same or better in most tests.

[informal]

The andandtech's 2.66vs2.66Ghz test was done on the system level,not VRM level... The difference is not small.

[Calmatory]

Some people are sure trying hard not to like i7. It has faster ipc and clocks higher all within the same power consumption, Some tests it uses more and others less, This also includes qpi so system power should be the same or better in most tests.

I dislike it for it's price. For the price of the platform. For the fact that I could get almost the same performance with C2Q CPU. Yes, it is better, but does it justify the price? IMO not and thus my opinion is that it has failed to take over the crown from Core2 arch at this moment.

[Hornet331]

The andandtech's 2.66vs2.66Ghz test was done on the system level,not VRM level... The difference is not small.

yes but the 920 performance = approx. QX9770.

So in theory you could clock down the 920 to a certain degree (lets say 2,3ghz) to match the performance of a 9450. Wonder what the 920 would than consume.

[Rammsteiner]

What leakage? Are you saying that they're not comparable at 3,2GHz because Nehalem is new or something? We are not talking about 65nm AMD process here

Another round of hypocrists please...

I should have missed the first hypocrist round then? Or do you recall the GPU+CPU argument, which was nothing more than crap anyway?

I dislike it for it's price. For the price of the platform. For the fact that I could get almost the same performance with C2Q CPU. Yes, it is better, but does it justify the price? IMO not and thus my opinion is that it has failed to take over the crown from Core2 arch at this moment.

Indeed. It's no news I like AMD a lot, especially K10 vs Yorkfield. K10 had a lot more things to tweak, cheaper etc. But, at this point MOSFETs blew out with my new 9950BE which I bought to fix the idle freezing of the 9850BE, where I thought the SB750 would fix that, which it didnt.

The, so to say, impopularity of AMD is slighlty showing in my experience with motherboards. I started to look around for an alternative, and as I really loved Asus' ROG boards for the Intel platform I decided to give that a try.

I started to compare prices where at some point I noticed the price of a 775 platform vs 1366 ain't that big, if you chose the best Yorkfield though. So at this point Im about to order some 1366 stuff since when you look at the enthusiast prices (apart from CPU's) the difference ain't that big compared to 775.

However, if you currently just got a new 775 system, or something similar with huge clocks, it's far from worth it to dump it and get Nehalem at all, in that matter you're very right.

AMD still is a very nice product for me, but shame the motherboards ain't that good compared to motherboards for the Intel platform. Maybe I just had bad luck, but too much badluck ain't an excuse to try it once again for me.

[Jacky]

And thus?

That makes no sense.

No matter what, I do think Intel messed something up with Nehalem regarding power usage. First K10 is called Prescott because of its power usage, now Nehalem uses the same architecture and it's all good?

There's a simple explanation I like, the i7 920 is the trash bin of Nehalem using enthusiast-voltages so yield will be no issue. I believe it, in dubio pro reo (as of now Intel has a good track-record). "Apples for Apples", highest vs highest end, Intel improved 30-40% in perf/W it seems, this bodes well for the future of i7.

We do not have enough datapoints to suport your conclusion as of now. K10 and Prescott needed to scale clock-wise but couldn't, Nehalem doesn't even need to scale as of now, because of it's incredible performance. So any such comparison is inappropriate.

[STaRGaZeR]

Read the post in which i asked for 2.66 vs 2.66Ghz comparison.Other user provided the link to the same anandtech website where 920 consumes a lot more than york @ 2.66Ghz.So picking up the most problematic part from power draw level(QX9770) just to paint a nice picture about i7 power draw is lame.
I provided the Lost Circuits measurement where QX9650 draws 20% less power while working at 11% higher clock.Shintai claimed the complete opposite-a lower power draw for same clocked core i7 and c2q

And have you think about how first Anand says X, and then Y? Read the review I posted, that covers overclocking and power consumption and explains why first X and then Y. Nehalem seems to be special and sensitive about this, can't wait to test it myself. That's why you see completely opposite PC measurements between reviews. Who's right? Still, in all reviews you can clearly see that perf/watt is greatly improved, even in your links. It sucks more power at times, but it destroys Penryn at the same time. Remember Intel's claims about a 2:1 ratio between perf and PC? That doesn't mean it won't consume less power, with seems to be the case regardless.

It's your choice: less power or more perfomance. Penryn vs Nehalem. But in perf/watt Nehalem wins hands down. If you have the money the choice is eaaaasy.

[Bellisimo]

Some people are sure trying hard not to like i7. It has faster ipc and clocks higher all within the same power consumption, Some tests it uses more and others less, This also includes qpi so system power should be the same or better in most tests.

i really like i7, but people dont have to exaggerate, nehalem is far from perfect

- it offers superior performance, but at a higher power consumption
- very expensive platform (X58 + DDR3 + CPU)
- poor extreme overclocking on normal chips (QPI limit) | they should introduce some black editions like AMD...
- Single-threaded not that fast

in my opinion you have to be realistic, no current CPU from AMD or Intel is perfect, nor is bad or crap
both AMD and Intel release wonderfull products, but this is quickly forgotten if one of the companies has a bit of a performance advantage

[Hornet331]

i really like i7, but people dont have to exaggerate, nehalem is far from perfect

- it offers superior performance, but at a higher power consumption
- very expensive platform (X58 + DDR3 + CPU)
- poor extreme overclocking on normal chips (QPI limit) | they should introduce some black editions like AMD...
- Single-threaded not that fast

in my opinion you have to be realistic, no current CPU from AMD or Intel is perfect, nor is bad or crap
both AMD and Intel release wonderfull products, but this is quickly forgotten if one of the companies has a bit of a performance advantage

ehem:
1) and lower powerconsumption if run at same performance
2) yes, but which generation change didn't do that (P4 to C2, etc.), the only real letdown are the boards, you cant get a board for less then 200€ (personaly i dont mind, but a lot of people do).
3) extreme ocerclocking has no "qpi" (more like blck) limit. The EE has unlocked multi so your up to go. 6ghz+ already happend on nehalem
4) Depends, but its not slower then the pervious generations so its good enough.

[STaRGaZeR]

- poor extreme overclocking on normal chips (QPI limit) | they should introduce some black editions like AMD...

4GHz or more is poor for you?

[Bellisimo]

4GHz or more is poor for you?

like i said, poor EXTREME overclocking on normal chips (aka NOT EXTREME versions)

ehem:
1) and lower powerconsumption if run at same performance
2) yes, but which generation change didn't do that (P4 to C2, etc.), the only real letdown are the boards, you cant get a board for less then 200€ (personaly i dont mind, but a lot of people do).
3) extreme ocerclocking has no "qpi" (more like blck) limit. The EE has unlocked multi so your up to go. 6ghz+ already happend on nehalem
4) Depends, but its not slower then the pervious generations so its good enough.

1) i know it has a performance advantage and a performance/watt advantage.... did i state otherwise?
2) not in the mood for research
3) i did say extreme overclocking using non extreme chips, hence the need for some black editions by intel
4) sometimes it is a little bit, but only a few percentages, but i agree on point 4

[Helmore]

4GHz or more is poor for you?

I guess some people aren't easily satisfied. Although Bellisimo is probably not properly informed yet. (Just a guess though.)

like i said, poor EXTREME overclocking....

Apparently it's that first category. Has there ever been a lower bin chip with good extreme overclocking? Core i7 920 does pretty good from an overclockers perspective.