LGA 1156 Core i7s & Core i5s Reviews

Linfield is a good performer, as expected. Imo, i5-750 is a best in performance/price area.
Noticed the following words from xbitlabs article:

Quote:

---

LGA1156 proves to be an excellent gaming platform. When we switched to Windows 7 operating system, all former issues with Core i7 performance in games vanished. As a result, even a $200 Core i5-750 is by far faster than Phenom II and Core 2 Quad. Core i7-870, however, quite successfully competes against the LGA1366 solution – Core i7-950.

---

http://xbitlabs.com/articles/cpu/dis...-i7-870_9.html

Intel Core i7 870 vs.

AMD Phenom II X4 955 BE
AMD Phenom II X3 720 BE
Intel Core 2 Duo E8400
Intel Core i7 920

Crysis
http://hwt.dk/images/literature/9710/crysis.jpg

3DMark Vantage
http://hwt.dk/images/literature/9710/3dmark_vantage.jpg

Sisoft Sandra
http://hwt.dk/images/literature/9710/sisoftsandra.jpg

Cinebench 10
http://hwt.dk/images/literature/9710/cinebench.jpg


Rest of the review (google-translated from danish: http://translate.google.com/translat...istory_state0= )

Quote:

---

Originally Posted by Utnorris

I want to see the overclocking potential.

---

http://www.fudzilla.com/content/view/15308/1/1/2/
Nothing to look forward to if you ask me.
Will improve over time I guess, though.

keep in mind these are stock cpu values in the benchmarks
just because a i5-750 out performs a i7-920, doesnt mean it will be the same when overclocked. turbo is annoying at this point

Quote:

---

Originally Posted by Mads321

Intel Core i7 870 vs.

AMD Phenom II X4 955 BE
AMD Phenom II X3 720 BE
Intel Core 2 Duo E8400
Intel Core i7 920

Crysis
http://hwt.dk/images/literature/9710/crysis.jpg



Rest of the review (google-translated from danish: http://translate.google.com/translat...istory_state0= )

---

Damn that Crysis chart looks weird,X4 II 955BE is a whooping 10FPS better than the I7 920?That's like more than 30%:confused:

Our french review :D

http://www.pcworld.fr/article/materi...0-banc/445781/
http://wd.ch-img.com/1059101-benchma...apitulatif.png

Quote:

---

Originally Posted by Newblar

keep in mind these are stock cpu values in the benchmarks just because a i5-750 out performs a i7-920, doesnt mean it will be the same when overclocked. turbo is annoying at this point

---

where do you see the i5-750 dominating the i7-920 ???

The I7-870 at stock clocks is winning but thats not clock 4 clock either.

i7-920 seems to be OCing a little better than the i5-750 and much better than the 870s but time will tell.

Ok now even at this point of time i am split between the two.

In anandtech's review in games first the 965 wins then i5 750 wins then i5 750 wins and last one also the 750 wins. But in most the difference is low. The general difference between the 965 and 750 is around 200-300mhz "Turbo mode gets the 750 to 3.2Ghz"
Also i am not that happy about 1.4v for 3.9Ghz, also CF takes a beating pointless in getting a 1156 for even 5870 CF.

I am think about getting a 920 before they go bye bye, but i will have to buy a top class mobo so that i can upgrade and OC a six core i9 later. On the other hand i can get a 955 OC it around 3.8Ghz "Most likely i will reach 3.8Ghz with a Asus M4A785TD-V EVO" and the 750 will probably go good till 3.9Ghz "Most likely i cant reach 3.9Ghz with a Asus P7P55D-LE"

I'm getting confused too. I keep seeing the 955 beat them on benches (albeit oc'd). What's the go here? I would have thought the 955 would be blown away by the i7's...

Turbo mode is really a saving grace for 750. Without it(3.2Ghz) it is slower in most cases than 965 and even 955. What matters is that most "advanced" desktop users won't use stock settings but will OC their systems to the max(or to the sweet spot for their chip). In this case there is no turbo to help 750 to boost its results since it's already OCed. At this point the difference is in price between AMD or intel rig(board,memory,cooling,crossfire support and scaling abilities).
I'm still to read a good i5/i7 1156 review that covers multi gpu scaling and the see how good (or bad) is the int. PCI-e controller.

Quote:

---

Originally Posted by ajaidev

I am think about getting a 920 before they go bye bye, but i will have to buy a top class mobo so that i can upgrade and OC a six core i9 later. On the other hand i can get a 955 OC it around 3.8Ghz "Most likely i will reach 3.8Ghz with a Asus M4A785TD-V EVO" and the 750 will probably go good till 3.9Ghz "Most likely i cant reach 3.9Ghz with a Asus P7P55D-LE"

---

if you are near a MicroCenter 199 for a 920 is impossible to beat

Iam sure a X58 Gigabyte would handle a i9 in the future and isn't too bad price wise.

I play all day at 4.3 ghz and 1.3v

Quote:

---

Originally Posted by Pt1t

Our french review :D

http://www.pcworld.fr/article/materi...0-banc/445781/

---

The Core i5 750 is weaker than Core 2 Quad Q9550? :D

Quote:

---

Originally Posted by ajaidev

Ok now even at this point of time i am split between the two.

In anandtech's review in games first the 965 wins then i5 750 wins then i5 750 wins and last one also the 750 wins. But in most the difference is low. The general difference between the 965 and 750 is around 200-300mhz "Turbo mode gets the 750 to 3.2Ghz"
Also i am not that happy about 1.4v for 3.9Ghz, also CF takes a beating pointless in getting a 1156 for even 5870 CF.

I am think about getting a 920 before they go bye bye, but i will have to buy a top class mobo so that i can upgrade and OC a six core i9 later. On the other hand i can get a 955 OC it around 3.8Ghz "Most likely i will reach 3.8Ghz with a Asus M4A785TD-V EVO" and the 750 will probably go good till 3.9Ghz "Most likely i cant reach 3.9Ghz with a Asus P7P55D-LE"

---

I think this review say pretty much everything (i´m also wondering 750 or 920/xeon, dont think that Xeon will disappear): http://74.125.39.132/translate_c?hl=...itt_einleitung

Quote:

---

Originally Posted by Migi06

I think this review say pretty much everything (i´m also wondering 750 or 920/xeon, dont think that Xeon will disappear): http://74.125.39.132/translate_c?hl=...itt_einleitung

---

Thanks for posting the link. From the conclusion :
http://img171.imageshack.us/img171/1194/30195239.png
Sums it up perfectly.It's from a range of 26 applications(includes synthetics,office and multimedia and lastly games).

Quote:

---

Originally Posted by informal

Turbo mode is really a saving grace for 750. Without it(3.2Ghz) it is slower in most cases than 965 and even 955. What matters is that most "advanced" desktop users won't use stock settings but will OC their systems to the max(or to the sweet spot for their chip). In this case there is no turbo to help 750 to boost its results since it's already OCed. At this point the difference is in price between AMD or intel rig(board,memory,cooling,crossfire support and scaling abilities).
I'm still to read a good i5/i7 1156 review that covers multi gpu scaling and the see how good (or bad) is the int. PCI-e controller.

---

For i5-750 TB in multithreaded apps is only 2.93GHz. I'm sure one more gigaherz will be felt very well.

Quote:

---

Thanks for posting the link. From the conclusion :
http://img171.imageshack.us/img171/1194/30195239.png
Sums it up perfectly.It's from a range of 26 applications(includes synthetics,office and multimedia and lastly games).

---

I think that games have too much weight in the final sum, espetialy when 9800GTX is used for the games.

There are still some singlethreaded apps used in tests on many review sites,so the 3.2Ghz is important Turbo mode target. MT is still kinda new,especially programs that can effectively utilize more than 2 threads(and not distribute 2 or 3 threads across 4 cores so that it "looks" like it is well threaded while in reality the total CPU utilization is still ~50%)..

One more chart from another review(thanks to vardirox for posting it):
http://img98.imageshack.us/img98/9015/perfratingc.jpg

I would certainly hope an AMD processor that is currently quite a bit more expensive than a lower level and lower cost Intel processor, the i5 750 vs. the 965, would perform better. And its (the 965) speed differential does help in a lot of situations.

But, at least to me, the "value" graphs are rather meaningless. What weight is put into what metric? How were they chosen? It's just too easy to skew the results in the "value" graphs to show anything you want....give me pure performance graphs, period, and forget the "value" graphing.

Taking into account that you've got a cpu that's $50 cheaper (i5 750 @ $179 vs. 965 @ $229), motherboards are in the same price range, and it performs at least on par with the 965 in most worst cases, flat out beating it in best cases, I cannot see how anyone can consider an AMD platform today.

(And by beating it, I refer to the majority of content creation benchmarks like second pass video encoding, WinRAR encoding, Sony Vegas Blu-Ray disc creation, Excel spreadsheet, Blender, SPECapc 3dsmax CPU rendering test, DivX / XMpeg 5.03 1080p encoding, Photoshop CS4 performance....there is more to computing than just gaming. And the i5 750 is ultra competitive in gaming, too, much to the embarrassment of the 965.)

Oh well, guess the AMD fanbois will have to clutch at any staws they can reach right now. :)

Those are pure performance graphs,both of them(average from the whole battery of tests). They are only presented in percentage since some tests are time to bench complete and some are "points". None of the 2 charts is value. And boards are not priced the same,you can always go cheaper with 140W supporting 785G board that is packed with features. And yes, 750 has a turbo function so in majority of cases it doesn't even run at 2.66Ghz(which is not bad,but is a bit misleading, at least while you try to figure how it stacks against phenoms and c2qs).

Quote:

---

Originally Posted by Humminn55

I cannot see how anyone can consider an AMD platform today.

---

Those who run more than one application simultaneously.
Prefetchers work well when there is a margin, no margin and they don't kick in.
There you need cache with high hitrates

Quote:

---

Originally Posted by informal

And yes, 750 has a turbo function so in majority of cases it doesn't even run at 2.66Ghz(which is not bad,but is a bit misleading, at least while you try to figure how it stacks against phenoms and c2qs).

---

Its not misleading, its part of the normal functionality of the cpu itself, if you fire up the cpu without altering anything in the bios thats how it will function.

If you want to isolate the performance a certain clock speed then you would want to disable turboboost but other than that it is now part of intels normal cpu function.

If the benches show an advantage from the use of turboboost than I would say its safe to say turboboost is a success in delivering more real performance when you need it based on load type.

The idea behind turbo boost is essentially a refined power management of sorts and it does make sense for many reasons. I am pretty impressed with how it works with the 750 and 8xx cpus compared to the 9xxs however. Very aggressive yet still has respectable power draw and heat output.

I would like to see them take things a step further and have turbo boost function on separate cores for even more aggressive clocking for single threaded apps (similar to how AMD can clock individual cores on their newer cpus ). It will only offer further benefits for your average user and make overclocking less attractive / required for general usage.

As far as it being misleading, I feel it kind of is (comparison wise) as unless you document it in a review with each test, you can't always say when the multi has been increased or is at stock for a given benchmark / comparison. As long as it is properly noted I don't take issue with it.

Like I said it's not a bad functionality,it's quite useful. But it can make one believe 750 does the same numbers in test while running at 2.66Ghz,which is not the case. What I'm trying to say is that it's harder to evaluate the possible performance of OCed 750 while looking at defaults out of the box numbers with turbo on(since while OCed the turbo won't be able to help it and the scaling from turboed clock to say 3.8 or 4Ghz would yield lower increase than one would expect-from 2.66Ghz as the box stamp says).

And there is still the unknown factor of CF/SLI on the new platform when the latest and greatest AMD/NV GPUs arrive.Will the integrated pci-e controller allow for the same scaling as off-die approach? Is there any review with present high end cards in CF/SLI done on s1156 boards?

Quote:

---

Originally Posted by informal

And there is still the unknown factor of CF/SLI on the new platform when the latest and greatest AMD/NV GPUs arrive.Will the integrated pci-e controller allow for the same scaling as off-die approach? Is there any review with present high end cards in CF/SLI done on s1156 boards?

---

Lynnfield in sli: http://www.pcper.com/article.php?aid...=expert&pid=11

Quote:

---

Originally Posted by informal

What I'm trying to say is that it's harder to evaluate the possible performance of OCed 750 while looking at defaults out of the box numbers with turbo on(since while OCed the turbo won't be able to help it and the scaling from turboed clock to say 3.8 or 4Ghz would yield lower increase than one would expect-from 2.66Ghz as the box stamp says).

---

What's the problem?
I suppose you've missed this link:
http://lab501.ro/procesoare-chipsetu...eld-in-actiune

Quote:

---

Originally Posted by highoctane

Lynnfield in sli: http://www.pcper.com/article.php?aid...=expert&pid=11

---

Thanks for the link,will read it now :).Still,I'd like to see HD58xx in CF on s1156.

Another review for the list on the 1st page:
http://www.phoronix.com/scan.php?pag...ynnfield&num=1

This one is a review under Linux and results are quite surprising(no amd quad core used,but instead a X3 Deneb)

Quote:

---

Originally Posted by informal

One more chart from another review(thanks to vardirox for posting it):
http://img98.imageshack.us/img98/9015/perfratingc.jpg

---

I would not call this thing "review". Theirs performance sum based only on 4 tests - winrar, 3dmark, vantage and superpi (and seems that winrar have the most weight in the sum). I suppose we should keep the thread clean from such "reviews"
.

Quote:

---

Originally Posted by informal

Thanks for the link,will read it now :).Still,I'd like to see HD58xx in CF on s1156.

---

Yeah, I'd be satisfied with just seeing 58xx review, should be very soon I guess!

Quote:

---

Originally Posted by kl0012

I would not call this thing "review". Theirs performance sum based only on 4 tests - winrar, 3dmark, vantage and superpi (and seems that winrar have the most weight in the sum). I suppose we should keep the thread clean from such "reviews"
.

---

I posted another one with a 26 applications showing almost the same numbers.

BTW ,under linux,even x3 710(2.6Ghz) is holding its own while going against the new i5 and i7-look at phoronix review under ubuntu(that's real world workloads). That's 3 cores @ 2.6GHz against 4c/8t @ 2.66-3.46Ghz :)

Quote:

---

Originally Posted by kl0012

I would not call this thing "review". Theirs performance sum based only on 4 tests - winrar, 3dmark, vantage and superpi (and seems that winrar have the most weight in the sum). I suppose we should keep the thread clean from such "reviews"
.

---

I have to agree, "reviews" that gives the perception that PIIs are better than what it is... and for some reason these "reviews" are only put up by a few users who have anti-Intel agendas. :rolleyes:

Ok as it stands i think the performance in high bandwidth CF "Like 4870 x2 CF" will be:-

i7+x58>PhII+790>i5+P55

For someone looking to build a screaming mid-range machine, it's hard to argue with i5 and P55. Tons of performance, value etc.

Also, the motherboard selection is fantastic!

Quote:

---

Originally Posted by informal

I posted another one with a 26 applications showing almost the same numbers.

---

Can't agree. The second diagram (which excludes games tested on high res with 9800GTX) actually shows that i5-750 overal faster then Pii-965.

Quote:

---

BTW ,under linux,even x3 710(2.6Ghz) is holding its own while going against the new i5 and i7-look at phoronix review under ubuntu(that's real world workloads). That's 3 cores @ 2.6GHz against 4c/8t @ 2.66-3.46Ghz :)

---

Hm... I wonder why they used Ubuntu 9.10 Alpha release. Alpha releases usually includes many debug code and has unpredictable performance on different systems.

Quote:

---

Originally Posted by kl0012

Can't agree. The second diagram (which excludes games tested on high res with 9800GTX) actually shows that i5-750 overal faster then Pii-965.


Hm... I wonder why they used Ubuntu 9.10 Alpha release. Alpha releases usually includes many debug code and has unpredictable performance on different systems.

---

Well we can go all day long finding errors in testing or test choices. 750 due to its high turbo mode looks faster than 965BE,but without it it's a bit slower.Also with turbo on,750 is a bit faster or on par with 860 i7 too... The difference between i5/965 and top of the line 870 with turbo and smt on is 16 percent in office and multimedia apps... No gaming involved. 16% is decent but nothing to write home about,especially if you consider the difference in price.

Phoronix website is quite professional all around. There is a reason they used that release.They tested the 920 i7 quite a long while ago and were quite pleased with its performance. You can always ask them though :).

pfff Kyle tested Lynnfield under gaming @ 640x480 what a champ!
:down:
http://hardocp.com/article/2009/09/0...7_processors/5

Quote:

---

Originally Posted by Nedjo

pfff Kyle tested Lynnfield under gaming @ 640x480 what a champ!
:down:
http://hardocp.com/article/2009/09/0...7_processors/5

---

Yea, what a champ... this is a CPU review not a GPU review. :rolleyes:

Quote:

---

Originally Posted by Clairvoyant129

Yea, what a champ... this is a CPU review not a GPU review. :rolleyes:

---

So you suggest that car reviewers pull out engine out of two ton car and put it on bench table to give you acceleration performance torque ratio, fuel efficiency etc... :rolleyes:

that page is about "Lynnfield under gaming", and people in 2009 don't game on SMARTPHONE resolution!!

Kyle, and obviosly you don't have clue about CPU role in gaming... if you did have clue you would know that true stress for CPU under gaming is with more detail!

The point is to find sweet spot that would stress CPU, and still not represent GPU bottleneck!

Ideal CPU under gaming conditions would be testing with Dual GPU (1 gig cards) setup (driver overhead comes as a burden for CPU) @ 1680x1050 Medium/High settings!

Quote:

---

Originally Posted by Clairvoyant129

Yea, what a champ... this is a CPU review not a GPU review. :rolleyes:

---

So u want to get the i5/i7 and play at 640x480??? Ohh okk so the logic is you get more frames...ohh :up:



EDIT: Where the hell are all the mobo reviews?? Wanted a review of P7P55D Plain and P7P55D LE soo bad.....

Quote:

---

Originally Posted by informal

Well we can go all day long finding errors in testing or test choices. 750 due to its high turbo mode looks faster than 965BE,but without it it's a bit slower..

---

Well in real function that would make the cpu faster, once again turboboost is an integral function of the cpu now as much as speedstep to lower the speed. If people are looking at the review to decide which out of the box cpu is faster then it is what it is, what they see is what they get.

Do you expect them to clock the i5 at the same speed as a comparably priced AMD cpu to be fair...

Quote:

---

Originally Posted by informal

Phoronix website is quite professional all around. There is a reason they used that release.They tested the 920 i7 quite a long while ago and were quite pleased with its performance. You can always ask them though :).

---

I don't think it is even worth to discus that. I prefer to wait for official release of Ubuntu 9.10 (somewhere in October) with all bugs filtered out (especialy those affected new hardware). I see no point in testing cpus on different alpha version of unstable software.

Quote:

---

Originally Posted by Nedjo

Ideal CPU under gaming conditions would be testing with Dual GPU (1 gig cards) setup (driver overhead comes as a burden for CPU) @ 1680x1050 Medium/High settings!

---

Is this what you're looking for?
http://xbitlabs.com/articles/cpu/dis...0_9.html#sect0

Quote:

---

Originally Posted by Nedjo

So you suggest that car reviewers pull out engine out of two ton car and put it on bench table to give you acceleration performance torque ratio, fuel efficiency etc... :rolleyes:

that page is about "Lynnfield under gaming", and people in 2009 don't game on SMARTPHONE resolution!!

Kyle, and obviosly you don't have clue about CPU role in gaming... if you did have clue you would know that true stress for CPU under gaming is with more detail!

The point is to find sweet spot that would stress CPU, and still not represent GPU bottleneck!

Ideal CPU under gaming conditions would be testing with Dual GPU (1 gig cards) setup (driver overhead comes as a burden for CPU) @ 1680x1050 Medium/High settings!

---

Are you serious? As mentioned, he's not stressing the GPU. The lower the resolution, the more CPU bound the game is.

If it were up to you, you would probably want to test it with AA also.:rolleyes:

Gaming with more detail is more stressful for the GPU, not so much the CPU. In case you didn't understand, he's trying to isolate the CPU performance.

Quote:

---

Originally Posted by Nedjo

pfff Kyle tested Lynnfield under gaming @ 640x480 what a champ!
:down:
http://hardocp.com/article/2009/09/0...7_processors/5

---

Jeez... did you even read the text of the article?

I quote Kyle (bold added by me):

Quote:

---

As always, these benchmarks in no way represent real-world gameplay. They are all run at very low resolutions to try our best to remove the video card as a bottleneck. I will not hesitate to say that anyone spouting these types of framerate measurements as a true measuring tool in today’s climate is not servicing your needs or telling you the real truth.

The gaming tests below have been put together to focus on the processor power exhibited by each system. All the tests below consist of custom time demos built with stressing the CPU in mind.

---

You may not find the information useful, others might. Personally, I don't care as I don't game, but putting down a testing procedure without at least some basic research is wrong in my opinion. Not only does Kyle state his goal very clearly and percisely, he also raises the necessary caveats on how the data can be used.. Again, if you don't find the data useful move on. However, I don't know of any other way to get the answer to the question he is raising. Which is, with as little GPU bottleneck as possible, just how fast can the CPU go on a particular game?

Quote:

---

Originally Posted by Nedjo

pfff Kyle tested Lynnfield under gaming @ 640x480 what a champ!
:down:
http://hardocp.com/article/2009/09/0...7_processors/5

---

Its only game when HT will gain some speed (multicore game..). Like someone said its good way to check CPU power with low res...

hmm so i5 has made AMD's mid range processors incompetent once again. I wonder what AMD will do to counter this? I hope whatever they do is better for themselves and consumers as well.

Quote:

---

Originally Posted by highoctane

Lynnfield in sli: http://www.pcper.com/article.php?aid...=expert&pid=11

---

interesting. nehalem is 10% faster clock for clock in memory intensive apps.

Quote:

---

Originally Posted by Clairvoyant129

:rolleyes: You have no idea what you're talking about gosh, just stop your anti-Intel nonsense

---

Why don't you explain how prefetchers work and when they work? :rolleyes:
While doing that you can also explain the differences in how the cache work on i5/i7 vs Phenom II

It is one thing to just say someone don't know, but to explain is another ballgame

Quote:

---

Originally Posted by Salavat23

Are you serious? As mentioned, he's not stressing the GPU. The lower the resolution, the more CPU bound the game is.

---

Once again: as the scene that's being rendered simpler - there's less work to be done for CPU - less geometry to feed GPU, less traficking over the PEG, less driver overhead to deal with...

I agree that there's a point in testing CPU's where GPU is becoming bottleneck and it shouldn't be passed for the sake of shadowing CPU performance, but going so low as Kyle did is ewen worse 'cos it misleads readers about true potential and usability of CPUs in gaming scenarios

Quote:

---

If it were up to you, you would probably want to test it with AA also.:rolleyes:

---

ABSOLUTELY! If I'm spending serious money on new powerful HW I want to see how it performs on my big 30" TFT, and apparently here's the answer:
http://www.anandtech.com/cpuchipsets...spx?i=3634&p=9
2560x1600:
http://www.abload.de/img/anandtechintelscorei78iueo.jpg

Quote:

---

Gaming with more detail is more stressful for the GPU, not so much the CPU. In case you didn't understand, he's trying to isolate the CPU performance.

---

He don't have clue... or maybe does... anyhow he's misleading readers! If he wanted to stress out CPU in gaming he should have run software rendering of Unreal 1 and Quake 1 in 640x480!! :rolleyes:

Insulating CPU in trying to portrait gaming performance is by definition nonsense, but you can make of it something if you know how to do it, and I've suggested how! And as someone pointed xbit is using that approach, as well as anandetch... http://www.anandtech.com/cpuchipsets...spx?i=3634&p=8
but yeah Kyle is smartest of them all :rolleyes:

Quote:

---

Originally Posted by Nedjo

Once again: as the scene that's being rendered simpler - there's less work to be done for CPU - less geometry to feed GPU, less traficking over the PEG, less driver overhead to deal with...

I agree that there's a point in testing CPU's where GPU is becoming bottleneck and it shouldn't be passed for the sake of shadowing CPU performance, but going so low as Kyle did is ewen worse 'cos it misleads readers about true potential and usability of CPUs in gaming scenarios

ABSOLUTELY! If I'm spending serious money on new powerful HW I want to see how it performs on my big 30" TFT, and apparently here's the answer:
http://www.anandtech.com/cpuchipsets...spx?i=3634&p=9
2560x1600:
http://www.abload.de/img/anandtechintelscorei78iueo.jpg

He don't have clue... or maybe does... anyhow he's misleading readers! If he wanted to stress out CPU in gaming he should have run software rendering of Unreal 1 and Quake 1 in 640x480!! :rolleyes:

Insulating CPU in trying to portrait gaming performance is by definition nonsense, but you can make of it something if you know how to do it, and I've suggested how! And as someone pointed xbit is using that approach, as well as anandetch... http://www.anandtech.com/cpuchipsets...spx?i=3634&p=8
but yeah Kyle is smartest of them all :rolleyes:

---

How is anand misleading? At that res you get very GPU limited very quickly. Ok it's Crysis but you know the further down the ladder of graphical fidelity you go the more cpu intensive it is, hello Source engine, hello COD4/5! It's a strange world in that if you have a 30" you are better off spending your cash on graphic card upgrades rather than cpu upgrades. The smaller your monitor the more cpu makes a difference, the bigger it is the less it does. So at the highest res available to consumers, you're fine with a high clocking e8x00. The sweet spot is 1920x1200, the cross section where you get most bang for buck from both. I've been considering downgrading for some time, but hopefully the 5xxx series and GT300 will change the noodle. :)

Quote:

---

Originally Posted by tdream

How is anand misleading?

---

who said anand is misleading? read again my post... anand's and xbit's approach is much more informative than {H}ocp's

the smaller the monitor - smaller is resolution so my advice to people with smaller monitors is go with AA as high as possible!

I'd agree with you that this and next year will establish 1080p as a mainstream resolution that's covered with a range from 22 'till 26 inch monitors, and that's a resolution that should be targeted in reviewing!

When it comes to methodology, my POV is to tune down any Shader, or other special effects, to leave any physics-type of effects as highest as possible...

point is to burden CPU as much as possible with all the tasks it should be doing to the max, and on the other side not to bottleneck graphics card... going on VGA and SVGA resolutions isn't answer... as simple as that!

Answer is MultiGPU system with tuned low Shader effects, and upped high geometry complexity ( 'cos CPU feeds geometry) and texture quality (modern engines are streaming engines so PCI Express traffic, and implementation counts) in 1080 rez!

All my respect onethreehill :up:

It's been a while since I've banned anyone, and I'm itching...

Quote:

---

Originally Posted by Nedjo

who said anand is misleading? read again my post...

---

Quote:

---

Originally Posted by Nedjo

Once again: as the scene that's being rendered simpler - there's less work to be done for CPU - less geometry to feed GPU, less traficking over the PEG, less driver overhead to deal with...

I agree that there's a point in testing CPU's where GPU is becoming bottleneck and it shouldn't be passed for the sake of shadowing CPU performance, but going so low as Kyle did is ewen worse 'cos it misleads readers about true potential and usability of CPUs in gaming scenarios

ABSOLUTELY! If I'm spending serious money on new powerful HW I want to see how it performs on my big 30" TFT, and apparently here's the answer:
http://www.anandtech.com/cpuchipsets...spx?i=3634&p=9
2560x1600:
http://www.abload.de/img/anandtechintelscorei78iueo.jpg

He don't have clue... or maybe does... anyhow he's misleading readers! If he wanted to stress out CPU in gaming he should have run software rendering of Unreal 1 and Quake 1 in 640x480!! :rolleyes:

Insulating CPU in trying to portrait gaming performance is by definition nonsense, but you can make of it something if you know how to do it, and I've suggested how! And as someone pointed xbit is using that approach, as well as anandetch... http://www.anandtech.com/cpuchipsets...spx?i=3634&p=8
but yeah Kyle is smartest of them all :rolleyes:

---

Eh you did. :ROTF:

Quote:

---

Originally Posted by Humminn55

...

I cannot see how anyone can consider an AMD platform today.

Oh well, guess the AMD fanbois will have to clutch at any staws they can reach right now. :)

---

Some of us would rather go without a CPU, (if AMD went under) than give our cash to Intel. :yepp:

No flames intended...

Quote:

---

Originally Posted by gosh

Why don't you explain how prefetchers work and when they work? :rolleyes:
While doing that you can also explain the differences in how the cache work on i5/i7 vs Phenom II

It is one thing to just say someone don't know, but to explain is another ballgame

---

Just love it when you try to flamebite people with the same stuff over and over again... and its esepcial funny since, most of your questions where already answerd in a thread you posted....

Quote:

---

Originally Posted by Hornet331

Just love it when you try to flamebite people with the same stuff over and over again... and its esepcial funny since, most of your questions where already answerd in a thread you posted....

---

... gosh is simply amazing... his question has been answered a thousand times and he has been proven wrong a thousand times but I guess the green smoke is too blinding for him.

Hopefully someone shows what the core i5 750 can hit with reasonable volts before long.... Something like lowest stable volts for 3.6g , 3.8g and 4.0g while still being stable on all course would be great.

This oh we need 1.5 volts for 3.6g is ridiculous and so is the fact that everyones pushing it to 4.2-4.5 at 1.4+ volts theres no way that will be stable specially since not all the reviewers had aftermarket heatsinks for it. Bnechmarking at that level on the stock heatsink is asking for blue screens, crashes and piossibly a fried chip in my opinion. 90 + degrees celcius no thank you.

Quote:

---

Originally Posted by Salavat23

Are you serious? As mentioned, he's not stressing the GPU. The lower the resolution, the more CPU bound the game is.

If it were up to you, you would probably want to test it with AA also.:rolleyes:

Gaming with more detail is more stressful for the GPU, not so much the CPU. In case you didn't understand, he's trying to isolate the CPU performance.

---

That doesn't mean that gaming tests at real world resolutions (i.e. 1280x1024 and 1680x1050) fails to show the performance difference wrt different CPU's.

That's why we have other benchmark test suites to test just the CPU. When it comes to testing CPU's and games, I'd like to see what difference a faster CPU can do for my graphics card with the settings I run my game at. No one on this forum (or any other PC hardware forum) runs their games at 640x480 and with minimum settings.

Quote:

---

Some of us would rather go without a CPU, (if AMD went under) than give our cash to Intel.

No flames intended...

---

I suggest you throw out most of the components in your PC then.

You do know that Intel collects royalties from the various technologies that they've invented other than the CPU itself? PCI Express, JEDEC, form factors, SATA, various I/O interconnects, CPU instructions, operating systems and circuitry are just some of the areas that Intel has either pioneered or contributed significant investment and reasearch into developing.

Quote:

---

Originally Posted by freeloader

Some of us would rather go without a CPU, (if AMD went under) than give our cash to Intel. :yepp:

---

Everytime you buy an AMD CPU you give cash to Intel. :up:

Quote:

---

Originally Posted by Chad Boga

Everytime you buy an AMD CPU you give cash to Intel. :up:

---

Yeah, yeah, you got me there! Darn licensing...:clap:

Quote:

---

Originally Posted by ajaidev

So u want to get the i5/i7 and play at 640x480??? Ohh okk so the logic is you get more frames...ohh :up:



EDIT: Where the hell are all the mobo reviews?? Wanted a review of P7P55D Plain and P7P55D LE soo bad.....

---

Here you go :)

ASUS P7P55D Motherboard Review
EliteBastards

MSI P55-GD80 Review
Guru3D

A look at Asus' P7P55D Deluxe, Gigabyte's GA-P55-UD6, and MSI's P55-GD65
The Tech Report

MSI P55-GD80 Motherboard Review
[H]ard|OCP

Which of the Asus boards, not including the maximus formula III is the best.

Quote:

---

Originally Posted by tdream

Eh you did. :ROTF:

---

if you've read my post correctly you would notice that my comment about clueless is about Kyle's approach, not Anand's :rolleyes:

Do people give any weighting to the min frame rates that the processors were showing in Dawn of War, Sacred 2, WoW and Prototype in Anand's review?

The i7-920 was a bit disappointing in 2 of those 4 games, compared to the i7-870, which seemed to be the best balanced of the group that includes the i7-870, i7-920, i5-750 and PHII965.

Quote:

---

Originally Posted by onethreehill

Here you go :)

ASUS P7P55D Motherboard Review
EliteBastards

MSI P55-GD80 Review
Guru3D

A look at Asus' P7P55D Deluxe, Gigabyte's GA-P55-UD6, and MSI's P55-GD65
The Tech Report

MSI P55-GD80 Motherboard Review
[H]ard|OCP

---

Thanks for the Asus P7P55D review. Waiting for the P7P55D LE one tough :yepp: also does anyone know if P7P55D plain has a option in the BIOS to force the second slot into 4x mode. What i am looking for is running 16x and 4x "16x for the 5870 x2/GT300 and 4x for a RAID card, but the card is 8x"

In case of P7P55D LE the second slot is able to deliver only 4x and cant split the 2 slots into 8x/8x mode only 16x/4x is possible.

Intel Core i5 750 and Core i7 870 Processors
http://www.legitreviews.com/article/1060/1/

Intel Core i7 870 Processor
http://benchmarkreviews.com/index.ph...=362&Itemid=63

Quote:

---

Originally Posted by gosh

Why don't you explain how prefetchers work and when they work? :rolleyes:
While doing that you can also explain the differences in how the cache work on i5/i7 vs Phenom II

It is one thing to just say someone don't know, but to explain is another ballgame

---

You have some graphs (Phenom II cache)^^ ?

I've read so many reviews, it's hard to keep them all in my head... ouch!




on topic:

The i7-860 looks to be a winner!

Sure, you give up a little memory bandwidth, but gain slightly in everything else, over the i7-920. (at lower wattage too)

I'm looking to stick with low db air cooled gamer rig. OC, and 27/7 operation. I think the i7-860 is the real winner in the enthusiast market. We'll see this chip @ $219 before the Superbowl.

The P55 seems to be kind to memory, so a nice 8GB, 7-7-7-20 @ 1866Mhz seems easily doable for a near quite system. (Raven 2 case)

Strictly gaming and video encoding, etc...

Quote:

---

Originally Posted by Xoulz

You have some graphs (Phenom II cache)^^ ?

I've read so many reviews, it's hard to keep them all in my head... ouch!

---

http://www.phoronix.com/scan.php?pag...ntel_lynnfield

most reviews are testing same applicatioins, you don't need to keep them in your head.
i5 is very good and at last applications can move to threaded workloads. i7 did not sell enough and core 2 isn't an option

^
-_- oooookayyy thers something quite wrong with this review....
870 can't be slower then the 920 in non bandwidth critical situations.... (audio encoding, pov and a few others....)

There are situations where the benches make sense (870 a tiniy bit faster then the 920 due to higher stock frequency), but some benches are just plain unbelivable (lame and flac comes to mind, just to mention a few examples).

Quote:

---

Originally Posted by gosh

http://www.phoronix.com/scan.php?pag...ntel_lynnfield

most reviews are testing same applicatioins, you don't need to keep them in your head.
i5 is very good and at last applications can move to threaded workloads. i7 did not sell enough and core 2 isn't an option

---

Those graphs are skewed or something. How does the i5 750 constantly beat the i7 870 in many of those benches ..?

Makes no sense!

Quote:

---

Originally Posted by Hornet331

^
-_- oooookayyy thers something quite wrong with this review....
870 can't be slower then the 920 in non bandwidth critical situations.... (audio encoding, pov and a few others....)

---

What is the difference in speed for the L3 cache in 920 vs 870?

Quote:

---

Originally Posted by Xoulz

Those graphs are skewed or something. How does the i5 750 constantly beat the i7 870 in many of those benches ..?

Makes no sense!

---

"constantly"?
there was a test where hyperthreading might have been a problem

Quote:

---

Originally Posted by gosh

http://www.phoronix.com/scan.php?pag...ntel_lynnfield

---

Even author by self admits that there is something wrong with the benches, which is not surprise since an unstable alpha version of OS was used and probably this version know nothing about the new hardware:

What is important to keep in mind though is that Intel Turbo Boost Technology was disabled on the processors during testing, since this functionality had not worked under Linux for increasing the clock frequency but instead appeared to cause some sporadic performance problems...

What is also missing from the equation is any thermal monitoring support for the P55 / Lynnfield processors using LM_Sensors on Linux, though that should come with time. Update: after starting to see a flow of Windows-based reviews today, it looks like there are some more serious Linux + Lynnfield problems at hand, which we are currently investigating.

Quote:

---

Originally Posted by gosh

What is the difference in speed for the L3 cache in 920 vs 870?

---

2,13ghz vs 2,4ghz.... :rolleyes:

The L3 speed of the 870 is faster then that of the 920... (at stock)

Also apps that don't really have needs for huge memory bandwidth don't profit that much from faster L3.

Quote:

---

Originally Posted by gosh

"constantly"?
there was a test where hyperthreading might have been a problem

---

And why does the 920 then performes better (also with HT) in contrast to the 870, which is basically the same cpu, just faster and performes worse.

Best example for that are again the audio encoding benchmarks.... how can a 870 perform worse then a 750 when the 920 is slower then the 870...


edit:
thx Kl0012 didn't noticed that, so basically that review is worthless :p:

Quote:

---

Originally Posted by kl0012

What is also missing from the equation is any thermal monitoring support for the P55 / Lynnfield processors using LM_Sensors on Linux, though that should come with time. Update: after starting to see a flow of Windows-based reviews today, it looks like there are some more serious Linux + Lynnfield problems at hand, which we are currently investigating.[/I]

---

Well I also think it is a bit strange that phenom II 710 did perform that good vs i5/i7.
But what is good is that there is at least one site that do test other different software. Doing that it might be problems because Intel may not have prepared it ;) (a joke)
You need only read one (or two) tests on sites that test the standard tests as everyone else (almost) so you know how they work.

The best thing with i5 is that I will soon have a portable that I can work on :up:

Russian review of i5-750 Turbo on/Turbo off/overclock:
http://www.3dnews.ru/cpu/intel_core_i5_750/index3.htm

Don't know if the Chile HW review was added,but here is the link :
http://www.chw.net/2009/09/intel-core-i5-750/5/
Pretty good review,pits the 750 against 920 i7 and 955BE(which it ties almost most of the time both in games and non-gaming usage).A lots of charts,good one chile hw.

As usual,IMO the Lost Circuits review is probably the best one.

Quote:

---

Originally Posted by informal

As usual,IMO the Lost Circuits review is probably the best one.

---

Not a bad review...quite good explaination of the core and motherboard.

But what's with the snide comment on page 6,

Quote:

---

Pretty much everybody should be familiar with the "unfair" practice by Intel to overclock one or several cores on demand.

---

Unfair? Now that's a laugh. Oh well, trudging onward....

It's too bad the review reuses the Phenom II 965 review's text on the Lynnfield on page 20, guess someone needs to do a little better proofing before publishing.

Quote:

---

Originally Posted by Humminn55

Unfair? Now that's a laugh.

---

Haha, yeah, I suppose they forgot to mention the unfair practice by AMD of directly clocking their processors at a higher frequency than Intel for the same market segments, or the unfair practice by both companies of underclocking and undervolting the cores when not in use to make them consume less... :D

Seriously though, I don't get why anybody could consider an advertised technological feature as "unfair". Are they clocking higher than usual some cores when not every of them is in full use? Hell, yeah, and that's a good feature for most of people :yepp:: You have a more performant CPU in some situations thanks to it. And it's not like if they're doing it in a hidden way to cause false interpretations or so, given that is one of their main marketing points...

Lost Circuits, Funny thing with them is that they have more reviews of AMD cpus, and more threads about AMD in complete contrast to every other site, due to Intel having 80% market share and more products you would think the natural flow would be more articles and threads on Intel, Unless there is a slight bias towards AMD?. And it also the site AMD fanboys always profess as the best? Many strange coincidences?

Their article list.

cpu

* intel ( 6 items )
* via ( 1 item )
* HP PA RISC ( 1 item )
* amd ( 9 items )

Graphics

* ATI / AMD ( 4 items )
* nVidia ( 1 item )

Also noticing a lot of AMDzone members are members there also, Yet another coincidence?

Quote:

---

Originally Posted by gallag

Lost Circuits, Funny thing with them is that they have more reviews of AMD cpus, and more threads about AMD in complete contrast to every other site, due to Intel having 80% market share and more products you would think the natural flow would be more articles and threads on Intel, Unless there is a slight bias towards AMD?. And it also the site AMD fanboys always profess as the best? Many strange coincidences?

Their article list.

cpu

* intel ( 6 items )
* via ( 1 item )
* HP PA RISC ( 1 item )
* amd ( 9 items )

Graphics

* ATI / AMD ( 4 items )
* nVidia ( 1 item )

Also noticing a lot of AMDzone members are members there also, Yet another coincidence?

---

Get a clue ,M.S. is a VP at OCZ tech. and he does the reviews .If you read the i7 review you can see he actually praises the tech a lot. The fact they have listed more AMD hw than intel is that the website was down and they lost a lot of their data a while back. LS actually have the most accurate data in their reviews. KTE ,a former member of XS who had a tremendous knowledge of hardware but left XS a while ago(a huge loss for XS),picked LS forum and is posting there regularly.

And gallag,what's with you and the AMDzone comments?Why you constantly mention that website? Did they do something bad to you so you hate them so much or what? You just can't get amdzone out of your head man.... Is that website so relevant(i highly doubt this) so it deserves so much of your attention?

Quote:

---

Originally Posted by gallag

Lost Circuits, Funny thing with them is that they have more reviews of AMD cpus, and more threads about AMD in complete contrast to every other site, due to Intel having 80% market share and more products you would think the natural flow would be more articles and threads on Intel, Unless there is a slight bias towards AMD?. And it also the site AMD fanboys always profess as the best? Many strange coincidences?

---

Maybe you can discuss the technology without needing graphs there?

According to the french site Hardware.fr, the second PCIe port is linked in 4X PCIe 2.0 (P55 chipset lines and not PCIe Cpu controller lines) in some mobos.

http://tof.canardpc.com/preview2/77e...782bf03030.gif

Somebody can confirm ?

edit : English Goggle Translate

6 pictures, rest read here http://lab501.ro/procesoare-chipsetu...d-in-actiune/3
made by Monstru In overclocking Intel all 4ghz, X4 965-3.8GHZ.
It's clear I5 750 overtakes X4 965 in many situations... Where not it can easily surpass Phenom under overclocking..:up:
In tests where 750 beats X4 965 , during overclocking the distance grows even more.:up:
So the concluzion...X4 965/955/945/940/920< I5 750....
If we add the fact that there are cheap P55 boards 100-130$(it will appear an Asrock board:P55M-PRO at 90$) than AM2+/AM3 is no more an more "economic platform" than Intel P55.:up:
And performance less...
http://lab501.ro/wp-content/uploads/2009/09/DivX.jpg
http://lab501.ro/wp-content/uploads/2009/09/DivX-OC.jpg
http://lab501.ro/wp-content/uploads/2009/09/WinRar.jpg
http://lab501.ro/wp-content/uploads/.../WinRar-OC.jpg
http://lab501.ro/wp-content/uploads/...9/FC2-1600.jpg
http://lab501.ro/wp-content/uploads/...C2-1600-OC.jpg

Quote:

---

Originally Posted by informal

If you read the i7 review you can see he actually praises the tech a lot.

---

Quite true and a good job was done. I just find it a little weird that the reviewer would call the Turbo Boost feature an "unfair" practice. Seems a tad biased in that left-handed comment.

But mostly, I was really struck by the obvious cut-and-paste job on page 20. The text is exactly the same as in the AMD Phenom X4 965 review, just with a new chart inserted into the two places, reflecting the new cpus. Seems like he probably meant to have text reflecting the Lynnfield review but instead has Phenom text.

And please don't label me an Intel fanboi......I've used AMD for well over a decade, beginning probably before most here with a 386DX40 cpu from AMD, and purchased it just after its release to market way back in the very early '90's....think like '91 or thereabouts.

But, I've always been dismayed that AMD seems to have difficulty keeping up with Intel. The Athlon X64 leadership AMD has for a couple of years seems to have evaporated and is again playing catch up, as usual.


Of course, the constantly changing sockets hasn't helped AMD either. Consider that during Intel's Socket 775 run, which debuted in 2004, AMD has had four different sockets.....939, which debuted with 775 and was supposed to last to give years and years of upgrade path; AM2 released 2 years after 939 and killed 939's upgrade path; AM2+....that was a nice upgrade with its split power plane....released one year after AM2; then AM3 which released 2 years after AM2+....and caused confusion as some AM3 cpus would work on AM2+ socketed motherboards, but not all---depended upon the mb manufacturer and BIOS, and AM2+ cpus would not function on AM3 boards at all....not exactly the best upgrade path again.

And now there's another new AMD socket on the horizon....from the roumors I've read....for Magny-Cours/Bulldozer.

In the meantime, Intel used 775 for 5 years, albeit with advancing chipsets. But the majority of the chipsets, around the 945 chipset, would allow the user to use older 775 cpus on newer boards, again, like AMD, depending upon BIOS. I've got an old 965 chipset Gigabyte motherboard that's running a Q9550 with no problems at all, even overclocks it pretty well.


I've lived through AMD's continuous socket changes and gave up after they killed 939. Of course, that was around the time C2D hit...and I have yet to see AMD produce anything that makes me want to revert back to AMD, at least for the current time. Maybe when Magny-Cours/Bulldozer hits, but by then, Sandy Bridge should be around, so I don't know.

Believe me, I root for AMD...I even buy their video cards, but for cpus in the last 3 years, AMD has just been a step or two behind. And unfortunate for AMD, they're relegated to fighting on price alone as their performance seems to be just behind Intel's performance.

Look at this release. Just when AMD is matching or beating the Q9650, Intel releases its new socket and processor for the masses, and it's undercutting the AMD prices (I'm using MicroCenter prices as they are most relevant to me as I have two and two Fry's near me).....i5 750 for $180 vs. X4 965 for $230. The motherboard prices are essentially the same and since both platforms use the same memory, that cost is irrelevant.

So, to buy a new platform today.....buy a socket that has what appears a very limited life now or a platform that will probably live on quite a number of years. In my case, until AMD severly undercuts Intel's prices or gets performance on par with Intel, it'll have to be competition on price which just kills AMD with each processor sold. Margins are thin for AMD already and having to cut the price of their top processor just to stay in competition with Intel's low end i5 is a losing proposition.

And for "I'll never give my money to Intel" sentiments, great. Don't. Buy AMD. We do need AMD to live to provide competition. It's just with my money, I'm not in the position to give charity when I buy computer components; instead, I look for max. performance for the dollar. AMD's 965 was decent and close, but even before i5 was released, I could buy a Q9550 for $160 vs. the $230 for AMD's 965, and use the $70 left for a better gpu.

To my view, the release of Socket 1156 is just another nail in relegating AMD to being the "value leader" instead of the performance/innovation leader.

Just my view.

Quote:

---

Originally Posted by Olivon

According to the french site Hardware.fr, the second PCIe port is linked in 4X PCIe 2.0 (P55 chipset lines and not PCIe Cpu controller lines) in some mobos.

http://tof.canardpc.com/preview2/77e...782bf03030.gif

Somebody can confirm ?

edit : English Goggle Translate

---

I know they're incorrect on the Asus P7P55D EVO and the Asus P7P55D PRO. Both of those boards are x16 for a single card, X8, x8 for dual cards. BUT, the caveat is you must use the blue and black PCI-e x16 slots for the x8, x8 function. The white PCI-e x16 slot on both of those boards function as an x4 electrically. Maybe that's where they're making their mistake....using the wrong slot for Crossfire/SLI.

Latest OC value comparison/review between i7 920 and Deneb :
http://www.pureoverclock.com/article809.html

Granted it's still not Lynnfield but the results should be close since they test the fixed OC frequency(3.9Ghz for both,both stock and i7 @ 4.5Ghz udner water-Deneb hit the wall @ 4.04GHz).Difference is not that big all around and SMT was ON in the tests.

Anyhow,what surprised me with Lynnfield is its die size of ~300mm2,which is actually a die size we would expect to see in MPU targeted at server market and not value desktop(750 model). The days of cheap MCM Yorkfields are gone for intel,now they fight value quad core segment with a much bigger die than 2xPenryn. This thing is faster than Yorkfield so people who considered value intel quad core will ditch aging s775 platform for the same priced s1156 platform(good for buyers ,bad for intel's margins).

Quote:

---

Originally Posted by informal

Latest OC value comparison/review between i7 920 and Deneb :
http://www.pureoverclock.com/article809.html

Granted it's still not Lynnfield but the results should be close since they test the fixed OC frequency(3.9Ghz for both,both stock and i7 @ 4.5Ghz udner water-Deneb hit the wall @ 4.04GHz).Difference is not that big all around and SMT was ON in the tests.

Anyhow,what surprised me with Lynnfield is its die size of ~300mm2,which is actually a die size we would expect to see in MPU targeted at server market and not value desktop(750 model). The days of cheap MCM Yorkfields are gone for intel,now they fight value quad core segment with a much bigger die than 2xPenryn. This thing is faster than Yorkfield so people who considered value intel quad core will ditch aging s775 platform for the same priced s1156 platform(good for buyers ,bad for intel's margins).

---

Its that big because of the PCIe controller other than that there are several things that go in production value other than the die size i am sure intel is making profit out of these in someway. :shrug:

+1 for the review its something i am looking forward to with a i5 750 and a PhII 955/965

Quote:

---

Originally Posted by Humminn55

I know they're incorrect on the Asus P7P55D EVO and the Asus P7P55D PRO. Both of those boards are x16 for a single card, X8, x8 for dual cards. BUT, the caveat is you must use the blue and black PCI-e x16 slots for the x8, x8 function. The white PCI-e x16 slot on both of those boards function as an x4 electrically. Maybe that's where they're making their mistake....using the wrong slot for Crossfire/SLI.

---

Thx for the enlightment Humminn55 ;)

Quote:

---

Originally Posted by ajaidev

Its that big because of the PCIe controller other than that there are several things that go in production value other than the die size i am sure intel is making profit out of these in someway. :shrug:

+1 for the review its something i am looking forward to with a i5 750 and a PhII 955/965

---

Yep i know its the integr. PCI-e controller's "fault",but still the days of cheap(from production POV) MCM Yorks are gone. This thing is noticeably more expensive than Penryn die,which really was designed primarily for mobile/Desktop and scaled up for servers.

PureOC article is something we need more often. They tested both at clocks that regular/more advanced DT users could achieve with aftermarket air coolers(and WC kits).

The CPU die size is bigger but the NB disappeared. P55's price ~ P45 + ICH10R. They are still making profit.

http://i279.photobucket.com/albums/k...p45_ich10r.png

Yes of course they make a profit,but bigger die on a wafer is still bigger die... Bigger die-> lower die to wafer ratio.

Quote:

---

Originally Posted by informal

Yep i know its the integr. PCI-e controller's "fault",but still the days of cheap(from production POV) MCM Yorks are gone. This thing is noticeably more expensive than Penryn die,which really was designed primarily for mobile/Desktop and scaled up for servers.

---

We don't have a clue what Intels production costs are to make that kind of judgement really, nvidia uses a godzilla die but somehow they manage.

Quote:

---

Originally Posted by highoctane

We don't have a clue what Intels production costs are to make that kind of judgement really, nvidia uses a godzilla die but somehow they manage.

---

296mm2 for 1 Lynnfield Vs 2 x 107mm2 for 1 Yorkfield,plus factor in the die to wafer ratio that is worse in former's case and you have a considerably pricier die.
NV is not a manufacturing company but a design one.They don't need to worry about costs of the production since they invest zero in that segment. As long as they sell the GPUs to their partners with a close to targeted profit they are good(of course competition's much smaller and excellent performing GPUs can put a dent into that plan though :) ).

Quote:

---

Originally Posted by informal

296mm2 for 1 Lynnfield Vs 2 x 107mm2 for 1 Yorkfield,plus factor in the die to wafer ratio that is worse in former's case and you have a considerably pricier die.
NV is not a manufacturing company but a design one.They don't need to worry about costs of the production since they invest zero in that segment. As long as they sell the GPUs to their partners with a close to targeted profit they are good(of course competition's much smaller and excellent performing GPUs can put a dent into that plan though :) ).

---

I just can't wrap my head around this "CONSIDERABLY pricier die", there can be plenty of variables to different die nesting, sure you can get more smaller dies but also how many more of the smaller dies are lost at the perimeter, how much margin is there between the dies, as in lost area between the many smaller dies vs less margin loss with fewer single larger dies, etc...

Sure the dies are bigger but it can amount to 1 cent or less a piece cost difference as far as production cost is concerned. Its an issue that simply doesn't matter to anybody but Intel and we all know Intel doesn't give up their gross margin easily.

We also have the move to 32nm steam rolling on the heels of these chips, hell Lynnfield is bigger than bloomfield but its a non issue really because nobody is buying a cpu based on the raw manufacturers cost.

Its just a silly thing to try to claim "considerably pricier die" when its totally up to ones imagination what that cost might be, do you want to at least throw a value on "considerable" what .000099...

hehe, Funny how suddenly a 50% larger die doesn't cost Intel any more in some peoples eyes, but with Agena's large die it was a case of " omg AMD must be losing money on these chips.. the're so big" etc etc.

Whilst this isn't another case of Agena (Since it underperformed as well as being expensively sized) I think it IS fair to say i5 750 probably isn't the biggest margin-giver for intel. It's considerably lower priced than it's HT-enabled brothers, and suffers in performance (reviews) without HT.

As people have rightly pointed out though, Intel make up for the big die costs some what by charging P45 chipset price just for a little southbridge (essentially). I don't know what the die size of P45 northbridge was, with that info you could at least get an idea.. provided the fact it's built on cheaper older process is taken into account .

And after all, for every i5 they sell, they sell a $40 P55 MCH

I think it's just a market share stealer, and the fact they probably don't make what they'd like to out of them is probably not a big deal so long as they're moving lots of them by stealing Phenom II market.

As for turbo mode on all 1156's . It's a great feature, I don't really think cheating is the word. It does mean though is a couple of things:

A. It's not a <insert box label>Ghz processor when enabled, often a lot higher even with a good load. Bonus for end user, but can cause issues as below..
B. Reviewers results will vary.. If a reviewer straps on a 5kg peice of copper + wind tunnel fan to the top of it for reviews (as many do) They may indeed get more performance than a review with the std HSF.
C. Non overclocked systems running mediocre cooling may suffer performance degredation over time and in poorly vented cases where full load temps throttle turbo mode back as low as the 'minimum' clockspeed on the box.

It would be interesting to test out point 3 in controlled conditions tbh. With such an aggressive turbo like that it might actually make a difference (unlike original core i7)

Again though, it should be mentioned. Turbo mode a great feature, and something AMD really stumbled on. They could have implemented it in K10 based arch with enough fwd thinking to help in the wait for there next gen uARCH. As John Conner would say.. Easyyy money :p

Quote:

---

Originally Posted by informal

And gallag,what's with you and the AMDzone comments?Why you constantly mention that website? Did they do something bad to you so you hate them so much or what? You just can't get amdzone out of your head man.... Is that website so relevant(i highly doubt this) so it deserves so much of your attention?

---

gallag probably has a low tolerance for the mentally deranged.

Quote:

---

Originally Posted by mAJORD

hehe, Funny how suddenly a 50% larger die doesn't cost Intel any more in some peoples eyes, but with Agena's large die it was a case of " omg AMD must be losing money on these chips.. the're so big" etc etc.

---

It wouldn't surprise me that a combination of; savings from eliminating a discrete north bridge (costs being die, packaging and testing), the design being much more tuned to the 45nm node improving parametric yields, and manufacturing on essentially amortized equipment make Lynefields equally if not more profitable than Yorkfields.

i5 750 vs i7 870 vs i7 920 vs X4 965
http://www.techspot.com/review/195-m...u-performance/

Core i5 and i7 and P55 Lab Update
http://anandtech.com/weblog/showpost.aspx?i=642

Lynnfield PCI-Express Gaming Performance
http://www.bit-tech.net/hardware/cpu...-performance/1