Printable View
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.
Not a bad review...quite good explaination of the core and motherboard.Quote:
Originally Posted by informal
But what's with the snide comment on page 6,Unfair? Now that's a laugh. Oh well, trudging onward....Quote:
Pretty much everybody should be familiar with the "unfair" practice by Intel to overclock one or several cores on demand.
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.
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... :DQuote:
Originally Posted by Humminn55
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?
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.Quote:
Originally Posted by gallag
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?
Maybe you can discuss the technology without needing graphs there?Quote:
Originally Posted by gallag
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
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.Quote:
Originally Posted by informal
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
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).
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:Quote:
Originally Posted by informal
+1 for the review its something i am looking forward to with a i5 750 and a PhII 955/965
Thx for the enlightment Humminn55 ;)Quote:
Originally Posted by Humminn55
Quote:
Originally Posted by ajaidev
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.
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 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.Quote:
Originally Posted by highoctane
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
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
gallag probably has a low tolerance for the mentally deranged.Quote:
Originally Posted by informal
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.Quote:
Originally Posted by mAJORD
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