Unlocked multies is for wussies... real men overclock the fsb.. or qpi :fact:
;)
Printable View
Unlocked multies is for wussies... real men overclock the fsb.. or qpi :fact:
;)
LOL he is in fact, but that's besides the point, he speaks the truth.
Yorkfields are clocked higher than Phenoms too remember, we're talking about a ~15% higher IPC and 3GHz+ vs 2.6GHz max, so yes, 30% is quite accurate if we're taking the top chips from Intel and AMD. Oh and please don't get started about how we're comparing a $1k CPU to a $200 one... that's really not the point here.
Well that sounds very promising! Given i d be happy with 3.6ghz ish, as with the perf gain of 5% at least, its almost a 3.8 yorkie.
i920 seems like it could be a fun choice now :) if msi can keep that promise.. OBR said his mobo died on flash so , i guess i ll have to wait until nda to know. Do you know the retail of that board.. under 300$ i hope?
AudiFreak already got the shipping info, bring it on :up:
:rofl:
Jk, I know I wouldn't mind having an unlocked multi 775 as I'm very fsb limited on my old p5wdh but .. it's the piece of hw I paid the most for iirc in my whole rig so it's staying till it breaks :)
Axp days are over? turbo is bios controlled who knows what the communities brightest can do :shrug:
I'm not holding my breath for it but I wouldn't faint when I hear it's done ( or.. I'll faint becuase I want one )
Well unless you are benching, i can actually see the powersaving and turbo mode features coming in useful, if I have my quad clocked at 3.8 - 4.0 at decent load and idle temps on air, i dont mind it powering down cores to save power and temps (wear and tear) as well as boosting up to 4.1, 4.2 if needed.
As far as i understood nehalems turbo, its is controlled by the PCU and bios has only a switch to enable/disable it.
I doubt someone is brave enough to mess with the firmware of the PCU, hell i doubt its even possible to manipulate it. :p:
Btw. question regarding the PCU, since its controlling all the whole powerconsumption of the cpu and gets all the Informations about Temps/Voltages, is there going to be a whitepaper on how to use that data?
:) :) I don't mean to pick on you, but this is an oxymoron. So Intel lost high end market share to who? Itself? :) :) Just sounds weird.
I think you meant to say that they just froze their current high end sales through an Osbourn effect.
What it does, though, is replaces their high end offering with a new high end offering, the current high end offering will slide down into the mid range, and the mid range will slide down into the low end... this is called progress.
...*heresy*...
Actually, it is true ... but under conditions of pure CPU limited measurements, for example, if you compare Phenom to Yorkfield at the same clock speeds at 1920x1200 using a nVidia 8500GT ... they will appear equal, but if you measure at 1280x1024 using a 280GT Yorksfield, on average, will be 20-30% faster clock for clock.
This does not make Phenom a cruddy gaming CPU, it can easily support high frame rates to make an enjoyable game... however, at an architectural level, the Yorkfield IPC is just much higher running gaming code.
I understand many people who are attached to their product preference do not accept this or like the result, but it is pretty much the case. RWT did an outstanding article to understand why this is the case:
http://www.realworldtech.com/page.cf...WT102808015436 here they compare an older K8 overall, but the general idea is the same. Most of the good performance in the core architecture arises from efficient use of cache and, most importantly, very strong branch prediction -- Intel's branch prediction just overwhelms AMD's implementation.
go through the page here: http://www.matbe.com/articles/lire/1...eurs/page9.php
translated to english : http://translate.google.com/translat...hl=en&ie=UTF-8
http://www.matbe.com/images/biblio/a...0000077877.pnghttp://www.matbe.com/images/biblio/a...0000077852.png
100 CPU tested! And that look more than 30% to me ... :rofl: :clap: :clap: :clap: humm hummm ... ok, from 26% to higher ! :P
what about this:
http://www.matbe.com/images/biblio/a...0000077856.png
Howcome you don't link to the overall performance graph which averages over all the tests? i.e. this one ("Une moyenne des 21 tests est ensuite calculée et reporté dans le graphe ci-dessous."):
2.5GHz Phenom 114.4
2.5GHz Yorkfield 139.9
18.2% or 22.3% difference depending on what you take as the base. Most would take that as ballpark 20%. 30%? Only Intel employees and Shintai :rolleyes:
I don't have a problem with company reps like Intel or OCZ posting here but trash-talking business rivals and using fanboi-math is distasteful and inappropriate. Getting too cosy eventually corrupts - remember Novak or Judith Miller? Or the newly-saccharine Bob Woodward?
When a large company uses these types of forums to spread the word about their products, it's a privilege for them to be allowed to post, not a favor they bestow on the forum.
http://www.matbe.com/images/biblio/a...0000077926.png