Sign in with
Sign up | Sign in
Your question

4 Ghz Core i7 920

Last response: in CPUs
Share
a b à CPUs
September 28, 2008 11:19:49 PM
a b à CPUs
September 28, 2008 11:25:10 PM

Wonder if that chip will make it a week at 1.47 volts.
September 28, 2008 11:27:22 PM

1.47 volt will likely last for a couple months at least, depends on the stepping.

EDIT: But I guess it is established that Nehalem can indeed hit 4.0Ghz on air. Now let's see if Deneb can do the same, like those from AMDzone claimed :D .
Related resources
a b à CPUs
September 28, 2008 11:45:58 PM

What really got me was the Everest score
September 29, 2008 12:06:32 AM

Not try to discredit this, but whats the deal with the asian sites always? Isnt Intel an american company? Why is it always asians that seem to "come across" these chips?
September 29, 2008 12:20:32 AM

Because most motherboard manufacturers are based in Asia, and they are the ones who get the engineering samples.
September 29, 2008 12:22:24 AM

Malovane said:
Because most motherboard manufacturers are based in Asia, and they are the ones who get the engineering samples.


Well, that completely went past my head. :pt1cable: 
September 29, 2008 1:15:08 AM

7K for a CPU score.

Holy Crap! I think i will get a Deneb upgrade as i bet Intel will milk the $hit out of these chips and motherboards. And frankly, i don't really need all that performance...

I never though i would ever say that. : (
September 29, 2008 1:30:49 AM

hey don't forget, that's the lowrange cpu at that
Anonymous
a b à CPUs
September 29, 2008 1:31:17 AM

whats the need for nehalem...

the pentium to core 2 jump was massive and was needed as it demonstrated a large performance increase in basically everything...

now an oced core 2 processor is still up there... and you won't see any real world advantages to having nehalem over a oced core 2 duo... theres just no reason... especially for people here who like to use their computers for gaming... and so far there has been no proof that the jump from core 2 to nehalem brings anything new to the gaming table
September 29, 2008 1:53:24 AM

So, a dual core at 4.5 will beat out a slowly burning out i7 at 4Ghz in alot of games. Hmmmm
September 29, 2008 1:55:13 AM

Quote:
so far there has been no proof that the jump from core 2 to nehalem brings anything new to the gaming table


Aside from a significant increase in performance, anyway.

The only real question is how long games will need to take advantage of being able to run eight threads simultaneously; hopefully newer engines will be able to divide the work into as many threads as the CPU can support.
September 29, 2008 1:57:16 AM

Considering 99.9% of the games out there dont even take advantage of TWO/FOUR cores yet, 8 is really pushing it.
September 29, 2008 2:06:10 AM

MarkG said:
Aside from a significant increase in performance, anyway.

The only real question is how long games will need to take advantage of being able to run eight threads simultaneously; hopefully newer engines will be able to divide the work into as many threads as the CPU can support.


Actually, some benchmarks floating around seem to indicate that i7 will likely have less performance clock per clock in games than Yorkfield, even in some multi-threaded games. For many other tasks (especially server related which was the focus of the architecture), it should do much better. It's unclear why there may be lower performance, though it could be smaller cache sizes to blame.
September 29, 2008 2:06:23 AM

Thus far, you have to look to find where quads help that much in games, unless its a x2 or sli/cf rig. Thats just because the cpus cant keep up with those setups, and have very litlle to do with gaming performance itself as to how the game "uses" a quad. Eight? Lets see, i7 will be i8 or 9 before we see that happen.
a c 127 à CPUs
September 29, 2008 2:08:58 AM

JAYDEEJOHN said:
So, a dual core at 4.5 will beat out a slowly burning out i7 at 4Ghz in alot of games. Hmmmm


Thats how it is either way. Faster CPU normaly means better FPS.

And its funny about the games and multicore. DX10 was built around multithreaded gaming hence why DX10 was supposed to be faster. But most game designers are too lazy to use that feature to its fullest, yet Microsoft did it easily in FSX which can use up to 16 cores.

It makes me think the game designers don't wan't to take advantage of the CPU power because they have a deal with the GPU people so that people will buy nw GPUs to play the newest games.
September 29, 2008 2:32:31 AM

Maybe its nVidia, as we see ATI heading in a multu core gpu direction, which obviously benefits from quads. nVidia has alot of pull when it comes to gaming, as seen by the non use of DX10.1 as well. This could be the woopass being handed out that we dont see, but is there nonetheless
September 29, 2008 2:41:35 AM

Malovane said:
Actually, some benchmarks floating around seem to indicate that i7 will likely have less performance clock per clock in games than Yorkfield, even in some multi-threaded games.


But they seem dubious compared to the performance gains seen elsewhere. It's certainly possible that the real chips running real drivers on real motherboards will end up slower, but it's at least as possible that the graphics drivers are defaulting to old and slow code paths because they don't recognise the CPU, or the motherboards aren't properly optimised yet.

We know the fundamental core performance is significantly higher and that the memory interface is dramatically faster; so unless those games are heavily cache-limited it's hard to see what could legitimately cause lower performance. And if those benchmarks are real, then Intel have seriously stuffed things up; which seems somewhat unlikely given their performance over the last few years.
September 29, 2008 2:50:31 AM

Considering the launch is a few weeks (?) away, I would damn sure hope that motherboards are "optimized" by now. i7 will die overnight if X58 turns into another 700i, but thats going a little overboard. Maybe.
September 29, 2008 3:05:24 AM

And is fairly consistant with what Annand found as well. Maybe 5% at best, nothing big, and its been written time and again, that i7 just doesnt perform in games
September 29, 2008 3:07:36 AM

JAYDEEJOHN said:
So, a dual core at 4.5 will beat out a slowly burning out i7 at 4Ghz in alot of games. Hmmmm


If you're purchasing Nehalem, you're either 1. rich enthusiast, or 2. someone who will utilize Nehalem's feature to achieve greater performance.

Other than that I agree with you. Nehalem is simply not a CPU for mainstream gamers and users.
September 29, 2008 3:10:23 AM

Well, to be honest targeting the gaming community would be a bad decision for Intel, considering PC gaming machines are just a tiny fraction of the market. The bigger consumers are servers, media machines, and home use. However considering the tech/hype and money, i7 should be doing alot better than what these benchmarks are indicating, but its just not.
September 29, 2008 3:21:41 AM

Intel has never marketed Nehalem for the gaming community, but for the enthusiast community - someone who would go out and spend big bucks to get the newest and the latest.

But for mainstream gamers, a fast dual core is plenty enough.

And of course, Nehalem's design is more suitable for server applications, but that would also mean pushing AMD out of its current market. Now the last thing Intel needs is an anti-trust case.

I think Core i7 performed superbly, and even more so when server benchmarks surface.
September 29, 2008 3:30:38 AM

To me, i7 is similar to what nVidia is doing with CUDA. Theyre both dependent on the future of how SW will be done. Performance was minilmalized for the sake of heading in each of their respective directions, and improving on that, instaed of traditional areas. I guess if you dont game, and do use multithreaded apps, youre feeling the same nVidia fans do currently. What Im wondering is, what will the next gen bring? nVidia will have its out first, and maybe thatll be a primer as to what to expect from Intel as well. I know it wont be a direct comparison, but both have wondered away from their traditional areas, seeking performance in newer areas, and how they respond going back to their older more traditional areas of performance may be seen as equals IMO. Tho, I suspect Intel has thrown everything into this direction (multicore,multithreading) with little alternative of IPC developement as seen in the past, as well as core speeds.
September 29, 2008 3:46:30 AM

spathotan said:
Well, to be honest targeting the gaming community would be a bad decision for Intel, considering PC gaming machines are just a tiny fraction of the market.


But how many people other than gamers, video editors and 3D animators actually need a quad-core CPU?

Sure, I've been using a dual quad-core server at work, but that's not exactly intended for word processing, email and web browsing; the only real reason for Joe Sixpack to buy anything faster than a dual-core Atom is to play games.
September 29, 2008 4:02:27 AM

And with CUDA, you can maybe eliminate 3D animators and video editing
Anonymous
a b à CPUs
September 29, 2008 4:03:45 AM

nehalem just isn't needed by anyone on this forum who actually uses their computer for games

people on this forum who are looking to build servers and work machines... often don't stay on the forums for any length of time...

TBH... nehalem is not needed except in the server area....

theres nothing an oced core 2 duo can't already do... and a oced core 2 quad just pushes the usability of this "old" generation further into the future...

I just am not understanding why this nehalem is "another core 2" as some sites publish... i mean sure it does improve in some areas and stuff... but who actually uses those things extensively... besides that... it just doesn't bring the kind of overall jump in performance as the core 2 jump did

I'm not really seeing a lot in terms of CPU mainstream advancement... I mean theres nehalem... which doesn't really bring much to the table that core 2 doesn't... and then after that theres westemere... which just adds more cores to the already non multi threaded generation... (hopefully this will change) and then theres sandy bridge... which just adds nodes to the mix, more cores, and a smaller die...

There doesn't really seem like there is a lot of advancement for the average hardware enthusiest...

for the average joe system builder... Sandy Bridge will be a break through as it will allow better overall system performance at a lower cost... but for someone who already has a high power dedicated gpu... and cpu... its not needed...

and on AMDs side... Deneb I don't really see doing much except MAYBE matching 45nm core 2's... and thats a stretch as it shows its coming close to the old core 2's... and farther along in the road for AMD... theres just not enough data to make any conclusions... but it doesn't look good for them... they haven't made ANYTHING that rivals the performance of the core 2 line... over 2 years later.... and nehalem is yet to release another "core 2"

Unless AMD pulls a magic CPU out of its a** I don't really see any need for anyone with a high power system (over 3 ghz core 2 system) to upgrade in the foreseeable future... to me it just would seem like a waste of money to get an extra 5 % performance



anyway my gibberish for now is complete :D 
September 29, 2008 4:14:41 AM

Maybe it is a lack of SW dev. Even games today, except for Crysis, theres not alot out there challenging todays gpus. Hopefully, this is a lull before the storm
September 29, 2008 4:17:55 AM

And Crytek games are nothing but tech demos anyways, so thats not saying much about the state of the software industry.
a b à CPUs
September 29, 2008 9:34:05 AM

I wonder how well it will fold...
September 29, 2008 9:59:43 AM

be interesting to see the final version of the CPU
September 29, 2008 10:23:57 AM

Quote:
nehalem just isn't needed by anyone on this forum who actually uses their computer for games

people on this forum who are looking to build servers and work machines... often don't stay on the forums for any length of time...

TBH... nehalem is not needed except in the server area....

theres nothing an oced core 2 duo can't already do... and a oced core 2 quad just pushes the usability of this "old" generation further into the future...


It depends on the game you play. FSX for example, no CPU can currently max it out, even Nehalem @ 4GHz with 8 threads won't come close.

Theres another game called Football Manager (or Worldwide Soccer Manager in the US) that can easily bring any CPU to its knees if you choose too many leagues to simulate, especially at detailed settings.

Now I realise these are somewhat 'niche' games (FM actually has a huge following outside of the US, its the top selling game in the UK for example) but it goes to show not 'all' games conform to the typical 'dual cores are plenty' adage.
Anonymous
a b à CPUs
September 29, 2008 11:35:49 AM

True, and I hope more games are soon to follow... but lately it just seems developers have just been making half-a**ed games to get money... they don't even try to be revolutionary...

We need a game... that not only uses 4 or more cores, has very high graphical detail at less of a performance hit than crysis, and is a GREAT game. A game like this would catapult the industry forward... the state of pc gaming has been stale for several years now... there haven't been a lot of good games that have come out.
a b à CPUs
September 29, 2008 12:55:26 PM

sheila = Ms. Bytch lol.
a b à CPUs
September 29, 2008 12:59:14 PM

oh and the world doesn't revolve around gaming. The world depends on servers. Lets see what Cray has to say about nehalem.
September 29, 2008 1:39:55 PM

I'm picking one up around Christmas, just because I'm a heavy Photoshop user, and my Socket 939 is having troubles keeping up.
September 29, 2008 2:07:54 PM

If you have adobe and a nvidia gpu, you may not have to, tho who knows for sure
September 29, 2008 5:08:50 PM

For gamers, it is hard to justify Core i7 RIGHT NOW. Who knows what will happen in the future; right now games aren't fully optimized for quad cores since they aren't the majority of the market (no statistics just what I have seen based on LAN parties and friends). If Core i7 and quad cores are really pushed for mainstream gaming than this could change and I have a feeling more games will be coming out with better quad support.

For the server and "special" gaming market (FSX and those "games" previously mentioned) I agree with most of the opinions expressed that it will be pretty good for these purposes.

As a heavy photoshop user I think my quad handles it just fine. My biggest issue is ram, 4gb just isn't enough for me especially if I have things running in the background. Core i7 won't help this initially since DDR3 prices are crazy atm and buying more than 4gb of ram (well since it is tri-channel 3gb or 6gb makes more sense) is beyond my budget. Although, a lot of crazy photoshopers will probably be able to pony up it is hard to justify it for students (me) and non-heavy users.

Also, something that has yet to be mentioned. This is an ES chip. Usually ES chips have unlocked multipliers so who knows if that 4.0ghz OC is even achievable with a normal 920.
September 30, 2008 1:37:19 AM

Like I said in my earlier post MU, these asian benchmark screenshots are always shady. We already know what the multipliers are, why mark it out?
September 30, 2008 1:40:35 AM

Coolaler is far from a "shady" source.

As for the scratched out multiplier and bus speeds...could simply be due to NDA of those particular areas. Who knows.

But I would believe Coolaler's screenshots over most "leaked" screenshots.
September 30, 2008 2:02:09 AM

Im begining to think with all these teases that i7 is more talk than show. Too much mystery. Too much of "leaked" shots that gives us nothing concrete. Poor showing by Intel. If they have something, then show it, dont keep building the hype. Do it or get off the pot
September 30, 2008 2:10:42 AM

JAYDEEJOHN said:
Im begining to think with all these teases that i7 is more talk than show. Too much mystery. Too much of "leaked" shots that gives us nothing concrete. Poor showing by Intel. If they have something, then show it, dont keep building the hype. Do it or get off the pot


So, because someone leaks screenshots, it's Intel's fault? I'm sorry, but that makes no sense.

Did any of these "leaks" have Intel's watermark on any of the screenshots? If not, then how can you claim that Intel is hyping i7? If you have an issue with leaked info, go to the source of the leak, and tell them how you feel, and not blame a company that is not causing the leak.

I don't see you complaining about the "leaked" Deneb screenshots, or telling AMD to get off the pot? Double standards?
September 30, 2008 2:32:57 AM

Nobody "leaked" anything that Intel doesnt want out, and we already know this, as well as the results. The onus is on whats NOT being shown, which is petty. If they had any cahones, theyd have shown the whole thing. But no, they play ball with Intel and the hype
September 30, 2008 2:42:50 AM

JAYDEEJOHN said:
Nobody "leaked" anything that Intel doesnt want out, and we already know this, as well as the results. The onus is on whats NOT being shown, which is petty. If they had any cahones, theyd have shown the whole thing. But no, they play ball with Intel and the hype


I guess you have never signed an NDA in your life, then, have you. It's not about cahones, it's about building a reputation, where a company will allow you to continue to test, or preview items, before their release. You sign an NDA and decide to ignore it, or as you think - "show their cahones", there can be reprecussions of that sort of action. It's funny that you think that Intel is forcing them to "play ball" and hype their product.

Go look up NDA, and see if it's reasonable to sign an NDA, then ignore the signed NDA.
September 30, 2008 2:44:45 AM

I said it was too hyped up a few weeks ago, and EVERYONE came out of the woodwork claiming, no, i7 isnt being hyoed at all. If this isnt hype, what is it? As far as Deneb goes, if theyre doing the same thing, shame on them as well. You cant go around saying that i7 isnt being overhyped, and thus causing the idea of something great coming from Intel as not going to happen, and then say this isnt hype. Thats all it is, and we dont no anything more than we already know, its just more hype, and I wont swallow that its not. You may agree with me, but still the questions remain: How well will it oc? CAN it be oceed? and stably? This isnt a AMD vs Intel thing, just because I dont like what Intels doing here doesnt mean Im favoring AMD, thats just plain fanboy paranoia. It doesnt mean Im not looking fowards to Intels offerings. It doesnt mean I wont be buying a Intel cpu. I dont like the hype, and showing us nothing we dont already know
September 30, 2008 2:53:39 AM

And as Ive said, its hype coming straight out of Intel. NDAs are NDAs. Having all this hype about this mystery, this unknown about its overclockability is pointless. Did we have this with C2D? I remember seeing alot different back in those days, all NDAs asside. This isnt how it was handled in the past, and that in itself shows somethings different here. Why all the sudden is Intel being coy? or Shy? Is it pure hype? Or is their cpus not oceable?
September 30, 2008 2:59:39 AM

Hype would be the company releasing it, claiming stuff that just doesn't seem possible. Want an example of hype? "We expect across a wide variety of workloads for Barcelona to outperform Clovertown by 40 percent," Allen said (Randy Allen, AMD corp. VP). That is hyping a product.

Showing a screenshot of a CPU hitting 4GHz, is not hyping a product. If anything, it shows that a supposed non-overclockable CPU, can overclock (although, it remains to be seen, if this is an unlocked ES sample).

You can claim that review sites' performance estimates are hype, but that is speculation by that review site. The closest hype you can claim Intel has done was during IDF. Otherwise, it's all speculation and "leaked" screenshots, that does not equal "Intel hyping" a product. If anything, it could be a reviewer pushing the limit of the NDA.
September 30, 2008 3:04:52 AM

JAYDEEJOHN said:
And as Ive said, its hype coming straight out of Intel. NDAs are NDAs. Having all this hype about this mystery, this unknown about its overclockability is pointless. Did we have this with C2D? I remember seeing alot different back in those days, all NDAs asside. This isnt how it was handled in the past, and that in itself shows somethings different here. Why all the sudden is Intel being coy? or Shy? Is it pure hype? Or is their cpus not oceable?


Actually, yes. The same thing happened during C2D. In fact, there was a huge uproar of unbelievability at this forum during that time, when leaked benchmarks were shown. Some believed it whole heartedly, others did not. Most did not just ignore them, but waited for more information.

There was leaked screenshots of X6800 months before it's release. There were leaked benchmarks, which showed scores that were impossible.

It was not hype back then, either. And Intel was very quiet during that time, too. That was what was so un-Intel like. Everyone expected a "see what we can do" attitude, but instead they laid low, until launch, then started talking after release.
a c 102 à CPUs
September 30, 2008 3:11:17 AM

NMDante said:
Coolaler is far from a "shady" source.

As for the scratched out multiplier and bus speeds...could simply be due to NDA of those particular areas. Who knows.

But I would believe Coolaler's screenshots over most "leaked" screenshots.


Well, it could be three things:

1. Multiplier of 20, bus speed of 200.5 MHz
2. Multiplier of <20, bus speed > 200.5 MHz
3. Multiplier >20, bus speed < 200.5 MHz.

The first two are simply bus reference clock overclocks like we've been doing on K8s for the last five years. That's nothing special *except* that it shows that the whole Intel spiel of "don't increase the QPI reference clock speed above 133 MHz- it can be dangerous" and the "unlocked Bloomfields won't OC" are both quite wrong. The last one could be interesting or not, depending on the exact chip in question as it would denote an (upwardly) unlocked multiplier, which was supposed to be only on EE chips. If it's just an unlocked engineering sample, big deal, lots of those are unlocked. But if it's a retail i7 920 with an unlocked multiplier, it's a big deal as few will buy the 940 and 960 EE models since the cheaper i920 has the goods.

What do I think? I can't tell. 4010 MHz rounds out very nicely to a bus speed of 200.5 MHz with the stock 20x multiplier, which is probably as close to 200 MHz exactly as you can get on that board. He has memory rated at DDR3-1600 in that board, which run at 200 MHz. So he likely overclocked the reference clock to 200 MHz to run his RAM at the full DDR3-1600 speed. If the 920 can only handle up to DDR3-1066, this makes perfect sense since the QPI:RAM strap is 1:4 in both cases. He could also have a multiplier-unlocked chip running at 30x multiplier on the stock QPI speed. 4010 divided by 30 gives a base clock of 133.7 MHz, which is suspicious for a multiplier overclock as it also comes out so cleanly. But then why the DDR3-1600 instead of DDR3-1066? I suppose it could just be what he had...

After looking at this, did this quy pick 4.0 GHz as a target and blank out the QPI clock, multiplier, and strap just so we can't tell unlocked multi vs. good overclock on the QPI bus by the core speed? Or did he just get to the magical 4.0 GHz figure and stop, then got lucky that 20 and 133 both went evenly into 4000? If it's the former, oooh, he's tricky...
!