Core i7 Mobile product names revealed

Here are the names for upcoming CPUs that will be used in Calpella, Clarksfield.

Core i7 920XM-2.0GHz Base Clock/3.2GHz Turbo/55W TDP(45W+10W Northbridge)/8MB L3

Core i7 820QM-1.73GHz Base Clock/3.06GHz Turbo/45W TDP(35W+10W Northbridge/8MB L3

Core i7 720QM-1.6GHz Base Clock/2.8GHz Turbo/45W TDP(35W+10W Northbridge)/6MB L3

*The Turbo clocks are the maximum speed in single core operation.*

 

Wow, those are some high TDPs for notebooks.

 

Well they're basically quad core chips, which kinda makes sense seeing as the chip itself includes the northbridge.

 

low frequencies... we'll have to see where this goes...

 

It shouldn't be bad when unused cores are turned off on the fly.

Gotta consider the per-clock increase in performance over an equivalent Core 2, though.

 

It might be more efficient clock for clock, but the fact is you need to measure its capabilities at load (to see its full capabilities). Even if it's 25% faster clock for clock, the overall clockspeed is only 80% of the current quads, which means it will only equal the top end QX9300.

 

When i had heard these CPU's were going to be 35watts i was SO excited... now they actually appear to be about even or maybe even a little worse than the core2quads currently available.

This makes me sad again.

Even the eXtreme version isn't that impressive. 2.0 quad 2.6 dual or 3.2 single Its nice and flexible, but still seems a disappointment for a 55watt total TDP.

Maybe the number tell a different story when they are released though. I've been waiting for this release eagerly to now have it "seemingly" crushed by very high TDP's and very low clocks.

 

They look absurdly crap.. Need way better clock speeds. QX9300 might pwen them...

 

If Core i7 is 25% faster than Core 2 and is operating at 80% of the speed, the two extreme models will be about equal. I doubt the Core 2 will pwn the Core i7s but the i7s won't bring much of an improvement compared to previous releases. We'll have to see how they OC and how much power they actually consume to truly get a good comparison, but the numbers don't look good ATM.

 

That TDP will be with hyperthreading meaning 8 threads executed at max.

 

Clock speed is not everything. These processors have features like Turbo Boost that can make them faster.

 

I just hope Intel doesn't get out of hand with the multicore trend like they did with the gigahertz trend back in the P4 days.

 

Turbo Boost only works by shutting down other cores to boost speed on one or possibly 2 cores. When you are at max load, those (low) clocks speeds are what you're going see. While it's true clock speed isn't everything, it is significant. As I said, the clock for clock advantage will probably cancel out the difference in clock speed.

 

I personally tihnk that turbo mode is overrated by the way intel describes its use. The premise is that if a program is single threaded and your CPU is not using all its cores, it will shut down a core (or more than 1) to overclock a single core for that application. While this might be good in say SuperPI, when doing real work, most are multithreaded and hence it will run at the clock speed mentioned.

 

Intel seems to have covered both grounds fairly well with these, not perfectly, but fairly well ( thought it seems you guys arent reading the fine print ).

For example, lets say you are using a single threaded application: use turbo mode and you will now have what is essentially, a dual core processor ( one core with HT ) running at 2.8Ghz that is about 20% ( this is taken from desktops ) faster than the older Core2 architecture.

Now, for comparison, lets say you are using a multicore/multithreaded app, you will have what is essentially an 8 core processor running at 1.6Ghz, which still leaves out the increase in efficiency per clock.

So basically, unless you are using an application that can only take advantage of 3 or 4 cores, and are using the Q9100 or higher ( Im betting that the i7-720 beats the Q9000 ), the mobile i7s will win out.

 

As I said before, comparing clock for clock, 20% better is not enough since the clock speed is lower. These tests include the extra thread. If you do encoding for example, you will utilize all 8 threads, but hyperthreading doesn't mean it's 100% better (ie. it's not 8 cores). The hyperthreading is where it exceeds the Core 2s (hence why it's better clock for clock). Overall if it encodes 20% faster (clock for clock), it won't be better than the current faster clocked models it is replacing.

 

Well, I do quite alot of encoding, and when I upgraded from a Q6600 to an i7 920, the fps almost went up 3 times of what I was getting before. I dont have much knowledge as to how hyperthreading really works, but 8 threads do help alot, esp during h.264/x264 encoding where all threads are stressed to ~100%. If Calpella is similar to Bloomfield, I would love to get a well-priced i7 notebook

 

Did you overclock your CPUs? The Q6600 is 2.4GHz, while i7 920 is 2.66GHz. Consider the QX9300 at 2.53GHz and the new mobile extreme i7 at 2GHz. To get the same ratio of clock speeds as your desktop comparison, it would be like a Core 2 at 3.3-3.4GHz to the 920 at 2.66GHz. If your Q6600 ran at that speed, encoding would be increased significantly. Not to say that the i7 isn't better, because for encoding, i7 is the most efficient. But for general users, they won't notice much difference in most other applications such as gaming.

 

That is dependant on how well the HT works. For example, since as you said, the HT is not the same as a single core, lets say each HT thread, is the same as half of a core. So then this would be 6 ( 4 cores + 2 half cores, and yes I know CPUs dont scale linearly, but his is for ease of math ) threads @ 1.6 Ghz ( before the 20% increase in efficiency ) vs 4 threads at 2.0Ghz. If we include the increase in efficiency, the 1.6Ghz becomes, about 1.9Ghz, but will also have two extra threads to execute on. You could cut down those 4 extra threads into one ( making each of them a quarter of a core ) and the new mobile i7 would STILL beat the Q9000, since it would be 5 threads at 1.9Ghz vs 4 threads at 2Ghz.

Also, i7 isnt faster than Core2 because of HT. i7 is faster than Core2 because its architecturally superior.

 

Hopefully, a program like RMClock will allow us to manually choose what mode the CPU is in. That would be the best answer. As of right now, I'm not impressed either.

 

My bad didnt mention it before, I had the Q6600 upto 3GHz, and 920 around 3GHz as well most of the time. And I use my desktop mostly for encoding, so yea, I did see a huge increase in performance when I upgraded to the i7.

 

Will be interesting to see how the numbers pan out, I also prefer battery life over increased power and I would really like to see hyper threaded dualcores at higher clocks and lower TDPs.

 

What I'd like to say is this:

I see the inherent joy of "turbo mode" in the current P and T 9xxx generation of Core 2 Duo mobile CPUs (my P9700 will singlecore at 2.93GHz, and dual-core at 2.8GHz, while only consuming 25w TDP). As it's more or less established that the upper non-OC limit of CPUs seems to be around 4GHz, and more cores = more lanes for "traffic" to go through, so it is possible, to some extent to substitute raw speed for cores, although 2 cores at 2ghz will not equal a 4ghz singlecore, assuming all other aspects (cache, fsb, etc.) are equal. As far as the Core 2 quad vs. i7 debates, the reason the i7 demolishes the core 2 quad is because while both have 4 cores, the i7 has 8 threads. (Remember the P4 HT was one core with 2 threads.) These threads are "logical cores", which means that they act like real cores, but there's only 4 that are real. As far as the current trend by Intel and others for substituting raw speed for multiple cores, I'm happy that i7-style technology is getting to laptops, however I am concerned that there is so much compromise towards "look how many cores" vs. actual speed.

At least for the time being, dual-core seems to be the sweet spot for most "everyday programs", with quad-core and beyond reserved for heavy-duty art and science and math programs, and games that are poorly ported (GTA4, anyone?) Now upcoming OS's such as windows 7, and many future programs will be quad optimized and then some, but at this stage, how many programs actually can make legitimate use of more than 2 cores for the everyday user? And laptop core i7 has 4 across the board, with varying levels of MHz as core #'s used drops. So 4 cores may be 2ghz, 2 cores is 3ghz, and 1 is 4ghz (just as examples).... that one core at 4ghz may be a VERY fast core, but the CPU bandwidth is less than even the 2 at 3GHz (IIRC, the multiplication factor is somewhere between 1.7 and 1.8... that is 2 cores at 3ghz = 1 core at like 5.5ghz). At least with the core 2 duos of the current generation, the discrepancy in CPU from 2 core to 1 core is not so great... though that can be construed as a good or a bad thing, depending on your point of view.

What concerns me more is the fact that this i7 uses much more wattage than even the gen 1 mobile core 2 quads. (45w for c2q vs. up to 55w for i7-mobile.) Those of us with high-end cpus in our 15" and 17" gaming machines can attest to the fact that our respective batteries range in size from big to "small-town power-plant", and the battery life is shorter than say, a mainstream studio xps with the same cpu. Now I will concede that part of this is attributed to the SLi setups and GTX cards (both in single and dual configurations), however every watt consumed and resulting in an increase of total watt usage, results in a reduction of battery life. With 55w being demanded from the CPU, what will have to happen is either lower-performance GPUs will be needed, more energy-efficient (and probably costly) components will need to be developed, and SSDs will have to become more in the forefront to compensate for the higher TDPs, or Core i7 Mobile will have to be further optimized to run at lower TDPs (even the extreme) before it becomes production-ready. Otherwise, other than for the high-performance users... why leave a multi-core processing system that has proven to work, is fairly efficient, and is fairly battery efficient for its performance level? For the average user, and even most gamers... some battery life preservation is a good thing.

I'm not intending to offend anyone, and if someone has more information on i7-m than what is currently available, please feel free to show me my faults (I'd love to know where my fallacies are)... As much as I want to believe that i7-m will be revolutionary, there are too many "whatifs" for me to really say that it will completely leave all core 2s in the dust. If you have to cut cores to get the same MHz speed as a core 2, you can only OC a single or dual core so far in "turbo core" before you hit the "cpu's limit" and that's that. Also, will the new "more efficient components" be able to cancel out the increased TDP of the new cpus... there's just so many questions intel has to answer and we have to piece together before we can completely write off core 2 as "archaic technology" for high-end users. Hopefully, in the ensuing weeks and months we'll get more concrete answers and less speculation in this topic... I'd love to see just how awesome the new i7-m is... I just want to have cautious optimism until I see the finished product (or at least some ES reviews)

Jason

 

While I agree that i7 is better now just because of HT, the thing is, we know how i7 performs against Core 2 from desktop CPU reviews. Extrapolating that data will show how the performance difference for notebooks won't be all that. I'm not saying the i7 will be worse, but it won't be much better than the current Core 2's.

 

wow... so is that going to be in the $3000+ laptops which are nearly 20+ lb desktop replacements with battery lives of up to 20 whopping minutes with a huge 12 cell battery???
(exxagerated in everyway possible)

 

It's not going to be bad at all. I assume a +3 grade speed boost at quad core(+400MHz), which would result a 2.4GHz Turbo operation for the 920XM, and should be faster than the 2.53GHz, QX9300. All other operations it'll be faster since the dual core and single core mode will be higher.

 

Yep, it'll require one hell of a cooling system, which will add significant weight to any mobile i7 system.

Though TDP is proportional to power consumption, they are not the same. The TDP of a CPU describes the thermal dissipation required by a given cooling solution. This is a distinction that I think is worth mentioning, though your observation is nevertheless correct.

How will AMD respond to this, I wonder? Anybody know?

 

Even without SMT, i7 > Core2. i7 is architecturally superior to Core2, SMT just makes the difference larger. For example, check out the Super PI benchmark here. Notice how the i7s are near the top, then after the i7-920 comes the E8600, which means either A) Super PI is singlethreaded, and gains advantages from pure clock speed and efficiency, or B) Super PI can use up to two cores but still gains an advantage from clocks and efficiency. Either way, its still i7 > Core2 even without SMT, as shown there. Here is another benchmark ( the second one ) that is single threaded, and shows that overall, i7 > Core2. And since it is single threaded, this can not take advantage of SMT.

I dont believe their is a multiplication factor to make several cores into one since different programs will act different, and currently, processors do not scale linearly with cores; think of it as SLI. Some games give 180% performance, others give none.

First, if youre not looking for higher end performance, why even bother getting a quad core CPU? I mean, really, most people will do fine with the current 25W Dual cores currently on the market. Also note two things, these quads are temporary, and in 6-8months after their release, they will be replace with more power efficient and faster processors, since that is how technology is. That is how its been. Also, like I said, theres no point in buying a high end CPU, if youre not really going to be using it. Its just like someone who gets a 9800 series GPU in their laptop. When they get it, they should already know not to expect good batterylife. Or, if youre not even going to use the power said GPU provides, then there is no point in getting it in the first place.

As for the high TDP, Im not sure what you guys were really expecting. I mean, think for a second. These processors are what, the mobile version of nehalem right? Well guess what, the desktop i7 series CPUs all have a TDP of 130W as compared to the other quads with a TDP of 95W ( it should also be mentioned that the some of the Core2 quads have received reduced TDPs; from 95W to 65W ).That right there, shouldve hinted that these CPUs will follow a similar trend. I mean, think about it, its basically the Q9000 ( a 35W CPU ) with an IMC on it. Also, IIRC, the Q9000 is the only mobile quad with a TDP 35W. The Q9100 has a TDP of 45W, and so do CPUs higher than it.

Im in the same boat as you, Im simply waiting for these mobile i7s as well. But I personally do not think Intel would release these CPUs if they couldnt beat their predecessors. But hey who knows, I could very well be wrong, I doubt it though.

@ Gogeta - I take it that would be due to clockspeed right? For example, would your stance be different it they were 2Ghz, 2.26Ghz, and 2.53Ghz?

 

With a groan, most likely.

 

Yes, I agree that i7 is superior clock for clock but the lower clock speeds negate a lot of the benefit.

Btw the Q9000 is still 45W:

http://ark.intel.com/Product.aspx?id=40480&processor=Q9000&spec-codes=SLGEJ

And SuperPI *is* singlethreaded.

 

Yes, that or a white flag!

 

I think they have already had the white flag in mobile platform..

(just my opinion, not trolling)

--

 

the price on the i7 are high... common AMD enought screwing around get working on a mobile processor, the PII isn't to bad at all, now get that in our laptops!

 

And throw it in people's faces?! ROFL. j/k, I just want a slice to eat.

 

Well, from major website reviews on the desktop i7s, the clock for clock difference between i7 and c2 is about 12%.

If the new notebook parts offer something more than that, I'll be happy. I'm willing to bet that performance at full load will be slightly better than current c2q chips using the same power and on light loads the processors will have better tools to scale back power consumption.

Also, since hyper threading is back, I wouldn't mind grabbing a dual core i5 (if that's what they call it).

One more thing: I don't like pie. I'll buy ice cream instead.

 

one thing that i would like to add though is the speculation that nvidia will NOT be able to produce gpu's for the mobile i7 platform. i would rather have a "lesser" processor and a dedicated gpu than a "better" processor running integrated graphics. maybe i'm wrong but that's just what i heard.

 

http://forum.notebookreview.com/showpost.php?p=5086073&postcount=36

From the review thread, check out the roadmap and TDP map.

 

I have cleaned up this thread and dealt with the appropriate members. Please stay on-topic from this point out, thanks.

 

I would gladly trade out my QX9300 for the new 920XM. I'm confident Intel will surprise with the new CPUs, even if their lowly clockspeeds indicate otherwise.

 

To bad it's going to require a whole new laptop, or I would say the same thing

 

Yes, yes I know....

 

SuperPi is single-cored. I do recall that when I ran a P4 and a P4 HT of the same everything (other than HT), the HT one ran a bit faster. Perhaps it's limited to one physical core, and all the threads it can handle. There is a "benchmarking thread" on NBR somewhere that shows superpi is single-core oriented... whereas wPrime is multi-thread / core oriented. Also, the i7-920 has a turbo mode where it can overclock ONE core, but both logical threads within that core can be overclocked... effectively allowing more "bandwidth" within that core. Thus, the i7 has the advantage there.

Yes, but on average, multi-cores tend to be a good bit faster than their single-cored brethren of the same clockspeed, as there are simply more cores to handle the load. It depends on the application, its optimizations, etc. Since you brought up SLi and multiplication factor, one notable example is WoW. I have had several friends of mine state that even with SLi on WoW and a quadcore, that only 2 of the cores are used, and it only really employs one card. The rest is essentially wasted. Other games, such as GTA4 NEED as much video card and cpu power as they can get their mitts on, and thus the multicore configuration is a must. (I know, it's a horrid port, but since it's brought up so often, I'm going to run with it.) The only reason I brought up 180% was because it seems to be the general average people provide with most applications. It wasn't intended to be an end-all, arbitrary amount, but something that is workable mentally, and a reasonable up=boost of the extra core at that speed vs. 1 core.

Precisely. No argument here.

What I'm concerned with was the idea that as far as laptops go, even the highest current end MOBILE cpu only draws 45w at its peak. The new core i7-ms draw up to 55w at their peak. Now while I will admit that going from 130w in the desktops to 55w in the mobile variants is a commendable down-grade, what I want to see is this: you can only stuff so many cells and so much watt-hours into a battery, and this applies even to large 17 and 18 inch notebooks. There's only a finite amount of energy that can be stored, and while the i7-m will down-TDP over time, how many early adapters are going to cry when they see 12 months down the line that they can get the same performance with double the battery life and less tdp? As for the Q9000, intel also lists the TDP at 45w, the reason it works with sagers and clevos of the 15" size is because it's a slower clock speed / core, and has less cache. The extra .26ghz / core and more cache makes too much heat for clevo-sager to plop it into the np8662, though in its 17 and higher models, you'll find it.

Northwood to Prescott.. from a cool CPU to a CPU that could heat a small house, for what? an extra 200MHz... in realworld use, it had to downclock to stay stable in a laptop (in it's desktop form.) I had an XPS GEN I with a prescott, I ultimately found its northwood equal, and while it was older, it ran MUCH MUCH cooler.

Jason

 

I'd wait for the next die-shrink on these. Also did Intel/Nvidia settle their differences on chipsets for the i7? I wonder what Apple is going to do as their currently married to nvidia's chipsets right now.

 

Nothing is different with Core i7M. Gosh, you guys do realize the CPU TDP number will now count towards CPU+Northbridge right??!? Battery life on these 55W Core i7M's will be same or better as the 45W quad cores and the 45W Core i7M's will be same or better than the 35W Core 2 Duos.

These are the same things the ARM supporters make against the Atom. They say the ARM has a 2mm2 die and uses 100mW of power blah... blah... but those cores don't even include FPU and caches unlike the Atom which includes both.

The new power management features of the Core i7 like the PCU will allow not only better power management at the same TDP, but integration of Northbridge will allow fine grained management even for that chip.

Core i7M 55W+Ibexpeak-M

=

Core 2 Quad 45W+P55+ICH9M