2920XM beats the 980X!!

Man oh man ... I would really like to see a M15X-R2 with this baby in it.

Mobile Gaming: Can Core i7-2920XM Beat Desktop Core i7-980X? : So, You Thought Notebooks Were Weak?

 

... is this a joke

 

fortunately I don't think so a mobile CPU with 1% more power at gaming than the king of desktop CPU's

 

Only at gaming because of optimisation. If we look at all the raw encoding benchmarks and stuff the 980x still pulls ahead. That being said, I'm sure the 2920xm is a beast, but does it have unlocked multipliers?

 

@ widezu69: Yes it has... and TS already supports SB afaik (I guess you know why I'm smiling...)

 

Guess the word now is...*a taboo term that is a synonym to the word faeces* now I really feel obsolete.

 

Overkill... unless you feel the absolute need to be the king of the hill for a few minutes this cpu will probably be well above and beyond any games or apps out there.

It is exciting that once again a notebook is just as powerful if not more than standard desktops.

Good post,

StevenX

 

Looking at the benchmarks, as expected, expensive CPUs are a waste of money for games since they're almost all GPU limited. Instead of wasting your money on this, the cost can go into getting a better GPU. Also, remember that despite the cost of the 980X, it uses the last generation Nehalem (die shrunk to Westmere) architecture and I'm sure the new high end desktop SB will smoke the 2920XM. Notebooks will never come close to the equivalent current generation desktop performance mostly due to superior cooling of desktops, which leads to much better clock speeds and can be pushed even further with overclocking. If they had thrown in the i7-2600K, you'd see how much of a difference same architecture vs. clock speed makes.

 

in the case of sandy bridge because the laptop auto set it's turbo according to heat/consumsion some laptop with a real coling could hand the desktop superiority thier

 

But here's the thing: there are reviews on the web of both the 2600K and the 2820QM and the latter does in fact come close. Of course, they will never equal desktop performance, but unlike Clarksfield vs. Nehalem, they're not that far off.

 

I just wish they'd done the smart thing and compared with the same GPU, but it doesn't look as sensational.
When you extrapolate the game data, you realise the 2920 from the 940XM makes perhaps 5% difference at most, based on the difference on a system with 940xm and 980x and the 980x and 2920xm, and using the 980x as the comparison point, with the 940xm often being about 1.5-2% behind in games, but the 2920 being 1-2% faster than the 980x.

For encoding, the new CPU is actually a nice step up (oh well, can't have everything, especially not given how much I've spent here...which really shouldnt have got me the machine I have), but unsurprisingly the desktop walks all over it.

So what did I really learn? Theres only a few percent of gpu bottleneck in there, and quite frankly, if I upgrade my cards but not CPU, Im still going to get 95+% of the performance, at least for gaming, and lets be fair compared to previous entries, the 940XM is still a pretty damned fast laptop processor.

 

With a desktop cpu you just plug it into the wall while a notebook requires batteries...........

 

Let's see:

1. Sandy bridge is significantly more powerful clock for clock than Nehalem and clock speeds are very close.

2. Games don't use multiple cores very effectively.

Also desktop GPU's are going to slaughter laptop GPU's anyway........ If you have a 980X on the desktop, you are not going to pair it with something as crappy (relatively) as the 6970M.

No way. They tested that processor at its full turbo speed. There is no way it would have performed the way it did compared to the 980X otherwise.

 

This is a rigged benchmark. If you look at the Sandra numbers where all cores/threads are utilized 100%, the 980x pulls ahead by a large margin. Just because a game doesn't utilize 12 threads doesn't suddenly make a weaker CPU "more powerful." I bet a Core 2 Duo could probably reach similar performance levels in many of the games tested, especially at higher settings where the GPU is taxed more heavily.

Compare the two in Handbrake or Vegas and then try to tell me the 2920xm still comes out ahead.

 

And notice the title of the article...2920XM beats 980X in GAMING. I'm sure the 980X would slaughter a notebook extreme processor in every other benchmark, Gulftowns are nothing to sneeze at (I have one too )

 

Also,

Isn't the 980X a 6-core processor, whereas the 2920XM is a 4-core processor?

 

Exactly, they aren't comparing things properly. Some benchmarks I expect will be running on only four cores because that is its limit. The Gulftown is still faster but the key phrase is 'when all cores are utilised'. I theorise that its slowness is the uneven distribution of 8 threads running on 6 cores that makes it a tad slower in some cases.

 

SB architecture is more efficient than Nehalem, which accounts for the entire difference. Games do not respond well to more cores at all. By 4 cores, you are facing greatly diminished returns. And since very few people have more than 4 cores anyway, there is really little to no incentive for games to take advantage of more even if they could be easily made to do so, which I doubt.

 

I guess the multi-million dollar task is to create an SDK for developers to create games that utilize the maximum number of threads without question and take the maximum clock speed per core/thread and distribute the processes evenly successfully. So if you have obscure ones like tri-cores and hex-cores they could still scale well.

 

Ironically it's the opposite I think. It's been acknowledged Sandy Bridge has an average 10% advantage over Nehalem, which isn't huge. Possibly AVX will bring some more changes. On the other hand stock clocks are quite a bit higher at a comparable price (i5-760 vs i5-2500 = +500mhz = +18%, i7-720qm vs i7-2720qm : +600mhz = +37,5% ! & much more room for turbo, even with all 4 active cores. Just about what you'd expect from 45nm -> 32nm). If Westmere 32nm quad-cores had been released Sandy Bridge would have been deemed considerably less impressive I think especially on the mobile side because its perceived improvements would have relied only on a small IPC increase, better power management & the IGP (turbo 2.0 too).

 

I don't think its quite so easy to spread the love around.

Your conclusion doesn't make sense. The 980X was running just a little faster than the 2920XM, within about 6%, and the 2920XM scored anywhere between 1-5% better than the 980X, so your 10% advantage seems to fully explain the difference.

Yeah, this is probably right. Only releasing 45nm Nehalem quad cores for consumers definitely made it easier on Intel for releasing SB.

 

Quality compilers, those that can generate efficient multi-threaded code, have always been the bug-a-boo of cpu makers. Part of it also depends on the cpu thread dispatcher in the OS core/kernel.

The best of the lot is probably the native C compiler for AIX on Power. In this regard x86 compilers are generally thought to be well behind what IBM, SGI/Cray, HP/Compaq/DEC, and Oracle/Sun have available.

It's also why Itanium held on as long as it did. HP and Intel had great multi-threaded compile capability for that CPU.

 

I guess your fears that the 920XM/940XM are much faster when overclocked were put to rest (or perhaps confirmed, thus making you happier with your current machine)?

 

I don't understand why there were game benchmarks on that review since most of the games on that review are GPU dependent.

 

I guess to prove that the 980X and 2920XM are equally large wastes of money, for gamers.

@ 1080p, the 2630QM preforms on level with the 940XM, at less than 1/3 of the price of the 2920XM. But we'll still see the XM chips, in sigs all over NBR....

 

I forgot CPUs were the most important hardware component when gaming..

That is one of the most useless articles I have spent time reading.

 

980x is an encoding king not a gaming king... also once you start to overclock the 980x will just exponentially take off.

 

both 2920x and 980x are far cry........

 

Aye, if you read between the lines, you'd realise that *for gaming*, theres only a few percent between the old and new processors. Hardly a massive bottleneck.

 

No kidding, the whole purpose of that processor is to push the limits. It is like only using the first three gears in Ferrari, oh so silly.

 

What does that mean? Are you referring to turbo boost? That doesn't really matter.

 

how arent both at their limits?

How do you know their limits?

What do you know of the performance of either chip?

 

Do you know what the 980x is? It is an extreme series desktop chip, that thing can go up to 4.0Ghz and beyond quite easily, point being you wouldn't pay the premium to acquire it if you were not going to overclock it, that is the niche the chip fills. Had they even slightly overclocked it there would have been a massacre.

 

I know that... I have friend that has one, it reached 5.2ghz with liquid cooling.

I dont pay premium for the extreme processors for 2 reasons:

1) I dont have desktops anymre for quite sometime.

2) I dont buy gaming machines, so the cooling wont handle the tdp well.

I was asking for fulcrum if he does know anything that he says, or he just pulls those things out of...

 

Point in case, they are comparing apples to oranges.

 

How is that different from the 2920XM, another extreme edition processor?

 

The 980x has a much higher overclocking ceiling, my point is had both processors been used to their full advantage the 980x would have pulled ahead. Unfortunately in a laptop the 980x most likely won't be used to it's full potential.