Is the GTX 965M Memory bus bandwidth sufficient?

From what I've read, the problem witht the early 2015 models was the power brick not being sufficient which a 240w brick fixed, and messed up bios which apparently the A06 release also fixed, plus the processors on the early models were haswell and broadwell, the new ones are skylake so idk.

 

the 3720QM i had has the same advertised TDP as the 6700HQ @45w

 

Please read the mobile i7 CPU guide in my signature for an explanation as to why the 3720QM is INFINITELY better than the 6700HQ. =D.

 

i appreciate the guide, and i know everything you've mentioned in it already.
But so far everybody is throwing a fit at those BGA chips, meanwhile all the benchmarks i've seen show that the 6700HQ is superior (though only slightly) to the 3720QM despite being soldered.

Is there REALLY any down particular downside other than "no more upgradeability"??, i know that the 4 core clock of the 6700HQ is only 3.1ghz as opposed to the 3720QM's 3.3, reportedly though the per ghz performance is improved on the skylake.

I'm not being biased to try comfort myself that what i'm getting is bad, i'm trying to rely on hard and factual information as much as i can and not on personal opinions about manufacturer policies and practices.

 

from a practical perspective, though - if it throttles anyway, wouldn't you expect the skylake variant to keep it's boost longer? And then basically be able to stay at a higher clock on average, after you reach "working temp"? Or am I just plain wrong on this one?

 

The 3720QM can be clocked at 3.7GHz however, and will hold its TDP indefinitely in a laptop that doesn't artificially limit the user. The 6700HQ is TDP limited and from what I have seen so far, is stuck at 3.1GHz.

 

on my 3720qm and during full load, all the cores hovered at 3.3ghz no problems at all, and the temps were around 91~97 depending on the core.

idk about the 6700HQ but it's supposedly more efficient and should net less heat and the same performance.

 

The 3720QM i had clocked at 3.6ghz max for 1 core, 3.5ghz for 2 cores, and 3.3ghz for 4 cores, i never OC'ed it and i will not OC the 6700HQ either.

and as i outlined in my previous comment "idk about the 6700HQ but it's supposedly more efficient and should net less heat and the same performance."

 

well up to you. If it functions fine for you, then ok. If you had an alienware before, I'm sure you could have gotten the 3720QM to run cooler somehow, but that's in the past.

The statement I made before still stands though. If you want the best CPU performance, the desktop CPU using models from Clevo is the way to go. If you're fine with what you have and have made no attempts to get the most out of it, then that's fine too. It's not fine for me, but I'm not you.

 

don't get me wrong, i'd rather get the best out there, but what i have here is Dell offering to compensate me for a laptop that uses components that are at least 3 years old (the 675m is actually a rebranded 580m so 4 years old there).

and while the performance may look the same at first glance on many components, the architectures is the latest there is with regards to the cpu and gpu, and even if the performance is only slightly better than what i had, it still is a better deal, all i really want is for them not to have throttling issues otherwise this whole thing is pointless.

 

[IMG width=100] https://no.wikipedia.org/wiki/Skrik...h_-_The_Scream_-_Google_Art_Project.jpg[/IMG]

Seriously, though - I'm no expert(and not usually given to encourage people to follow Intel's upgrade paths), but from what I've read, Skylake does offer some gains on your typical hyperthreading friendly scenarios, and should have at least the same or better power-control scheme that haswell had (which you should notice on battery very fast - largely software and updated embedded firmware, apparently, but still..). So if you don't have a modded or overclocked ivy bridge qm, and don't have the cooling or psu needed to keep it on higher clocks - a skylake chip seems like a good way to lower the watt-drain slightly in practice, without losing performance. New mainboard config hopefully without the mistakes from panther point as well doesn't seem like a bad idea.

My guess is that the desktop variants have practically no gains over the previous generation in the typical tests. But that if you compare the scenarios where you normally run into soft throttling (which you do very quickly when reaching normal temps - that the boost will trigger, but not last very long), this should be where you get something useful. Which should be on these laptops in /practical/ situations, after a few hours run-time, that you typically don't see in a laptop tests when running 3dmark on a cold start, so to speak.

Btw, really nice thread. Learned a lot.

 

i'm glad someone got something positive out of my worries xD