Hello everyone, new user (but been watching the crew since the start, and a big fan of the technical stuff since long before the split).
So I just ordered a new laptop last night, Inspiron 15 Gaming. The new one with the Core i7 7700-HQ processor (don't worry, I sprung for the 3840x2160 IPS screen, not the dreadful TN one). It has a 512GB SSD, 15.6" screen, 16GB of RAM, GTX 1050ti GPU, backlit KB. It cost $1299.
It's replacing an HP DV7T Quad Edition that had a Core i7 3610QM processor. That laptop has a GeForce 650M, 1TB HDD, 8GB RAM, 17.3" 1080P LCD, backlit KB. I bought that HP laptop in June of 2012 for $1030.
So my question is - when I look up CPU benchmarks on these two CPU's, the CPU Mark score on the 7700HQ is 8896. The CPU Mark score on the 3610QM is 7460. That's almost 85% of the score of a brand new 7th gen Core i7 chip from a chip that is 4.5 years old. They are both 45W TDP chips with 4 physical cores and 2 logical cores per physical core. The 7700HQ runs natively at 2.8Ghz vs 2.3 on the 3610QM and boosts to 3.8Ghz vs 3.3GHz on the 3610QM.
What am I missing here? Or why isn't a 4.5 year old chip substantially outperformed by the brand new one? Is CPU Mark just an unreliable metric for comparison? I don't imagine my 3610QM was a beast back in the day considering it was actually in a cheaper laptop. I always seek out good bang-for-buck in terms of performance but still, it was 4.5 years ago. So I am thinking I must just not be understanding the true differences in the chips and those differences are not reflected in things the CPU Mark score.
I would be very appreciative if someone who is more knowledgeable were willing to drop some knowledge on the subject!
I'd say two major factors contribute to what you're seeing here.
While Intel makes performance improvements on every generation on desktop CPU's, it is of course far more important to do so for laptop CPU's. So in terms of computer science, such improvements are impressive even if it isn't much as you say.
Intel has had very little competition in the CPU space (hopefully that is changing with Ryzen) and whatever monopoly they held on the desktop market is even more drastic with the mobile laptop space, where AMD is almost non-existant (On the performance/gaming level). If competition is weak, there is less reason in Intel's eyes to achieve heavy innovation.
Hope that makes sense, and this is just my two cents, nothing official/I don't have sources for any of this.
For the last six years or so, Intel have been focusing more on improving power consumption and integrated graphics, because they haven't had anyone pushing them to dramatically increase IPC. Generally there's been a 5-10% performance improvement per generation, so I'm surprised to see the two chips benchmark so closely. This either comes from clock speed increases, or architecture changes/optimisations.
Yes, both chips have a 45W TDP, but this doesn't translate into the same power consumption. The 7700HQ is a far more efficient chip, which will output less heat with the same amount of power.
As far as benchmarks go, Cinebench probably gives the best idea of relative performance.
Cpu mark isn't the best metric for comparing raw cpu performance. Personally, I prefer cinebench R15, which really does show the difference. A 3610QM does about 560 in cinebench R15, and a 7700HQ does about 735. That means the 7700HQ is 1.3x faster than the 3610QM in this test, or to put it a different way, the 3610QM is 76.2% the speed of the 7700HQ in cinebench r15.
That right there is the problem. It's a synthetic benchmark that isn't particularly reliable. Furthermore, if you don't know the testing methodology, it's difficult to compare numbers. This is especially true in laptops where you have limited thermal headroom. In laptops, it isn't uncommon for a slower CPU to outperform an faster one (e.g. an Core i5 vs Core i7) due to thermal throttling. So thermals alone can cause massive discrepancies if not accounted for. Other factors include OS, other active processes, power management, etc. So ultimately, a PassMark score alone isn't going to tell you much of anything. But even so, the performance is more than you think - that's nearly
Notebookcheck can give you a better idea. Based on Cinebench, H.264 encoding, and 3DMark, the 7700HQ is closer to 30% faster than the 3610QM, which is pretty substantial.
Intel cpus haven't improved much since sandy bridge. Sandy was the last big improvement with a whopping 30% over Bloomfield. It was about 7-10% to Ivy, then another 7-10% to Haswell, 3% to skylake and 0% to kaby lake.