I've got two laptops, one with an Intel 6700HQ and the newer one with a 7700HQ.
I've been benchmarking the integrated graphics by disabling discrete graphics through the Nvidia control panel, basically saying to use internal graphics for all tasks. It's obvious internal is being used as the results from the benchmarks are much slower than the Nvidia cards.
In every test the Intel HD 630 graphics on the 7700HQ are slower than the 530 on the 6700HQ. Both have the latest drivers from Intel installed. Windows 10, all latest updates installed. I have 16GB DDR4 RAM @ 2400MHz on both laptops, both display adapters show as 128mb dedicated video memory, though it doesn't appear this can be changed.
Really not sure what is going on. In most other general CPU benchmarks the 7700HQ is slightly ahead as I expected, but for the graphics tests it seems behind when everything I read online shows that the 630 is a slight (~10%) improvement, but I'm not seeing that?
In the power options high performance mode is enabled, the specific option for the integrated GPU is "maximum performance", other processor specific options are on 100% when plugged in, and I am testing while plugged into power.
Is that something I have any ability to change? I don't think I saw anything when looking through there earlier. If not, then I guess I'm stuck with this?
Basically I was going to do a write up comparing the two, and it seems a bit silly if the 630 doesn't perform any better, unless of course that result is consistent but I'm guessing it's not, especially if it's specific to what ever the vendor defined.
Sometimes you can alter it in the UEFI, more commonly on desktops but a few laptops have the options available. Run GPU-Z in the background and see what the IGP turbos to, compare to Intel's spec. If it's lower, you've got your answer.
@Fouquin I've looked at GPU-Z, interestingly, on the 530 the GPU core clock is sitting on 350MHz when doing nothing, it changes to 200Hz when I start running Heaven benchmark. The memory clock stays on 1066.7Mhz, and the memory usage increases from 79MB to ~1200MB.
With the 630 though doing the same thing, the GPU core clock stays on 4250MHz with a consistent 1200MHz memory clock. The memory usage increases from 71MB to ~1200MB here as well.
I'm not sure I understand this, the Intel spec for both says the base frequency is 350MHz, so why is the 630 so much higher?
I've got screenshots of GPU-Z showing this if it helps.
@ipushpeople Drivers for the 630 are fully up to date, version number in the above screenshot, though it was extremely hard to find these. The official Intel site didn't even have them, I found people complaining about it and it does seem strange, people were resorting to getting generic drivers from providers like Dell/HP.. Mine ended up coming through Windows update, however I also tried one of the generic drivers from HP but it made absolutely no difference.
GPU-Z is not reading the HD 630's clock values correctly at all. No consumer chip is going to be running at 12.75GHz in any capacity.
Check it again after a restart and look at the sensors tab. When you put a load on it the graph will show clock fluctuations. (Supposing it reads the clockrates properly.)
I've tried shutting down and powering up twice, same results. Here's the sensors tab after running a benchmark. I had it logging out to a file last time and the clock speed was the same the entire time which just seems completely wrong :/
Alright so that's a dead end. I'll talk to w1zzard (the author) and let him know about this bug. What laptop are you using to test with? Is there any other documentation online that might give clues to what the clockrates are, whether the chips TDP has been altered in any way?
Have you monitored the clocks and temperatures of both chips? It might be that the 630 is downclocking to cope with overheating or something. I saw somewhere that 630 is supposed to be rather capable, so I don't imagine that it should be performing worse.
The one thing that immediately apparent to me from the screenshots is that the HD630 is on a 64bit wide memory bus, while the HD530 is on a 128bit bus. There are 64bit versions of that GPU for notebooks. I don't know why but there are and they're just savagely bottlenecked. I never understand intels logic behind making such rubbish decisions.
As for the Core Clocks, GPU-z is clearly not yet updated to include the new CPU's and GPU's so it's reading the wrong sensor address (that of the CPU clock).
Aside from that I suspect that the HD630 is likely running on a 300Mhz clock actually (not 12750 XD ) your pixel fillrate and texture fillrate is determined by the memory clock, core clock plus bus width so it's completely wrong for the HD630.
You can try use something like AIDA64 to see if it says the correct values.
