Author Topic: ParkControl v2.1.6 (Read 22074 times)

Jeremy Collake · « **on:** August 15, 2014, 12:46:09 PM »

BenYeeHua · « **Reply #1 on:** August 15, 2014, 02:45:49 PM »

It is funny to see 3.52 Ghz of 3.53 Ghz, even I know it is caused by the Motherboard Spread Spectrum function, or other thing.

I wonder did someone will like, what!? It parked 0.01 Ghz of my core!?

edkiefer · « **Reply #2 on:** August 15, 2014, 04:45:55 PM »

yes, software can be funny , playing BF4 today HWinfo64 says my cores went to 4341 and 4318mhz . not sure where it gets those number but all other HW apps show true 4300mhz (43x *100mhz bus) .

Jeremy Collake · « **Reply #3 on:** August 15, 2014, 07:06:17 PM »

Spread spectrum is probably the cause.

Also note that technologies like AMD and Intel's TurboBoost and TurboCore, respectively, can cause individual cores to temporarily exceed the maximum reported CPU frequency.

edkiefer · « **Reply #4 on:** August 16, 2014, 09:16:06 AM »

since I am OC, I have Spread spectrum disabled now . I used to have it enabled with small OC (4.0ghz) . What I notice is CPU-Z with it enabled show bus jumping between 99.9mhz and 100mhz , with it disabled bus reports 99.98mhz .
I have all cores running same freq , 43x , 43x ,43x ,43x

Edit: just tried HWinfo64 again and its not reading high on mhz , saying 4300 so it might of been a conflict running more than one HW monitor or those Asus sensors , either way back on topic .

edkiefer · « **Reply #5 on:** August 16, 2014, 11:56:28 AM »

I just check latest park control . the max CPU freq is only 3.32 which is still way below 3.4 (stock non turbo modes) . This i5 3570K stock turbo is 38x,38x,37x, 3.6x .
Idle say 1.56 when it should be 1600 (1.6ghz )

Jeremy Collake · « **Reply #6 on:** August 16, 2014, 12:37:18 PM »

The issue is that I divide the OS reported Mhz by 1024, which causes the discrepancy. Apparently it should be divided by 1000.

I had just assumed that megahertz/gigahertz follows the same rules as megabytes/gigabytes, being multiples of 1024. Oops.

Jeremy Collake · « **Reply #7 on:** August 16, 2014, 12:41:56 PM »

I uploaded ParkControl v2.1.8 with this simple adjustment (divide by 1000 instead of 1024).

Give 'er a try. It should show the correct CPU frequencies.

edkiefer · « **Reply #8 on:** August 16, 2014, 01:38:25 PM »

that works, 3.4ghz, no turbo modes though .

thats a 32bit exe so can't replace parkcontrol.exe in 64bit install .
will have to wait on next 64bit update ?

Jeremy Collake · « **Reply #9 on:** August 16, 2014, 01:49:29 PM »

I only list the 32-bit download these days, since it works on both platforms.

Since there are no performance concerns with this utility, it is unnecessary to distribute a 64-bit build.

However, the 64-bit build is also available for download. https://bitsum.com/files/parkcontroldist64.exe

For Lasso, the new ParkControl will come with the next update.

edkiefer · « **Reply #10 on:** August 16, 2014, 02:01:57 PM »

yes, what i meant I didn't think I could place that exe in the 64bit PL path where other older version is, thats all .

I can't even use parkcontrol with i5 so no biggy . I'll just wait for next update .

BenYeeHua · « **Reply #11 on:** August 16, 2014, 11:47:25 PM »

For the difference of 32-bit and 64-bit, I guess it can be just the DLL issues, which somehow 64-bit software loaded 32-bit dll because of the User has damaged system files, and it can't found 64-bit dll.

And ya, I wonder how is the Windows 8+ Task Manager calculate the CPU frequency, it look like a average of the frequency that it get from x second, which showing like 1.48, then 1.50 Ghz etc.

edkiefer · « **Reply #12 on:** August 17, 2014, 06:27:43 AM »

Its reading what windows uses for CPU max amount and then scales it, I guess .
Windows think I have 3.4 , doesn't matter if you OC, it stays the same .

CPU-Z reads real physical speed

Jeremy Collake · « **Reply #13 on:** August 17, 2014, 01:53:32 PM »

It's funny how complicated this discussion can become given per-core dynamic frequency scaling often beyond the reported max available CPU frequency.

What I typically do to calculate a single current CPU frequency is average the frequency of all cores at the time the sample is taken.

edkiefer · « **Reply #14 on:** August 17, 2014, 02:02:30 PM »

I guess the question is why does windows report only base clock and none of turbo clocks .
example i5-3570k max base clock 3.4 but the lowest turbo clock (all cores running full ) is 3.6 and highest single core is 3.8

Jeremy Collake · « **Reply #15 on:** August 17, 2014, 02:11:53 PM »

Probably for design simplicity. Plus, it makes some sense to show the 'base frequency', rather than the boosted frequencies.

Of course, the real answer is probably less about rational decision making and more about the Task Manager and other OS components having been developed prior to TurboBoost and TurboCore, then not updated there-after for the above rationale.

edkiefer · « **Reply #16 on:** August 17, 2014, 02:19:07 PM »

You know part of problem is Intel , if you look at a 4770 , Intels market it as 3.5ghz , even though turbo goes to 3.9 .
So it could be a simple I bought x CPU and windows should match that, for the non techy .

Jeremy Collake · « **Reply #17 on:** August 17, 2014, 02:32:13 PM »

That's very true.

BenYeeHua · « **Reply #18 on:** August 18, 2014, 06:24:31 AM »

So I guess it means no Wintel!

----
Anyways, it is funny that for the Turbo frequency, it is to reduce the money for cooling capacity of laptop(Acer love to use Ultrabook for laptop too), or Ultrabook.
So when you are just normal use like browsing, then it can turbo to max easier, but when something like converting video that using AVX, or gaming which start using the GPU, you will start to see it will use only base frequency which is much lower for Ultra CPU(xxxx-U, not xxxx-M), including the Boost frequency for Nvidia/AMD, because of the bad cooling capacity.

Of cause this will be reducing as Intel and Nvidia are good at reducing the power consumption for next generation, but you can't said that most Laptop will not reducing the cooling capacity too.

edkiefer · « **Reply #19 on:** August 18, 2014, 07:54:19 AM »

Yes, because of laptops reduced TDP as it can only get rid of limited amount of heat once sustained load saturates the TDP limit it must throttle .

So a lot of boost freq used are just that limited time in boost , then back down .

As they move to smaller processes this maybe will be relieved a bit but then they will boost frequency more again

BenYeeHua · « **Reply #20 on:** August 18, 2014, 02:25:42 PM »

And yup, boost has difference TDP as well, so they can also set it higher for Turbo, and you can check for it by using HWiNFo etc to see how long it can be.

Quote

CPU Thermal Design Power (TDP): 35.0 W
CPU IA Cores Thermal Design Current (TDC): 112.0 A
CPU GT Cores Thermal Design Current (TDC): 46.0 A
CPU Power Limits (Max): Power = Unlimited, Time = 64.00 sec
CPU Power Limit 1 (Long Duration): Power = 35.00 W, Time = 32.00 sec [Locked]
CPU Power Limit 2 (Short Duration): Power = 43.75 W, Time = Unlimited [Locked]

This should be set by the BIOS, if I am right, because I saw similar settings on the BIOS hidden settings.

The temp is always the issues here, as if the temp is higher, the current is higher as well.
And you can workaround it by getting a laptop that having dGPU, and not using the iGPU, so you just reduce the current used for CPU, but I don't recommend it as it increase heat and power consumption, unless it(dGPU) has 200% of your iGPU performance.

---
And ya, I still think that, Windows should still showing the Base frequency as Max frequency, as it is the max frequency it should be.
As you know, there are difference Turbo frequency based on how many Core is parked/unused, so if it is not using max frequency when it is 100% load, the user will be feel strange too.

News:

Author Topic: ParkControl v2.1.6 (Read 22074 times)