W230SD 4700mq + 960m problem

Hi all
I recently upgraded my w230sd with 4700mq but now I am having a problem
the processor is getting too hot(It is very hot in her)so I disabled turbo in throttlestop
but the problem is whenever I load the CPU and GPU (like gaming) or even simple benchmarks like throttlestop ts bench+GPU-Z render test the gpu clock start fluctuating between 135mhz and 1200mhz(this is with cpu temp reaching 93 max with turbo off and gpu reaching only 56 max)
I recently repasted both CPU AND GPU with Gelid Extreme and I am using the original 120 watt psu.
thank you for your help

 

Do you have some screenshots of the temperature graphs from some normal gaming showing from idle to throttle temps?

 

Hi Meaker
thanks for the quick reply
as I said I am using throttlestop ts bench+GPU-Z render test so I am observing the temps in real time and the gpu never reaches 60°
and the weird thing is that GPU-Z says perf cap reason is idle
which is clearly not true
Edit:I am using win 10 with the latest update and using every high performance option i can find.

 

This is probably caused by your heatsink and fan. Dissassemble your heatsink and fan, remove all the dust that filled your heatsink and fan. Also make sure to reseat your Heatsink properly to your fan, then tell us how the new temps are.

 

Hi Danishblunt
I recently repasted My CPU and GPU but it is too hot here(ambient temp about 30°)so I disabled CPU turbo
but as I said the GPU temp is very good(never reaches 60° under load according to GPU-Z)so I don't know why the GPU is throttling(It is really fluctuating between 135 and 1200 and not throttling)
GPU-Z says perf cap reason is idle
could it be a power issue?
Edit:I also wanted to add that it never does this when loading the GPU alone and the CPU is running at full 2400mhz even though it is reaching 93° so the only problem is with GPU when loading both CPU and GPU.

 

That's why I want you to open the case, and reseat your cooling. Check if you put enough paste on the CPU etc. Even tho your cooling is absolute pathethic, it should somewhat be able to keep the CPU without going beserk.

Also you can also do some undervolting, the 4700MQ on adaptive voltage do give way to much voltage to the CPU for no apparent reason.

 

I am going to do that which will hopefully decrease the temp of the CPU
but the problem of the GPU still remains as it does not look like a heat issue(It's temp is very good)
My problem now is the GPU fluctuating clock as the CPU is stable even though it is reaching a high temp.

 

Well yeah, this is simple math tho. You have a 120W PSU, a CPU that takes 50W, a GPU that takes 75W.
50 + 75 = 125W

Then the rest of the system such as Screen, RAM, Mainboard etc. also need power. There is your answer.

 

Well this is the real question as I read others who have the same specs don't have a problem with the 120 watt psu
And also after disabling turbo the CPU is only using 23 watts according to throttlestop.
That is why I hope others who have similar specs chime in.
Edit:I also read HTWingNut review and he said that the 120 watt psu handled 4710mq well.

 

23 Watts on stock non turbo @ 2.4ghz, over 90C? I kinda have a hard time believing that. If you did some massive undervolt, yes, but stock, no way. I don't see how a GTX 960M and a 4700MQ on full speed on a 120W PSU would run fine. If possible try a 180W PSU and see if it still throttles, obviously after you fixed the temps.

Maybe you got a friend with an MSI notebook or something, the PSU's are plug and play.

 

Hostility #1... It may not be a tri-fan, six-heat pipe solution, but it keeps the GPU to 75–80°C, and the CPU to around 85–90°C on a 13.3" notebook... That's not bad.

Yes. The CPU really does draw that little power at 2.4 GHz (I had a 4710MQ, was 2.5 GHz maximum non-Turbo) on all four cores. It's called quadratic scaling—power use scales with the square of the voltage, and as frequency increases, the voltage supplied will increase proportionally to keep the CPU stable. Hence a i7-4XXXMQ CPU draws some 60 W running at 3.7 GHz on all four cores (tried and tested with Prime95), while only drawing ~25W at 2.5 GHz. Therein lies the efficiency of notebook CPUs.

Yes way... Perhaps you should purchase a Haswell notebook yourself and try it out.

They run fine. I owned this notebook for three years with a 120 W CPU, and I frequently ran both the CPU and GPU to full throttle playing games and transcoding video. The PSU got nice and toasty (but then again, like OP, where I live is also very toasty), but it never complained.

I did blow the PSU—but only when I downgraded the CPU microcode, attempted to push my sad little 4710MQ CPU to 4.0 GHz, ended up drawing 90 W and poof, it just quit. Obviously my laptop shut down, because the battery wasn't plugged in.

@sharkam52, I suggest replacing your heatsink on your notebook—but before that, remove it, clean off the excess thermal paste on your chips, and reapply something decent, like Gelid GC Extreme, IC Diamond or Thermal Grizzly Kryonaut.

 

Triggered #1, it's still a bad design, the reason why the CPU most likely is heating up is due to the terrible design of the heatsink. It's behaving like the unified alienware heatsinks, where it takes the GPU heat onto the CPU instead of the vaporation point of the heatsink, due to bad thermal transfer of the CPU.

I recently had a 4710MQ in a P170SM notebook which drew around high 30s - low 40s Watts under stresstest on 2.5ghz. So I find it hard to believe that yours did draw low 20s, could be a misreading from HWinfo or whatever you used, which isn't that uncommon. The U series intel CPUs with dualcore run around 15 - 20Watts on at high 2ghz / low 3ghz clock, so it's quite hard for a stock 4710MQ to reach that low with twice the ammount of cores despite being on a 2.5ghz, but that the 4710MQ runs in the high 60s when OCed to 3.6ghz is correct.

I got multiple haswell notebooks with different CPU's (4700mq, 4710mq, 4810mq, 4940MX), I know how they behave. That's why I also use static voltage to keep it low voltage and stable under overclock. You cannot let haswell run on their own multiplier unless you want to run them underclocked.

You do realize that the setup would run the PSU to its limit right which is pretty riksy and can break the PSU and/or notebook as you have already experienced yourself. Depending on the EC and PSU it would behave differently, I had a PSU that got damaged differently than yours, it would work but not provide the wattage it could anymore (degraded from 180W to ~120 or something) which causes my GPU to underperform. It's not that unlikely that the OP when doing tests has simply broken the PSU due to the hardware stressing the PSU and degrading it, as you demonstrated with your post very well, you're stressing that thing to the limit, which is never healthy for a PSU, getting a powerbrick that doesn't run near or at 100% is a must, unless you want to try your luck and see if you can damage your notebook and/or PSU.

 

Teardown, remove the heatsink, dont clean anything, post a photo on here showing the cpu and gpu dies and how the paste is spread.

 

Sounds like the GPU driver isn't switching to dGPU while CPU is active.
First update the Intel GPU driver and if that isn't enough also manually select to run your stress test on dGPU via right click and selecting the NVIDA GPU.

As for the CPU heat issue, just undervolt core and cache etc by like 75mV-.

It is a great system and may have been the last 13" with replaceable CPU ever made...
Haven't upgraded ever since and still using my W230Sx as a daily driver.

I also tripped the AC Adapter once at a 4702MQ 4.5Ghz run, which then killed the battery...

 

Thank you all so much for your input.
Much appreciated
but still the problem is NOT related to heat(unless the GPU sensor is faulty or GPU-Z is not reading temp correctly)
the problem is when using both CPU and GPU (even with simple loads and with CPU turbo disabled) the GPU clock fluctuates between 135 and 1200mz which is the weird thing.
as I said before GPU-Z says that the reason for throttle is idle
@Prema I did select the nvidia GPU and it is using it(it is just GPU-Z render test)the problem is the GPU clock is fluctuating
Thank you so much for your reply!
Edit:I even start the GPU-Z render test before throttlestop ts bench and its okay
the moment I start ts bench the GPU clock starts fluctuating.

 

As I said update the Intel GPU driver or the system won't use the NVIDA GPU for many things...

 

So I just updated the Intel GPU driver and still the same problem
here is a picture of what is happening

And here is the GPU-Z log file

 

WHen running GPU only it boosts normally?

 

Yes.

 

Run something which puts some actual GPU load on and doesn't allow to switch GPUs, like AIDA64 combined load test.
The stock BIOS is set to throttle the CPU and not the GPU unless it reaches its own power or temp limits.

 

I tried the AIDA64 system stability test(is it the same as the combined test?)and chose both CPU and GPU but it does not look like it is using the 960m and only using the intel GPU even though I selected the nvidia GPU as the preferred GPU and it is detecting it as I can bench the 960m in the GPGPU benchmark
Edit:I opened GPU-Z at the same time and it looks like the the 960m is being utilized after all and not fluctuating like it was.Thank you Prema!
Which brings me to my real issue,while playing GTA V after a period of time the frame rate starts to drop very low and the game stutters very bad.
I thought it was a temp or power issue but after playing in windowed mode and monitoring both CPU and GPU it looks like the GPU utilization starts to drop after a while from about 50% to near zero for no apparent reason which causes the frame rate drop and stuttering as you can see near the end of GPU-Z log.
what could be the cause of this?

 

Still the same problem, your driver is dropping back to iGPU and not using the dGPU. Make sure to install the latest compatible Intel GPU HD4600 driver you can find.
Check station-drivers website:



http://www.station-drivers.com/index.php?option=com_remository&Itemid=353&func=select&id=108&lang=en

 

Hi Prema
I installed the latest Intel driver from Intel's site
and it begins doing this after about 20 minutes of running GTA V
the first 20 minutes of playing there is no problem and the game runs fine
Thank you so much for your help!

 

Intel's site never offers the latest driver...

 

https://downloadcenter.intel.com/pr...-i7-4700MQ-Processor-6M-Cache-up-to-3-40-GHz-
It is dated 5/4/2018
Is it possible that there is a newer version?

 

If you tell the driver to use the dGPU for everything, then restart, does it do the same thing?

 

If the CPU is still getting super hot, it could be a heatpipe failure.

Mine failed in my W230SS a few years back. A crack developed on the CPU pipe so the CPU basically wasn't being cooled.
The most obvious sign of the failure was under a CPU only benchmark (GPU idle) the fan would blow out cold air.

Got a new one shipped out and it was back to normal.

 

I installed windows 7 and the CPU temp improved greatly
now on throttlestop ts bench the temp is only reaching 85 when it was reaching 93 in windows 10(in both cases turbo is disabled)
About GTA V I am starting to think it is a memory leak issue as the GPU usage drops only happen when RAM usage reaches 90%(I have only 4 GB of RAM but I am playing at the lowest settings and there was no problems in the past)
I have read that many users experienced memory leaks with GTA V and so far I haven't found a solution.
Thank you all for your help especially Prema!

 

optimus does use ram to store the completed image but the IGP doing the processing is even more intensive, very odd. If you can get 4GB more then 8GB is the sweet spot atm.