High temperatures during GTA V. Suggest a cooler.

[GammaLazer]

My first time on forums of this website.

Hi my CPU temperature goes upto 70-75 C while playing GTA V. I live at place where summers are already soaring and temperature outside is around 35C. My table setup make CPU lie in a box which is not much airy.
So please suggest me if I should go for a CPU cooler or should try something else. Also my GPU temperature soars upto 67C.

I have short listed following:

1. Deepcool Neptwin V2
2. Deepcool Lucifer V2
3. CM Hyper 212x
4. CM Hyper TPC 812

Thanks

 

[lehtv]

Post your specs. Particularly your CPU.

 

[GammaLazer]

Post your specs. Particularly your CPU.

CPU: Intel i5-3570k (stock)
GPU: Asus AMD 7970 HD DCII 3 GB
RAM: 2 X 4GB G.Skill RipJaws
MoBo: Asus P8Z77-M Pro
Case: CM HAF 912
SSD : Samsung 840 Evo 120 GB
HDD1: WD Enterprise Black 1TB
HDD2: Seagate 5900 RPM Green

 

[lehtv]

75C is fine for i5-3570K, and 67C is more than fine for a 7970. These temperature simply aren't cause for concern.

What does Task Manager's Performance tab look like while playing GTA V? That should tell you roughly how much the CPU is doing work. Using the stock cooler, 75-80C is normal during heavy load even at 20-25C ambient. What exactly is your ambient temperature?

Also, download Prime95, run "In-place Large FFTs" stress test for 10-15 minutes. It will create an extreme load that should tell you how much the CPU will heat up in the absolute worst case scenario. If that temperature is too much, then you have a good reason to upgrade your cooling. (Another good reason is if you just want a quieter computer.)

Out of the coolers you listed, the cheapest one (212) will be fine for reducing noise and temperatures. Only consider a more expensive cooler if you want to have proper headroom for overclocking. About 4.3-4.5GHz should be doable with higher end air cooling, and 4.0-4.2 GHz is easy to achieve with minimal fiddling of settings. DeepCool Lucifer is quite a good cooler, but I'm wondering if there are other options, such as Thermalright HR-02 Macho or True Spirit 140, or Scythe Mugen MAX?

 

[GammaLazer]

75C is fine for i5-3570K, and 67C is more than fine for a 7970. These temperature simply aren't cause for concern.

What does Task Manager's Performance tab look like while playing GTA V? That should tell you roughly how much the CPU is doing work. Using the stock cooler, 75-80C is normal during heavy load even at 20-25C ambient. What exactly is your ambient temperature?

Also, download Prime95, run "In-place Large FFTs" stress test for 10-15 minutes. It will create an extreme load that should tell you how much the CPU will heat up in the absolute worst case scenario. If that temperature is too much, then you have a good reason to upgrade your cooling. (Another good reason is if you just want a quieter computer.)

Out of the coolers you listed, the cheapest one (212) will be fine for reducing noise and temperatures. Only consider a more expensive cooler if you want to have proper headroom for overclocking. About 4.3-4.5GHz should be doable with higher end air cooling, and 4.0-4.2 GHz is easy to achieve with minimal fiddling of settings. DeepCool Lucifer is quite a good cooler, but I'm wondering if there are other options, such as Thermalright HR-02 Macho or True Spirit 140, or Scythe Mugen MAX?

I performed the Prime95 test just now with small FFTs and the temperatures shot up to 99 C after which I terminated the test. My idle temperatures are around 47-51 C with ambient being 34 C right now. During GTA V, CPU load is between 15% - 30% and temperatures around 65+C on all cores.

I can buy any cooler which performs the best, the only problem is availability in India. Also the prices of most PC peripherals here are around15-20 USD costlier than what are in US.
[EDIT]
Can you please reply quickly as I am right in a city for few hours where I could buy it.
Thanks

 

[BonzaiDuck]

I performed the Prime95 test just now with small FFTs and the temperatures shot up to 99 C after which I terminated the test. My idle temperatures are around 47-51 C with ambient being 34 C right now. During GTA V, CPU load is between 15% - 30% and temperatures around 65+C on all cores.

I can buy any cooler which performs the best, the only problem is availability in India. Also the prices of most PC peripherals here are around15-20 USD costlier than what are in US.
[EDIT]
Can you please reply quickly as I am right in a city for few hours where I could buy it.
Thanks

I just put together an i5-3570K system for my brother, with a 12-year-old computer case with worse airflow potential than your HAF 912 -- a lot worse, actually.

With just the CoolerMaster EVO 212 -- which you say is available, I think your severe load temperatures should drop at least 15C -- maybe 20C -- I'd have to check some review comparisons. To clarify, I mean better by that many degrees over the stock Intel cooler.

My Bro's system deploys the EVO 212 with a $4 ThermalRight accordion duct to the rear exhaust fan of the case. Here's the item as sold from an outfit in Florida (USA):

http://www.performance-pcs.com/ther...r-01x-120mm-tower-cooler-fan-duct-for-k8.html

If you can't find one in India, you can build your own. Pick up a panel of foam art-board at an arts and crafts store, design your duct-box to fit the fan and cooler with some tabs to hold it in place.

The ducting in my own systems has been known to decrease CPU temperatures by another 5C degrees.

There are better coolers than the EVO: The Noctua NH-D14 performs about 5C cooler; the EVGA ACX (formerly named "SuperClock") performs a full 11C better. But the EVO should do the trick for you.

Also, be aware that certain stress tests with the latest AVX2 features will really toast your processor. Ashamed to say it, but I'm not sure whether this is common between the Ivy Bridge and the Haswell cores, or just the Haswells. But you'd also do well to download OCCT and stress-test with the OCCT:CPU test as well as the built-in LinPack:CPU test.

 

[BonzaiDuck]

Generally, I look for reviews published or linked by Frosty Tech, and many of the comprehensive comparisons are posted by Hardware Secrets:

http://www.hardwaresecrets.com/article/Cooler-Master-Hyper-212-EVO-CPU-Cooler-Review/1407/6

I find it odd that they'd show the EVO outperforming the NH-D14 by some 3C degrees, when I'd seen similar comparisons with the D14 outperforming the EVO by 5C.

And I say this, directing your attention to the EVGA ACX/Superclock performance in the same table: just what I said it would be. Fans, maybe? I can't say.

So there is at least some indication that the 212 EVO will serve even better.

EVEN SO. The reason I lost myself in this digression about reviews: I wanted to find a comparison between the EVO and some other coolers against the Intel stock cooler. Maybe the test bed doesn't quite match an Ivy Bridge core, but it's there -- at the top of the list: 90C at full load, and a difference of what looks to be about 20C higher than the performance of the EVO.

I rest my case, counselor. The Intel cooler is guilty! Guilty! Guilty!

 

[GammaLazer]

Thanks for the reply. Actually almost all brands and coolers are available either through online retailers from country or newegg. I am getting 212x for about $47 and lucifer & neptwin both for $55. Actually I can buy any from cooler master as they are even available in local retail shops. Budget :$65.
Which one should I choose taking my ambient temperature into consideration.

 

[GammaLazer]

I don't think test benches match actual conditions where the processors run. Most of the reviews and benchmarks over the net show that hyper 212x beats many top ranked coolers.

 

[BonzaiDuck]

I don't think test benches match actual conditions where the processors run. Most of the reviews and benchmarks over the net show that hyper 212x beats many top ranked coolers.

Don't want to get into a p***ing contest about this. But! . .

I ran my own tests on the D14 and the ACX/Superclock coolers. I was able to replicate the relative performance against a handful of performance comparison reviews.

It would depend on the thermal wattage of the test-bed. For instance, either of my overclocked i7-2x00K processors reach 137W under severe load and peak "turbo" VCORE readings; they're both set to reach voltages of 1.35+V for the 2700K and 1.37V for the 2600K.

The i5-3570K has a stock TDP of 77W, while my Sandy Bridge cores at stock clocks are about 95W. But the IHS on the IB is installed with a thermal interface putty, while the SB's are affixed with indium solder.

The Ivy Bridge cores may therefore reach higher temperatures on the core sensors with the same or lower wattage, and they vary more in that regard because of variation in the gap between the IHS and silicon die.

I'd have to study that Hardware Secrets review some more. There may be a difference in fan deployment. You can squeeze more performance from the D14 with fans offering better CFM airflow, and you can pick those fans to be just about as quiet as the original Noctua fans. But . . what's good for the goose is sauce for the gander, and you can do the same thing with the other coolers.

Just to top off the discussion, though . . Choice of "best" or second-best TIM between HSF and IHS can mean a few C difference, so I use nano-diamond paste; others use Coolaboratory Liquid Pro, which beats the diamond. If you lap all the nickel-plate off the IHS (and the cooler base if it isn't bare copper), you get a few more degrees improvement -- beyond just making the surface flat.

But even those results will vary with the thermal wattage of the test-bed.

I understand the climate in which you live. So I can appreciate your concerns with this. I think you'll do fine with your choices on this -- as good as possible given ambient room temperatures of your computing environment.

 

[GammaLazer]

That was some real reply. Wow and thanks. I am still going through hardware secrets reviews to find best one for me.

 

[lehtv]

I wouldn't sweat too much on finding the best one, almost any 120mm+ tower will make your CPU run at acceptable temps at stock. At the prices you posted I think DeepCool Lucifer is excellent value, 212 not so much.

With your cramped case setup, there's also an argument there for a 120mm liquid cooler, like a Cooler Master Nepton or a Corsair Hydro. The benefit there would be to immediately exhaust heat generated by the CPU, so that heat wouldn't build up inside the case so much.

 

[GammaLazer]

I wouldn't sweat too much on finding the best one, almost any 120mm+ tower will make your CPU run at acceptable temps at stock. At the prices you posted I think DeepCool Lucifer is excellent value, 212 not so much.

With your cramped case setup, there's also an argument there for a 120mm liquid cooler, like a Cooler Master Nepton or a Corsair Hydro. The benefit there would be to immediately exhaust heat generated by the CPU, so that heat wouldn't build up inside the case so much.

Thanks for the reply. I'm more of air cooled system guy. So i need quick final words why lucifer over neptwin when latter has twin towers and is TPC 812 of any competition here.
Thanks again!

 

[lehtv]

I would prefer one 140mm fan over two 120mm fans, assuming similar heat sinks. Hence Lucifer V2 over Neptwin. It's should be quieter, but performance will be roughly the same.

TPC-812 has great cooling capacity but that comes at the cost of noise. It uses one 120mm fan at max. 2400 RPM, while Neptwin's max RPM is 1500. See Hardwaresecrets. TPC-812 is 2C cooler than Neptwin, but 5dBa louder - which is a huge difference, as the decibel scale is logarithmic, and an increase of 3 dBa means doubling in intensity.

If you ran TPC-812 with the low noise adapter at max 1600 RPM, the noise level would be the same or possibly slightly better (since Neptwin uses two fans), but it wouldn't be competitive with Neptwin's temperatures, instead it'd dwindle down to the level of Arctic Freezer i30 which is 5C hotter in the Hardwaresecrets chart.

The chart is also a good reference for why almost any decent cooler will be enough for your needs. The hottest 120mm tower in the tests is True Spirit 120 (which is odd as it should be similar to 212 EVO), but that's still 23C cooler than the stock cooler. The CPU used is i5-2500k.

 

[MongGrel]

I like my D14 still,but they are big, but there is nothing wrong with a EVO 212.

Decent paste helps, there are a lot of things out there but I like the NT-H1 also still.

It sounds you have a smaller case, so other recommendations probably better, I like big towers still.

I really haven't been trying to shave off degrees in awhile to be real I guess.

The X5680 runs well in it as is, I have never even pushed it that hard past 4.5 in reality.

 

[lehtv]

It sounds you have a smaller case, so other recommendations probably better, I like big towers still.

According to specs, HAF 912 can fit up to 175mm tall coolers... which includes basically any air cooler in the market. The only potential problem here is not being able to use the side fan slot. Width and length shouldn't cause issues in an ATX mid tower.

 

[BonzaiDuck]

According to specs, HAF 912 can fit up to 175mm tall coolers... which includes basically any air cooler in the market. The only potential problem here is not being able to use the side fan slot. Width and length shouldn't cause issues in an ATX mid tower.

Wey-uull . . . I've been bragging about my i5-3570K replacement for my brother's old system, also the occasion for me to retrieve my CM Stacker midtower case and replace it with an old-style ATX midtower CM Centurion.

I keep thinking I should take some pictures. I gave bro a total of 750GB of local storage: 256GB for the boot SSD, and 500GB of a cached, 2.5" laptop WD Blue. The three 2.5" drives could all fit in my wallet.

So I turned most of the 5.25" bay space into a wind tunnel -- put a 140mm Akasa Viper in there so there are two intake fans. And I just blocked off the side-panel vent, which had originally been designed for an 80mm or single 120mm fan. So there's the Akasa and a Cougar Vortex below it in the lower front-panel position intended for a 120mm fan.

And -- it's fitted with a 212 EVO, TR accordion-duct -- one pusher fan and the exhaust at the other end of the duct. Seems to run cool enough. Not sure if we're going to overclock it much. Maybe later.

 

[GammaLazer]

Okay thanks guys. Any suggestions for thermal paste? Should I use the one that comes with the cooler or get any other?
I was thinking of Arctic silver 5.

 

[lehtv]

Okay thanks guys. Any suggestions for thermal paste? Should I use the one that comes with the cooler or get any other?
I was thinking of Arctic silver 5.

You'll be fine with whatever comes with the cooler, IMO. Extra paste is useful if you need to seat the cooler multiple times

 

[GammaLazer]

Okay thanks everyone. I will post the results as soon as I get my hands on the cooler.

 

[GammaLazer]

One Week use:
Ambient temperatures are around 27 C atm.

Stock Cooler:

Idle Temperatures:
Min: 49 C
Max: 55 C
Average: 53 C

Playing GTA V:
Min: 70 C
Max: 85 C
Average: 78 C

DeepCool Lucifer Cooler:

Idle Temperatures:
Min: 31 C
Max: 41 C
Average: 34 C

Playing GTA V:
Min: 45 C
Max: 54 C
Average: 49 C


That's a hell lot of difference imo. I used the thermal paste that came with the cooler with pea shape.
But there is one problem I am facing that the Fan that came with the cooler being PWM works at max 400RPM. Any solutions?

 

[guskline]

GammaLazer: GOOD JOB. With that difference in temps, don't touch a thing! Looks like it's working very well.

As for the fan at 400rpm, How did you measure that speed? I just read the specs for the fan and the speed should vary from @700rpm up to @1380rpm.

What fan header on the mb do you have it plugged into?

 

[GammaLazer]

GammaLazer: GOOD JOB. With that difference in temps, don't touch a thing! Looks like it's working very well.

As for the fan at 400rpm, How did you measure that speed? I just read the specs for the fan and the speed should vary from @700rpm up to @1380rpm.

What fan header on the mb do you have it plugged into?

I have plugged the fan into CPU_FAN port. For fan speed system gives warning on start for low fan speed which I have turned off. I have checked fan speed in BIOS, speed fan and HWMonitor

 

[lehtv]

As for the fan at 400rpm, How did you measure that speed? I just read the specs for the fan and the speed should vary from @700rpm up to @1380rpm.

DeepCool Lucifer is 300±150 ~ 1400±10%RPM.

The motherboard controls the RPM according to temperture... Sounds to me like the CPU is simply not running warm enough for the RPM to need to ramp up.

You can try setting the PWM mode to "turbo" or whatever the maximum performance mode is called, see if it runs any faster (not that it needs to - just to find out).

 

[BonzaiDuck]

DeepCool Lucifer is 300±150 ~ 1400±10%RPM.

The motherboard controls the RPM according to temperture... Sounds to me like the CPU is simply not running warm enough for the RPM to need to ramp up.

You can try setting the PWM mode to "turbo" or whatever the maximum performance mode is called, see if it runs any faster (not that it needs to - just to find out).

The ASUS Z77 board should have at least three pre-sets in the BIOS: "Turbo," "Standard," "Silent" etc.

For my brother's i7-3570K Asus Z77-A rig, I set it to "Turbo" intending to define a custom fan curve later, and I'd assumed that Turbo would be too noisy. But it wasn't -- perhaps given the fact that I'd chosen quiet PWM fans. And the fact is -- those fans could run at slower RPMs.

In your case, something else may be effective. Maybe it's set to "Silent" mode.

Others could discourage it, but I would install the Ai Suite software provided for that motherboard -- a download from the ASUS website. You can turn off the overclocking tools, but leave the sensor-monitoring and fan control features active. Then you should be able to define fan-curves -- schedules of % duty-cycle according to CPU temperature -- for at least two and possibly three different fan-ports: CPU_FAN, CHA_FAN1 and possibly CHA_FAN2.

You can change the threshold of "warning" for the CPU_FAN in BIOS. Once you define these fan curves with the software, you should re-enter the BIOS and select "User" ("Custom" or whatever it's called) as opposed to the other presets.

Also -- DO NOT run two ( or more) sensor monitoring programs simultaneously.* If you choose to use the ASUS "Fan Xpert" or whatever it's called, you don't need SpeedFan. If you use SpeedFan -- you don't need Fan Xpert. But I recommend the ASUS software over anything else. If you want to monitor with HWMonitor, then close/exit AI Suite completely. The fan control settings will continue to be effective. In fact, I believe you could completely uninstall the ASUS software and the "User" settings defined with it would continue to determine fan speed according to temperature.

* Reason: I think there's a consensus here at the forums, especially among the overclocking enthusiasts, that two programs simultaneously polling the same sensors may cause the system to crash. Take for instance my 2x GTX 970 SLI graphics. I had them set at the default stock speeds. I chose to run Heaven Benchmark, which monitors GPU speeds and temperatures by itself. I had left NVidia-Inspector's monitoring software running, and the system crashed about 5 minutes into the Heaven Benchmark run -- with Stop Code 124. After rebooting, turning AI Suite and everything else off, I ran Heaven again and then fell asleep. It was still running tip-top six hours later. There is only one assignable cause for the original BSOD.