Fixed a 2500$ laptop, but it runs super hot. Is this normal?

[iamdarkyoshi]

Its a dell precision m6800 with an i7 4800MQ and Nvidia Quadro K3100M. When on battery, the CPU runs at 2.7ghz. Plugged in, it runs at 3.0ghz.

 

Running aida64 (GPU/CPU/RAM) with it plugged in, I managed to get the CPU to NINETY SEVEN degrees! All we have on hand is shit thermal paste. Is this issue 100% the paste's fault, or is the cooling just poorly designed?

[Lynks]

I think that's normal for a gaming laptop on heavy load. The paste wouldn't make too much of a difference unless you applied WAY too little. You can try getting one of those cooler which go underneath the laptop. They help a little.

[Julian5]

whats the TJ max? it might be the paste

[AresKrieger]

I vote both, the cooler likely sucks but bad paste surely isn't helping

[iamdarkyoshi]

Just now, Julian5 said:

whats the TJ max? it might be the paste

It is unbranded white thermal paste...

[AresKrieger]

Just now, iamdarkyoshi said:

It is unbranded white thermal paste...

well then

[don_svetlio]

13 minutes ago, iamdarkyoshi said:

Its a dell precision m6800 with an i7 4800MQ and Nvidia Quadro K3100M. When on battery, the CPU runs at 2.7ghz. Plugged in, it runs at 3.0ghz.

 

Running aida64 (GPU/CPU/RAM) with it plugged in, I managed to get the CPU to NINETY SEVEN degrees! All we have on hand is shit thermal paste. Is this issue 100% the paste's fault, or is the cooling just poorly designed?

It's the pastejob

 

5 minutes ago, AresKrieger said:

I vote both, the cooler likely sucks but bad paste surely isn't helping

Cooler is quite good actually. Check google.

 

6 minutes ago, Lynks said:

I think that's normal for a gaming laptop on heavy load. The paste wouldn't make too much of a difference unless you applied WAY too little. You can try getting one of those cooler which go underneath the laptop. They help a little.

That is not a gaming laptop nor is 97*C normal for a gaming laptop. Paste can have a BIG difference. Those cooler are also shit in 50% of the cases due to the bottom grill not being under the fans

[iamdarkyoshi]

Just now, don_svetlio said:

It's the pastejob

 

Cooler is quite good actually. Check google.

 

That is not a gaming laptop nor is 97*C normal for a gaming laptop. Paste can have a BIG difference. Those cooler are also shit in 50% of the cases due to the bottom grill not being under the fans

I might have to convince my boss to get some decent paste...

[don_svetlio]

1 minute ago, iamdarkyoshi said:

I might have to convince my boss to get some decent paste...

Also make sure the mounting pressure is enough.

[AresKrieger]

Just now, don_svetlio said:

It's the pastejob

 

Cooler is quite good actually. Check google.

 

That is not a gaming laptop nor is 97*C normal for a gaming laptop. Paste can have a BIG difference. Those cooler are also shit in 50% of the cases due to the bottom grill not being under the fans

I did and it can get warm if you throw it to 100% at 3.0Ghz, but usually it should stay around 85 max

[don_svetlio]

1 minute ago, AresKrieger said:

I did and it can get warm if you throw it to 100% at 3.0Ghz, but usually it should stay around 85 max

From NBC, P95 + Furmark gets them 95*C so I'd imagine Aida64 CPU-only would need to be around 80-85 yes.

[AresKrieger]

5 minutes ago, don_svetlio said:

From NBC, P95 + Furmark gets them 95*C so I'd imagine Aida64 CPU-only would need to be around 80-85 yes.

Yeah Furmark is a brutal test, especially on a laptop, though normally if this laptop hit's 90C it should try and down clock to 2.8Ghz, as for Aida I don't know but an XTU test supposedly got someones between 85 and 90 (downclocking as it got to 90 then going back down to 85 just to up clock)

 

[iamdarkyoshi]

1 minute ago, AresKrieger said:

Yeah Furmark is a brutal test, especially on a laptop, though normally if this laptop hit's 90C it should try and down clock to 2.8Ghz, as for Aida I don't know but an XTU test supposedly got someones between 85 and 90 (downclocking as it got to 90 then going back don to 85 just to up clock)

 

It DID actually start throttling, 4% max

[AresKrieger]

Just now, iamdarkyoshi said:

It DID actually start throttling, 4% max

Well at least that is working mostly

[don_svetlio]

2 minutes ago, AresKrieger said:

Yeah Furmark is a brutal test, especially on a laptop, though normally if this laptop hit's 90C it should try and down clock to 2.8Ghz, as for Aida I don't know but an XTU test supposedly got someones between 85 and 90 (downclocking as it got to 90 then going back don to 85 just to up clock)

 

Meh, 85*C is fine. Even 90*C is fine for a stress test.

1 minute ago, iamdarkyoshi said:

It DID actually start throttling, 4% max

Paste

[D2ultima]

2 hours ago, Lynks said:

I think that's normal for a gaming laptop on heavy load. The paste wouldn't make too much of a difference unless you applied WAY too little. You can try getting one of those cooler which go underneath the laptop. They help a little.

Not even close... that's EXTREMELY abnormally high temperatures, even for that machine.

2 hours ago, AresKrieger said:

I did and it can get warm if you throw it to 100% at 3.0Ghz, but usually it should stay around 85 max

Still nope.

 

That machine is so far overheating it's not funny, something has to be wrong somewhere. It should have a better CPU cooling setup than I do too.

 

2 hours ago, iamdarkyoshi said:

Plugged in, it runs at 3.0ghz.

This is supposed to be 3.5GHz to 3.7GHz at stock settings, depending on the type of load (heavy single thread pushes to 3.7, heavy dual-thread to 3.6, heavy tri-thread or quad-thread to 3.5). You should get that one looked at, and call Dell support on the matter if you cannot get it fixed.

[AresKrieger]

13 minutes ago, D2ultima said:

snip

Wrong machine if yours runs 3.5-3.7, this laptop is more limited than the full potential of the chip inside, it is limted to 3.0Ghz max if under a multiple thread load  and can throttle down

Quote

In the Cinebench R10 64-bit, we measure 6,665 points (single thread) and 21,426 points (multithread), which is quite a bit lower than the 7,112 points (single thread) and 27,537 points (multithread) of the i7-4900MQ in the Dell Precision M4800. This is a much larger difference than we would expect to see as the core clock speed differs by a mere 100 MHz. The Alienware 17 scores much better with the Intel Core i7-4800MQ: 7,051 points (single thread) and 25,111 points (multithread). This is in the region of what we would expect from our test model.

The HWinfo64 tool offers a possible explanation: as soon as the CPU processes multithreaded tasks, the core temperature reaches a constant 98 °C (208.4 °F) and is tagged with the throttling icon. In the CPU load test using Prime95, the clock speeds fluctuate right off the bat between 2.4 - 2.8 GHz (later with FurMark: between 2.1 and 2.5 GHz). We used the "Ultra Performance" profile (max fan levels) for our tests. It appears the cooling system is not capable of keeping the CPU temperature in control. This is a little surprising as the Precision M4800 with the Intel Core i7-4900MQ (47 Watt TDP) and the predecessor the Precision M6700 with the Intel Core i7-3920XM (55 Watt TDP) do not have such problems. We believe that this issue could be limited to our test model or the cooling system is not working as expected. If possible, we will check this again using a different configuration in a test update.

 

[D2ultima]

1 hour ago, AresKrieger said:

Wrong machine if yours runs 3.5-3.7, this laptop is more limited than the full potential of the chip inside, it is limted to 3.0Ghz max if under a multiple thread load  and can throttle down

 

I have the wrong machine, but the same CPU.

 

If a 4800MQ is not running at 3.5GHz under 3 or 4 core loads in ANY LAPTOP, the CPU is not functioning correctly. This is what you need to understand.

 

If it's overheating, that's one thing. But if he isn't overheating and it's pegged at 3GHz, then the laptop is by-design limiting his potential, and that's just awful. I do not consider that to be "working" and as a business laptop, I would INSTANTLY have them look to fix that under warranty. That's what those expensive business-class laptops are for.

[iamdarkyoshi]

1 hour ago, D2ultima said:

I have the wrong machine, but the same CPU.

 

If a 4800MQ is not running at 3.5GHz under 3 or 4 core loads in ANY LAPTOP, the CPU is not functioning correctly. This is what you need to understand.

 

If it's overheating, that's one thing. But if he isn't overheating and it's pegged at 3GHz, then the laptop is by-design limiting his potential, and that's just awful. I do not consider that to be "working" and as a business laptop, I would INSTANTLY have them look to fix that under warranty. That's what those expensive business-class laptops are for.

I did component level repair on the motherboard

 

But thats not whats causing it to go at a lower speed. Maybe its because it turbos one core to 3.5 but when all four cores are loaded down, it goes to 3.0ghz?

[D2ultima]

Just now, iamdarkyoshi said:

I did component level repair on the motherboard

 

But thats not whats causing it to go at a lower speed. Maybe its because it turbos one core to 3.5 but when all four cores are loaded down, it goes to 3.0ghz?

No... the i7-4800MQ's specs are:

 

2.7GHz base clock, 1-4 core load.

+1GHz boost clock, 1-core load. (3.7GHz)

+900MHz boost clock, 2-core load. (3.6GHz)

+800MHz boost clock, 3-core & 4-core load. (3.5GHz)

+400MHz overclock bins to all cores, resulting in 4.1GHz 1-core, 4GHz 2-core and 3.9GHz 3-core & 4-core maximum overclocked speeds.

 

If your 4800MQ CPU is not turboing to any of those speeds, and cannot overclock to any of those speeds, then your machine is one of the following:

- Broken (functioning incorrectly; physical issue).

- Designed to be broken (artificially limited by manufacturer in some way).

- Has a malfunctioning EC or BIOS (requiring an update, re-flash, or reset trick to fix).

 

Hence why I said, contact support and make them give you a working system. If your machine is BY DEFAULT not using the 4800MQ in the way I described, something is wrong. Please note, turboing to those speeds I consider independent of overheating, or TDP limitation... those should limit your chip after it's been under load for some time, not before. I'm describing a system that cannot, even ice cold, achieve those speeds even for a short amount of time.