First Overclock Went Well: Watercooled 5820K MSI Gaming 9 ACK

[Benji_w]

Just now, Schoolofmonkey said:

This is why I keep suggesting OCCT, I find it better and safer than Prime95, if I have an unstable overclock OCCT will find it within 1 - 10 minutes, if you pass an hour you're set.

4.3Ghz@1.25v is the max I can do with OCCT, any higher and I will fail, but I can pass Realbench fine.

 

I know people say it puts "unrealistic" loads on the CPU, but I would prefer that than not knowing my overclock is 100% stable.. 

The whole point of stress testing is to find any form of instability regardless of realistic loads or not. So people who say that pretty much get written off by me as people who don't know diddly squat about overclocking.

[Schoolofmonkey]

1 minute ago, Benji_w said:

The whole point of stress testing is to find any form of instability regardless of realistic loads or not. So people who say that pretty much get written off by me as people who don't know diddly squat about overclocking.

And that's why I continue to use it, and not the crappy Intel Stress test or Realbench for that matter...lol

Over at Overclock.net many of the "elite" would throw that around, "you don't really need to run OCCT, as long at it's Realbench, x264 Benchmark stable its fine", yeah right.

[Deli]

1 hour ago, Schoolofmonkey said:

This is why I keep suggesting OCCT, I find it better and safer than Prime95, if I have an unstable overclock OCCT will find it within 1 - 10 minutes, if you pass an hour you're set.

4.3Ghz@1.25v is the max I can do with OCCT, any higher and I will fail, but I can pass Realbench fine.

 

I know people say it puts "unrealistic" loads on the CPU, but I would prefer that than not knowing my overclock is 100% stable.. 

 

Yeah the GTX570 shouldn't heat up the case that much.

 

1.35v creates a lot of heat, even mine will hit 79c when using the same voltage, try lowering the voltage to 1.25v with the same overclock and see what happens.

That's a little odd. My CPU passed OCCT linpack with 1.28v for 12 hours, but failed Realbench.

 

At the end, I passed AIDA 64, Intel Burn test and OCCT. But crashed when using Handbrake.

[Benji_w]

2 minutes ago, Deli said:

That's a little odd. My CPU passed OCCT linpack with 1.28v for 12 hours, but failed Realbench.

 

At the end, I passed AIDA 64, Intel Burn test and OCCT. But crashed when using Handbrake.

IBT AVX will often find an unstable OC incredibly quick.

[Schoolofmonkey]

1 minute ago, Deli said:

That's a little odd. My CPU passed OCCT linpack with 1.28v for 12 hours, but failed Realbench.

Yes but OCCT is testing the CPU, realbench is testing the CPU/Ram/Cache, it could of picked up a voltage issue due to the cache not getting enough 

And we all know that even though you increase the ring voltage if you don't have enough core voltage you can have Realbench fail, it's something I've personally encountered and tested.

Dropping ring clocks back to default without making any changes to the core clock, ring or core voltages lead to system stability.

 

X99 is a finicky beast, where the impossible is possible..lol

[Deli]

1 hour ago, Benji_w said:

IBT AVX will often find an unstable OC incredibly quick.

When I turn on AVX, my CPU will run 15C hotter. Which is insane.

 

I've yet to see any real world application that can stress the CPU, so the temperature will rise higher than in AIDA 64. After the crash in Handbrake. I upped the vcore to 1.295v from 1.290v. And it's been stable since, for like 8 months now.

[Benji_w]

2 minutes ago, Deli said:

When I turn on AVX, my CPU will run 15C hotter. Which is insane.

 

I've yet to see any real world application that can stress the CPU, so the temperature will rise higher than in AIDA 64. After the crash in Handbrake. I upped the vcore to 1.295v from 1.290v. And it's been stable since, for like 8 months now.

That's the whole point, it's stressing your overclock more...

[Deli]

1 hour ago, Benji_w said:

That's the whole point, it's stressing your overclock more...

I don't feel comfortable to run the CPU at those temp(85C) for 12 hours. Thanks God, I don't have to stress test any CPU for at least a couple of years.

[Benji_w]

1 minute ago, Deli said:

I don't feel comfortable to run the CPU at those temp(85C) for 12 hours. Thanks God, I don't have to stress test any CPU for at least a couple of years.

If you're comfortable not knowing if the data being written to your storage is correct or not that's up to you. I'd rather know my windows isn't going to randomly fail to load.

[NavXIII]

3 hours ago, NumLock21 said:

Here is LTT's OC database. Can compare your 5820K to other members here with the same CPU.

 

 

So I compared my to someone with a DH14 air cooler and an H105 240mm liquid cooler. Both are about 4.5GHZ and both hit around 79C under max load. Mine runs at 83C under max. It just seems natural to think that a 280mm water cooler should run cooler than a 240mm and an air cooler.

[Deli]

1 hour ago, Benji_w said:

If you're comfortable not knowing if the data being written to your storage is correct or not that's up to you. I'd rather know my windows isn't going to randomly fail to load.

Yes, that's my choice. And I'm comfortable with my choice.

[NavXIII]

16 hours ago, Schoolofmonkey said:

OCCT is a stress testing app, like Prime95.

http://www.ocbase.com/

 

I actually have the Business edition of AIDA64, but while you're in the stress test windows select preferences you'll be able to select what you want to display in the statistics window.

 

I was also thinking, have you set your pump to full speed in the Corsair software and seen if that makes a difference.

I'm going to conclude this thread with one last question. In AIDA64 should I be looking at the CPU temps (77C), CPU package temps (83C), the average temps of all 6 cores (74C), or the temps of the highest core (81C)? Depending on the answer I could or couldn't overclock a bit more.

Also the temps still fluctuates in AIDA64 like +/- 7C.

[iain_incredible]

On 3/5/2016 at 3:43 PM, NavXIII said:

Check out the pictures. Can't quite tell if the plate is making contact with the CPU. I think it is.

Also after reseating, I get idle temps of 35C on the package and under the stress test it's 70C. Also it fluctuates a lot under the stress test. Every 2 seconds it would go from 65 to 70 and back to 65, etc.

 

Wouldn't my 280mm cooler run colder than your 240mm cooler? If we're talking about hottest core then I get about 74C.

I only got 83C on 1.35V, which wasn't needed for 4.5GHZ. I get about 74C on 1.3V

BTW, You might want to change that thermal paste, looks like you have had it on really long.

[NavXIII]

2 minutes ago, iain_incredible said:

BTW, You might want to change that thermal paste, looks like you have had it on really long.

I had it on for a week.