Raising throttling temp?

[Somerandomtechyboi]

My p6x58d-e seems to throttle at 95c but my old p6t deluxe throttles at 100c so is there a way aside from just crossflashing the bios to raise throttling temp to like 100-105c? Cooler is kinda dinky but it wont really be running at 100c constantly, just need the extra temp headroom so less throttling happens when the thing is being very heavily stressed

 

Currently ive disabled the tm function so itll run hotter but i dont know if itll auto shutdown at say 107.5c like the p5q or at a higher temp like 120c

 

 

Currently im using prime95 smallest ffts to validate cpu stability so itd be nice if there was another program that i could use to validate cpu stability but doesnt run the cpu as hot as smallest ffts

[lordmogul]

Which CPU are using on it? That is where the throttle limit is set. Most of the LGA1366 i7 have their limit at 100°C

 

As for different stability tests. prime95 is obviously very hard, and small FFTs fit inside the cache and will be extra hot.

• You could run large FFTs, which should give less heat and also stress the memory controller, which might be useful if you want to test more than just the CPU. Or just set a custom FFT size.
• OCCT has a scanner
• AIDA64 has a test. The FPU test gets hotter than the "normal" CPU test
• Intel Burn Test is an option. (It runs Linpack)
• Unigine Valley and Heaven are also options. More of a GPU stress test, but they give consistent load and if you put the settings all the way down, you should crunch enough frames to give the CPU something to do.
• Things like 7-zip compression benchmark
• And the most important one: The tasks you actually do. Synthetic benchmarks and stress tests have their use, but you aren't planning on only bench on the system, right? So start your most demanding typical task and test with that. A good example are CPU heavy games, like simulations. Things like Cities Skylines or Kerbal Space Program. And since both are singleplayer, you won't annoy your teammates when your system keeps crashing.

[Somerandomtechyboi]

31 minutes ago, lordmogul said:

Which CPU are using on it? That is where the throttle limit is set. Most of the LGA1366 i7 have their limit at 100°C

 

As for different stability tests. prime95 is obviously very hard, and small FFTs fit inside the cache and will be extra hot.

• You could run large FFTs, which should give less heat and also stress the memory controller, which might be useful if you want to test more than just the CPU. Or just set a custom FFT size.
• OCCT has a scanner
• AIDA64 has a test. The FPU test gets hotter than the "normal" CPU test
• Intel Burn Test is an option. (It runs Linpack)
• Unigine Valley and Heaven are also options. More of a GPU stress test, but they give consistent load and if you put the settings all the way down, you should crunch enough frames to give the CPU something to do.
• Things like 7-zip compression benchmark
• And the most important one: The tasks you actually do. Synthetic benchmarks and stress tests have their use, but you aren't planning on only bench on the system, right? So start your most demanding typical task and test with that. A good example are CPU heavy games, like simulations. Things like Cities Skylines or Kerbal Space Program. And since both are singleplayer, you won't annoy your teammates when your system keeps crashing.

I use an x5650

The old p6t deluxe with an x5660 only started throttling at 100c

 

currently im in the process of tuning it up and suprisingly this dinky 92mm alseye is able to run the cpu at 4.2ghz with some gd007 paste so i think i got a bad batch of gd900, now im onto tuning the rams so itd be nice to have some sort of benchmark to see if im making any sort of progress cause the only ideas i have for benchmarking ram would be spi or wprime

[lordmogul]

You can always see if you can improve cooling. Maybe all it takes is a new fan.

 

I'm running a X5460 @ 4.1 with a Hyper 212 Evo and don't have any thermal issues there. Running in the low 80s with >85% all core load during summer.

Sure, it's only a quad, but also on a bigger manufacturing process.

 

For RAM benchmarking you could go with simple transfer tests like from AIDA64. But it's still only a theoretical test.

[Somerandomtechyboi]

11 minutes ago, lordmogul said:

You can always see if you can improve cooling. Maybe all it takes is a new fan.

 

I'm running a X5460 @ 4.1 with a Hyper 212 Evo and don't have any thermal issues there. Running in the low 80s with >85% all core load during summer.

Sure, it's only a quad, but also on a bigger manufacturing process.

 

For RAM benchmarking you could go with simple transfer tests like from AIDA64. But it's still only a theoretical test.

That cpu draws less power

wolfdale/kentsfield have garbage die -> ihs transfer so that explains the temps

 

15 minutes ago, lordmogul said:

You can always see if you can improve cooling. Maybe all it takes is a new fan.

Definitely needs a better cooler but 4.3ghz needs like 1.4v so itd be running well out of v/f efficiency so i think ill just leave the cooler alone

 

Though i will go and buy a 775 copper core to stick on the rx580 cause the cooler is such dogshit that it can only keep the gpu around 90w, copper core will definitely perform better

[lordmogul]

Yup, those 1.4v at >4 GHz put out some heat and are way beyond the best efficiency. I'd say that is somewhere around 3.4-3.6 GHz should be the sweet spot, in same cases even more efficient than stock in terms of power/performance. Penryn/Nehalem/Gulftown aren't that different in terms of voltage/clock curve, but the newer chips do obviously a bit better.

I wouldn't even say it's the die -> IHS transfer but the IHS -> cooler transfer. The IHS on those chips was often curved and people lapped their Q6600 to improve cooling. Some manufacturers went so far and curved the cooler as well.

 

Going from a 92mm cooler to a 120mm does a lot. Only reason I got the Hyper 212 was the price. And it does well enough for a quad like that, even with the 120W TDP + oc. Otherwise I'd throw my HR-02 Macho on there.

 

43 minutes ago, Somerandomtechyboi said:

Though i will go and buy a 775 copper core to stick on the rx580 cause the cooler is such dogshit that it can only keep the gpu around 90w, copper core will definitely perform better

Might be worth giving more fans a try. Back on my 8800 GT I just removed the metal shroud from the cooler and strapped a pair of case fans on. Dropped temps by almost 30°C on that tiny cooler.

If your RX580 has a metal backplate might give it a try and use that for cooling. Some thermal pads to connect it to the hotspots on the boards and a fan directed towards it. Getting into ghetto cooling territory there.

[Somerandomtechyboi]

38 minutes ago, lordmogul said:

I wouldn't even say it's the die -> IHS transfer but the IHS -> cooler transfer. The IHS on those chips was often curved and people lapped their Q6600 to improve cooling. Some manufacturers went so far and curved the cooler as well.

So its just garbage ihs issue

ill see what i can do for my golden e5400 sample aside from just delidding and running direct water over die cooling

 

40 minutes ago, lordmogul said:

Might be worth giving more fans a try. Back on my 8800 GT I just removed the metal shroud from the cooler and strapped a pair of case fans on. Dropped temps by almost 30°C on that tiny cooler.

If your RX580 has a metal backplate might give it a try and use that for cooling. Some thermal pads to connect it to the hotspots on the boards and a fan directed towards it. Getting into ghetto cooling territory there.

Already tried with a fucking 120mm server fan that runs upto 3500 ish rpm alongside a 92mm fan that also runs to around that speed and still 90c temps

 

So yea need to ziptie a copper core on it and that should fix temps

[lordmogul]

 

16 hours ago, lordmogul said:

 

I'm running a X5460 @ 4.1 with a Hyper 212 Evo and don't have any thermal issues there. Running in the low 80s with >85% all core load during summer.

Sure, it's only a quad, but also on a bigger manufacturing process.

 

Oh, turns out there was lots of dust buildup over the winter. Needed some cleaning and it's down to the mid 50-low 60s now. Perfectly fine.

 

But yeah, to come back to the CPU throttling situation. It's obviously not just the fan but also the cooler itself. Oh, and maybe a second fan at the back of the cooler to pull the air out. ymmv obviously. And airflow around the cooler, enough intake and outgoing air and not having the system in one of these enclosures some desks have.

[Somerandomtechyboi]

2 hours ago, lordmogul said:

 

Oh, turns out there was lots of dust buildup over the winter. Needed some cleaning and it's down to the mid 50-low 60s now. Perfectly fine.

 

But yeah, to come back to the CPU throttling situation. It's obviously not just the fan but also the cooler itself. Oh, and maybe a second fan at the back of the cooler to pull the air out. ymmv obviously. And airflow around the cooler, enough intake and outgoing air and not having the system in one of these enclosures some desks have.

Cooler has a 3.5k rpm foxconn fan so that should not be an issue