Contribute
Register

Core i7-3970X, GA-X79-UD5 Monster SSD Build, Geekbench Score: 20313

Status
Not open for further replies.
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Heya morganb -

Thanks! :)

What is CPU cooler are you using? Also, keep in mind every CPU is different too. And yeah, I don't see a meaningful reason to go really high and risk stability. For me, stability is key, and everyday I have thoughts of bringing my CPU back to stock clock frequencies lol -- although it is completely stable at 4.2GHz.

The audio popping issue is indeed solvable with the onboard audio. You just have to use a DSDT with a layout that forces ML into using ID 898 instead of ID 1. Indeed the Gigabyte boards are the most hassle free boards for hackintoshing purposes.

Nice speakers BTW. :)

/ gx


Hey, Im on a corsair H100 but my gut feeling is that its high and that I have done a bad job with the thermal grease. I have tried various BIOS and various DSDT. I have been unable to load a DSDT. Every time I have tried, it has failed and resulted in an unrecoverable mess (for me at least). Anyway, if it had have worked then I would not have explored the external DAC option and and ended up with the Aktimate's.

I just chose user DSDT but without a DSDT, hnak ethernet, 10.8.1+ Trim, HW monitors, NullCPU..., INtel/Jmicron/GSata drivers, MacPro 5,1 and then added SleepEnabler. Its really REALLY slow to boot but flies when its up.
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Hey, Im on a corsair H100 but my gut feeling is that its high and that I have done a bad job with the thermal grease. I have tried various BIOS and various DSDT. I have been unable to load a DSDT. Every time I have tried, it has failed and resulted in an unrecoverable mess (for me at least). Anyway, if it had have worked then I would not have explored the external DAC option and and ended up with the Aktimate's.

I just chose user DSDT but without a DSDT, hnak ethernet, 10.8.1+ Trim, HW monitors, NullCPU..., INtel/Jmicron/GSata drivers, MacPro 5,1 and then added SleepEnabler. Its really REALLY slow to boot but flies when its up.

Heya morganb -

How much thermal paste did you put on? I would put no more than a blob the size of a single grain of rice in the center of the CPU.

What are your temps at idle and full load?

/ gx
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Heya morganb -

How much thermal paste did you put on? I would put no more than a blob the size of a single grain of rice in the center of the CPU.

What are your temps at idle and full load?

/ gx

38 -> 76, haven't allowed it to get higher than that. Will redo the paste. It was a bigger die so I used a bigger blob. Too much I think. I also don't think it was the best paste.
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

38 -> 76, haven't allowed it to get higher than that. Will redo the paste. It was a bigger die so I used a bigger blob. Too much I think. I also don't think it was the best paste.

The grain-of-rice method is enough actually. It's enough for a 3970X and that's the biggest thing out there. :)

Your 38 to 76 is honestly not abnormal. Your CPU, with load, will kick into the highest ratio you defined, which I think you said was running at 4.5GHz. That's gotta be hot, and I am sure you had to increase VCORE a few steps to achieve that.

Your CPU's TDP of 130W is actually the same as the fastest Sandy-Bridge Xeon CPUs.

I'm sure with optimal thermal paste application, and a cleaning of your H100's radiator, you might be able to bring the CPU temp down 2C or so. But beyond that, not much. It's already running at expected temperature ranges for your overclock, and H100 setup.

What I'd check:

1/ Thermal paste amount
2/ Type of paste
3/ Radiator cleanliness (a vacuum greatly helps)
> A thorough rinse of the radiator will help too, but don't forget to remove your fans from the radiator
4/ Radiator fan speeds (maybe one or both fans are dirty and slowing down or going bad and slowing down?)

Let me know how it goes. :)

/ gx
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

The grain-of-rice method is enough actually. It's enough for a 3970X and that's the biggest thing out there. :)

Your 38 to 76 is honestly not abnormal. Your CPU, with load, will kick into the highest ratio you defined, which I think you said was running at 4.5GHz. That's gotta be hot, and I am sure you had to increase VCORE a few steps to achieve that.

Your CPU's TDP of 130W is actually the same as the fastest Sandy-Bridge Xeon CPUs.

I'm sure with optimal thermal paste application, and a cleaning of your H100's radiator, you might be able to bring the CPU temp down 2C or so. But beyond that, not much. It's already running at expected temperature ranges for your overclock, and H100 setup.

What I'd check:

1/ Thermal paste amount
2/ Type of paste
3/ Radiator cleanliness (a vacuum greatly helps)
> A thorough rinse of the radiator will help too, but don't forget to remove your fans from the radiator
4/ Radiator fan speeds (maybe one or both fans are dirty and slowing down or going bad and slowing down?)

Let me know how it goes. :)

/ gx



Its the rate it gets there that is the concern. If it stabilised there then yeah, I would say situation normal and carry on knowing that IRL it would never stay at those temps. However, its the speed it gets to which I feel is abnormal - it climbs to 71 almost immediately and keeps going. I have to quit p95 after 2 cycles as I dont want to start degrading my chip.

Rad is pretty new. At this stage Im betting on poor paste applied too heavily.

Vcore at 1.55, LLC at Turbo - wont hold voltage otherwise.

Got 9-11,10,24 at T1 out of some 9,11,12,28 T2 2133 RAM though with no extra voltage so Im stoked with that.
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Its the rate it gets there that is the concern. If it stabilised there then yeah, I would say situation normal and carry on knowing that IRL it would never stay at those temps. However, its the speed it gets to which I feel is abnormal - it climbs to 71 almost immediately and keeps going. I have to quit p95 after 2 cycles as I dont want to start degrading my chip.

Rad is pretty new. At this stage Im betting on poor paste applied too heavily.

Vcore at 1.55, LLC at Turbo - wont hold voltage otherwise.

Got 9-11,10,24 at T1 out of some 9,11,12,28 T2 2133 RAM though with no extra voltage so Im stoked with that.

What will it top out at?
If you top out at 72C, I would say that's fairly normal.

If you open Terminal, and run 12 instances back to back of "yes > /dev/null &" my CPU at 38C jumps within 1 or 2 seconds to mid 60's, and then climbs to 70 within 2 or 3 seconds after that. I top out at 74C with all 6 cores pegged, after no more than 10 to 15 seconds.

Since the thermal diode resides in the CPU die, it detects thermal changes very rapidly. And your CPU at idle is probably running at no more than 1.2GHz (mine idles at 1.2GHz), and jumps to 4.5GHz + your VCORE increase (I'm using a voltage offset), to be honest, that seems normal to me.

The rate at which it climbs to 72C that you speak of, is definitely normal. Conversely, the rate at which it drops will be dramatic as well, once you issue a "killall yes" command. Again, this is because the CPU core will gain and lose heat very quickly, due to it gaining and losing 3,300,000,000 on-off cycles per second--from 1.2GHz idle to 4.5GHz in full load, all in a split second in your case. :)

Writing the frequency jump out like that, I think should let you rest easy now. :)

/ gx
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Thx. It doesnt stop it just keeps climbing. its on p95. I just cancel it when it gets too high. Not had the balls to see where it stops with a $600 chip. For the sake of not degrading the chip I have just stopped it.

Will get paste sorted at some point but right now I have a stupid amount of projects building up so need to get across some of them ASAP.

Also, given that you seem to know your way around a hack, would you have any idea why I cant get a DSDT top work with it? I cant even extract one, I just get a spinning beach ball. Not that Im unhappy but Im just curious - could it be a subtle BIOS fault? Ive already RMA'd this board once.
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Thx. It doesnt stop it just keeps climbing. its on p95. I just cancel it when it gets too high. Not had the balls to see where it stops with a $600 chip. For the sake of not degrading the chip I have just stopped it.

Will get paste sorted at some point but right now I have a stupid amount of projects building up so need to get across some of them ASAP.

Also, given that you seem to know your way around a hack, would you have any idea why I cant get a DSDT top work with it? I cant even extract one, I just get a spinning beach ball. Not that Im unhappy but Im just curious - could it be a subtle BIOS fault? Ive already RMA'd this board once.

Hi morganb -

So I downloaded Prime 95 and ran it in "Blend" torture mode, my CPU, depending on the phase of the test, hits as high as 81C, but for 99% of the test, hovers around the low 70's.

When you say it "gets too high" -- how high is too high? :)

Even if you install too much thermal paste, there's no way the CPU is going to get damaged, unless there's no contact, or very, very poor contact between the cold-plate surface on your H100 and the CPU. But it would have gone higher from turning on the PC, and not just hover at 38C on idle. And all modern CPUs have overheat protection, which will force it into Halt state and shut down. To be honest, your CPU can actually operate at 100C without dying.

As for DSDT, you actually don't need a DSDT, since you board uses UEFI.
I can't tell from your spinning beachball state what's wrong with the system. You would have to boot into verbose mode by using a -v flat at the Chimera/Chameleon boot selection screen.

But I am sure it's related to "Waiting for root device...".
See my comment on the original post about AHCI:

With many X79 motherboards, you will not be able to boot into Mountain Lion with UseKernelCache=Yes. You will receive the "waiting for root device" message and it will sit there indefinitely. This has nothing to do with the way your boot drive is installed onto the motherboard's SATA ports or boot priority settings. It all has to do with the way AHCI is enumerated during boot time. To fix this, you MUST boot with UseKernelCache=No. Unfortunately, this slows down boot time because OS X must reload all kexts each boot. To get around this, you muse patch IOAHCIBlockStorage.kext. To do so, Google for bcc9's AHCI patch. It is simple to do (execute his patch script in Terminal). After this, you may enable UseKernelCache to "Yes".

Anyway, back to your Prime 95 + heat concern: For the most part, your CPU's floating point cores will be off, unless you have a program that explicitly calls for FP processing. This maintains the CPU at a relatively cool state. Prime 95 literally does torture your floating point units, and will run a CPU very close to max operating temperatures. Prime 95 is designed to do this.

With your H100 and your CPU, your temps should climb no higher than 85C -- unless you're heavily affected by ambient temperatures.

Hope that helps! :)

/ gx
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Impressive build!

However, I'm a bit confused about your claim:

Fujitsu DYNA-MO 1300 U2 Magneto-Optical Drive
Note: You can only find the Fujitsu DYNA-MO 1300 U2 on eBay nowadays. For archiving important files, you really can't beat Magneto-Optical discs. They are based on laser-magnetic curing, and cannot be damaged by magnets, lasers, and heat. Ultra long shelf life. Caveat: expensive.


there's no hacking or installing of modded kext files (save for audio).
Are you sure???

you will need NullCPUPowerManagement.kext to get your system to boot.
MultiBeast Settings:

User DSDT (no DSDT)
TRIM Enabler Patch (for 10.8.1+)
AppleIntelE1000 Ethernet
FakeSMC with plugins
HWMonitor App
MultiBeast is installing the modded kext files for you. Sure it's been made easy for you, but I don't think your assertion is correct.
you muse patch IOAHCIBlockStorage.kext. To do so, Google for bcc9's AHCI patch. It is simple to do (execute his patch script in Terminal). After this, you may enable UseKernelCache to "Yes".
Hmmmm. :)


Regarding backups, you wrote:
Fujitsu DYNA-MO 1300 U2 Magneto-Optical Drive
Note: You can only find the Fujitsu DYNA-MO 1300 U2 on eBay nowadays. For archiving important files, you really can't beat Magneto-Optical discs.
So you have more than one of these drives? I would feel uncomfortable archiving important files on media that I could only access with second-hand gear which I could hopefully find if my only drive failed. It's great to have reliable disks, but they're not the only point of failure. :)


But once again: very impressed!
 
Core i7-3970X Monster SSD Build, Geekbench Score: 20313

Impressive build!

However, I'm a bit confused about your claim:

Are you sure???

MultiBeast is installing the modded kext files for you. Sure it's been made easy for you, but I don't think your assertion is correct.

Hmmmm. :)

Regarding backups, you wrote:
So you have more than one of these drives? I would feel uncomfortable archiving important files on media that I could only access with second-hand gear which I could hopefully find if my only drive failed. It's great to have reliable disks, but they're not the only point of failure. :)

But once again: very impressed!

Heya DBP -

Thanks! :D

Maybe I should have been a bit clearer. :) In regards to the MO part, I meant in reason. I'm sure a 300mw laser or something will damage the discs lol.

In regards to not having to installed modded kext files, perhaps I was a bit stretching, but I meant compared to hackintoshing of a while back. But you're right. I think I should change it to "minimal." As for the the AHCI bit: if people don't want to modify the kext file, they could technically boot with kext caching off hehe.

I have two of those drives actually, one of them is still in a box, but confirmed working. You're absolutely right that the drive can be a point of failure, but my meaning was relative in regards to the media. :)

/ gx
 
Status
Not open for further replies.
Back
Top