Core I7, shoudl hyperthreading be disable ?

Sojuuk said:
does turning off HT lower the temperatures enough to overclock significantly more?
Click to expand...

In most cases turning off HT does enable you to reach higher overclocks. For gaming this may be of some consequence, but I think that if you do other things with your PC then you'll lose more by turning off HT than you would be leaving it on at a lower clock speed.
 
Dan D said:
The only reason to turn Hyperthreading off on any Core i7 system is if you are a benchmark whore. The "negative impact" that HT has on any game's performance is negligible. The difference isn't something you are likely to feel in games or even see. Only people interested in benchmarking performance will see a difference. I think the performance gain in applications offsets any performance lost that might be experienced in games. Even with HT on the performance in games is unmatched by anything else.
Click to expand...

You're not taking into account overclocking though. Having HT on is better than having HT off at the same clock speed, obviously. However, if turning off HT reduces your load temps by 5-10C and thus allows you to increase BCLK by like 10MHz, then you end up with higher clock but HT off vs lower clock with HT on.

It's that situation where I think having HT off is probably better for gaming, not that I'd know.
 
Dan_D said:
In most cases turning off HT does enable you to reach higher overclocks. For gaming this may be of some consequence, but I think that if you do other things with your PC then you'll lose more by turning off HT than you would be leaving it on at a lower clock speed.
Click to expand...

DragonQ said:
You're not taking into account overclocking though. Having HT on is better than having HT off at the same clock speed, obviously. However, if turning off HT reduces your load temps by 5-10C and thus allows you to increase BCLK by like 10MHz, then you end up with higher clock but HT off vs lower clock with HT on.

It's that situation where I think having HT off is probably better for gaming, not that I'd know.
Click to expand...

I addressed this in my above post. I don't believe that gains in games will be that significant from a 10MHz BCLK increase. When you are already at 4.2GHz, a 200MHz increase to 4.4GHz won't change the fact that you are most likely severely GPU limited in most games. For the guys running high end multi-GPU configurations at resolutions above 1920x1200, there may be an increase on a technical level but my experience tells me that even in those situations the difference will be negligible. So while you may be technically correct, I think that if you are already pushing a 4GHz clock speed, anything you do (increase BCLK, CPU clock, turn off HT, turn on HT) will have a minimal impact at the most.

It isn't as if a game that is unplayable will suddenly become playable because you turned off HT on your CPU and raised the CPU clock speed 200MHz. Again, if you are a benchmark whore there may be some difference in the numbers, but in the real world I think your efforts will produce meaningless results.
 
Yes, 10-15C on average. When you consider that good air cooling on OC'd i7's, load temps run 70-80C with HT off, HT on will cause you to BSOD. I'm talking LinX testing here, not P95, which won't max out your heat.
 
rendering in 3dsMAX/vray is faster with hyperthreading. i am really liking this i7, it beat our dual xeons at work with the same file.
 
MassiveOverkill said:
Yes, 10-15C on average. When you consider that good air cooling on OC'd i7's, load temps run 70-80C with HT off, HT on will cause you to BSOD. I'm talking LinX testing here, not P95, which won't max out your heat.
Click to expand...
It isn't the fault of HT when your system throws a BSOD. The problem is instability caused by the overclock.
 
rflcptr said:
Heat from HT on top of heat from the overclock...which do you think is the greater contributer?
Click to expand...

Dude, do you even have an i7?

My 3.8 Ghz OC is fully Linx stable. It's not until I turn on HT that my system crashes and that's only when Real Temp spikes above 85C. EVERYONE who KNOWS about I7 KNOWS that HT creates a crap load more heat. Now.........If my system maxes out at 80-85C at full load with HT off, and HT adds another 10-15C to your max load temps and 100C is the max these processors are rated for, what do YOU think is the reason for the unstability with HT turned on?

Why don't you read the article Kyle did about the I7 and temps?
 
I have an i7 920 D0 @ 4GHz w/ HT on, on Air Cooling.... And its stable. My temps are like 80C full load.
 
USMC2Hard4U said:
I have an i7 920 D0 @ 4GHz w/ HT on, on Air Cooling.... And its stable. My temps are like 80C full load.
Click to expand...

Similar to me, except mine is 3.9GHz except when it goes to 22x for a core or two. What's your VCore? Mine's at 1.224V, lowest it'll go.
 
MassiveOverkill said:
Dude, do you even have an i7?
Click to expand...
Has nothing to do with this?
My 3.8 Ghz OC is fully Linx stable. It's not until I turn on HT that my system crashes and that's only when Real Temp spikes above 85C. EVERYONE who KNOWS about I7 KNOWS that HT creates a crap load more heat. Now.........If my system maxes out at 80-85C at full load with HT off, and HT adds another 10-15C to your max load temps and 100C is the max these processors are rated for, what do YOU think is the reason for the unstability with HT turned on?
Click to expand...
Yes, we know that HT creates more heat than when it is disabled. But since the default state for HT is it being activated, overclocking is the reason for the BSOD. Disabling HT and preventing the BSOD doesn't mean HT was at fault. It means you've disabled a prominent feature of the CPU to alleviate heat output.
 
rflcptr said:
Has nothing to do with this?

Yes, we know that HT creates more heat than when it is disabled. But since the default state for HT is it being activated, overclocking is the reason for the BSOD. Disabling HT and preventing the BSOD doesn't mean HT was at fault. It means you've disabled a prominent feature of the CPU to alleviate heat output.
Click to expand...

It most certainly does. I'm done talking to you. When you do own one, I'll more than happy to debate with you.
 
MassiveOverkill said:
It most certainly does. I'm done talking to you. When you do own one, I'll more than happy to debate with you.
Click to expand...
What does ownership have to do with being able to discuss this? Or can you not think of a meaningful response?
 
MassiveOverkill said:
What's your voltage? D0's are known to run higher OC's with lower voltage, Are those temps with P95 or LinX?
Click to expand...

DragonQ said:
Similar to me, except mine is 3.9GHz except when it goes to 22x for a core or two. What's your VCore? Mine's at 1.224V, lowest it'll go.
Click to expand...

I am at 1.325 setting in the Bios, but 1.29 is what it actually reports.

My Uncore is at 3.6GHz, so my Voltage needs to remain here otherwise its unstable.

And the Temp is 24 hrs with LinX @ 80 - 81C
 
MassiveOverkill said:
My 3.8 Ghz OC is fully Linx stable. It's not until I turn on HT that my system crashes and that's only when Real Temp spikes above 85C. EVERYONE who KNOWS about I7 KNOWS that HT creates a crap load more heat. Now.........If my system maxes out at 80-85C at full load with HT off, and HT adds another 10-15C to your max load temps and 100C is the max these processors are rated for, what do YOU think is the reason for the unstability with HT turned on?

Why don't you read the article Kyle did about the I7 and temps?
Click to expand...
Reaching the Tjmax temperature will cause your CPU to throttle its speed. It will not cause a BSOD. If you had a BSOD, it means that your CPU was not stable with HT enabled. HT makes your CPU do more work and puts it under more stress (hence the higher temps), so your system simply wasn't stable under that kind of load. It had nothing to do with the heat.
 
Zero82z said:
Reaching the Tjmax temperature will cause your CPU to throttle its speed. It will not cause a BSOD. If you had a BSOD, it means that your CPU was not stable with HT enabled. HT makes your CPU do more work and puts it under more stress (hence the higher temps), so your system simply wasn't stable under that kind of load. It had nothing to do with the heat.
Click to expand...

So go ahead and remove the heat sink from your Quad Core and boot your computer up and let it run all day. The thermal throttling should save your CPU, I promise.
 
Here's an easier to read table from the Xbit article linked earlier, that includes a whole list of games / apps and the effect of HT on them.

http://www.xbitlabs.com/articles/cpu/display/intel-core-i7_8.html#sect0


Since I game a lot, HT is off. Xbit and other sites have noted WORSE game performance with HT on.
Also, my highest stable overclock was 200 mhz slower with HT enabled, as it added 15 deg C when LInX testing for an hour or more.
 
USMC2Hard4U said:
Yeah is that bad? I ran it for 24 hrs, it was like 500 runs or w/e it just kept doing it for all day until I stopped it 1 day later.
Click to expand...

Can someone else with HT on run at the 4.0 Ghz range run 1 pass of LinX for us?

It takes me 10 minutes to run 1 pass of LinX. 10 minutes x 500 passes = 5000 minutes. 5000 minutes / (60 minutes x 24 hours) = 83.33_ hours or 3.472_ days
 
DragonQ said:
At what problem size? My 3.9GHz i7 does the default 11530 test in ~22.4s.
Click to expand...

Well I do have 12 gigs of RAM, I'll try it with only 6 gig problem size.

Yep, only took half the time, 5 minutes..........both were the 64-bit test btw.
 
Last edited:
MassiveOverkill said:
Can someone else with HT on run at the 4.0 Ghz range run 1 pass of LinX for us?

It takes me 10 minutes to run 1 pass of LinX. 10 minutes x 500 passes = 5000 minutes. 5000 minutes / (60 minutes x 24 hours) = 83.33_ hours or 3.472_ days
Click to expand...

It takes me 212 seconds to run 1 pass of LinX @ 4GHz w/ HT (3.6GHz Uncore/DDR3 1600 7-7-7).... Whatever that is in minutes.
 
Makes sense.........you all really should be maxxing out your available memory size when doing stability testing though. 64 bit test should be done, but I can understand if you don't have a 64-bit OS.

This could also explain why some stable overclocks are higher than others.
 
I have Windows 7 x64 but I'm using XP at the moment cos I haven't installed Office on 7 yet and I'm doing a report at the moment. :)

I tried maxing out the memory once in Intel Burn Test and I just got some DLL error complaining about memory being moved or something. Nothing on the internet about it. :/
 
Yes, I am using the latest version of LinX now (0.6 I believe?). That's the one that took 22.4s.
 
MassiveOverkill said:
Makes sense.........you all really should be maxxing out your available memory size when doing stability testing though. 64 bit test should be done, but I can understand if you don't have a 64-bit OS.

This could also explain why some stable overclocks are higher than others.
Click to expand...

I am, Max availabily which is usualy about 5000MBs
 
Dan_D said:
The only reason to turn Hyperthreading off on any Core i7 system is if you are a benchmark whore. The "negative impact" that HT has on any game's performance is negligible. The difference isn't something you are likely to feel in games or even see. Only people interested in benchmarking performance will see a difference. I think the performance gain in applications offsets any performance lost that might be experienced in games. Even with HT on the performance in games is unmatched by anything else.
Click to expand...

This.

/end thread
 
MassiveOverkill said:
So go ahead and remove the heat sink from your Quad Core and boot your computer up and let it run all day. The thermal throttling should save your CPU, I promise.
Click to expand...
Taking off the heatsink would cause the CPU to reach the thermal shutoff point and turn off. However, that is a completely different situation. You said that HT raises your temps by 15C. What's 85C + 15C? 100C. Which is below the thermal shutoff point. So the thermal protection (aside from the throttling that occurs at about 97C) wouldn't kick in. By the way, I've run my CPU all the way up to 100C without crashing it, and although it's not an i7 and doesn't have HT, the thermal protection built into 45nm C2Qs is exactly the same system as the one built into the i7 chips. So if my CPU didn't BSOD at 100C and yours did, it means that the temperature was not the cause. So go ahead and actually read the Intel datasheets like I have before being obnoxious and making more assumptions.
 
Last edited:
Zero82z said:
Taking off the heatsink would cause the CPU to reach the thermal shutoff point and turn off. However, that is a completely different situation. You said that HT raises your temps by 15C. What's 85C + 15C? 100C. Which is below the thermal shutoff point. So the thermal protection (aside from the throttling that occurs at about 97C) wouldn't kick in. By the way, I've run my CPU all the way up to 100C without crashing it, and although it's not an i7 and doesn't have HT, the thermal protection built into 45nm C2Qs is exactly the same system as the one built into the i7 chips. So if my CPU didn't BSOD at 100C and yours did, it means that the temperature was not the cause. So go ahead and actually read the Intel datasheets like I have before being obnoxious and making more assumptions.
Click to expand...

Just realized you don't have an i7 either, and I'm done talking to you as well ;)
 
MassiveOverkill said:
Just realized you don't have an i7 either, and I'm done talking to you as well ;)
Click to expand...
Like I explained already, the thermal management system built into the i7 CPUs is basically identical to the one in 45nm Core 2 CPUs. And I may not own an i7, but I apparently understand more about them than you do.
 
I'm under the impression that HT was badass when there were only single core CPU's, hell I had a HT p4 and it was wicked. With 2 or more cores I don't really see a point in it.
 
bigdogchris said:
I'm under the impression that HT was badass when there were only single core CPU's, hell I had a HT p4 and it was wicked. With 2 or more cores I don't really see a point in it.
Click to expand...

For desktops the applications for more than 4 cores and hyperthreading are limited. Outside of video encoding there isn't much you can do at home to really make use of that many threads. This will change as software becomes more able to utilize more simultaneous threads. With professional workstations used for 3D applications or in the server environment where virtualization can be leveraged the added thread capabilities are extremely valuable.
 
Dan_D said:
For desktops the applications for more than 4 cores and hyperthreading are limited. Outside of video encoding there isn't much you can do at home to really make use of that many threads. This will change as software becomes more able to utilize more simultaneous threads. With professional workstations used for 3D applications or in the server environment where virtualization can be leveraged the added thread capabilities are extremely valuable.
Click to expand...
Anyone who runs a lot of VMs and does a lot of multitasking will benefit a lot from HT. I have two VMs running for Folding@Home and I could definitely use the extra threads that an i7 would provide me with. When I'm running a lot of programs my CPU definitely gets bogged down, and it's running a healthy 1GHz overclock.
 
Zero82z said:
Anyone who runs a lot of VMs and does a lot of multitasking will benefit a lot from HT. I have two VMs running for Folding@Home and I could definitely use the extra threads that an i7 would provide me with.
Click to expand...

I too use VM's at home for testing and for work purposes. Aside from that I wouldn't benefit from it too much. Though I do multi-task fairly heavily, nothing I use is normally all that demanding outside of gaming.
 
You must log in or register to reply here.
Back
Top