Overclock.net banner

Intel Turbo Boost caused freezing/lockups

10K views 48 replies 8 participants last post by  hhuey5 
#1 ·
So Intel Turbo Boost was causing my computer to freeze and lockup under cpu stress, fans would go max RPM, power to usb ports were gone, even prevented me from accessing bios for a bit, did a cmos battery reset and restored defaults, disabled the turbo boost and stress tests were fine.

Rig is 990x, dx58og mobo, 16g ddr3 ballistix (2 sticks of 8), gtx 1070 ftw acx 3.0, tx850w corsair, h60 corsair liquid cooler. Got it from my pops who had it for a few years.

Should this be an indication that this cpu is unable to overclock? Or were the factory settings turbo boost was using wrong? I ask because I can attempt to manually overclock it but just want to make sure that if factory turbo boost settings failed then is that indicative of this cpu in it's last years.
 
#3 ·
#5 ·
Latest version.

I think I nailed the issue, northbridge sits at 60c in bios and jumps to 90c in OS. Got got arctic silver coming in today, going to apply. Noticed the tension springs on the screws holding the heatsink do not decompress, so will pull them out and stretch them out and add additional rubber spacers for tension support. Will have a 120mm fan directed at it, might get a 60mm.

If problem persists, then I'm just going to build a coffee lake setup. 8400. Debating between getting the z370 apro or wait for Intel's H370 in March. Any suggestions?
 
#6 ·
Quote:
Originally Posted by ThrashZone View Post

Hi,
What oc were you using exactly ?
H60 is a pretty small cooler do you monitor your temperatures and what were they ?
http://www.majorgeeks.com/files/details/core_temp.html

http://www.techpowerup.com/downloads/2089/real-temp-3-70/mirrors
The cpu temps are fine. 40-50c under load. The h60 is small but it definitely handles stock factory turbo boosts.

The stock factory turbo boosts were going to 3.74ghz 28/28/27/27/27/27 multipliers at 133bclk. TDP 130w, 1.2v core, low droop, 1.1v IOH (chipset), qpi date rate 6.4GT/s.
 
#7 ·
update: I took the heatsink off, cleaned every little remains of the old TIM, applied arctic silvers, added the spacers, tightened the screws very well, have that fan blowing directly on it at 600 rpm. temps are still the same (90c). It wasn't the heatsink.

Something is causing my chipset to go from 60c to 90c from bios to startup....
 
#9 ·
Quote:
Originally Posted by tsodinson View Post

The cpu temps are fine. 40-50c under load. The h60 is small but it definitely handles stock factory turbo boosts.

The stock factory turbo boosts were going to 3.74ghz 28/28/27/27/27/27 multipliers at 133bclk. TDP 130w, 1.2v core, low droop, 1.1v IOH (chipset), qpi date rate 6.4GT/s.
Hi,
Why are you using 133 bclk ?
 
#10 ·
Quote:
Originally Posted by ThrashZone View Post

Hi,
Why are you using 133 bclk ?
Stock settings. The freezing seems to be caused by higher multipliers under stress as I raised voltages and found no issues, 26x non turbo has no issues. Are you suggesting that I increase bclk to see if I can get a stable overclock? I just tested turbo with same multipliers as non-turbo, 26x. Everything was fine. So this is a problem with the multipliers or the higher frequency. I can test raising the blck to see if I get the same results, if so then it pinpoints the problem to higher frequencies.

Ran intel processor diagnostic tool, gave my cpu the pass.

William, I think you're right and this mobo is giving up. It handled turbo fine before, now it can't. This board is discontinued, I'm not going to go through the headache of the RMA. Tried flashing to older versions of bios didn't work. Just going to sell the 990x to help fund a coffee lake build if a windows reinstall won't help. 8400, msi z370 apro, 2x8gb 2666 ddr4 vengeance lpx.
 
#11 ·
Hi,
No actually it seems very high at 133 :/
I always just use default of 100
If I want a hair more to get just above my multiplier clock I just enter 100.1
That keep say 4.5 a hair above otherwise If I use cpu-z cpu verify the clock speed is just above 4.8 it will show above that say 4806.69
https://valid.x86.fr/myc03b
Otherwise it may show 479?.?? if you see what I'm saying
smile.gif
 
#12 ·
Quote:
Originally Posted by ThrashZone View Post

Hi,
No actually it seems very high at 133 :/
I always just use default of 100
If I want a hair more to get just above my multiplier clock I just enter 100.1
That keep say 4.5 a hair above otherwise If I use cpu-z cpu verify the clock speed is just above 4.8 it will show above that say 4806.69
https://valid.x86.fr/myc03b
Otherwise it may show 479?.?? if you see what I'm saying
smile.gif
the stock setting is 133. you should check what platform he's on.
wink.gif


oh yeah, what i was going to post:

tsodinson, check your power delivery, mobo VRMS. seems the issue is when the cpu needs lots o' power quickly.
 
#13 ·
Quote:
Originally Posted by looniam View Post

tsodinson, check your power delivery, mobo VRMS. seems the issue is when the cpu needs lots o' power quickly.
130w tdp, low droop, 1.2v dynamic (setting is off to let processor manage when not testing,) c1e & c states disabled (when testing they are re-enabled while I'm not turbo boosting), speedstep cannot be disabled when turbo is on, 1.1v IOH (x58 chipset northbridge), vreg temps are 36c in bios can't identify the monitor in HWMonitor.

v12.0 --- 12.125v
v5.0 --- 5.051v
v3.3 --- 3.231
v-memory --- 1.510 (balliatix sport drr3 1600 specs)
v-processor core --- 1.136 (this is without dynamic 1.2v on)

QPI / uncore voltage 1.225v

I tried this angle, I kept the vcore set at 1.2v and everything was fine non-turbo. Are you suggesting that it could a bad capacitor?
 
#14 ·
#15 ·
great u changed out paste
have you given memory test to look for bad memory stick

my 4790K
I had it at 4.3 but had to bring it down to 4.24 after like 2 yrs
around the time I did memory test ; ran prime95 lol my cpu cooler didnt like it
havent re up'd the cpu after rma the memory
 
#16 ·
Quote:
Originally Posted by looniam View Post

i meant more of cpu power management (not speed step! and all that) than voltage settings. having gone from conroe/wolf to ivy not sure where in the bios it would be and i'm sure that intel board has a sparse one. looking at the manual wasn't helpful:
https://www.intel.com/content/dam/support/us/en/documents/boardsandkits/desktop-boards/X58/DX58OG/DX58OG_ProductGuide01.pdf

but it might be a wild goose chase . .
redface.gif
Yeah, I'm getting limited information. We are dealing with an 8 year old bios setup, and my board itself is 5 years old with 3 years of on and off use. Still a great mobo of it's time, should be a wall mount if this doesn't work out.

hhuey5, got a usb drive coming in the mail for this particular reason. Looking forward to running memtest.

Come to think of it, this all really started when I updated my bios to latest to change my memory timing and multipliers. I flash back to the version that was working and put the memory settings back to where they were before (88820 timing, 20 uncore multipliers and 10 memory multipliers which results in 1333) and still had the same issues so I sort of it ruled out (perhaps prematurely as there is most likely a relation.)

Can upping the memory multipliers to it's manufacture recommended specs and the timing on old sticks that have been running on a lower multiplier more or less damage the sticks?

Maybe it's a timing issue? I have no idea how to tune sticks for looser timings. 99924 is where it's at currently.

If the sticks are bad, don't know weather to get new sticks and maybe try triple channel or again go for a coffee lake build.
 
#17 ·
Quote:
Originally Posted by looniam View Post

the stock setting is 133. you should check what platform he's on.
wink.gif


oh yeah, what i was going to post:

tsodinson, check your power delivery, mobo VRMS. seems the issue is when the cpu needs lots o' power quickly.
Hi,
Thanks yeah I saw asrock but dang that's a lot of blck lol
biggrin.gif
 
#18 ·
Quote:
Originally Posted by tsodinson View Post

Yeah, I'm getting limited information. We are dealing with an 8 year old bios setup, and my board itself is 5 years old with 3 years of on and off use. Still a great mobo of it's time, should be a wall mount if this doesn't work out.

hhuey5, got a usb drive coming in the mail for this particular reason. Looking forward to running memtest.

Come to think of it, this all really started when I updated my bios to latest to change my memory timing and multipliers. I flash back to the version that was working and put the memory settings back to where they were before (88820 timing, 20 uncore multipliers and 10 memory multipliers which results in 1333) and still had the same issues so I sort of it ruled out (perhaps prematurely as there is most likely a relation.)

Can upping the memory multipliers to it's manufacture recommended specs and the timing on old sticks that have been running on a lower multiplier more or less damage the sticks?

Maybe it's a timing issue? I have no idea how to tune sticks for looser timings. 99924 is where it's at currently.

If the sticks are bad, don't know weather to get new sticks and maybe try triple channel or again go for a coffee lake build.
I spent 350 for 32gb before it went down to 235 now its back up in 300s

doesnt the memory have lifetime warranty?
mem company needed me to prove its bad stick
I had to keep notes of all my test to prove to them I located the bad stick

i was changing out and shuffling the memory sticks to find that one stick
I was worried - am i wearing out all that gold?

I tested just in XPS and default sweet spot mode until I found the bad stick
with default memory voltages
 
#19 ·
Quote:
Originally Posted by ThrashZone View Post

Quote:
Originally Posted by looniam View Post

the stock setting is 133. you should check what platform he's on.
wink.gif


oh yeah, what i was going to post:

tsodinson, check your power delivery, mobo VRMS. seems the issue is when the cpu needs lots o' power quickly.
Hi,
Thanks yeah I saw asrock but dang that's a lot of blck lol
biggrin.gif
he has an intel board. but that has nothing to do with it. older platforms had much much higher base clocks. coreduos were anywhere from ~350 to 450 with cpu mulitpliers being much lower (x9 - x22).

that changed to ~100mhz when intel got rid of the northbridge and moved it on the chip w/sandy.
 
#20 ·
Quote:
Originally Posted by hhuey5 View Post

I spent 350 for 32gb before it went down to 235 now its back up in 300s

doesnt the memory have lifetime warranty?
mem company needed me to prove its bad stick
I had to keep notes of all my test to prove to them I located the bad stick

i was changing out and shuffling the memory sticks to find that one stick
I was worried - am i wearing out all that gold?

I tested just in XPS and default sweet spot mode until I found the bad stick
with default memory voltages
Yeah just read an article why ram is so expensive now, low supply and waiting for new manufacturing implementations next year. Part of the reason why I'm considering throwing in the towel if it's the ram. 110$ for triple channel 12gb or 150$ for dual channel 16gb ddr3, and won't have assurance that it'll fix the issue. 300$ more I can build a modern setup that'll blow the 990x away. Seems like an odd issue if it's the ram. Can that really be responsible for elevated northbridge temps and/or preventing the cpu from running at higher frequencies? Like I said, there is definitely a relation since this only started when I set the ram timings and frequency to manufacture specs (as it was running off auto.) Then again, I put it all back to default to where it was and still have the same issue.
 
#21 ·
Quote:
Originally Posted by tsodinson View Post

Yeah just read an article why ram is so expensive now, low supply and waiting for new manufacturing implementations next year. Part of the reason why I'm considering throwing in the towel if it's the ram. 110$ for triple channel 12gb or 150$ for dual channel 16gb ddr3, and won't have assurance that it'll fix the issue. 300$ more I can build a modern setup that'll blow the 990x away. Seems like an odd issue if it's the ram. Can that really be responsible for elevated northbridge temps and/or preventing the cpu from running at higher frequencies? Like I said, there is definitely a relation since this only started when I set the ram timings and frequency to manufacture specs (as it was running off auto.) Then again, I put it all back to default to where it was and still have the same issue.
yeah I get that

months before I was bad ram stick hunting
I tried changing cpu speed ... ram speed ... its then when program accessing bad ram ... the computer freezes or reboots
i doubt its the reason for prime95 test to trigger thermal shutdown LOL

sure in your case even if u have life time warranty ... it be question if they still have ddr3 ram in stock
 
#22 ·
Quote:
Originally Posted by hhuey5 View Post

yeah I get that

months before I was bad ram stick hunting
I tried changing cpu speed ... ram speed ... its then when program accessing bad ram ... the computer freezes or reboots
i doubt its the reason for prime95 test to trigger thermal shutdown LOL

sure in your case even if u have life time warranty ... it be question if they still have ddr3 ram in stock
ok you're definitely onto something with the ram. I lowered the ram multipliers from 20 to 18 to get 1333, kept the timings at 99924. Turned on turbo,only bumped it by x1, so 27x multiplier. Stable. Did 28x, unstable, won't start OS. Yeah when that usb gets in tonight, memtest will reveal whats going on. Increasing the frequency to 1600 by doing a 20x multiplier must've did something to the hardware. I'll see if moving them to a different channel will help as well.

Northbridge still in the low 90c but I'm just going to accept this.
 
#23 ·
Quote:
Originally Posted by tsodinson View Post

Quote:
Originally Posted by hhuey5 View Post

yeah I get that

months before I was bad ram stick hunting
I tried changing cpu speed ... ram speed ... its then when program accessing bad ram ... the computer freezes or reboots
i doubt its the reason for prime95 test to trigger thermal shutdown LOL

sure in your case even if u have life time warranty ... it be question if they still have ddr3 ram in stock
ok you're definitely onto something with the ram. I lowered the ram multipliers from 20 to 18 to get 1333, kept the timings at 99924. Turned on turbo,only bumped it by x1, so 27x multiplier. Stable. Did 28x, unstable, won't start OS. Yeah when that usb gets in tonight, memtest will reveal whats going on. Increasing the frequency to 1600 by doing a 20x multiplier must've did something to the hardware. I'll see if moving them to a different channel will help as well.
Northbridge still in the low 90c but I'm just going to accept this.
ya know i found this before, real similar to yours and it mentions "living with" that toasty northbridge.
http://www.overclock.net/t/1381718/overclocking-on-am-intel-dx58og/0_50
 
#24 ·
Quote:
Originally Posted by looniam View Post

ya know i found this before, real similar to yours and it mentions "living with" that toasty northbridge.
http://www.overclock.net/t/1381718/overclocking-on-am-intel-dx58og/0_50
Wow, word for word issue. Even his max stable boost/OC was identical to mine, 3.6 And fails under the same conditions. Like mass effect andromeda and battlefront 2 are cpu/memory intensive during loading and freeze the system. But when I do turbo boost for doom or shadows of war, smooth as silk.

Seems it is a bad stick/s. Set to 1333 in bios then ran the tests and passed for an hour and half and stopped because I was curious about 1600. Changed to 1600 and did not pass. Something like 6500+ errors on test 9 (first test I wanted to try) and then memtest got stuck like a system hang.

I will contact crucial to see if they'll cover it, doubt it. If so, great. If not, coffee lake it is. Debating about a case, the corsair carbide 270r has my eye.

Thanks hhuey5, I should've suspected the ram at first
jealoussmiley.gif
 
#26 ·
Quote:
Originally Posted by hhuey5 View Post

your welcome
for me it took over year before I ran into the memory issues and these where new memory
its was multiple ***? moments ' i thought it was cpu LOL

now I has to test new ddr4 memory on new build
dont want to get burnt again LOL
Yeah, will do the same. What frequency did you get? I'm getting conflicting reviews that 2400-2600 is the sweet spot, 3000-3400 is the sweet spot, coffee lake benefits more with 3k+, etc...
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top