Personal computing discussed

Moderators: Flying Fox, morphine

 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

This G3258 goes to 1.7V

Fri Aug 04, 2017 1:53 am

No actually useful hardware was harmed in the making of this post (probably). The mobo already killed a 4690K, and damaged this G3258 enough to make it clear it isn't worth saving. So what do I do with hardware already destined for the garbage? See just how fast it'll go, of course.

Once, this board/CPU combo would do 4.3 at 1.33V. That degraded quickly to more like 4.1 at 1.36V, at which point I started doing manual clock control and not letting it run over a volt unless necessary, which seems to have halted its degradation well. Those numbers are with all the power management and miscellaneous settings at stock/efficient/practical values. For this, I disabled one core, disabled all the power-saving features and safeties, boosted lots of miscellaneous voltages by 0.05V, threw most of the switches ASRock said might help with overclocking, and clocked the uncore and RAM at 3.2 and 1600 without turning down their voltages. Testing was 3-5 minutes of prime95 in-place large FFTs, because I didn't want to spend all day on this.

freq   Vcore   Vin   status
4.4   1.40   1.90   pass
4.5   1.40   1.90   pass
4.6   1.40   1.90   pass
4.7   1.40   1.90   epic fail, had to use CMOS clear button
4.7   1.45   1.95   failed at Gnome initialization
4.7   1.50   2.00   pass
4.8   1.57   2.07   failed at early boot
4.8   1.60   2.10   failed at Gnome initialization
4.8   1.63   2.13   failed at early boot
4.8   1.66   2.16   failed at Gnome initialization
4.8   1.70   2.20   failed at Gnome initialization

I called it at that point because I don't want a literal halt-and-catch-fire, and that RAM is still useful and in the potential blast radius of such high-voltage shenanigans. I also wasn't stopping to check thermals, but I disabled that limiter and extrapolation says that if I got prime95 to run at 1.75+V it might well break 100C.

I was testing with a stock Antergos image, hence the Gnome. The real kicker here is that on 1C1T at 4.7 GHz with all the power management disabled, Gnome *finally* runs passably fast. :lol:
 
Noinoi
Gerbil Team Leader
Posts: 223
Joined: Fri Jun 26, 2015 11:31 pm
Location: Somewhere, anywhere!

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 2:27 am

There's probably a story to the motherboard killing CPUs despite normal nominal voltages... did you ever find out the cause?

I do wonder if it's theoretically possible to instantly ruin a CPU by extreme overvolting. My Z97 board allows me to set the voltage to 1.92V, throwing caution to the wind, I'd rather not do that... Or would something else stop it?
i5-4590 | Kingston 2x8GB | Asus Strix GTX 970 | Asus Z97-Pro Gamer | Kingston Fury 240GB + WD Black 2TB + Blue 2TB | Win 10 FCU
 
Chrispy_
Gold subscriber
Maximum Gerbil
Posts: 4025
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 5:29 am

I've just thrown out a 4790K this week (was running at stock) but the memory controller can no longer power four DIMM slots and now requires extra voltage at stock to stay stable for >24H of 100% load doing software renders. I assumed the instability was board, PSU or RAM but after replacing all those to no avail I swapped the CPUs in two identical PCs and the problem stayed with that CPU.

The Ryzen 7 that replaced it is already munching through tasks almost twice as fast, but I am curious why the CPU died. I almost never encounter dead CPUs, especially when they've been run at stock voltages, so I'd be curious to hear your thoughts....
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 6:21 am

synthtel2 wrote:
I called it at that point because I don't want a literal halt-and-catch-fire, and that RAM is still useful and in the potential blast radius of such high-voltage shenanigans. I also wasn't stopping to check thermals, but I disabled that limiter and extrapolation says that if I got prime95 to run at 1.75+V it might well break 100C.

Hey, I accidentally ran the Phenom 9550 in my web server at 100C for a while (due to an unnoticed dead CPU fan). It suffered no discernible long-term damage. After replacing the fan, it went on to run for years after that, until I retired it.
Nostalgia isn't what it used to be.
 
Duct Tape Dude
Gold subscriber
Gerbil Elite
Posts: 565
Joined: Thu May 02, 2013 12:37 pm

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 6:56 am

just brew it! wrote:
Hey, I accidentally ran the Phenom 9550 in my web server at 100C for a while (due to an unnoticed dead CPU fan). It suffered no discernible long-term damage. After replacing the fan, it went on to run for years after that, until I retired it.
I think heat death is grossly overhyped. I've run a GTX 260m at 110C-120C for years of gaming sessions and everything still works fine. Voltage, however... I know that slowly electromigrates your chip into retirement. 1.7V seems insane for any duration.
 
derFunkenstein
Gold subscriber
Gerbil God
Posts: 23630
Joined: Fri Feb 21, 2003 9:13 pm
Location: Comin' to you directly from the Mothership

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 8:21 am

Heat death was a real thing back in 1999, since you could fry Athlons. Since then, AMD has learned its lesson and the CPU is much more resilient.
"And and if you start to bleed, stop wiping." -whm1974
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 8:38 am

derFunkenstein wrote:
Heat death was a real thing back in 1999, since you could fry Athlons. Since then, AMD has learned its lesson and the CPU is much more resilient.

I was still surprised that the Phenom didn't start throttling before it hit 100C.
Nostalgia isn't what it used to be.
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1980
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 10:08 am

Duct Tape Dude wrote:
just brew it! wrote:
Hey, I accidentally ran the Phenom 9550 in my web server at 100C for a while (due to an unnoticed dead CPU fan). It suffered no discernible long-term damage. After replacing the fan, it went on to run for years after that, until I retired it.
I think heat death is grossly overhyped. I've run a GTX 260m at 110C-120C for years of gaming sessions and everything still works fine. Voltage, however... I know that slowly electromigrates your chip into retirement. 1.7V seems insane for any duration.

Agreed.

People love to complain about "hot" chips...but if they aren't throttling, they'll live essentially forever.
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | Samsung 4K 40" | 480 + 240 + LSI 9207-8i (128x8) SSDs
 
Duct Tape Dude
Gold subscriber
Gerbil Elite
Posts: 565
Joined: Thu May 02, 2013 12:37 pm

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 10:38 am

derFunkenstein wrote:
Heat death was a real thing back in 1999, since you could fry Athlons. Since then, AMD has learned its lesson and the CPU is much more resilient.
I thought that was because of a lack of thermal control. Today's chips have several throttling temperatures and a shutoff temperature as a last resort.
 
Duct Tape Dude
Gold subscriber
Gerbil Elite
Posts: 565
Joined: Thu May 02, 2013 12:37 pm

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 10:39 am

Waco wrote:
Agreed.

People love to complain about "hot" chips...but if they aren't throttling, they'll live essentially forever.
Right, and even if they throttle, if the chip can stay below the shutoff temp it's probably fine.
EDIT: Sorry for the double post, hit the wrong button :(
 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

Re: This G3258 goes to 1.7V

Fri Aug 04, 2017 2:48 pm

Noinoi wrote:
There's probably a story to the motherboard killing CPUs despite normal nominal voltages... did you ever find out the cause?

I still have no idea what's actually wrong with the motherboard. I'm just thankful this G3258 didn't wreak any software-side havoc on its way out like the 4690K did.

Chrispy_ wrote:
I've just thrown out a 4790K this week (was running at stock) but the memory controller can no longer power four DIMM slots and now requires extra voltage at stock to stay stable for >24H of 100% load doing software renders. I assumed the instability was board, PSU or RAM but after replacing all those to no avail I swapped the CPUs in two identical PCs and the problem stayed with that CPU.

The Ryzen 7 that replaced it is already munching through tasks almost twice as fast, but I am curious why the CPU died. I almost never encounter dead CPUs, especially when they've been run at stock voltages, so I'd be curious to hear your thoughts....

At stock? Huh. Did it spend most of its ilfe under heavy load, like that rendering? That would definitely wear some stuff faster than a typical desktop duty cycle of <25%, but that is still weird.

just brew it! wrote:
Hey, I accidentally ran the Phenom 9550 in my web server at 100C for a while (due to an unnoticed dead CPU fan). It suffered no discernible long-term damage. After replacing the fan, it went on to run for years after that, until I retired it.
Duct Tape Dude wrote:
I think heat death is grossly overhyped. I've run a GTX 260m at 110C-120C for years of gaming sessions and everything still works fine. Voltage, however... I know that slowly electromigrates your chip into retirement. 1.7V seems insane for any duration.
Waco wrote:
Agreed.

People love to complain about "hot" chips...but if they aren't throttling, they'll live essentially forever.

Heat and voltage aren't independent causes of wear. AFAICT, every 10C increase basically doubles whatever wear would be caused by voltage/current. See also how LN2 allows over two volts, which would be insta-frying stuff on normal cooling. 100C at normal voltage is fine, and 1.5V at 50C would probably be mostly fine, but 1.7V at 105C is just :o :o :o .
 
Chrispy_
Gold subscriber
Maximum Gerbil
Posts: 4025
Joined: Fri Apr 09, 2004 3:49 pm
Location: Europe, most frequently London.

Re: This G3258 goes to 1.7V

Sat Aug 05, 2017 9:06 am

synthtel2 wrote:
At stock? Huh. Did it spend most of its ilfe under heavy load, like that rendering? That would definitely wear some stuff faster than a typical desktop duty cycle of <25%, but that is still weird.


Oh, absolutely. It was a number-cruncher for VRay software rendering. I doubt it even had enough downtime to apply OS updates. 100% across 8 threads 24/7/365 since installed four years ago or whenever it was. Still, it was in a server room with conditioned UPS power and dedicated AC so it was running at stock in ideal conditions.

It was one of a 16-node farm so there are 15 other identical machines from the same cabinet running the same work at the same time for the same duration under the same conditions. I've just never had a CPU die under these conditions before (I think that room's farm count is up to 72 mixed nodes in total now) so I thought it was worthy of mention.
Congratulations, you've noticed that this year's signature is based on outdated internet memes; CLICK HERE NOW to experience this unforgettable phenomenon. This sentence is just filler and as irrelevant as my signature.
 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

Re: This G3258 goes to 1.7V

Sat Aug 05, 2017 11:34 pm

All that comes to mind is Intel's statement that "Long term reliability cannot be assured unless all the Low-Power Idle States are enabled", implying that they're not designed for 100% duty cycle anymore. That is a pretty obnoxious situation though, especially for applications like a render farm.

It definitely ties into my worries about high factory voltages on modern CPUs. If that can happen in this short timespan to a 4790K, what about a 6700K on a smaller process pushing 0.1V more?
 
Noinoi
Gerbil Team Leader
Posts: 223
Joined: Fri Jun 26, 2015 11:31 pm
Location: Somewhere, anywhere!

Re: This G3258 goes to 1.7V

Sat Aug 05, 2017 11:52 pm

synthtel2 wrote:
All that comes to mind is Intel's statement that "Long term reliability cannot be assured unless all the Low-Power Idle States are enabled", implying that they're not designed for 100% duty cycle anymore. That is a pretty obnoxious situation though, especially for applications like a render farm.

It definitely ties into my worries about high factory voltages on modern CPUs. If that can happen in this short timespan to a 4790K, what about a 6700K on a smaller process pushing 0.1V more?


I wouldn't be too worried: that 4790K was one out of many in a farm. Perhaps the CPU was a marginal die that passed QC but isn't exactly long for the world.
i5-4590 | Kingston 2x8GB | Asus Strix GTX 970 | Asus Z97-Pro Gamer | Kingston Fury 240GB + WD Black 2TB + Blue 2TB | Win 10 FCU
 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 12:47 am

It mostly isn't Haswells I'm worried about. What I am worried about is that five years ago, wear-related CPU deaths hardly happened unless you were pushing the overclocks very hard. A (subjectively / IMHO) much greater proportion of moderately-OCed Sandys have died before they outlived their usefulness, Haswell seems to be on track for as much or more of that despite typical OC voltages lower than Sandy's, and 14nm often brings stock voltages up to levels that only OCers (and fairly heavy OCers at that) would push through Haswell. It's a bad trend.
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1980
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 2:05 am

synthtel2 wrote:
All that comes to mind is Intel's statement that "Long term reliability cannot be assured unless all the Low-Power Idle States are enabled", implying that they're not designed for 100% duty cycle anymore. That is a pretty obnoxious situation though, especially for applications like a render farm.

It definitely ties into my worries about high factory voltages on modern CPUs. If that can happen in this short timespan to a 4790K, what about a 6700K on a smaller process pushing 0.1V more?

They warranty them for 24/7 full load on the Xeon side of the house.
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | Samsung 4K 40" | 480 + 240 + LSI 9207-8i (128x8) SSDs
 
Ryu Connor
Gold subscriber
Global Moderator
Posts: 4204
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 3:03 am

The sample size is a million in total, but certain studies required narrowing down to hundreds of thousands of machines. Regardless, a sufficient sample size to avoid P hacking problems. This is all laid out within the study, but nobody is gonna actually read that. So I'll just summarize the interesting points here.

Defective CPUs aren't all that uncommon.

4.1 CPU subsystem (MCE) failures
As shown in the first row of Figure 2, machines that accumulate at least 5 days of TACT (Total Accumulated CPU Time) have a 1 in 330 chance of crashing due to an MCE during the 8 month observation period. After a machine has crashed once, its crash probability increases by a factor of 100, and the probability continues to increase with subsequent crashes. The second row in the figure shows that the same trend holds for machines with at least 30 days of TACT, but the initial probability of failure is higher. Further analysis shows that, of the machines with at least 5 days of TACT that experience a recurring crash from an MCE, 84% of machines experience a recurrence within 10 days of TACT, and 97% of machines experience a recurrence within a month.


Overlocking wrecks CPUs. An overclock is 5% or more outside of rated speed. AMD and Intel, in 2008 (the study was for 8 months in 2008) had a huge difference in how resilient their CPUs were to failure. Who was better? The study refuses to name names.

5.1.2 Overclocking analysis
Figure 3 shows the impact of an overclocked device on the probability that a machine will crash from a machine check exception. For brevity, results are shown only for machines with at least 5 days of TACT; similar trends hold for machines with at least 30 days of TACT. We have divided the analysis between two CPU vendors, labeled “Vendor A” and “Vendor B.” The table shows that CPUs from Vendor A are nearly 20x as likely to crash a machine during the 8 month observation period when they are overclocked, and CPUs from Vendor B are over 4x as likely. After a failure occurs, all machines, irrespective of CPU vendor or overclocking, are significantly more likely to crash from additional machine check exceptions.


Underclocking greatly increase stability.

5.2 Effects of underclocking
As noted in Section 5.1.1, CPUs that are not explicitly overclocked by users still run slightly above or below their rated frequency. We were interested to see whether there was any variance in reliability among non-overclocked machines. Therefore, we further partitioned the non-overclocked machines into underclocked machines, which run below their rated frequency (65% of machines), and rated machines, which run at or no more than 0.5% above their rated frequency (32% of machines). As shown in Figure 5, underclocked machines are between 39% and 80% less likely to crash during the 8 month observation period than machines with CPUs running at their rated frequency.
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 3:13 am

Waco wrote:
They warranty them for 24/7 full load on the Xeon side of the house.

They also clearly aren't willing to clock Xeons quite as high as they clock consumer parts, despite the money they charge for top Xeon bins being more than enough to get the very best dies. Everyone calls market segmentation BS on that, but I think there might be a more concrete reason.

If the E3-1280v6 clocked like a 7700K, gamers and other non-professionals still wouldn't be the ones buying it. It'd basically be ECC versus overclocking, with ECC carrying a $270 premium. Still a pretty raw deal, if you ask me. Power use shouldn't explain that either. If you're paying that kind of money for the very fastest E3, is another 20W really likely to stop you? As is, nobody's paying the premium for that part (and the volume has got to be so low Intel hardly cares either), because there's no point. If they could make something worthy of that premium, I think they would.
 
Froz
Gerbil
Posts: 38
Joined: Tue Mar 04, 2014 7:24 am

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 4:22 am

synthtel2 wrote:
Heat and voltage aren't independent causes of wear. AFAICT, every 10C increase basically doubles whatever wear would be caused by voltage/current. See also how LN2 allows over two volts, which would be insta-frying stuff on normal cooling. 100C at normal voltage is fine, and 1.5V at 50C would probably be mostly fine, but 1.7V at 105C is just :o :o :o .

Hmm, is it really just the temperature itself? I recall reading somewhere (sorry, I couldn't find the source) that the damage is done mostly by change in temperature. So 24/7 constant 100% load might actually be less damaging than typical desktop usage which jumps between idle and full load all the time and so has many and big changes in temperature.
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 6:29 am

Froz wrote:
synthtel2 wrote:
Heat and voltage aren't independent causes of wear. AFAICT, every 10C increase basically doubles whatever wear would be caused by voltage/current. See also how LN2 allows over two volts, which would be insta-frying stuff on normal cooling. 100C at normal voltage is fine, and 1.5V at 50C would probably be mostly fine, but 1.7V at 105C is just :o :o :o .

Hmm, is it really just the temperature itself? I recall reading somewhere (sorry, I couldn't find the source) that the damage is done mostly by change in temperature. So 24/7 constant 100% load might actually be less damaging than typical desktop usage which jumps between idle and full load all the time and so has many and big changes in temperature.

I don't think there's a clear answer to that. Cycling will indeed be more likely to cause failures due to thermal expansion/contraction, but electromigration is affected just by temperature and voltage (not cycling).
Nostalgia isn't what it used to be.
 
jihadjoe
Gerbil Elite
Posts: 597
Joined: Mon Dec 06, 2010 11:34 am

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 7:12 am

I just wanna say you are a brave dude. I wouldn't have tried that even on 65nm.
 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 1:49 pm

Froz wrote:
Hmm, is it really just the temperature itself? I recall reading somewhere (sorry, I couldn't find the source) that the damage is done mostly by change in temperature. So 24/7 constant 100% load might actually be less damaging than typical desktop usage which jumps between idle and full load all the time and so has many and big changes in temperature.

As JBI said, cycling has got to be an independent cause, but I still wouldn't want to go 30C -> 100C -> 30C constantly with a soldered IHS. Fortunately for those of us with soldered IHSes, solder makes it easy to keep the temperature swing to half that. 8)

More than just overall power use, Intel does seem to have much higher power density (especially under AVX load). It seems plausible enough to me that their use of TIM is related to that, and if so, it's a shame that the fix for that problem makes the electromigration etc problem so much more difficult.

jihadjoe wrote:
I just wanna say you are a brave dude. I wouldn't have tried that even on 65nm.

The hardware no longer needing to last beyond the next line on the OC log made it easier. :wink:
 
DrDominodog51
Gerbil Elite
Posts: 655
Joined: Sat Jan 17, 2015 12:23 pm
Location: Silicon Valley

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 4:28 pm

When I was initially overclocking my z97 system, I misread 1.35V as 3.5V.

I booted my Pentium AE with 2V vCore and 2.4V VCCIN at 4.4 GHz and ran it for ~40 minutes before I smelt the motherboard burning.

Both the Pentium AE and motherboard survived this. The Pentium continued to run at 4.4 GHz with around 1.35V vCore until I replaced it a year later with a 4690k.
 
synthtel2
Gold subscriber
Gerbil Elite
Topic Author
Posts: 603
Joined: Mon Nov 16, 2015 10:30 am

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 6:14 pm

Now *that's* impressive. That's the same board you're still using now? Seems like power limiters might have saved the day, but even with them, that's some fine luck.
 
DrDominodog51
Gerbil Elite
Posts: 655
Joined: Sat Jan 17, 2015 12:23 pm
Location: Silicon Valley

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 7:50 pm

synthtel2 wrote:
Now *that's* impressive. That's the same board you're still using now? Seems like power limiters might have saved the day, but even with them, that's some fine luck.

Yip. I'm still using the same z97 Pro4 although I hate it a lot more these days.
 
Forge
Lord High Gerbil
Posts: 8217
Joined: Wed Dec 26, 2001 7:00 pm
Location: SouthEast PA

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 11:04 pm

derFunkenstein wrote:
Heat death was a real thing back in 1999, since you could fry Athlons. Since then, AMD has learned its lesson and the CPU is much more resilient.


Athlon MP 1.2GHz, ran with a failed fan for between 2 and 3 months at 120-140C throughout. Only noticed when smell of carbonizing dust became notable. Fan later replaced, and CPU continued for several years of issue-free useful work.

I flatly deny the THG-inspired stories of heat death! Barring removal of the heatsink, Athlons simply did not cook to death! If removing the heatsink was allowed, there are LOTS of P3s and P4s that would also fry to death, thermistor or no.
Behemoth - Plex server
Galactica - FreeNAS
Pegasus - R710 ESX host
fsociety - MBP 15
Rocinante - E5530 testing machine
Saratoga, Ticonderoga, BunkerHill - the Hydra desktop PC-thing.
 
Forge
Lord High Gerbil
Posts: 8217
Joined: Wed Dec 26, 2001 7:00 pm
Location: SouthEast PA

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 11:13 pm

Duct Tape Dude wrote:
derFunkenstein wrote:
Heat death was a real thing back in 1999, since you could fry Athlons. Since then, AMD has learned its lesson and the CPU is much more resilient.
I thought that was because of a lack of thermal control. Today's chips have several throttling temperatures and a shutoff temperature as a last resort.


Quite a few current CPUs can be spiked past their hard cutoff temps and damaged before they self-power-off, if you completely remove the heatsink. It's just power density/thermal output versus surface area. If you remove the HSF, you remove 99% of the heat dispersal, so even that last couple of cycles before the limiter cuts power, it generates heat that has to go somewhere. With no HSF present, it's all dumped into anything in physical contact, meaning the die itself spikes, then the IHS spikes, then the substrate temp spikes, and it'll conduct down into the board and out into the air as best it can, but it can take a long time to disperse, and even a few seconds at 150-200C can be Bad For Things. I saved an i7-4800HQ from certain death by sacrificing the pad of one thumb. The marks have only just finished disappearing, 2-3 years later. A heatsink I am not.
Behemoth - Plex server
Galactica - FreeNAS
Pegasus - R710 ESX host
fsociety - MBP 15
Rocinante - E5530 testing machine
Saratoga, Ticonderoga, BunkerHill - the Hydra desktop PC-thing.
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 11:18 pm

Forge wrote:
I saved an i7-4800HQ from certain death by sacrificing the pad of one thumb. The marks have only just finished disappearing, 2-3 years later. A heatsink I am not.

Now THAT'S dedication! (Or perhaps a split-second "is the skin on my thumb worth $350" decision which you now regret... :lol:)
Nostalgia isn't what it used to be.
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1980
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 11:23 pm

just brew it! wrote:
Forge wrote:
I saved an i7-4800HQ from certain death by sacrificing the pad of one thumb. The marks have only just finished disappearing, 2-3 years later. A heatsink I am not.

Now THAT'S dedication! (Or perhaps a split-second "is the skin on my thumb worth $350" decision which you now regret... :lol:)

:lol: WITNESS!

http://www.youtube.com/watch?v=x2CeDY9Ywhs
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | Samsung 4K 40" | 480 + 240 + LSI 9207-8i (128x8) SSDs
 
just brew it!
Gold subscriber
Administrator
Posts: 48982
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: This G3258 goes to 1.7V

Sun Aug 06, 2017 11:25 pm

Going a little further back, I also had a K6-III+ that ran with a failed fan for a number of days/weeks/months (really have no idea now long it was), without showing any signs of trouble. I only noticed the fan was completely seized up when I opened the side of the case to install additional RAM. IIRC it was even running the distributed.net client at the time, so 100% CPU load 24x7.
Nostalgia isn't what it used to be.

Who is online

Users browsing this forum: No registered users and 1 guest