Overclock.net banner

"Unstable" 5820K - help! (?)

5K views 21 replies 9 participants last post by  Wireline 
#1 ·
Hello there

Can anyone help with 5820K OC stability? My chip seems to be a bit of a dog:

i7-5820K, 4.3GHz unstable (1.32V), 4.0GHz stable (1.268V)
ASUS X99 Deluxe Motherboard
Corsair Dominator DDR4 2666MHz (not currently running the XMP yet so its 2133MHz)
PSU: 1000W Superflower Leadex Platinum
H105 cooler (240mm rad)

Target OC = 4.3GHz.
 
#2 ·
One of the three 5960X I had would not do LinX 0.6.5 stable at stock. It might be worth an RMA.

You could also try upping VCCIN / Input CPU Voltage to 1.9 or a tiny bit higher. Tuning System Agent voltage can also help but that is more likely to be an issue with a RAM and/or cache overclock. Have you tried higher V-core? I think some Haswell-E really are simply very weak and you might have gotten one of the bottom 10%.
 
#3 ·
Quote:
Originally Posted by Asmodian View Post

One of the three 5960X I had would not do LinX 0.6.5 stable at stock. It might be worth an RMA.

You could also try upping VCCIN / Input CPU Voltage to 1.9 or a tiny bit higher. Tuning System Agent voltage can also help but that is more likely to be an issue with a RAM and/or cache overclock. Have you tried higher V-core? I think some Haswell-E really are simply very weak and you might have gotten one of the bottom 10%.
Hi there

Thank you for replying
smile.gif
Can you actually RMA a CPU to Intel on the basis of it being a terrible overclocker? I took out the OC insurance, I'd love it if it covered that. It does seem to be stable at stock speeds, but those are rubbish
frown.gif


I am also having enormous troubles with the X99 Deluxe its sitting on, and will be chatting with ASUS about it, so at the moment its diffiicult to tell whats causing it. In answer to your questions, I went up to 1.3V and didn;t want to go higher because ASUS and Intel had recommended that as a safe maximum, VCCIN was on auto and showing as 1.9, but I did not have the confidence to adjust this.
 
#4 ·
Well with the OC insurance they say hassle free so I imagine they won't look too closely if you simply say it isn't stable.

I recently had really bad experiences with two X99-Deluxe motherboards but that is with the 5960X and it may have been simply bad luck. I don't want to go too negative and many others are running fine. I will not use an X99-Deluxe for a 5960X anymore but I really liked the boards before they died and took the CPUs with them (or maybe the CPU died and took the board with it?).
 
#5 ·
Quote:
Originally Posted by Wireline View Post

i7-5820K, 4.3GHz unstable (1.32V), 4.0GHz stable (1.268V)
Those are some mighty high voltages for those clocks. I have my 4670K running stable at 4.5Ghz and 1.252V, and back when I was at 4.2Ghz it ran fine with 1.2V, though I didn't stress test that clock too much. It seems strange to me that your chip would require a whole 0.06V more than mine at the same clock, even if they are different chips.
 
#6 ·
Quote:
Originally Posted by Dinosaur Brutus View Post

Those are some mighty high voltages for those clocks. I have my 4670K running stable at 4.5Ghz and 1.252V, and back when I was at 4.2Ghz it ran fine with 1.2V, though I didn't stress test that clock too much. It seems strange to me that your chip would require a whole 0.06V more than mine at the same clock, even if they are different chips.
I think I might have a bad chip too because mine does 1.29V @4.2 GHz... Or the Haswell-Es just require more voltage one of the two. I have a 5820K BTW.
 
#7 ·
Quote:
Originally Posted by ThornTwist View Post

I think I might have a bad chip too because mine does 1.29V @4.2 GHz... Or the Haswell-Es just require more voltage one of the two. I have a 5820K BTW.
Don't take my numbers as too accurate as I didn't do a whole lot of testing at 4.2Ghz/1.2V. As well, the 4670K is a 4 core while the 5820K is a 6 core with hyperthreading (I'm not too clear as to what sort of power requirement impact these features would have on the chip). The other thing with overclocking chips is that the voltage requirements do not climb linearly, and require progressively more power for each bump in the multiplier. This, combined with the additional cores, might have an effect on the way voltage needs to be applied to overclock the chip.

I am not too well versed in the mechanics of effectively applying voltage and the way power is used by the chips, so I could be completely wrong. That aside, however, it does strike me as odd that you need to apply 0.09V more for the same clock than I do.
 
#8 ·
Quote:
Originally Posted by Dinosaur Brutus View Post

Don't take my numbers as too accurate as I didn't do a whole lot of testing at 4.2Ghz/1.2V. As well, the 4670K is a 4 core while the 5820K is a 6 core with hyperthreading (I'm not too clear as to what sort of power requirement impact these features would have on the chip). The other thing with overclocking chips is that the voltage requirements do not climb linearly, and require progressively more power for each bump in the multiplier. This, combined with the additional cores, might have an effect on the way voltage needs to be applied to overclock the chip.

I am not too well versed in the mechanics of effectively applying voltage and the way power is used by the chips, so I could be completely wrong. That aside, however, it does strike me as odd that you need to apply 0.09V more for the same clock than I do.
Wattage limits for my chip are 140W, but if you're OCing it could go over that. What's you chips wattage supposed to be?
 
#10 ·
Quote:
Originally Posted by Dinosaur Brutus View Post

The i5-4670K has a wattage limit of 84W.
That would explain the increase in voltage between the two chipsets because V (voltage) squared * I (current) = W (wattage)
 
#11 ·
Quote:
Originally Posted by Dinosaur Brutus View Post

Those are some mighty high voltages for those clocks. I have my 4670K running stable at 4.5Ghz and 1.252V, and back when I was at 4.2Ghz it ran fine with 1.2V, though I didn't stress test that clock too much. It seems strange to me that your chip would require a whole 0.06V more than mine at the same clock, even if they are different chips.
mmm... I do not think it is accurate to compare a 4670K to a 5820K especially since the 5820K has 12 threads vs 4 threads of 4670k. Has 6 core vs. 4 cores - 140W TDP vs the 84W TDP, L2/L3 cache is x2 on the 5820K than on the 4670k
Quote:
Originally Posted by ThornTwist View Post

Wattage limits for my chip are 140W, but if you're OCing it could go over that. What's you chips wattage supposed to be?
I am not sure if this is right but, when you overclock you produce more TDP (hopefully this answers what you are asking)

Also from what I gathered in 5 minutes, your voltage requirement is higher than on average - most chips can reach 4.3 (5820K) @ 1.25-1.275 - without knowing what other settings you are using in the Bios and tests, it is harder for us to help you.
 
#12 ·
Quote:
Originally Posted by soulwrath View Post

mmm... I do not think it is accurate to compare a 4670K to a 5820K especially since the 5820K has 12 threads vs 4 threads of 4670k. Has 6 core vs. 4 cores - 140W TDP vs the 84W TDP, L2/L3 cache is x2 on the 5820K than on the 4670k
I am not sure if this is right but, when you overclock you produce more TDP (hopefully this answers what you are asking)

Also from what I gathered in 5 minutes, your voltage requirement is higher than on average - most chips can reach 4.3 (5820K) @ 1.25-1.275 - without knowing what other settings you are using in the Bios and tests, it is harder for us to help you.
I used AISuite 3 5WO. I got 4.3 the first time I did it, forgot the voltage. But I got 4.2 GHz the next times, which is weird right? I'm trying to stay away form manually OCing my CPU because I already ruined a Mobo because of this. I need more info on how to manually OC before I can feel confident to do it again. Loosing $219 + $189 was a lot of money to me at the time so that has caused me to be cautious. Current setting are 42 x 100, 167W, 1.269V, Temp went to 70 during OC. Don't know why AIS3 stopped the OC. *shrugg*
 
#13 ·
Quote:
Originally Posted by soulwrath View Post

mmm... I do not think it is accurate to compare a 4670K to a 5820K especially since the 5820K has 12 threads vs 4 threads of 4670k. Has 6 core vs. 4 cores - 140W TDP vs the 84W TDP, L2/L3 cache is x2 on the 5820K than on the 4670k
I am not sure if this is right but, when you overclock you produce more TDP (hopefully this answers what you are asking)
I mentioned this in another comment of mine, actually. I just didn't think that there would be such a drastic difference in voltages.
 
#14 ·
Quote:
Originally Posted by ThornTwist View Post

I used AISuite 3 5WO. I got 4.3 the first time I did it, forgot the voltage. But I got 4.2 GHz the next times, which is weird right? I'm trying to stay away form manually OCing my CPU because I already ruined a Mobo because of this. I need more info on how to manually OC before I can feel confident to do it again. Loosing $219 + $189 was a lot of money to me at the time so that has caused me to be cautious. Current setting are 42 x 100, 167W, 1.269V, Temp went to 70 during OC. Don't know why AIS3 stopped the OC. *shrugg*
This is probably more of the culprit, always best to manually OC there should be some guides up or if not up soon.
 
#15 ·
#16 ·
Did you try at 4.2? It is a very decent OC anyways 700mhz above stock, right? I guess 4.3 is where your CPU hits its ceiling and would require a huge ammount of Voltage to be stable. I don't think you can RMA a chip like that lol. It's not a bad OC at all anyways. Don't even compare it to other quadcores because their stock frequency is much higher than urs.
 
#17 ·
#18 ·
try 1.3vcore, vccin 2.1v, 1.15v vring
my 5820k isnt that great either but its stable at 4.4ghz with the above volts.

the max i would go is 1.35vcore 2.25vccin, 1.25vring for 24/7 usage with insane watercooling/ambient temps. if you have to go above 1.325vcore for 4.3ghz then i would just stay at 4.2
 
#19 ·
Quote:
Originally Posted by e6800xe View Post

try 1.3vcore, vccin 2.1v, 1.15v vring
my 5820k isnt that great either but its stable at 4.4ghz with the above volts.

the max i would go is 1.35vcore 2.25vccin, 1.25vring for 24/7 usage with insane watercooling/ambient temps. if you have to go above 1.325vcore for 4.3ghz then i would just stay at 4.2
Hey thanks man, having some numbers is useful. I just RMA'd my X99 Deluxe as it stopped working and developed all sorts of ****ty problems, got a refund in the end so now I am looking for a new mobo. Avoiding ASUS at the mo as they seem to be struggling the most,
 
#20 ·
Quote:
Originally Posted by Wireline View Post

Hey thanks man, having some numbers is useful. I just RMA'd my X99 Deluxe as it stopped working and developed all sorts of ****ty problems, got a refund in the end so now I am looking for a new mobo. Avoiding ASUS at the mo as they seem to be struggling the most,
I love my MSI Mpower , I mainly bought asus motherboards in the past but IMO they my have their support tech guy here so they don't lose anymore sales but like I said that's my opinion
 
#21 ·
Quote:
Originally Posted by HyperC View Post

I love my MSI Mpower , I mainly bought asus motherboards in the past but IMO they my have their support tech guy here so they don't lose anymore sales but like I said that's my opinion
Right now I feel quite driven by which manufacturer offers a decent returns policy, which TBH seems to be none of them. I think most of the time people don't have problems and so the shiny happy view of the company is preserved, but then you come up against their warranty claims dept and its Jekyll and Hyde time.

Ever sent anything back to MSI at all? At the moment neither ASUS EU nor EVGA EU are looking that great, so its coming down to a race between Gigabyte and MSI. Gigabyte boards are fuggington mc fugly though
sicksmiley.png
:
 
#22 ·
By way of update, just had confirmation that my X99 Deluxe was faulty, got it refunded via an Amazon seller but they weren't able to say what was wrong with it. ASUS had suggested a fritzed BIOS chip. The thing had only been switched on a week before
frown.gif
So yeah, I doubt any of that had a positive impact on my overclocking result, so I will probably go for another Deluxe and see how we get on. There's just been a "stability" update with the latest BIOS too.

In all honesty I haven't seen any X99 board that has screamed "obvious alternative", they all seem to have similar or issues of their own (such as MSI XPower's daft SLI slot placement when you have a 28 lane CPU, will force me to vent one of my EVGA ACX 980's onto its brother).

So maybe my 5820K has a bit more up its sleeve than originally hoped
smile.gif
 
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