X99 Club

[lexluthermiester]

Back in the day 5930k 4.5 at 1.35v was considered a darn good chip

Holy crap that's high voltage! I would NEVER run a Haswell that high!

 

[ThrashZone]

Holy crap that's high voltage! I would NEVER run a Haswell that high!

Hi,
Have sign into google to read the guide now

The Haswell-E Overclocking Thread and Overclocking Guide

Starting off with a basic guide of relating to the 5960X on ASUS motherboards: https://drive.google.com/file/d/0Bz2VRRbLPrZnMXpJY3k5Vk8zSVU/edit?usp=sharing Latest UEFI Build for R5E: RAMPAGE V EXTREME UEFI 0801 0801 Download Mirror

rog.asus.com

5820k at 1.17v has to be a typo under load

 

[johnspack]

Good god no... mine is at like just over 1v stock. I'm going to crank my cache from 3.8 to 4 and see how goes. Think this chip will do 4.1 cache....

 

[lexluthermiester]

Have sign into google to read the guide now

The Haswell-E Overclocking Thread and Overclocking Guide

Starting off with a basic guide of relating to the 5960X on ASUS motherboards: https://drive.google.com/file/d/0Bz2VRRbLPrZnMXpJY3k5Vk8zSVU/edit?usp=sharing Latest UEFI Build for R5E: RAMPAGE V EXTREME UEFI 0801 0801 Download Mirror

rog.asus.com

I'm not reading through that. Haswell CPU's default to 1.00v or lower, so 1.35V is crazy high.

 

[johnspack]

I'll do more tomorrow. I'm tired. This cpu will actually do over 4.7.

Intel Core i7 5820K @ 4497.34 MHz - CPU-Z VALIDATOR

[m3srnv] Validated Dump by DESKTOP-R524VHQ (2023-02-01 00:28:51) - MB: ASRock X99 Extreme4/3.1 - RAM: 32768 MB

valid.x86.fr

 

[ThrashZone]

I'll do more tomorrow. I'm tired. This cpu will actually do over 4.7.

Intel Core i7 5820K @ 4497.34 MHz - CPU-Z VALIDATOR

[m3srnv] Validated Dump by DESKTOP-R524VHQ (2023-02-01 00:28:51) - MB: ASRock X99 Extreme4/3.1 - RAM: 32768 MB

valid.x86.fr

Hi,
That isn't a stress at all so yeah at least r23.

 

[johnspack]

Don't know why I have to prove it... but here ya go:

 

[ThrashZone]

I'm not reading through that. Haswell CPU's default to 1.00v or lower, so 1.35V is crazy high.

Hi,
Raja oc guide so weren't you defending him when he got demoted recently ?
Either way i believe it was in the beginning comment on it but what ever

Don't know why I have to prove it... but here ya go:

Well you said something unbelievable frankly so I just asked to run a very quick test thanks for entertaining the request we are done
It's not like I asked to run p95 for an hour like some others around here

 

[Fouquin]

Holy crap that's high voltage! I would NEVER run a Haswell that high!

It's pretty standard, really. 1.4v is the maximum on Haswell-E but with any Intel 22nm chip temperature is the problem 99.9% of the time. Haswell-E also has a relatively (to desktop) low clock ceiling anyway and slamming it with voltage won't gain you much. It is generally agreed among those that habitually overclock Haswell that 1.35v is the 24/7 set point on desktop, and 1.25v is the 24/7 set point on HEDT temperatures permitting.

Now, important note, Haswell and Haswell-E have different maximum voltage ratings. Haswell desktop's maximum is 1.52v.

Also stock voltage is whatever Intel sets at the factory for that chip. VID is not a constant value and Intel themselves state as such in their documentation. One chip may run stock clocks at 1v, and the next one in line may run 1.1v.

 

[johnspack]

I've looked all over the internet and ppl just throw 1.30-1.35 into these, and it's just not needed. I probably have a gold sample, so I shouldn't preach too much...
I'm going to try 4.6-7 tomorrow using same voltage. Bet it'll do it. I stay under 1.3 on my older 1680 v2, but it's just not as fast as this 5820k....
Oh and this chip's vid is 1.0.....

 

[Zyll Goliat]

I've looked all over the internet and ppl just throw 1.30-1.35 into these, and it's just not needed. I probably have a gold sample, so I shouldn't preach too much...
I'm going to try 4.6-7 tomorrow using same voltage. Bet it'll do it. I stay under 1.3 on my older 1680 v2, but it's just not as fast as this 5820k....
Oh and this chip's vid is 1.0.....

It`s not that people don't believe you but most of the time it's one thing to enter the windows or even do CPU-z or some other "light" benchmark and to be 24/7 100% fully stable.....saying this it's not unusual for Haswell-E to work on 1.1V when is "slightly" OC but soon as you pass that 4,3-4,5Ghz line majority of CPU's will ask for more Juice.........Well looks like you won the lottery with that 5820k.....

 

[Fouquin]

I've looked all over the internet and ppl just throw 1.30-1.35 into these, and it's just not needed.

It was needed when the chips and manufacturing process were newer. Intel made Haswell for about three years and many of the later production CPUs (and second generation X99 boards) just don't need as much to be stable. If you're looking at what people were doing in 2014 to keep their brand new 5820K at 4.5GHz, then yeah. My launch day 5820K requires 1.42v just to remain stable at 4.5GHz, it was never a peach for overclocking and it never managed to pull the numbers that chips made a year later could.



The maximum OC under 1.3v for this chip is only 4.3GHz. You got a good chip, enjoy it.

Here's a mid-2015 purchased 5960X for comparison on max clocks and voltage:



And a later production E5-1650 V3 for another comparison (notice especially the temperature difference on the Xeon):

 

[ThrashZone]

Hi,
Here's a good one for broadwell-e

It`s not that people don't believe you but most of the time it's one thing to enter the windows or even do CPU-z or some other "light" benchmark and to be 24/7 100% fully stable.....saying this it's not unusual for Haswell-E to work on 1.1V when is "slightly" OC but soon as you pass that 4,3-4,5Ghz line majority of CPU's will ask for more Juice.........Well looks like you won the lottery with that 5820k.....

Also my understanding is a 5820k is a low bin 5930k so that is why it's pretty unbelievable.

Here's a good one showing what happened to hasswell-e on newer broadwell-e bios and broadwell-e of course voltages just by activating xmp profiles Raja at work
Notice only pch stayed at it's default 1.05v cpu and system agent were zapped with an additional 0.25v

 

[lexluthermiester]

Raja oc guide so weren't you defending him when he got demoted recently ?

Like I said, I didn't read it. Didn't know they were the same person. Don't Care. The download link doesn't work anyway.

It's pretty standard, really. 1.4v is the maximum on Haswell-E but with any Intel 22nm chip temperature is the problem 99.9% of the time.

I'll take your word for that. I've never pushed any Haswell higher than 1.3v and considered that very high.

I probably have a gold sample, so I shouldn't preach too much...

Not from my experience. 4.5ghz at 1.17v is nice but not far from par for the course. 1.2v usually renders 4.3, 4.4 or 4.5.

My launch day 5820K requires 1.42v just to remain stable at 4.5GHz

That's insane! How has that CPU not died?

 

[Zyll Goliat]

Also my understanding is a 5820k is a low bin 5930k so that is why it's pretty unbelievable.

I kinda agree with that theory as Intel is well known to do "cherry picking"....'tho it was probably only in first few batches so it was plausible that later batches of 5820k could be also better binned CPU's....

 

[ThrashZone]

Like I said, I didn't read it. Didn't know they were the same person. Don't Care. The download link doesn't work anyway.


I'll take your word for that. I've never pushed any Haswell higher than 1.3v and considered that very high.


Not from my experience. 4.5ghz at 1.17v is nice but not far from par for the course. 1.2v usually renders 4.3, 4.4 or 4.5.


That's insane! How has that CPU not died?

Hi,
I wouldn't be surprised if Raja removed the guide since he's no longer with asus
Ran across this going down ocn memory lane on it's x99 mother board thread
4.6 on 5930k I did years ago unfortunately just showing someone that oc still worked on last haswell-e bios 2101 when broadwell-e ocing was killed on asus boards with a crappy intel micro code included

But notice vcore and vccsa voltages verses the other screen shots yellow highlighted ones on broardwell-e bios
Yes a very sharp difference doesn't show vccio cpu though but you can bet it was on default 1.05+-

 

[Fouquin]

That's insane! How has that CPU not died?

Because it's not wildly out of spec, and because the daily OC run on it until about four months ago was 4.4GHz at 1.35v. The key and determining factor in the chip surviving this entire time is that it was never run over 85C.

 

[lexluthermiester]

The key and determining factor in the chip surviving this entire time is that it was never run over 85C.

Heat is an important factor of the livelihood of CPUs(and electronics in general), however, pathway degradation from excessive voltage is very real problem. It's one of the reasons IC's malfunction. It is a known killer of CPU's, GPU's and other chipset IC's.

 

[bobbybluz]

Heat is an important factor of the livelihood of CPUs(and electronics in general), however, pathway degradation from excessive voltage is very real problem. It's one of the reasons IC's malfunction. It is a known killer of CPU's, GPU's and other chipset IC's.

I've degraded a few CPU's with core voltage during extended overclocking (a few years each) but never had any cooling issues contribute to their gradual decline. Loud yes, hot no.

 

[RJARRRPCGP]

It was needed when the chips and manufacturing process were newer. Intel made Haswell for about three years and many of the later production CPUs (and second generation X99 boards) just don't need as much to be stable. If you're looking at what people were doing in 2014 to keep their brand new 5820K at 4.5GHz, then yeah. My launch day 5820K requires 1.42v just to remain stable at 4.5GHz, it was never a peach for overclocking and it never managed to pull the numbers that chips made a year later could.

View attachment 281843

The maximum OC under 1.3v for this chip is only 4.3GHz. You got a good chip, enjoy it.

Here's a mid-2015 purchased 5960X for comparison on max clocks and voltage:

View attachment 281844

And a later production E5-1650 V3 for another comparison (notice especially the temperature difference on the Xeon):

View attachment 281845

My FX 8350 (made in 2014) possibly requires less than 1.35V @ 4.4 GHz. And FX-es usually require more volts, TMK. Well, I at least gave it a test at likely stock Vcore of only 1.287V, IIRC!
(2-pass x264, which does give cores a good workout)
(single-pass is in mostly RAM, OTOH)
And I suspect the 5820K was intentionally down-binned, like I would suspect for FX 8320s and Core i7 920s.

 

[Fouquin]

My FX 8350 (made in 2014) possibly requires less than 1.35V @ 4.4 GHz. And FX-es usually require more volts, TMK. Well, I at least gave it a test at likely stock Vcore of only 1.27V, IIRC!
(2-pass x264, which does give cores a good workout)
(single-pass is in mostly RAM, OTOH)
And I suspect the 5820K was intentionally down-binned, like I would suspect for FX 8320s and Core i7 920s.

Bulldozer is a totally different beast, the voltage requirements are not comparable to Haswell-E in any way, shape, or form. I expect every FX 8350 to do 4.4GHz at stock voltage, that is less than a 5% overclock over turbo.

 

[RJARRRPCGP]

Bulldozer is a totally different beast, the voltage requirements are not comparable to Haswell-E in any way, shape, or form. I expect every FX 8350 to do 4.4GHz at stock voltage, that is less than a 5% overclock over turbo.

Really? I thought that Bulldozer would require more than that. I do remember noticing that my FX 8350 had a lower VID than many, IIRC. Instead, mine acted kind of like a Ryzen and would heat up a lot at 1.28V-1.30V, LOL. It seemed that I had to do nothing more than increase the current limits to test 2-pass x264.

Yes, I know that FX was a radically different architecture, which had trouble with low-threaded tasks. I also thought that they usually required more voltage than Intel, especially at 4.5 GHz+.

That AMD arch, was AMD's "Pentium 4 moment".

 

[ThrashZone]

Hi,
I wouldn't be surprised if Raja removed the guide since he's no longer with asus
Ran across this going down ocn memory lane on it's x99 mother board thread
4.6 on 5930k I did years ago unfortunately just showing someone that oc still worked on last haswell-e bios 2101 when broadwell-e ocing was killed on asus boards with a crappy intel micro code included

But notice vcore and vccsa voltages verses the other screen shots yellow highlighted ones on broardwell-e bios
Yes a very sharp difference doesn't show vccio cpu though but you can bet it was on default 1.05+-

Hi,
My bad uploaded the wrong image there
But system agent is pretty different from the highlighted image
The all voltage list was taking later is why max system agent is different





Once again this on haswell-e bios not broadwell-e bios and is why voltages are different.

 

[johnspack]

Yep, 4.6 at same voltage. 4.7 might need 1.18 or so, oh well. Not sure how much more I'll be doing with this as I just inherited an x399 system
and a 1920x cpu. No ram though so my x99 might have to go down for awhile.....

 

[Pentium320]

Hi, and welcome to forum

Nice to see there are still some of us at X99. I'm rocking with Xeon E5-2697 v3 at ASRock X99 Extreme4, and honestly this is my favourite platform after LGA775 and 1366
Previously had i7-6800K, but wasn't really satisfied with it, OC was quite dire - maybe my sample wasn't that great, and Xeons are so cheap, that I've gone for that solution. Got some better GPU, and well - everything works perfectly. Not much to show-off when it comes about OC, but here are some of my results with turbo unlock mod @ 3.78GHz and -30mV.

Honestly, after grabbing 1440p monitor I regret getting "only" 6700 XT. 6800 XT or maybe even 6900 XT IMHO would also fit there perfectly for gaming. Unfortunately, ReBAR mod works only "halfly" at this motherboard.
What's not to like about this setup? Well, horrendous power consumption and generated heat of course, but can live with that, at least for a while