X79 and/or rampage IV OC'ing thread for those of us still left....

[TheRealTechGeek]

Straps are unavailable for locked xeons.
2667 and 2680v2 hit max turbo bins with no problems on light workloads.

Follow xCuri0's guide for above4g fix on x79...

It's a bit confusing and like I said just not even properly working there needs to be an video to be made.

 

[ZOX]

It's a bit confusing and like I said just not even properly working there needs to be an video to be made.

I tried to make a video guide, but had a problem with RIVE BIOS, so I put it on pause.
Will do it tomorrow for other BIOSes but I don't have the will to experiment with RIVE at the moment, although I have some ideas ...

 

[TheRealTechGeek]

I tried to make a video guide, but had a problem with RIVE BIOS, so I put it on pause.
Will do it tomorrow for other BIOSes but I don't have the will to experiment with RIVE at the moment, although I have some ideas ...

Hi thanks, I'm sorry too hear you're having issues with RIVE. You'll figure it out.
I want to get the Rampage Extreme Black Edition; though have to hack the bios for nvme. I'm on the Asus Deluxe. It's not the greatest overclocker maybe it's the chip might have to get another one though still don't want to give up on it is phenomenal as though i'm going too X99.

 

[Aquinus]

Follow xCuri0's guide for above4g fix on x79...

Easy recommendation when you're running Windows. I don't use Windows on this machine anymore. Only Linux. Either way, I don't think rebar is going to get me a whole lot with a Vega 64.

 

[ZOX]

Easy recommendation when you're running Windows. I don't use Windows on this machine anymore. Only Linux. Either way, I don't think rebar is going to get me a whole lot with a Vega 64.

Some tools have Linux version, but you could also always make live Win USB.
xCuri0 claims 12% uplift for RX580 so Vega should also see nice gains...

 

[maur0]

For those who use Rampage IV and want an updated network driver, as the Rampage 82579v network is no longer supported, the solution is to use the 82580 driver, which is still supported and the card works normally.

download last driver

adaptador Ethernet Intel® pacote de drivers completo

Este download instala a versão 29.0 do adaptador Ethernet Intel® Pacote de driver completo para as versões suportadas do SO.

www.intel.com.br

manually install select folder according to your OS here windows 11, select 82580 on list

work perfect new driver on rampage iv extreme old nic 82579v to 82580

 

[Good Guru]

Copied from another thread-
5930k @ 4.4 1.308v 1.91 vrin LLC 50% 68 celcius max temp usually ~62 or less with a Supernova 1260mm radiator
Ram 3000 mhz 16-18-18-38 2t 1.25v from 2400 mhz 15-15-15-35 2t 1.2v
Timings won't go lower even at 1.38v vdimm
ALso ring/cache will only go max 3700 mhz with 1.15v
SA is low 1.05v and i/o 1.05v
Are the memory timings descent?
Motherboard is MSI x99a Godlike Gaming quad channel

Now trying 15-18-18-38 2t @ 1.4v Forgot to mention its Kingston Fury 2400 15-15-15-35 1.2v has Micron chips. Nope Timespy froze. Wont do 16-16-16-36 @1.4V either

Is 16-18-18-36 timings not bad for 3000 mhz speed @ 1.25v?
Now i'm at 3200mhz cl17-19-19-38 2t 1.3v are those {ok} timings at that speed? I even had somewhat stability at 3333mhz but I could'nt get it stable, probably needs moe vdmm or Digital i/o voltage or both, had the vdmm at 1.4 when trying 3333 but could have needed more i/o up from 1.1 to 1.15. SA is at 1.1v and does'nt like much higher. Keeping the cpu at 4.4/3.7 ring cache. Its Kingston Hyper X Micron B-die ram rated for 2400mhz 15-15-15-35 @ 1.2v btw.

3000 cl16-18-18-36 1.25v or 3200 cl17-19-19-38 1.3v
Cas latency at 16 3000 =10.67
Cas latency at 17 3200 =10.63
Except i'm thinking 3000 cl16 would be better since I can lower the other 3 timings also 16-18-18-36
Which do you think generally will give the best performance?
Now I got em down to 16-19-19-36 at 3200 1.36v Hows that for timings at that speed? Good enough for gaming?

I read that Tertiary Timings can have a big impact with ddr4 gonna try lowering some of those.

 

[ZOX]

Guide:

Add suggestions if needed and sorry for bad english ...

 

[x79ftw]

I recently got a E5 1680 V2 for my Evga X79FTW, it seems like my max usable multiplier is 39x and i need to use a strap of 1.25x to get around the 4.5ghz mark. I've seen some users overclock their E5 1680 V2 using a multi over 39, do some E5 1680 V2's come with a higher usable multi than others or maybe its my motherboard limiting it?

 

[lexluthermiester]

I recently got a E5 1680 V2 for my Evga X79FTW, it seems like my max usable multiplier is 39x and i need to use a strap of 1.25x to get around the 4.5ghz mark. I've seen some users overclock their E5 1680 V2 using a multi over 39, do some E5 1680 V2's come with a higher usable multi than others or maybe its my motherboard limiting it?

That might be a BIOS limitation. The multiplier limit for X79/X99 is 50x. Presuming your BIOS is up to date?

 

[x79ftw]

That might be a BIOS limitation. The multiplier limit for X79/X99 is 50x. Presuming your BIOS is up to date?

I actually figured it out, my motherboard wasnt adjusting cores 7-8 even though i set the master multiplier ratio, so i just had to go there and repeat my settings. I got a pretty bad chip though, wont go past 4.3ghz, i can get it to boot 4.4ghz at 1.5v but not stable. My E5-2667v2 i was able to bclk overclock it to 4.1ghz all core at a much lower voltage than this E5 1680 v2, going to have to pick up another one, was hoping for 4.5/4.6ghz

 

[lexluthermiester]

I got a pretty bad chip though, wont go past 4.3ghz

In my experience, that's a decent performance. Not great, but I wouldn't call it "bad". I've seen some Xeon V2's that couldn't get passed 4.1ghz. 4.3ghz? I'd be happy with that on a daily basis.

My E5-2667v2 i was able to bclk overclock it to 4.1ghz all core at a much lower voltage than this E5 1680 v2

That's not surprising. The 2667V2's are binned in similar ways to the 1680V2's.

going to have to pick up another one, was hoping for 4.5/4.6ghz

If you're looking for a model to experiment and play with, ok go for it. If you're looking for a daily driver, don't waste your money. 4.3ghz is a solid OC and great for daily use.

 

[ZOX]

Can't find a similar topic in this thread, but have you tried messing with msi_util?
Those with NVMe BIOS driver should check their settings for controller.
For me, it defaulted to 16 IRQs/Threads and high priority.
It caused random latency spikes since NVMe drive would use all threads with high priority.
Changing max number to 8 (actual physical cores) and choosing "undefined" for priority completely changed OS behavior: everything is snappy and smooth ...

 

[Aquinus]

Quick question for anyone who has loaded up a modded BIOS to get PCIe bifurcation on an ASUS X79 board (not necessarily the same as my P9X79 Deluxe.) Did you ever hit a case where the settings didn't seem to work? Right now there is only a single PCIe slot that I was able to get bifurcation to work with in a 4/4 configuration, which unfortunately is one of the x8 slots that brings all 4 slots down to x8. I was hoping to use the second electrical x16 slot so my Vega 64 would continue to have 16 lanes, but even after setting it to be 4/4/4/4, it wouldn't detected anything except for the last NVMe drive.

 

[ZOX]

Quick question for anyone who has loaded up a modded BIOS to get PCIe bifurcation on an ASUS X79 board (not necessarily the same as my P9X79 Deluxe.) Did you ever hit a case where the settings didn't seem to work? Right now there is only a single PCIe slot that I was able to get bifurcation to work with in a 4/4 configuration, which unfortunately is one of the x8 slots that brings all 4 slots down to x8. I was hoping to use the second electrical x16 slot so my Vega 64 would continue to have 16 lanes, but even after setting it to be 4/4/4/4, it wouldn't detected anything except for the last NVMe drive.

As far as am aware, this a general x79 behavior. No 4/4/4/4 on x16 port.
Only one working 4/4 is last physical x8.
Check your topology ...

 

[Aquinus]

As far as am aware, this a general x79 behavior. No 4/4/4/4 on x16 port.
Only one working 4/4 is last physical x8.
Check your topology ...

That's in line with what I've tested. I read somewhere that it might have to do with how the PCIe lanes are switched between the ports. Not sure if I buy that, but it is what it is I guess. Either that or I dish out $130 USD on some Chinese made 4-port solution with a PLX bridge chip.

 

[MaksDampf]

A few posts ago i shared my Story about Turboboost unlock not working with an e2667v2 on an Asrock x79 xtreme3.

I suppose it is either a windows10 thing or Gigabyte somehow removed the Turbo unlock OC in the newer BIOSes as this was probably not intended by Intel. Does it requite intel Software to enable the Clockspeeds in windows, am i missing something? I also installed (old) Intel XTU and Multipliers are greyed out, but is says that it supports 40x Multi.

I now have a theory why that is:

I recently got an Asrock Z97 Anniversary with a non-K i7 4790 for free. I is a Multiplier-locked CPU the same way as the Xeon.
With the original BIOS on that board i was able to set the all-Core Turbo boost clock to 4Ghz either by using a fixed clock, but also with speedstep enabled. But the BIOS would not detect the nvme SSD, so updated it to the latest one.
With the new BIOS, the Turbo settings in BIOS as well as allcore Fixed Multiplier would have no effect in windows.

This is a know issue for LGA1150 mainboards using a CPU-Microcode above V7:
https://www.reddit.com/r/overclocking/comments/szco3g
Apparrently when Intel released all those Spectre and Meltdown Mitigations, they also added new microcode for Haswell, Broadwell that was silently disabling the turboboost unlock Loophole and slowing down turbo boost behaviour with the pretext of making the CPU "safer" as some rowhammer-like Cache-exploits worked better on CPUs with higher, almost unstable clocks.

My theory is that the same has been the case fo LGA2011 / x79 and Ivybridge-E, which is why my turbo boost settings don't work anymore on my x79 motherboard.

Both boards had an official BIOS update from Asrock in August 2018, which i have been using since both fixed or introduced NVME support officially.

I already patched the Z97 one with the old V7 microcode by using UBU. Now Turbobosst unlock is working again for the 4790.
I currently don't have access to the x79 system, but i will definitely try that later. Does anybody have experience with UBU and Microrode rollback on the x79 Platform?

I suppose most people in this thread did not come across this Problem because most x79 motherboards didn't have mid 2018 BIOS Updates or after Intel introduced the Microcode updates in early 2018 for that matter.

I actually figured it out, my motherboard wasnt adjusting cores 7-8 even though i set the master multiplier ratio, so i just had to go there and repeat my settings. I got a pretty bad chip though, wont go past 4.3ghz, i can get it to boot 4.4ghz at 1.5v but not stable. My E5-2667v2 i was able to bclk overclock it to 4.1ghz all core at a much lower voltage than this E5 1680 v2, going to have to pick up another one, was hoping for 4.5/4.6ghz

My 2667v2 is insanely well binned too.

It will only draw 95Watts at 8x 3,7Ghz and stay below 56°C while my 6-Core 1650V2 with half as much cache needed 150Watts at 4.2Ghz and was hitting 75Degrees on the same board and cooler. So the Cores of the 2667v2 MCC DIE were almost twice as efficient as the 1650v2s LCC DIE Cores.

Seeing the 4790 does also draw around 90Watts with Turboboost unlock and 4.1ghz allcore, one could say that twice as many ivybridge Cores and a huge 25mb cache equal half as many Haswell Cores with only 8M. I am quite impressed. Dunno about games, but i suppose in modern games more than four cores is more important than Haswell IPC.

Don't have the 1680v2 myself, but the 1000 series xeons seem to be of a lower bin considering the 1680 has 300mhz lower baseclock and 100mhz less turbo at the same 130W TDP as the 2667v2. A shame that the latter is locked and doesn't have straps available- would be the perfect LGA2011 CPU.

 

[Zyll Goliat]

A few posts ago i shared my Story about Turboboost unlock not working with an e2667v2 on an Asrock x79 xtreme3.



I now have a theory why that is:

I recently got an Asrock Z97 Anniversary with a non-K i7 4790 for free. I is a Multiplier-locked CPU the same way as the Xeon.
With the original BIOS on that board i was able to set the all-Core Turbo boost clock to 4Ghz either by using a fixed clock, but also with speedstep enabled. But the BIOS would not detect the nvme SSD, so updated it to the latest one.
With the new BIOS, the Turbo settings in BIOS as well as allcore Fixed Multiplier would have no effect in windows.

This is a know issue for LGA1150 mainboards using a CPU-Microcode above V7:
https://www.reddit.com/r/overclocking/comments/szco3g
Apparrently when Intel released all those Spectre and Meltdown Mitigations, they also added new microcode for Haswell, Broadwell that was silently disabling the turboboost unlock Loophole and slowing down turbo boost behaviour with the pretext of making the CPU "safer" as some rowhammer-like Cache-exploits worked better on CPUs with higher, almost unstable clocks.

My theory is that the same has been the case fo LGA2011 / x79 and Ivybridge-E, which is why my turbo boost settings don't work anymore on my x79 motherboard.

Both boards had an official BIOS update from Asrock in August 2018, which i have been using since both fixed or introduced NVME support officially.

I already patched the Z97 one with the old V7 microcode by using UBU. Now Turbobosst unlock is working again for the 4790.
I currently don't have access to the x79 system, but i will definitely try that later. Does anybody have experience with UBU and Microrode rollback on the x79 Platform?

I suppose most people in this thread did not come across this Problem because most x79 motherboards didn't have mid 2018 BIOS Updates or after Intel introduced the Microcode updates in early 2018 for that matter.



My 2667v2 is insanely well binned too.

It will only draw 95Watts at 8x 3,7Ghz and stay below 56°C while my 6-Core 1650V2 with half as much cache needed 150Watts at 4.2Ghz and was hitting 75Degrees on the same board and cooler. So the Cores of the 2667v2 MCC DIE were almost twice as efficient as the 1650v2s LCC DIE Cores.

Seeing the 4790 does also draw around 90Watts with Turboboost unlock and 4.1ghz allcore, one could say that twice as many ivybridge Cores and a huge 25mb cache equal half as many Haswell Cores with only 8M. I am quite impressed. Dunno about games, but i suppose in modern games more than four cores is more important than Haswell IPC.

Don't have the 1680v2 myself, but the 1000 series xeons seem to be of a lower bin considering the 1680 has 300mhz lower baseclock and 100mhz less turbo at the same 130W TDP as the 2667v2. A shame that the latter is locked and doesn't have straps available- would be the perfect LGA2011 CPU.

My X79 Platform with Xeon 2697 V2 and 64Gb still going strong as my daily driver I was considering to get something newer but few month ago I get myself RX 6700XT and now I really don't have any need to switch as everything works just fine.... anyway my machine is OC via BCLK(115) since the first day I build this RIG and never notice any problems with turbo issues currently working on W11 and other system is Linux Mint......

 

[ZOX]

A few posts ago i shared my Story about Turboboost unlock not working with an e2667v2 on an Asrock x79 xtreme3.



I now have a theory why that is:

I recently got an Asrock Z97 Anniversary with a non-K i7 4790 for free. I is a Multiplier-locked CPU the same way as the Xeon.
With the original BIOS on that board i was able to set the all-Core Turbo boost clock to 4Ghz either by using a fixed clock, but also with speedstep enabled. But the BIOS would not detect the nvme SSD, so updated it to the latest one.
With the new BIOS, the Turbo settings in BIOS as well as allcore Fixed Multiplier would have no effect in windows.

This is a know issue for LGA1150 mainboards using a CPU-Microcode above V7:
https://www.reddit.com/r/overclocking/comments/szco3g
Apparrently when Intel released all those Spectre and Meltdown Mitigations, they also added new microcode for Haswell, Broadwell that was silently disabling the turboboost unlock Loophole and slowing down turbo boost behaviour with the pretext of making the CPU "safer" as some rowhammer-like Cache-exploits worked better on CPUs with higher, almost unstable clocks.

My theory is that the same has been the case fo LGA2011 / x79 and Ivybridge-E, which is why my turbo boost settings don't work anymore on my x79 motherboard.

Both boards had an official BIOS update from Asrock in August 2018, which i have been using since both fixed or introduced NVME support officially.

I already patched the Z97 one with the old V7 microcode by using UBU. Now Turbobosst unlock is working again for the 4790.
I currently don't have access to the x79 system, but i will definitely try that later. Does anybody have experience with UBU and Microrode rollback on the x79 Platform?

I suppose most people in this thread did not come across this Problem because most x79 motherboards didn't have mid 2018 BIOS Updates or after Intel introduced the Microcode updates in early 2018 for that matter.



My 2667v2 is insanely well binned too.

It will only draw 95Watts at 8x 3,7Ghz and stay below 56°C while my 6-Core 1650V2 with half as much cache needed 150Watts at 4.2Ghz and was hitting 75Degrees on the same board and cooler. So the Cores of the 2667v2 MCC DIE were almost twice as efficient as the 1650v2s LCC DIE Cores.

Seeing the 4790 does also draw around 90Watts with Turboboost unlock and 4.1ghz allcore, one could say that twice as many ivybridge Cores and a huge 25mb cache equal half as many Haswell Cores with only 8M. I am quite impressed. Dunno about games, but i suppose in modern games more than four cores is more important than Haswell IPC.

Don't have the 1680v2 myself, but the 1000 series xeons seem to be of a lower bin considering the 1680 has 300mhz lower baseclock and 100mhz less turbo at the same 130W TDP as the 2667v2. A shame that the latter is locked and doesn't have straps available- would be the perfect LGA2011 CPU.

Windows microcode update breaks OC only for Haswell and newer platforms.
Why on Earth would you install Spectre/Meltdown mitigations BIOS on x79?
I have them disabled in Win also as it's pointless to give up 30% of perf for such a non-existent risk.

Turbo is one thing, OC is another. With any microcode revision, CPU has to boost because it is factory setting.
Maybe you didn't turn on C3 or C6 for CPU (not for package)?
As far as I've seen, all x79 BIOSes default to C3 disabled and C6 enabled but some boards don't have the option to change that and only amicbp helps.
AFAIK, x79 had last microcode update in 2019.

150W @ 4.2 is nothing strange for 1650. 2667 @ 4.0 will consume the same amount of power.
3930k @ 4.4 will eat 200W - the same as 1680 @ 4.4, despite having two more cores.
My 2667 has a ceiling at 4.1 after which voltage required for 100% stability climbs sharply.
On the other hand, 1680 scales linearly till 4.5.
It's just silicone lottery...

 

[MaksDampf]

Windows microcode update breaks OC only for Haswell and newer platforms.
Why on Earth would you install Spectre/Meltdown mitigations BIOS on x79?

I wouldn't. Why do you think i would?

latest BIOS says only "add NVME support" and "new CPU Microcode". I needed NVME support, how could i have known that it brakes Turbo?

I have them disabled in Win also as it's pointless to give up 30% of perf for such a non-existent risk.

Of course, me 2.

Turbo is one thing, OC is another. With any microcode revision, CPU has to boost because it is factory setting.

Dude, do i have to repeat the first part of the story in every post? Its like you try to misunderstand me on purpose. It doesn't turboboost like it should.
I saw 37x only for a split second with pure ST load and 38x would be active for a second when i manually disable 2 cores in the BIOS. 39x and 40 are never reached and the fixed multi setting (40max) in the bios has no effect, unlike any other x79 that i know of.

Maybe you didn't turn on C3 or C6 for CPU (not for package)?
As far as I've seen, all x79 BIOSes default to C3 disabled and C6 enabled but some boards don't have the option to change that and only amicbp helps.

I already tried every combination. currently have everything on i think. Rebootet to Bios and windows for hours, but it won't go beyond 36x on any setting.

150W @ 4.2 is nothing strange for 1650. 2667 @ 4.0 will consume the same amount of power.
3930k @ 4.4 will eat 200W - the same as 1680 @ 4.4, despite having two more cores.
My 2667 has a ceiling at 4.1 after which voltage required for 100% stability climbs sharply.
On the other hand, 1680 scales linearly till 4.5.
It's just silicone lottery...

Im was not talking about sandybridge. Those are hot fuckers for sure.
But 2667v2 and 2667 are different CPUs. I was referring to the Ivybridge-E Version with 8 Cores.

Usually clock scales linearly with voltage only for the lower part.
But while power draw scales linearly with Clockspeed it scales linearly with voltage at the same time. So usually we have exponential scaling for power draw on a CPU.

 

[k_9virus]

hi peeps does anyone tried this on our platform?

Resizable Bar (sandy-bridge/ Ivy-bridge)

 

[Zyll Goliat]

hi peeps does anyone tried this on our platform?

Resizable Bar (sandy-bridge/ Ivy-bridge)

Yeah and It's working fine......'tho I managed to have rebar size of 1gb and that's not that much complicated to implement (at least not for me) but to have more than this then you need to do more complex bios "patching" but I guess also depends of your motherboard bios......anyway it's working and you can read step by step instruction HERE

 

[k_9virus]

Yeah and It's working fine......'tho I managed to have rebar size of 1gb and that's not that much complicated to implement (at least not for me) but to have more than this then you need to do more complex bios "patching" but I guess also depends of your motherboard bios......anyway it's working and you can read step by step instruction HERE

thank you i have a 2080 on my 2697v2 this helps alot.

 

[maur0]

thank you i have a 2080 on my 2697v2 this helps alot.

Do you know if this rdimm memory works on the asus rampage iv with this xeon e5-2697v2?

 

[lexluthermiester]

Do you know if this rdimm memory works on the asus rampage iv with this xeon e5-2697v2?

Very unlikely. Not because of the CPU, but because ASUS does not support RDIMMs on their consumer line of boards.