Hitting 43x multiplier wall on Rampage IV Black edition - E5 1660 v1

[Mirkoskji]

Hello everyone. I'm about new to the forum.

Recently I've Bought a Xeon e5 1660 v1 for my RIVBE, because I wanted to do overclock for cheap (about 120 euros) and have a decent workstation setup.
After a long research over the internet I came to the conclusion that every xeon 16XX seems to have an unlocked multiplier (or I hope i've not stumbled over the sole 1660 with a locked one). My older cpu was a e5 1620 v1 for which I know that, as the i7 3820, it has a locked multiplier of 43x.

Today I've installed My new cpu and discovered that, like the older one, i cannot get over x43. Simply my motherboard wont POST.

I can choose a multiplier up to 57x (whereas for the 1620 I was limited to 44x) bu everything over 43x just don't post, even with 1,4 or 1,45v
Why my RIVBE cannot overclock, via multiplier or straps, over 43x?
Do I need to mod something maybe?

Does anyone have the same problem on this board here? Have you heard of similar problems?

Sorry if my english sounds a little baroque and thanks in advance for help

Mirko

 

[RejZoR]

That's 6 cores with 12 threads. Hitting 4.3GHz seems pretty reasonable for such chip imo. I mean, this is still 32nm. I'm hitting 4.5GHz limit (multi of 45) on my 5820K with same CPU/core arrangement and it's a newer 22nm chip. I can try everything beyond that point, but it's just not stable or even post-able. Mine does like really low voltages though, 4.5GHz at 1.15V is something hardly any 5820K achieves. I guess tradeoff for that is an OC limit at 4.5GHz...

Try the same as I have. So, 4.3GHz is your limit, now try decreasing Vcore and see how low it'll go. I've also started at 1.3V in the beginning and went all the way down to 1.15V...

 

[Komshija]

Some motherboards limit the multiplier to prevent possible damage to the CPU and there is also a possibility that CPU itself has such limiter. For E5 1660, clock of 4,3 GHz is excellent. Find the stable voltage for that setting; if it's too high, lower the multiplier to 42.

 

[RejZoR]

Considering it's a Rampage board, I highly doubt it's limit anything. This board is meant for overclocking.

 

[basco]

did you try to raise the bclk?
like: 104x43=4472mhz

some mainboards did not like to sit at exact 100bclk:
try 99 or 101+

and asus says it only supports the 1660 V2:
https://www.asus.com/de/Motherboards/RAMPAGE_IV_BLACK_EDITION/HelpDesk_CPU/

 

[Mirkoskji]

I have updates. I can get to 4.6 GHz using Intel extreme tuning utility with about 1,36v. And also with Asus aisuite (which I hate because it messes up my offset voltage an makes my PC unstable). The problem is: I have to set a fixed voltage in bios an then I have to change the frequency in windows, which is unpractical. My project was to use per core overclock to boost singe and multi threaded processes to 46-47x. With 43x it's stable at 1,26v

For Basco: I know the support list is only for ivy bridge ep. But I have an example of a rivbe overclocking a e5 1650 v1 to 5.0 GHz with only base clock oc. My board seems to recognize my CPU but it puts some sort of limit to multi and bclk.

 

[eidairaman1]

YMMV

 

[RejZoR]

You're at a point where you should ask yourself, is additional 0.1V worth the extra 300MHz and potential instability. Because 1.36V is pretty high. Not quite 1.4V yet, but very close... Or stick with 4.3GHz and try to get voltage even lower...

 

[Aquinus]

Because 1.36V is pretty high.

I wouldn't consider 1.36v for a Sandy Bridge CPU to be too high if temperatures are in check. It is a 32nm chip and not a 22nm or 14nm one.

 

[RejZoR]

In general, 1.4V is considered upper edge across the range of Intel processors. Sure, it'll work, but it also means it will most likely degrade faster and demand even more voltage or it'll become unstable at same clocks where it used to be stable before.

 

[Mirkoskji]

OK, but I know limits. The question is not whether I should or I shouldn't go over. That's my own risk. My question is: in windows I can set 44x multiplier. In bios I can't. With the same voltage and same contest. I'm trying to understand why

 

[jboydgolfer]

Are you sure it's actually being set to that 44 or is it just showing that? Personally I don't use the software OC Utilities

 

[Mirkoskji]

Cinebench shows higher score when I change it, hwinfo shows 4400mhz. Either I wouldn't mess with software oc. But it serves me as a proof of concept for my problem.

 

[sneekypeet]

Have you tried multiple motherboard BIOS to see if maybe one will do what you want?

 

[Mirkoskji]

I tried bios 801, bios 403, bios 701 and bios 603.

I wouldnt insist if i knew its an objective limit of my hardware. the problem is that i know of people, on HWBOT for example, who have no problem reaching 5.0ghz with only bclck overlclock with my board and the same xeon processor.

 

[sneekypeet]

I tried bios 801, bios 403, bios 701 and bios 603.

I wouldnt insist if i knew its an objective limit of my hardware. the problem is that i know of people, on HWBOT for example, who have no problem reaching 5.0ghz with only bclck overlclock with my board and the same xeon processor.

Well, you seem to be talking about two different things here! In your OP and in following posts, you are addressing the OC by using multipliers. In this last post you say that others can hit the 5GHz mark with the BLCK, which eliminated the multi cap you are experiencing.

Have you tried raising the BLCK to see if you can get further than 4300MHz?

 

[Random Murderer]

I tried bios 801, bios 403, bios 701 and bios 603.

I wouldnt insist if i knew its an objective limit of my hardware. the problem is that i know of people, on HWBOT for example, who have no problem reaching 5.0ghz with only bclck overlclock with my board and the same xeon processor.

Make sure you try the following settings:
DISABLE Anti-Surge Support (I cannot stress how important this is for overclocking past 4.3ish on the 6+ core CPUs, especially SB-E)
Set CPU Current capability to 180% (can be lowered later, realistically 140% or 150% should be good for up to ~5GHz)
Set CPU Power Duty Control to EXTREME and make sure you have good airflow on the VRM heatsink
Set the VCCSA Current Capability to 140%
ENABLE Extreme Tweaking
DISABLE ALL Spread Spectrums
DISABLE Limit CPUID Maximum

These settings helped me push my 3820 to 4.9GHz 24/7 stable on my Rampage IV Extreme. If your chip is voltage-hungry, you may need to set a (relatively) high CPU VCORE Boot-up Voltage. This setting can be found in the DIGI+ section. It allows you to set a specific voltage it will supply to VCORE on cold boots to help stubborn chips POST. Once it has POSTed, the VCORE will be set to whatever your VCORE is set to in the main voltage section.

 

[Mirkoskji]

Done. No 44x boot

 

[Random Murderer]

Done. No 44x boot

Alright, let's try something a bit easier. Set the multi to 36 and the BCLK to 125, that should net you 4.5GHz.
If you're relegated to 43x as a max multi, then start playing with the BCLK. You can still break 5GHz if the max multi you have is 43x, 40x 125 would get you 5GHz. My 3820 happily ran 38x 129 giving me 4.9GHz on the CPU and 2407 on the RAM.

 

[RejZoR]

Forgo to mention something worth a try. Try bumping up DRAM voltage a bit. This can sometimes help overclocking the CPU.

 

[Mirkoskji]

Already tried with bclck, cannot go over 100.9 mhz, it gets unstable. As for memories, I have G-skill Trident 2400mhz operating at 1.65v. So I think is already high voltage...

 

[Aquinus]

Alright, let's try something a bit easier. Set the multi to 36 and the BCLK to 125, that should net you 4.5GHz.
If you're relegated to 43x as a max multi, then start playing with the BCLK. You can still break 5GHz if the max multi you have is 43x, 40x 125 would get you 5GHz. My 3820 happily ran 38x 129 giving me 4.9GHz on the CPU and 2407 on the RAM.

This. Also, if the OP is running overclocked (> 1600Mhz) memory, then it might be wise to to try running it closer to what the CPU considers stock speeds. My 3820 gives me better overclocks on the cores when I'm not trying to push the IMC incredibly hard. My motherboard also acts incredibly funky with offset voltages using BCLK straps, so forcing a constant voltage is the only good way to go about it. Also, at clocks this high, my 3820 becomes more sensitive to the VCCSA voltage. I tend to scale the VCCSA to how much I change the core voltage. Running it too low in relation to the core voltage feels like it tends to add to instability.

Either way, the only way you're going to get this CPU over 4.4Ghz (well,) is to use the 125Mhz BCLK strap, if it's going to behave like the 3820 where it's acting like it has a limited unlocked multiplier.

 

[fullinfusion]

Done. No 44x boot

I bought an asus board some time ago and had an I7 2700K and it wouldn't do a x50 multi, It did once and then after a reboot it was a no go.

Ended up having to pop out the CMOS battery and reset the bios. After that it would do X50 + but after any power down or BSOD the board wouldn't set the multi.

Turned out it was a bum board and returned it and got something else and never had a problem after that.

 

[RejZoR]

Change the CPURAM ratio so that RAM speed is closer to stock speed or even below stock, just to see if RAM is limiting you or not.

 

[cdawall]

You're at a point where you should ask yourself, is additional 0.1V worth the extra 300MHz and potential instability. Because 1.36V is pretty high. Not quite 1.4V yet, but very close... Or stick with 4.3GHz and try to get voltage even lower...

I am at 1.42 on a 6850K they are fine at that voltage.