Is BCLK Overclocking back on LGA1151?

[Lotus]

Shortly before release, several articles popped up saying that since Intel moved a bunch of stuff off the chip and back into the motherboard and isolated the PCIE bus from the baseclock that BCLK overclocking on Skylake was back, even on multiplier locked CPUs. Now, there are no locked CPUs released, but has anyone tried a BCLK OC on their K series CPU just to see? I know it's pointless since you can just use the multiplier, but I'm sure lots of people who would be getting an i5-4460 or 4590 would like to know if it's worth it to wait and get the skylake equivalent if you can OC it via BCLK. I mean, that would probably make for a significant performance jump in the cheaper locked CPUs even if the most expensive CPUs didn't get a big jump.

 

Here's what I'm referring to:

http://www.overclock3d.net/reviews/cpu_mainboard/intel_skylake_i5_6600k_i7_6700k_1151_z170_review/4

[Aniallation]

What do you mean is back? It never left?

[Lotus]

What do you mean is back? It never left?

Yes it did. From Sandy Bridge to haswell you couldn't adjust the BCLK by more than +/- 5% (default BCLK = 100) without going unstable due to unstable buses, and even then a BCLK of 105 was dangerous and not always stable. I ran with a BCLK of 187 on my i7-920. That's from 2.66 stock speed to a 3.94 GHz overclocked speed on a locked CPU. You can't overclock locked haswell CPUs because the multiplier is locked and increasing the BCLK leads to horrible instability as mentioned. It's looking like with Skylake you can once again overclock non-K CPUs, but I want to be sure first.

[BradH]

yes it is oc3d tv did a video on it

[typographie]

What do you mean is back? It never left?

 

The Z170 PCIe and I/O frequencies are locked to 100 MHz independent of the BCLK, meaning you can now adjust the BCLK without affecting all the other dependent frequencies that made BCLK overclocking effectively impossible on Haswell.

[Lotus]

yes it is oc3d tv did a video on it

So basically once the Skylake locked i5 chips come out, they will significantly outclass their haswell counterparts once you overclock, depending on binning of course.

 

Plus this means we're back to overclockable Xeons! WOOOOOOOOOOOOOOOOOOO!!!!!!

[typographie]

So basically once the Skylake locked i5 chips come out, they will significantly outclass their haswell counterparts once you overclock, depending on binning of course.

 

Plus this means we're back to overclockable Xeons! WOOOOOOOOOOOOOOOOOOO!!!!!!

 

Depending on whether or not Intel has some other trick up their sleeve to prevent you from overclocking non-K processors. Which I expect they do.

[Megahurt]

You definetly can on unlocked cpus, I wonder if intel managed to lock it on their locked cpus.

[Lotus]

You definetly can on unlocked cpus, I wonder if intel managed to lock it on their locked cpus.

I don't see how the CPU could do that. BCLK is a function of the motherboard and chipset, and the CPU has no idea of how fast it's going (except in relation to the BCLK). That's the entire point of the baseclock: so the CPU doesn't have to keep its own time but can just reference the same clock the other components do. The only way I can see this getting locked down is with some EXTREMELY consumer unfriendly chipset shenanigans, which hopefully board partners will find a way to circumvent a-la non-Z OCing of the G3258.

[Imakuni]

I don't see how the CPU could do that. BCLK is a function of the motherboard and chipset, and the CPU has no idea of how fast it's going. That's the entire point of the baseclock: so the CPU doesn't have to keep its own time but can just reference the same clock the other components do. The only way I can see this getting locked down is with some EXTREMELY consumer unfriendly chipset shenanigans, which hopefully board partners will find a way to circumvent a-la non-Z OCing of the G3258.

They don't need to "lock" it. In fact, it wasn't locked on Haswell either. It was just that the CPU was unstable to run when BCLK OCed.

 

The same could be said about Skylake. They just need to get chips on the very edge of being unstable. If you try to OC, you'll just run into instability. Afterall, no matter how good your mobo is, if your chip is bad, you get a crap OC. Period.

[Lotus]

They don't need to "lock" it. In fact, it wasn't locked on Haswell either. It was just that the CPU was unstable to run when BCLK OCed.

 

The same could be said about Skylake. They just need to get chips on the very edge of being unstable. If you try to OC, you'll just run into instability. Afterall, no matter how good your mobo is, if your chip is bad, you get a crap OC. Period.

True, but that's highly unlikely that intel would risk selling chips that are just barely binned high enough to reach their stock speeds at high voltages. Remember, OCing is more than just adjusting timing but voltages too. Also, as has been said plenty, the reason why Haswell BCLK overclocking resulted in unstable chips (bus speeds being tied to the BCLK) is now gone, and it's more akin to the Nehalem days like my i7-920 (locked multiplier stock 2.66 GHz, I had mine at 3.94 GHz).

 

Basically, this is when we gamers are actually about to get an overclockable i3, and I don't see enough people mentioning it. This is big news and it's flying under the radar right now. Plus the big spenders are going to have access to overclockable Xeons once 2011-3 is replaced. These are all hugely significant IMO.

[Imakuni]

True, but that's highly unlikely that intel would risk selling chips that are just barely binned high enough to reach their stock speeds at high voltages. Remember, OCing is more than just adjusting timing but voltages too. Also, as has been said plenty, the reason why Haswell BCLK overclocking resulted in unstable chips (bus speeds being tied to the BCLK) is now gone, and it's more akin to the Nehalem days like my i7-920 (locked multiplier stock 2.66 GHz, I had mine at 3.94 GHz).

 

Basically, this is when we gamers are actually about to get an overclockable i3, and I don't see enough people mentioning it. This is big news and it's flying under the radar right now. Plus the big spenders are going to have access to overclockable Xeons once 2011-3 is replaced. These are all hugely significant IMO.

Honesly, with the Skylake shortage due to newer manufacturing process (Broadwell doesn't exist, also the next Gen will be on 14nm as well), I can actually see the chips being binned close to the limit, just to try and keep up with the demand. Besides, it seems that motherboards are also very unoptimized (dem crazy stock volages....).

[Lays]

Honesly, with the Skylake shortage due to newer manufacturing process (Broadwell doesn't exist, also the next Gen will be on 14nm as well), I can actually see the chips being binned close to the limit, just to try and keep up with the demand. Besides, it seems that motherboards are also very unoptimized (dem crazy stock volages....).

 

 

I don't think that has anything to do with motherboards being unoptimized, from what I've seen, Skylake has a higher voltage tolerance as well as higher stock voltage.

 

Intel was telling reviewers under 1.5v is perfectly fine apparently according to the Skylake talk on OCN.

[Imakuni]

I don't think that has anything to do with motherboards being unoptimized, from what I've seen, Skylake has a higher voltage tolerance as well as higher stock voltage.

 

Intel was telling reviewers under 1.5v is perfectly fine apparently according to the Skylake talk on OCN.

And AMD tells people that 105ºC on a an R9 290x is fine. Nvidia also states that up to 95ºC on Gtx 970 is A-OK.

 

I just can't believe 1.5v is acceptable.

[Raudi_]

BCLK overclocking never left. It was just harder......

[Raudi_]

And AMD tells people that 105ºC on a an R9 290x is fine. Nvidia also states that up to 95ºC on Gtx 970 is A-OK.

 

I just can't believe 1.5v is acceptable.

We are not dealing with Haswell anymore, this is Skylake its a sku of CPU's that will take voltage 10x better then Haswell ever has or will.

[Lotus]

And AMD tells people that 105ºC on a an R9 290x is fine. Nvidia also states that up to 95ºC on Gtx 970 is A-OK.

 

I just can't believe 1.5v is acceptable.

Uh, no offense, but how is someone supposed to refute your incredulity? Let's be honest here, when was the last time you say a chip, GPU or CPU, die due to thermals or volts? Disregarding insane circumstances of course. Chips are way more resilient than you give them credit for, and AMD has been running their chips at these voltages for years. It's all down to architecture.

[Lotus]

BCLK overclocking never left. It was just harder......

What the shit? Are there absolutely NO components connected via PCIe lanes? Pretty much every haswell cpu should crash instantly with anything close to that. I'm calling shenanigans. Something weird happened to get that.

 

edit: also, I wouldn't say 1.54 GHz is an "overclock" so clearly even if that did work, it wouldn't actually do anything positive.

Edited August 31, 2015 by Lotus

[Lays]

What the shit? Are there absolutely NO components connected via PCIe lanes? Pretty much every haswell cpu should crash instantly with anything close to that. I'm calling shenanigans. Something weird happened to get that.

 

edit: also, I wouldn't say 1.54 GHz is an "overclock" so clearly even if that did work, it wouldn't actually do anything positive.

 

 

It depends on the motherboard.  Some boards have PLX chips built in and you can BCLK OC on them like a boss if you move the GPU into a slot that runs off the PCH.

[Imakuni]

Uh, no offense, but how is someone supposed to refute your incredulity? Let's be honest here, when was the last time you say a chip, GPU or CPU, die due to thermals or volts? Disregarding insane circumstances of course. Chips are way more resilient than you give them credit for, and AMD has been running their chips at these voltages for years. It's all down to architecture.

I'll believe Skylake can handle 1.5v just fine once I see a 3rd party review / test that can run a chip a such high voltage for hours of stress test without a problem. And without LN2 or other crazy cooling setup, just some regular watercooling setup.

[Raudi_]

What the shit? Are there absolutely NO components connected via PCIe lanes? Pretty much every haswell cpu should crash instantly with anything close to that. I'm calling shenanigans. Something weird happened to get that.

 

edit: also, I wouldn't say 1.54 GHz is an "overclock" so clearly even if that did work, it wouldn't actually do anything positive.

Was just doing HWBot stuff this was my system set up at the time also the 290x Lighting was running on the ONLY 16x CPU line the board has:

[Lays]

And AMD tells people that 105ºC on a an R9 290x is fine. Nvidia also states that up to 95ºC on Gtx 970 is A-OK.

 

I just can't believe 1.5v is acceptable.

 

 

Why would you not trust what the manufacturer tells you is safe? They're the ones designing and engineering the product and they're the ones who do thorough testing to make sure they know what it's limits are.

[Megahurt]

I don't see how the CPU could do that. BCLK is a function of the motherboard and chipset, and the CPU has no idea of how fast it's going (except in relation to the BCLK). That's the entire point of the baseclock: so the CPU doesn't have to keep its own time but can just reference the same clock the other components do. The only way I can see this getting locked down is with some EXTREMELY consumer unfriendly chipset shenanigans, which hopefully board partners will find a way to circumvent a-la non-Z OCing of the G3258.

They could have an agreement with motherboard partners to lock it in the bios, or it could be hardwired in the Z170 chipset, I'm sure they don't lack imagination to find a way to limit the amount of fun we can have with our hardware.

[Django]

What the shit? Are there absolutely NO components connected via PCIe lanes? Pretty much every haswell cpu should crash instantly with anything close to that. I'm calling shenanigans. Something weird happened to get that.

 

edit: also, I wouldn't say 1.54 GHz is an "overclock" so clearly even if that did work, it wouldn't actually do anything positive.

 

Some Haswell CPU's have their own dividers such that when BCLK is say 167MHz then PCIE clocks are 100MHz, google "bclk straps".

 

Clocks will be controlled via Management Engine (ME) firmware which may block bclk overclocking on non-k processors.

 

Skylake Vcc specs.

 

Usually with semiconductors too much voltage will kill a chip instantly. Too much current can be a bit more subtle, anywhere from degradation to snowballing into oblivion. This can also be compounded by high temperatures. Of course the more core voltage one uses then typically more current is used.

[OhJeebus]

The reason the OCs were so unstable in Haswell is because the put the voltage regulator on the CPU instead of keeping on the Mobo.  With Skylake they took it back off the CPU and put it on the mobo.  Thats why a BCLK of ~105 was risky...  I plan on getting the 3.9 I3 and seeing how high I can OC it on water.  Here is a good reference  guide that takes you step by step how to OC using BCLK (some of it might not be needed if Z170 chipsets have locked the pcie and other I/O at 100.

 

http://www.masterslair.com/determining-your-maximum-bclk-base-clock-frequency-i7-i5-i3