Rejoice! Base Clock Overclocking to Make a Comeback with "Skylake"

[R0H1T]

I wonder if Intel is going to push a microcode update that breaks this, similar to what they did on the Pentium G3258.

IMO they will stomp this at the time of Kabylake launch, there is recent precedent wherein they blocked non Z OC on x87 & x97 chipsets firstly with MS' Windows updates & then BIOS updates from MB makers, that put a final stop to it. I'd say they're just looking to drive up the sale of Skylake & will eventually block this unless of course Zen is a major blockbuster

Obviously I'm hoping for the latter but even then there's a slim(mer) chance that Intel will curb BCLK OC just to push more Kabylakes out

 

[R-T-B]

I don't get the purpose of BCLK overclocking. I just slammed multi of 45 on my 5820K and be done with it running at 4.5 GHz. Why would you want to fiddle with the BCLK when overclocking using multi is so much easier and frankly not any worse? Cache is overclockable separately so you're also pushing the throughput faster.

The purpose here is to get around locked multis. Now you can aparently do it pretty well, and without wrecking the igpu too.

 

[geon2k2]

2 days later ... TechSpot had a review since Friday.

This is really good, and indeed a good reason to upgrade to Skylake.
That OC i3 from TechSpot had similar performance to an i5 Haswell.

 

[trog100]

some good overclocking fun for those on a tight budget..

it kind of reminds me of days gone past..

trog

 

[Jegergrim]

Could someone with insight tell me a few things about Skylakes BLCK overclocking (Wall of text Warning):
1) From my understanding BLCK overclocking in the past made multiple subsystems unstable, not just PCI-E/DMI - for instance it changed some relation with the SATA ports etc. Am I wrong, and if so what is the case now?

Also, could someone explain what DMI inherets in skylake? As I am hoping it is the part of the system with the Sata ports etc.

2) http://overclocking.guide/wp-content/uploads/2015/12/slide-3-1024x576.png
Shows that it is still tied to DDR and System Agent - what influence will BLCK have on these? As in if you for instance have DDR4-2133 RAM, and it is set to 2133Mhz in bios, how then is the RAM affected by upping the BLCK with 5points (100->105)? Will the RAM clocks change relative to the BLCK, perhaps exponentially? Would you then be required to start your RAM at much lower clocks?

3) Is upping the BLCK WITHOUT upping any voltages dangerous to the system? Disregarding the instability/crashes you'll get. For instance upping multipliers without changing voltages isn't hurting anything on Sandybridge, other than giving you an unstable system - is this the case with Skylake's non-K BLCK overclocking, or should you be careful increasing this as it might damage components in the system if set too high?

Perhaps more people than I wonder about the same things,
Thanks in advance

 

[trog100]

Could someone with insight tell me a few things about Skylakes BLCK overclocking (Wall of text Warning):
1) From my understanding BLCK overclocking in the past made multiple subsystems unstable, not just PCI-E/DMI - for instance it changed some relation with the SATA ports etc. Am I wrong, and if so what is the case now?

Also, could someone explain what DMI inherets in skylake? As I am hoping it is the part of the system with the Sata ports etc.

2) http://overclocking.guide/wp-content/uploads/2015/12/slide-3-1024x576.png
Shows that it is still tied to DDR and System Agent - what influence will BLCK have on these? As in if you for instance have DDR4-2133 RAM, and it is set to 2133Mhz in bios, how then is the RAM affected by upping the BLCK with 5points (100->105)? Will the RAM clocks change relative to the BLCK, perhaps exponentially? Would you then be required to start your RAM at much lower clocks?

3) Is upping the BLCK WITHOUT upping any voltages dangerous to the system? Disregarding the instability/crashes you'll get. For instance upping multipliers without changing voltages isn't hurting anything on Sandybridge, other than giving you an unstable system - is this the case with Skylake's non-K BLCK overclocking, or should you be careful increasing this as it might damage components in the system if set too high?

Perhaps more people than I wonder about the same things,
Thanks in advance

edit double post f-ck up

trog

 

[trog100]

Could someone with insight tell me a few things about Skylakes BLCK overclocking (Wall of text Warning):
1) From my understanding BLCK overclocking in the past made multiple subsystems unstable, not just PCI-E/DMI - for instance it changed some relation with the SATA ports etc. Am I wrong, and if so what is the case now?

Also, could someone explain what DMI inherets in skylake? As I am hoping it is the part of the system with the Sata ports etc.

2) http://overclocking.guide/wp-content/uploads/2015/12/slide-3-1024x576.png
Shows that it is still tied to DDR and System Agent - what influence will BLCK have on these? As in if you for instance have DDR4-2133 RAM, and it is set to 2133Mhz in bios, how then is the RAM affected by upping the BLCK with 5points (100->105)? Will the RAM clocks change relative to the BLCK, perhaps exponentially? Would you then be required to start your RAM at much lower clocks?

3) Is upping the BLCK WITHOUT upping any voltages dangerous to the system? Disregarding the instability/crashes you'll get. For instance upping multipliers without changing voltages isn't hurting anything on Sandybridge, other than giving you an unstable system - is this the case with Skylake's non-K BLCK overclocking, or should you be careful increasing this as it might damage components in the system if set too high?

Perhaps more people than I wonder about the same things,
Thanks in advance

i think the skylake base clock can be altered on its own simply to alter the cpu clock.. 100 x 40 equals 4 gig.. 125 x 40 equals something different.. in the past altering the base clock altered everything else.. now it dosnt..

in short it simply enables none K chips to be overclcocked without the need to change the multiplyer which intel have locked.. it should not alter anything else.. the same overclocking rules will apply.. the chip speed is just being arrived at in a different way..

its quite big news in its own way.. it kind of negates point of the more expensive K chips.. and pokes a big hole in intels speed regulated pricing structure on the none K chips....

we are talking basically the same chips being sold at different speeds at different prices all controlled by intel.. being able to overclock such chips is a big game changer..

trog

 

[Sempron Guy]

still needs a Z170 to adjust bclk right? Unless they'll make it available for "budget" skylake compatible board then it won't be even near the realms of "budget" overclocking we come to love back in the old days.

 

[Steven B]

Could someone with insight tell me a few things about Skylakes BLCK overclocking (Wall of text Warning):
1) From my understanding BLCK overclocking in the past made multiple subsystems unstable, not just PCI-E/DMI - for instance it changed some relation with the SATA ports etc. Am I wrong, and if so what is the case now?

Also, could someone explain what DMI inherets in skylake? As I am hoping it is the part of the system with the Sata ports etc.

2) http://overclocking.guide/wp-content/uploads/2015/12/slide-3-1024x576.png
Shows that it is still tied to DDR and System Agent - what influence will BLCK have on these? As in if you for instance have DDR4-2133 RAM, and it is set to 2133Mhz in bios, how then is the RAM affected by upping the BLCK with 5points (100->105)? Will the RAM clocks change relative to the BLCK, perhaps exponentially? Would you then be required to start your RAM at much lower clocks?

3) Is upping the BLCK WITHOUT upping any voltages dangerous to the system? Disregarding the instability/crashes you'll get. For instance upping multipliers without changing voltages isn't hurting anything on Sandybridge, other than giving you an unstable system - is this the case with Skylake's non-K BLCK overclocking, or should you be careful increasing this as it might damage components in the system if set too high?

Perhaps more people than I wonder about the same things,
Thanks in advance

1)Intel effectivly removed Skylake's BLCK for the CPU, Cache, FLCK, and memory from the DMI/PCI-E. So the CPU's PCI-E clock and the DMI run at a constant 100MHz. The rest of the domains can run higher without affecting those key domains.
2) A=If your BLCK is 125MHz and your CPU multiplier of 36x, uncore/cache/ring/whatever you want to call it of 36x, memory of 24x, and FLCK of 8x is equal to=
B=If your BLCK is 100MHz and your CPU multiplier of 45x, uncore/cache/ring/whatever you want to call it of 45x, memory of 30x, and FLCK of 10x is equal to

Both A and B will give you CPU @ 4.5GHz, uncore/cache/ring/whatever you want to call it of 4.5GHz, memory of 3GHz, and FLCK of 1GHz . However, in both cases your PCI-E and DMi are locked at 100Mhz.

3) While this is all great and all, the higher binned 6700K still seems to overclock better on average than the 6600K which would lead me to believe that they will continue to OC further. However, OCing non-K SKUs means you will OC a lot of dual core CPUs where you have less chance of getting a worse core, most dual cores should OC higher than quad cores just like in the past.

I want to add that the Skylake engineers worked quite hard to implement the BLCK OCing capabilities of SKylake from their own testimony during IDF, but most of us there were kind of like, well why? Multiplier OC is simpler and latency is usually better with multiplier OC (also higher BLCK = more jitter = less stability), but if they knew that this hack to the UEFi was possible and eventually the board maker's BIOS engineers would figure it out, their excitement was 100% warranted but i guess they couldn't tell us exactly why.

BTW manufacturers are already releasing their UEFIs, ASRock already released a UEFi to OC non-K SKUs on their OC Formula.

 

[rtwjunkie]

ask them also, if they feel cheated now, for paying 10% more for no reason at all?

How will they be cheated? K-series will still be the overclocking champs because of being able to use baseclock and multipliers to boost overclocking.

 

[TheinsanegamerN]

right. in what world being able to log into your desktop computer BIOS remotely, (while it's powered off even) is not good for enthusiasts? or take over the picture of your desktop computer on hardware level without any screen sharing software being installed is not beneficial to enthusiast? Or trusted execution for safer execution of shady stuff?

The only reason people bought K series was because they valued OC more than those features... didn't mean they didn't like those features.

For that matter why don't gamers buy CPUs with VT-D enabled, or computers with smart card and biometric readers built in?!?!? INTEL HAS STOLEN OUR CHOICE!

Or, in reality, these are all things that business users use. Gamers have no need for business features, remote security, they need raw performance. The few gamers who DO want this stuff buy xeons. Most people dont want to pay for those features that they dont use. Same reason gamers dont buy quadro or firepro cards, or ECC memory.

 

[wolar]

How will they be cheated? K-series will still be the overclocking champs because of being able to use baseclock and multipliers to boost overclocking.

Does it matter ? it stills feel like intel played this for many generations just to get more money with -k- processors being higher priced. Even though the other processors could overclock too , if intel actually wanted and make it easy.
Next cpu gonna be AMD for sure.

 

[rtwjunkie]

Does it matter ? it stills feel like intel played this for many generations just to get more money with -k- processors being higher priced. Even though the other processors could overclock too , if intel actually wanted and make it easy.
Next cpu gonna be AMD for sure.

Really? There has ALWAYS been a premium for the best binned, highly overclockable chips. Go back to QX or X days in the Core 2 era. If you want the best, you pay for it. It's pretty much like that with anything in the world you buy.

 

[R0H1T]

Really? There has ALWAYS been a premium for the best binned, highly overclockable chips. Go back to QX or X days in the Core 2 era. If you want the best, you pay for it. It's pretty much like that with anything in the world you buy.

Except that K parts aren't the best bins, T & S are

 

[Jegergrim]

1)Intel effectivly removed Skylake's BLCK for the CPU, Cache, FLCK, and memory from the DMI/PCI-E. So the CPU's PCI-E clock and the DMI run at a constant 100MHz. The rest of the domains can run higher without affecting those key domains.
2) A=If your BLCK is 125MHz and your CPU multiplier of 36x, uncore/cache/ring/whatever you want to call it of 36x, memory of 24x, and FLCK of 8x is equal to=
B=If your BLCK is 100MHz and your CPU multiplier of 45x, uncore/cache/ring/whatever you want to call it of 45x, memory of 30x, and FLCK of 10x is equal to

Both A and B will give you CPU @ 4.5GHz, uncore/cache/ring/whatever you want to call it of 4.5GHz, memory of 3GHz, and FLCK of 1GHz . However, in both cases your PCI-E and DMi are locked at 100Mhz.

3) While this is all great and all, the higher binned 6700K still seems to overclock better on average than the 6600K which would lead me to believe that they will continue to OC further. However, OCing non-K SKUs means you will OC a lot of dual core CPUs where you have less chance of getting a worse core, most dual cores should OC higher than quad cores just like in the past.

I want to add that the Skylake engineers worked quite hard to implement the BLCK OCing capabilities of SKylake from their own testimony during IDF, but most of us there were kind of like, well why? Multiplier OC is simpler and latency is usually better with multiplier OC (also higher BLCK = more jitter = less stability), but if they knew that this hack to the UEFi was possible and eventually the board maker's BIOS engineers would figure it out, their excitement was 100% warranted but i guess they couldn't tell us exactly why.

BTW manufacturers are already releasing their UEFIs, ASRock already released a UEFi to OC non-K SKUs on their OC Formula.

I see, how about the dangers of the overclock itself - is it as neutral as simply changing a multipler without touching voltages - or does it actually affect components dangerously?

i think the skylake base clock can be altered on its own simply to alter the cpu clock.. 100 x 40 equals 4 gig.. 125 x 40 equals something different.. in the past altering the base clock altered everything else.. now it dosnt..

in short it simply enables none K chips to be overclcocked without the need to change the multiplyer which intel have locked.. it should not alter anything else.. the same overclocking rules will apply.. the chip speed is just being arrived at in a different way..

its quite big news in its own way.. it kind of negates point of the more expensive K chips.. and pokes a big hole in intels speed regulated pricing structure on the none K chips....

we are talking basically the same chips being sold at different speeds at different prices all controlled by intel.. being able to overclock such chips is a big game changer..

trog

Thanks, that does indeed explain alot of my questions. The question about the safety of BLCK vs Multiplier, disregarding voltages though, any insight on that?

Thanks again

 

[Frick]

Really? There has ALWAYS been a premium for the best binned, highly overclockable chips. Go back to QX or X days in the Core 2 era. If you want the best, you pay for it. It's pretty much like that with anything in the world you buy.

But you could overclock the others chips. That was the point of it. Buy a lowly E4300, mod it and ramp it up to E6600 speeds. To me Intel definitely killed overclocking. Instead of making slow chips fast you make the fastest chips even faster.

 

[rtwjunkie]

Except that K parts aren't the best bins, T & S are

And you get this info from where? Endless discussions on numerous forums? Here is the info straight from Intel. They have different purposes, period.
http://www.intel.com/content/www/us/en/processors/processor-numbers.html

And some more info from Anand wihich explains the binning process and does testing on T and S. Nowhere is the assertion made that S and T are higher binned. http://www.anandtech.com/show/8774/...ew-core-i3-4130t-i5-4570s-and-i7-4790s-tested

But you could overclock the others chips. That was the point of it. Buy a lowly E4300, mod it and ramp it up to E6600 speeds. To me Intel definitely killed overclocking. Instead of making slow chips fast you make the fastest chips even faster.

Yes you could, but it was always a lot harder and a lot more juggling to do it, rather than the chips designated for overclocking making it a lot easier to get a stable overclock. I had, for example, Q9650 and QX9650. The QX was much easier to work with. Sure, I got an e8400 to almost 4Ghz, but damn, I had to really work at it!

 

[BarbaricSoul]

Yes you could, but it was always a lot harder and a lot more juggling to do it, rather than the chips designated for overclocking making it a lot easier to get a stable overclock. I had, for example, Q9650 and QX9650. The QX was much easier to work with. Sure, I got an e8400 to almost 4Ghz, but damn, I had to really work at it!

Damn man, the E8400 I had easily hit 4 GHz and was able to hit up to 4.25GHz at 24/7 safe voltages and temps. And that was with a Nvidia 750i chipset board (EVGA 750 FTW). The Q9650 I had with a P45 chipset Gigabyte board that everyone said was one of the best OC'ing boards for C2Q CPUs (UD3P iirc) was able to hit 4.5 GHz (I was running it at 4.25 GHz 24/7 when I upgraded to a 2600K). Keep in mind, I am far from being considered a OC'ing expert.

 

[rtwjunkie]

And that was with a Nvidia 750i chipset board (EVGA 750 FTW).

That's what I was using for a board on it! If your memory hasn't jaded you, that board was like trying to master witchcraft and Newtonian Physics! It was one of the crankiest, finicky boards that required a Voodoo Priestess to be present anytime you had to play with it and push it further. Don't get me wrong, I loved that board, but DAMN was it painful at the same time!

So you see why I only got to 3.9. I was not an overclocking expert...hell I was still a beginner, and had read all 180 pages of the thread on overclocking that board on OCN, and I probably didn't have the best binned chip either.

 

[Frick]

Yes you could, but it was always a lot harder and a lot more juggling to do it, rather than the chips designated for overclocking making it a lot easier to get a stable overclock. I had, for example, Q9650 and QX9650. The QX was much easier to work with. Sure, I got an e8400 to almost 4Ghz, but damn, I had to really work at it!

So you're saying they were less fun? If you wanted to truly max it out, sure you had to juggle (like you have to do with the K cpu's as well, as I understand it), but I bet you got it to 3.7 without much effort. The first thing I did with my €95 1.8ghz e4300 on my cheap motherboard was running it on 2.4Ghz. Ditto with the Venice Athlon I had. If you had a thousand euros to spend on an Extreme CPU, sure why not, but you could have just as much fun with a sub-hundred euro CPU, and overclock it so it was faster than stuff twice as expensive.

Things are different now, it wouldn't be the same thing even if you could overclock like you used to, but still. Intel killed overclocking.

 

[rtwjunkie]

@Frick, I have to admit, my most fun CPU to overclock was the e4600! All those 4 series were hidden gems!

But keep in mind, speed wasn't everything. There were still things those lowly overclcockers could not do well at a certain point.

 

[Parn]

But you could overclock the others chips. That was the point of it. Buy a lowly E4300, mod it and ramp it up to E6600 speeds. To me Intel definitely killed overclocking. Instead of making slow chips fast you make the fastest chips even faster.

Exactly.

I only do mild OC because most of the time the bottleneck of my system lies somewhere else (e.g. graphics card, hdd, etc.). I got my current 4690K and 6600K because only the K version allows OC (didn't know bclk OC is back with skylake). Back in the C2Q/Athlon era, I used to get a lower clocked CPU, OC it to match the more expensive parts (AXP 1700 to 2100, Q6600 to 3.33, Q9550 to 3.8), and then put the cash saved towards a better graphics card.

 

[RejZoR]

Also, when overclocking with BCLK you'll be constantly struggling with stability unless you pump higher voltage even for idle, you can't use power saving features, your idle clock will be higher and because of all this, your CPU will be eating tons more power with no realistic benefits.

How many people with unlocked CPU's also perform overclocking? Not many I'd say. They have crab motherboards that aren't designed for overclocking anyway, they don't have coolers capable for doing it anyway and thos who do already buy unlocked stuff.

 

[BarbaricSoul]

That's what I was using for a board on it! If your memory hasn't jaded you, that board was like trying to master witchcraft and Newtonian Physics! It was one of the crankiest, finicky boards that required a Voodoo Priestess to be present anytime you had to play with it and push it further. Don't get me wrong, I loved that board, but DAMN was it painful at the same time!

So you see why I only got to 3.9. I was not an overclocking expert...hell I was still a beginner, and had read all 180 pages of the thread on overclocking that board on OCN, and I probably didn't have the best binned chip either.

Yeah, I understand, the Nvidia chipset boards could be quite finicky to OC on, not to mention the FSB wall they all seemed to have. I think I just got lucky in the silicon lottery with the FTW board and E8400.

 

[Eiswolf93]

And what about Xeon V5?

If the upcomming Mainboards from Asus and Gigabyte with C232 Chipset get also a rebuild Bios the E3-1230 v5 could be very interesting! If someone needs Hyperthreading and ECC Memory.

E3-1230 v5 + C232 Mainbaord for around 370 Bucks. A fully unlocked Skylake Quad Core with more than 4,5 Ghz would be a solid system to build on.