Sandy-Bridge news update: Sub-6s 1M at 6250MHz and 7GHz theoretically possible!

The Coolaler forums are known for the pre-NDA news updates of upcoming hardware releases, especially when it comes to Intel hardware. With the upcoming Sandy-bridge, they have again leaked some results.

The most important results, for us overclockers, are without any doubt the 7.488s at 5GHz SuperPI 1M. This means that a result below 6 seconds could be possible at only 6.25GHz, which is significantly lower than the current 7GHz requirement.

Next to that, we see that the maximum CPU ratio is no longer the 57x we all feared, but has been increased to at least 65x. Taking into account the rather limited BCLK overclocking capabilities, we can estimate a theoretical maximum frequency of 65x110MHz = 7150MHz. Of course, we hope the BCLK frequency will not be limited as rumours seem to tell us.

Screenshots below.


20

Belgium Massman says:

Still hoping the BCLK limitations will get fixed :)

Croatia Bullet says:

6.25GHz for sub6sec? Yeah but without tweaks I guess. 6.1GHz would be enough for getting 5.xxx I think. Hope the same for BCLK btw but this looks good too.

Belgium Massman says:

Yeah, possibly even lower than 6.1G. It was a really rough estimation :)

Netherlands diezel says:

nice

Denmark noxon says:

So about 7-9% faster clock for clock than 980X in 1M.

unknown Hollywood says:

Massman said: Still hoping the BCLK limitations will get fixed :)


Yep! Nice Vcore btw... :)

Belgium Massman says:

I guess this might also change the game in 32M a bit. Will be interesting for sure.

Sweden Calathea says:

It looks too good to be true. If it is, poor AMD...

TaPaKaH says:

0.981 is the VID and not the actual voltage .. latest CPUz can't read Vcore with SB correctly

Russian Federation Antinomy says:

Massman said: Still hoping the BCLK limitations will get fixed :)
So, ATM overclocking for the masses is killed? I mean if using regular CPUs.

Are there any high-res pics of the boards?

Belgium Massman says:

Antinomy said: So, ATM overclocking for the masses is killed? I mean if using regular CPUs.

Are there any high-res pics of the boards?


There will be two K-series SB's, which are 'designed for overclocking'. Marketing folks are being overly enthusiastic about the upcoming K chips, because they think those are binned for overclocking and go totally nuts. Reality is that the problem lies in the BCLK overclocking limitations: for some reason, increasing the BCLK 10% higher than stock is already impressive.

Rumours go it's a DMI problem. Other people just say it's something inside the CPU that doesn't like a high frequency.

Russian Federation Antinomy says:

I've heard from Intel fellows that SB chipsets will have an integrated clock gen so it won't be possible to move the frequency even a bit. And you tell about up to 10%. So it can be changed, but there are some issues? That's already great news because it makes solving the problem much easier. I think it can be managed.

Russian Federation Antinomy says:

Kill the dupe, it's my Internet's fault :(

Belgium Massman says:

Antinomy said: I've heard from Intel fellows that SB chipsets will have an integrated clock gen so it won't be possible to move the frequency even a bit. And you tell about up to 10%. So it can be changed, but there are some issues? That's already great news because it makes solving the problem much easier.

I think it can be managed.


No it will be possible. The clockgen has indeed moved into the chipset and Intel is allowing changes through the bios (per 0.1MHz LOL!). There's just something holding back the overclocking really badly.

United States Bobnova says:

Didn't they link everything directly to the bclk? Sata/usb/pci/pcie along with the DMI (PCIe by another name) and QPI (if it still exists).
That's what was being said some months ago.

Russian Federation Antinomy says:

Massman, better than nothing. Looks like an interesting case to solve. Bobnova BCLCK and USB have different frequencies. I assume you meant that they are all running sync and pushing the BCLCK will raise all other frequencies. Well, this we've already seen on i9x5 chipsets. I'd like to see a photo of a mainboard and/or I2C registers dump.

Belgium Massman says:

Bobnova said: Didn't they link everything directly to the bclk? Sata/usb/pci/pcie along with the DMI (PCIe by another name) and QPI (if it still exists).
That's what was being said some months ago.


That's, as far as I know, how it still is.

Some people are pointing to the DMI protocol when blaming for the low BCLK overclocking capabilities. Although being very similar to PCIe, there are som extra bits which might be causing trouble.

The again, everything is derived from this one 100MHz BCLK frequency ... i can imagine on-chip USB, Sata, Audio will not like that.

If it really is the DMI, or something on-die, I wonder if it'd be possible just to disable all. Oh well, guess we'll have to wait and see.

Antinomy said: Bobnova BCLCK and USB have different frequencies. I assume you meant that they are all running sync and pushing the BCLCK will raise all other frequencies. Well, this we've already seen on i9x5 chipsets.
I'd like to see a photo of a mainboard and/or I2C registers dump.


Gigabyte has shown their P67 boards already ;)

Russian Federation Antinomy says:

I meant - good quality photos ;) The ones that were shown are not good enough.

Poland Yotomeczek says:

This one can help? http://pclab.pl/zdjecia/artykuly/yotomeczek/1155/ASUS_Pre/f/IMG_4992-big.jpg

FlanK3r says:

the highest busspeed what i seen is 108 MHz to time....Maybe 110 is really posible with a bit luck.

Please log in or register to comment.

Leave a Reply: (BBCODE allowed: [B], [QUOTE], [I], [URL], [IMG],...)