Question about AMD's ACC (Advanced Clock Calibration)

[265586888]

We know that AMD's ACC solution in SB710 and SB750 southbridges are very popular now, not only because it can boost the overclocking capabilities of the processor, but with the corresponding version AGESA code in the BIOS, CPUs which are having one or two cores disabled can be brought back to life.

So, as usual, I googled, and found one Anadtech preview and one TGDaily report and a report from AMDZone quoting an removed article from Overclockers.com, some highlights as usual:



TGDaily report:

Six “hidden pins” on the processors and chipsets are the secret, which, our sources told us, will enable simple overclocking through the southbridge – and accelerate the current processors on demand.

AMD's SB750: Enabling Higher Phenom Overclocks?:

The SB750 now has a direct 6-pin interface to the AM2+ socket on the motherboard, there are now pins on the Phenom CPU that connect directly to the SB750. These pins were previously unused and are now used as a means of communication between the South Bridge and the CPU. The SB750, in combination with an updated BIOS, can now override some of the CPU's internal settings which can potentially increase the overclocking headroom of the chip.

AMD says that the settings tweak doesn't impact performance and doesn't change thermals or voltages, it simply can allow a Phenom processor to clock higher when overclocking. The BIOS exposes the parameter being changed, which AMD refers to as the Advanced Clock Calibration (ACC) value. Typically this value has a range of -2 to 0, on motherboards with the SB750 that support ACC the value can be set from -12 to +12. Higher numbers should allow for higher clock speeds, while lower values should allow for lower voltages/lower power operation.

Even after further pressuring, AMD wouldn't tell us what this value actually adjusts - simply stating that it makes it easier for the CPU to run at higher speeds.

AMDZone report:

Overclockers believes (dead link) that AMD's new ACC overclocking feature in the SB750 is indeed clock skewing.

There are certain manual adjustments that can be made in the BIOS of certain mobos which can serve to adjust these timings and allow a CPU to run at a higher speed than it could otherwise. This is very similiar to loosening the timing on memory. These adjustments go under the name "clock skew," and you can read some more about one form of it here, and see an example of clock skew controls here.

And from a magazine I've read today (in Chinese, translated below),

As for the actual working principles of ACC, since AMD did not disclose much about it, so it's still a mystery, we only know that the southbridge has an integrated microcontroller. And as enthusiasts discover that ACC can make Phenom II X2/X3 to become a quad-core CPU, the functions of ACC became more bewildering.

Anyone had any ideas about how this ACC thing actually worked? Don't seem to find any further information about this over the net.

[charlie]

Originally Posted by 265586888

Anyone had any ideas about how this ACC thing actually worked? Don't seem to find any further information about this over the net.

I can't shed any more light on it than Anand did. If you think about it, OCing one part, IE the CPU, usually does not change the frequencies on other parts. The way I understood it is that it allows parts like the NB and the SB to change internal multipliers more than they normally would be able to so they won't be forced to speeds higher than they can take.

When you jack up the speed of an AMD core, if you have to keep the multiplier within a certain range or you push the HT links too high, and the machine locks. If you don't have enough of a multiplier, your OC becomes limited by the HT link even if your CPU can go higher.

If you substitute SB for HT, you kind of get the idea. Pushing the CPU too high may hit multiplier limits on the chipset, NB or SB, and force them to speeds that they are not meant to go.

The non-specific way ACC was explained is that it expands the range of internal multipliers so things that really don't matter can be kept in a sane range, and the machine does not lock. The direct link from the CPU to the SB allows them to be set directly and transparently.

I will try and get some more direct info when I get in front of AMDers next.

-Charlie

[265586888]

Thanks.

[y eye]

Originally Posted by 265586888

We know that AMD's ACC solution in SB710 and SB750 southbridges are very popular now, not only because it can boost the overclocking capabilities of the processor, but with the corresponding version AGESA code in the BIOS, CPUs which are having one or two cores disabled can be brought back to life.

So, as usual, I googled, and found one Anadtech preview and one TGDaily report and a report from AMDZone quoting an removed article from Overclockers.com, some highlights as usual:



TGDaily report:


AMD's SB750: Enabling Higher Phenom Overclocks?:


AMDZone report:


And from a magazine I've read today (in Chinese, translated below),


Anyone had any ideas about how this ACC thing actually worked? Don't seem to find any further information about this over the net.

[PCGH]: Phenom II AM3 CPU, MotherBoard Lists update ( 1 2)

[265586888]

Originally Posted by y eye

[PCGH]: Phenom II AM3 CPU, MotherBoard Lists update ( 1 2)

You know why I didn't include those BIOS update information/news in my post?
Because I don't see what I wanted to know in those articles/news.

What I want to know is the working principle of ACC, not what ACC can do for consumers.

But, anyway, thank you for letting me to know ACC can even unlock the multiplier and voltage restrictions on non-black edition processors, even though I knew it when the news came out on day one.

[Saaya]

i think its syncing the cores to each other and to the integrated northbridge/L3/imc and maybe also the external chipset...
signal termination/drive strength... something like that...

[rich wargo]

It looks like Charlie has the right scope on it. The ACC feature actually allows the CPU frequency to be boosted without needing to boost any other clock. The link to the southbridge allows the southbridge clock to be decoupled from the CPU clock to keep the southbridge stable.

[Saaya]

Originally Posted by rich wargo

It looks like Charlie has the right scope on it. The ACC feature actually allows the CPU frequency to be boosted without needing to boost any other clock. The link to the southbridge allows the southbridge clock to be decoupled from the CPU clock to keep the southbridge stable.

but afaik the southbridge isnt even connected to the cpu... its connected to the northbridge which is then connected to the southbridge...
so how exactly does the southbridge cause the cpu to crash at certain clockspeeds when its not even directly connected to it?

unless the southbridge generates or manipulates some clock signals for the cpu... but why would that be done in the sb and not cpu directly? wouldnt the best way to fix this then be to move this to the cpu like intel did instead of improving the connection between sb and cpu?

[265586888]

Originally Posted by Saaya

but afaik the southbridge isnt even connected to the cpu... its connected to the northbridge which is then connected to the southbridge...

unless the southbridge generates or manipulates some clock signals for the cpu... but why would that be done in the sb and not cpu directly? wouldnt the best way to fix this then be to move this to the cpu like intel did instead of improving the connection between sb and cpu?

ACC also has 6-pins between CPU and southbridge.

AMD's SB750: Enabling Higher Phenom Overclocks?:

The SB750 now has a direct 6-pin interface to the AM2+ socket on the motherboard, there are now pins on the Phenom CPU that connect directly to the SB750. These pins were previously unused and are now used as a means of communication between the South Bridge and the CPU. The SB750, in combination with an updated BIOS, can now override some of the CPU's internal settings which can potentially increase the overclocking headroom of the chip.

I think ACC is some dedicated hardware on the southbridge that monitors the internal multipliers or something like that in the CPU, and can send signals to the CPU to control the internal multipliers to a reasonable range, that's why the 6 previously unused pins. And the rest just like what Charlie said.

I think moving ACC to the CPU may need a complete redesign of the CPU, which is more risky for overclocking (more complex ICs tend to be harder to do overclocking).

[steven444]

265586888 u have a great knowledge about electronics thanks for your informative reply