Lets talk breaking the HTT 230 limit on k10 here

[lukija]

So each core is fed with 18.9A at that voltage?

Beats me if I know. It says expected, but it is not clear under what load. Maybe average, or typical. Another quote:

"The values are expressed in amps; they are not intended to convey final product power levels; they may not match the power levels specified in the Power and Thermal Datasheets. These fields, may be subsequently altered by software; they do not affect the hardware behavior. These fields are encoded as follows:

IddDiv Current Equation Current Range
00b IddValue / 1 A 0 to 255 A
01b IddValue / 10 A 0 to 25.5 A
10b IddValue / 100 A 0 to 2.55 A"

4x20A * 1.2V = 96W , sounds familiar ? It would be interesting to check these values for 9700.

[KTE]

Core VID is always 1.250 stock. If you change that manually, it'll not bootup Meaning it must set the Amps fed to each core like you say, yep. And multiply amps by the core voltage gives you the power consumed under maximum lab load. That's where they get the TDP value from then. Very nice.

However, mine has set 1.2000VID many times randomly and that makes you get less HT reference MHz.

[eXa]

Slightly ot?

The Sapphire PC-AM2RD790 aint no different from this board at all right? so if getting the sapphire may be easier than the dfi one it wont matter, right?

[Tony]

its the same board yes

[Tony]

Core VID is always 1.250 stock. If you change that manually, it'll not bootup Meaning it must set the Amps fed to each core like you say, yep. And multiply amps by the core voltage gives you the power consumed under maximum lab load. That's where they get the TDP value from then. Very nice.

However, mine has set 1.2000VID many times randomly and that makes you get less HT reference MHz.

Changed up or down?

If this is the case it would be best to leave VID at default...1.25V and use the "special add" option say in the DFI genie bios only.

Thing is i have VID to 1.5V+...used 1.4 etc with no issue so i am thinking you mean undervolting?

[lukija]

OK, new updated version, new defaults according to KTE screen shots.

[DevilCry]

its the same board yes

today i got my sapphire board(pics).Is it possible to flash it with dfi bios?Or we have to wait to do your magic, again Tony?

[Tony]

Tony, while you're at it, can you check what happens if you change CPU multi to 11.5x or 12x through P-States?

If it's anything like on MSI, then it'll just stay at the maximum stock multiplier (11x on 9500).

If I change NB to past 9x multi through P-States, the system will downclock the NB/CPU/HT to keep below 2GHz NB. Like this: http://img403.imageshack.us/img403/9...wnclocksr2.png

My board has issues with AOD, only 2.10 works with it and still freezes the system up at times, hence why I can't really oc. I have to keep reinstalling it to keep it working well or it bugs out after a while and the shortcut disappears. It does however report everything correctly.

EDIT: the support forum in your sig is down.

After some testing i am stuck with a NBcore clock around 1950ish max, not had any luck hitting 2000MHZ+.

this looks to be be my max that i have to work around using NB FID etc to keep the NBcore clock low enough to bring the HT Ref clock (HTT) higher.

Issue is 240 is my stable max at this time...looks like i need a better CPU if im to push higher.

[Tony]

today i got my sapphire board(pics).Is it possible to flash it with dfi bios?Or we have to wait to do your magic, again Tony?

it may have the same bios m8...Check before you try flashing the DFI to it

[KTE]

Changed up or down?

If this is the case it would be best to leave VID at default...1.25V and use the "special add" option say in the DFI genie bios only.

Thing is i have VID to 1.5V+...used 1.4 etc with no issue so i am thinking you mean undervolting?

I can't change it down but can change slightly up. CPU VID 28 is 1.250V, but sometimes it is 1.2000V randomly. I tried CPU VID 29 and it showed as 1.1875V in AOD. 27 has no POST and 29 has reboot at Windows loadup.

Macci said earlier: "note typically CPUs are locked to 1.25V or 1.30V VID value so trying to set higher than this might result in non-BOOT"

Run HT 9x and NB 9x, both CPU/NB dividers to 1 (which is 0 in your BIOS I think), HT ref. to 240MHz. Try feed it approximately 1.35V CPU, 1.25V HT, 1.3V NB.

If HT ref. 240 is stable, it should bootup I'm thinking but you may need to increase PCIe to 105-109 to get it stable and be running like so:

CPU: 2640MHz
NB: 2160MHz
HT: 2160MHz

Like this:


Is that not possible on yours?

[tictac]

I can't change it down but can change slightly up. CPU VID 28 is 1.250V, but sometimes it is 1.2000V randomly. I tried CPU VID 29 and it showed as 1.1875V in AOD. 27 has no POST and 29 has reboot at Windows loadup.

Macci said earlier: "note typically CPUs are locked to 1.25V or 1.30V VID value so trying to set higher than this might result in non-BOOT"

Run HT 9x and NB 9x, both CPU/NB dividers to 1 (which is 0 in your BIOS I think), HT ref. to 240MHz. Try feed it approximately 1.35V CPU, 1.25V HT, 1.3V NB.

If HT ref. 240 is stable, it should bootup I'm thinking but you may need to increase PCIe to 105-109 to get it stable and be running like so:

CPU: 2640MHz
NB: 2160MHz
HT: 2160MHz

Like this:


Is that not possible on yours?

some Vcore VID register... Hope this help...

[KTE]

Ah no wonder. Guess what mine read? 1.20V



Oh and tictac, don't even think if it helps. It always does!
Thank you.

Tony, I can't boot higher NB anymore either. Below 2GHz only. But before, I had run higher NB perfectly and here's a comparison result attached.

[Tony]

so you could run >2000 for a while and now<2000 only?

[KTE]

Done upto 3.1GHz on NB alone and it was a little unstable. That was using P0D BIOS but only single channel RAM. Now using 113 and I can't boot plus 2GHz NB.

However, in windows using AOD I can still get the above NB/HT speed.

[Tony]

Ok sounds like we have source code issues in bios, I know of another issue also and have spoken to AMD already.

I will retry some NB core clocking again tomorrow, however if the source award/AMI code has issues it may be worthless.

[KTE]

It probably is because I've done more than 50 benches and P95 stability testing from 2GHz-3.1GHz NB at different CPU speeds, but now I can't boot even 2GHz and the HT ref. max bootable has also dropped by about 15MHz. Regardless of cooling or voltages. It'll be good to have a workaround quick.

[bingo13]

Ok sounds like we have source code issues in bios, I know of another issue also and have spoken to AMD already.

I will retry some NB core clocking again tomorrow, however if the source award/AMI code has issues it may be worthless.

Hit 2220 NB with the new Gigabyte BIOS tonight, anything faster results in a nice black screen. Benchmark stable is around 2040 Max, without higher NB speeds, this CPU runs like a pig in deep mud.

[macci]

2.6G NB (with 1.5V CPU NB VID) - aircooled

[muzz]

Will DC run at that setting Macci?

[macci]

This is DC. Unganged (2x64bit) is just detected as Single.

[KTE]

Nope. Changing NB/CPU VID is not possible with mine.
Neither can you get above 2GHz bootable NB speed.
If you use OD for it, t freezes at ~2080MHz on NB speed with stock VID but stock or max voltages. This is not how it used to be.

Is this a screenshot of the settings by an unlocked Phenom at 3GHz w/CnQ? http://www.amd.com/us-en/assets/cont...ing_expert.jpg

Impressive gains there macci.

[muzz]

This is DC. Unganged (2x64bit) is just detected as Single.

O yeah, forgot about that 1.

Nevermind.....

[Aussie FX]

Great to see some sound thinking going into this.

Well done to those concerned and thanks.

[KTE]

Aussie FX: You have a 9500, can you boot above 2GHz NB speed? (look at OverDrive to see what your core speeds read please)

[Aussie FX]

Aussie FX: You have a 9500, can you boot above 2GHz NB speed? (look at OverDrive to see what your core speeds read please)

Sorry KTE my motherboard has died, so I'm back to using a 6400 and crosshair.

I need a 790 mobo the Xhair doesn't like the Phenom....yet.