OC potential of Phenom II on SB600

[Stuen4y]

We all see what is the overclock potential of the new 45nm shrink of the Phenom core, but the first thing that came to my mind when all those articles and validations started to show up in the IT community was: "Is the bug with the old SB fixed and could the users, who haven't yet bought a board with SB750(to define them SB600 owners/in particular me) , reach the same clocks like the guys with the new boards?". I haven't seen this theme discussed, so I have very simple question - Does anyone have any information about tests made on SB600 boards and what is the difference, if there exists some kind of difference, between the results.
Thanks in advance!

[Oese]

i can tell you nothing about 45nm on sb600, but about 65nm on sb600. there is no bug. the first chips simply didnt clock well.

good chips dont reach higher clocks on the sb750. but voltage sometimes can be set lower with ACC.

medium to low end clocking chips though reach higher clocks on sb750, so sb750 is the safer deal for phenom oc. but the sb750-boards available, excluded the expensive as hell foxconn, all have inferior power regulator systems compared to high end sb600 boards. well, due to acc on those sb750 boards high voltage is not needed so often and with 45nm the power consumption will come down a lot, but i would suppose keep your dq6 anyway until am3 boards come out finally...

[Stuen4y]

I was thinking of exactly the same thing - keeping DQ6 till AM3, but if there is no backwards compatibility between AM2+ processors(Phenom II 940 for example) and the AM3 mobos, I should have to wait till February to buy a new processor. Yes, it's not so much, but if I could run a 940 on AM3 mobo it will be perfect and everybody will be happy. This so called "bug" is not in the processor, it is in the SB. My question was, could it be prevented with changes in the CPU architecture or something like that or it will stay and the OC will suffer from that.
P.S. M3A79-T Deluxe has 10 phases and the same PCB like M3A32-MVP Deluxe, so there are existing 790FX/SB750 mobos .

[Oese]

the asus m3a79-t has 10 phases? well ok i didnt know but it has a more or less bad bios i think..

no, to my knowledge there are bugs in the sb600, but they do not hold oc back. that has to do with the pll circuits on the boards, which can be properly tuned with sb750, and the complex design of the k10 i addition. ACC is a "trick" or more like a feature to reach higher clocks due to better timed clock skew, but it does not work in all cases. if you own a good clocker, the difference will be nothing to 100mhz between sb600 and sb750. whatever, overall oc is slightly increased but not much.

i will buy an am2+ phenom II and clock it on my m2r and i think i'll reach good clocks. i get 3,56ghz out of my 9750 with htt overclocking with only water, and i think phenom II will push that to 4ghz easily if everything is real what we hear these days..

am3 i will wait for a decent dfi UT board and good am3 ram at a decent price.. no need to change whole system right now only because of DDR3...

[Stuen4y]

the asus m3a79-t has 10 phases? well ok i didnt know but it has a more or less bad bios i think..

8+2 to be correct. 8 for the cpu and 2 for the HT (I'm sure you know that, stop talking ironically ).

I'm not very well informed how to go round this "trick" and I'm definitely not concerned with my results with this MB and CPU(maybe the CPU is a very bad piece) - something like 3090MHz for screen, 3050 for tests and 2810 for 24/7. My next mobo will also be DFI, but don't you think that it will be better if you pay for the exact same product only once. There is no point to buy a 940 firstly and then buy 945 when you want to change the platform. Yes, there is another option, to buy FX processor for AM3, but the price is different there.

2 for the NB

It's almost the same . Whether it refers to the north bridge or to the HT which connects the NB the CPU and the DIMM slots I can hardly find any difference .

[Oliverda]

8+2 to be correct. 8 for the cpu and 2 for the HT (I'm sure you know that ).

2 for the NB

[SkullCracka]

the Asus boards are the exception not the rule it seems. love my MVP-Deluxe w0ot LOL

[Chicken Patty]

The difference with the SB600 and the SB750 is the ACC. I think if the SB750 had no ACC it would clock just as the SB600 would.

Another thing that improved at least in the ASUS boards was the PWM. The M3A79-T (SB750), has a much better PWM and was a great improvement over my M3A32 (SB600). So I think the Phenom II will get away with better clocks on the SB750 board, however, it should still perform very well on a SB600 board. Lets just wait and see, only time will tell

[Daveburt714]

Stuen4y: FWIW, I get the feeling that the PII 940 will clock pretty well using a board with the SB600.. Ofcourse at this point it's all specualtion, but here is the reason why.

My original and favorite board (before the Asus 79-T) was an MSI K9A2 with a SB600 southbridge. Here are my results with different X4's (all these were stable clocks and not pushed to the edge, I'm a wimp that way )...

9600 - Max OC 2.5Ghz (MSI)
9600BE - Max OC 2.65 (MSI)
9850BE - Max OC 2.82 (MSI) / 3.0 Max (Asus)
9950BE - Max OC 3.0Ghz (MSI) / 3.2 Max (Asus)

From my experience it looks like each revision of the core gets better performance (even on SB600), so if the numbers we're seeing turn out to be legit, I would expect the PII to hit 3.5+Ghz on an SB600 board.

I'm sure an SB750 would give you better results, but I'd be surprised if a PII wouldn't give at least decent clocks on the DQ6 (or K9A2 ).

BTW: I'm not really an Asus fan, but the 79-T is the best AM2+ board I've had so far! I had a 78-T as well and it was solid, but a terrible overclocker.
The MSI DKA790GX was a great OC'r, but the PWM section was a little weak. Mine died at 1.55 Vcore, and it was a little pricey.
Very nice board though and would probably be fine with a PII...

[Stuen4y]

The difference with the SB600 and the SB750 is the ACC. I think if the SB750 had no ACC it would clock just as the SB600 would.

Another thing that improved at least in the ASUS boards was the PWM. The M3A79-T (SB750), has a much better PWM and was a great improvement over my M3A32 (SB600). So I think the Phenom II will get away with better clocks on the SB750 board, however, it should still perform very well on a SB600 board. Lets just wait and see, only time will tell

Can you explain what exactly is this ACC and how it affects on the clock potential. We know that Phenoms can hardly overclock by HT link and with SB750 this still remains the same like the SB600.

From my experience it looks like each revision of the core gets better performance (even on SB600), so if the numbers we're seeing turn out to be legit, I would expect the PII to hit 3.5+Ghz on an SB600 board.

I'm sure an SB750 would give you better results, but I'd be surprised if a PII wouldn't give at least decent clocks on the DQ6 (or K9A2 ).

If the processor does 4+GHz with SB750, how would I feel? Screwed? Yes I would .

Thanks for the posts. I'll keep the thread open, so if some kind of info appears, please write here.

[Daveburt714]

If the processor does 4+GHz with SB750, how would I feel? Screwed? Yes I would .

Thanks for the posts. I'll keep the thread open, so if some kind of info appears, please write here.

Well, your original question seemed to be "How well will my DQ6 (SB600) do with a PII"... I tried to help answer that question for you...

If you've paid any attention at all, you should know that an SB750 board (W/ACC) will help your overclock... How does ACC work? Umm, not sure anyone can answer that question, but it does...

If you plan on buying an SB750 board (so you don't feel screwed)....
I'm not sure what the whole point of this thread was...

[Stuen4y]

I really don't want to buy an SB750 board and want to move to AM3 platform when it appears. A lot of people in the HW forum in my country are asking themselves the same question like me - "Should I buy an SB750 board or I can put my new upcoming processor in my old board, for which I have paid several hundred dollars and receive the bandwidth I'll receive without making more investments?", cause it's a problem here to sell all those high end parts all the more that AMD is with not so good reputation right now. I appreciate your wiling to help, but the argumentation is the weak point of your statement. Doing some contentions couldn't definitely assure the theory that the new processors will affect the same way as the old ones from this ACC. Do you understand my point now? I asked if there is some info, because there are a lot of men here, who have access to more spicy information than the others. They could not want to tell me their info but what costs me try asking politely and descriptive. And yes, it's all about money and budget, cause if the "news" are good I can save some money to buy better memory modules or processor with higher frequency instead of giving it for mobo. If I should, I'll do it but if I could save this spending? I meant I'll feel screwed if I have to buy new mobo.

You won't have to buy a new MB.
http://www.gigabyte.com.tw/News/Moth...px?NewsID=1423

It will work, but how well it will work noone knows at this point.
I also hope ACC solved some problem with the 65nm parts that isn't a problem anymore with their 45nm parts, but I think a 10% difference between highest overclock on SB600 and SB750 is a more realistic expectation at this point.

And this prediction was made inspired by what? I know my mobo can run a Deneb CPU without problem, the problem is OC and I think that per cent you've given is far-fetched and I can hardly believe it. We can spin yarns all day long.

[Miss Banana]

You won't have to buy a new MB.
http://www.gigabyte.com.tw/News/Moth...px?NewsID=1423

It will work, but how well it will work noone knows at this point.
I also hope ACC solved some problem with the 65nm parts that isn't a problem anymore with their 45nm parts, but I think a 10% difference between highest overclock on SB600 and SB750 is a more realistic expectation at this point.

[duploxxx]

Stuen4y: FWIW, I get the feeling that the PII 940 will clock pretty well using a board with the SB600.. Ofcourse at this point it's all specualtion, but here is the reason why.

My original and favorite board (before the Asus 79-T) was an MSI K9A2 with a SB600 southbridge. Here are my results with different X4's (all these were stable clocks and not pushed to the edge, I'm a wimp that way )...

9600 - Max OC 2.5Ghz (MSI)
9600BE - Max OC 2.65 (MSI)
9850BE - Max OC 2.82 (MSI) / 3.0 Max (Asus)
9950BE - Max OC 3.0Ghz (MSI) / 3.2 Max (Asus)

From my experience it looks like each revision of the core gets better performance (even on SB600), so if the numbers we're seeing turn out to be legit, I would expect the PII to hit 3.5+Ghz on an SB600 board.

I'm sure an SB750 would give you better results, but I'd be surprised if a PII wouldn't give at least decent clocks on the DQ6 (or K9A2 ).

BTW: I'm not really an Asus fan, but the 79-T is the best AM2+ board I've had so far! I had a 78-T as well and it was solid, but a terrible overclocker.
The MSI DKA790GX was a great OC'r, but the PWM section was a little weak. Mine died at 1.55 Vcore, and it was a little pricey.
Very nice board though and would probably be fine with a PII...

have the same boards msi k9a plat and the new asus 79-T

my 9850 does not oc better on the sb750 board then the sb600, but it gives me a bit lower vcore to do the same oc.

so in the end you might have a slight different oc then but if i would have known i would have never bought the 750. maybe phenom2 oc is better on 750 boards, but did gigabyte already released the sb750 on dq5-dq6? if not they were using sb600 in the amd labs

i have my board now and i'll wait a long time before to go to ddr3, there is no rush at all since we can just put any phenom2 in these boards.

[Stuen4y]

have the same boards msi k9a plat and the new asus 79-T

my 9850 does not oc better on the sb750 board then the sb600, but it gives me a bit lower vcore to do the same oc.

so in the end you might have a slight different oc then but if i would have known i would have never bought the 750. maybe phenom2 oc is better on 750 boards, but did gigabyte already released the sb750 on dq5-dq6? if not they were using sb600 in the amd labs

i have my board now and i'll wait a long time before to go to ddr3, there is no rush at all since we can just put any phenom2 in these boards.

They do their tests on 790GP-DS4H, which is a SB750 mobo . All these arguments are pointless if everybody are sharing only their experience with the old CPUs and made general conclusions on the basis of it.

[Miss Banana]

And this prediction was made inspired by what? I know my mobo can run a Deneb CPU without problem, the problem is OC and I think that per cent you've given is far-fetched and I can hardly believe it. We can spin yarns all day long.

What do you propose we do then? Noone has an answer for you at this point, so besides speculation there really is nothing anyone can say.
I think there will be a max OC difference, but I also think this difference will not be substantial enough for most to justify an upgrade.

Unless you want to achieve a world record overclock of course, but in this case you shouldn't have bought that MB in the first place.

[Lightman]

Well I have GA 790FX-DQ6 and all my previous Phenoms were clocking very badly, but 9850BE was doing 2.7GHz stable in 64bit mode and my new 9950BE is doing 3.2GHz stable in 64bit mode (1.4V).
32 bit mode gives me 3.33GHz stable using same voltage and all that on SB600... (3.4GHz bench stable at 1.44V)

As many before me said, SB600 boards will clock P II well if you're lucky to get very good chip. SB750 boards will clock well most of the chips. That's the difference.

If you're guys willing to wait, I'm getting P II ASAP and will test it for you. (I doubt you will wait TBH )

[justapost]

9600 - Max OC 2.5Ghz (MSI)
9600BE - Max OC 2.65 (MSI)
9850BE - Max OC 2.82 (MSI) / 3.0 Max (Asus)
9950BE - Max OC 3.0Ghz (MSI) / 3.2 Max (Asus)

Here
9600BE - Max OC 2.65 (DFI/Air) / 2.9 Max stable (AsRock/Air) / ~3.4GHz benchable
9850BE - Max OC 2.85 (DFI) / 2.82 Max (AsRock)
9950BE - Max OC 3.05Ghz (DFI) / 3.2 Max (AsRock/Air)

Only this early 9850BE does not benefit from ACC.

[Chicken Patty]

Can you explain what exactly is this ACC and how it affects on the clock potential. We know that Phenoms can hardly overclock by HT link and with SB750 this still remains the same like the SB600.

If the processor does 4+GHz with SB750, how would I feel? Screwed? Yes I would .

Thanks for the posts. I'll keep the thread open, so if some kind of info appears, please write here.

I dont think anybody knows exactly what ACC is. It was never made clear what it was, all I know is that with my Current Phenom, I cannot be stable over 3.1 Ghz with it on, when I enable ACC, I can go up to 3.3 GHz. On my 9850 which was a better chip, I went up to 3.55 @ only 1.456v. Got to validate at 3.6 Ghz also.

So its a heck of difference, but i'll let somebody else explain what it is if they have a better idea, but last time I checked, there was no real explanation to what it actually did.

[Oese]

htt oc IS working well on the phenoms. or my chip is the big exclusion from the rule.. i had it as high as 300mhz htt prime stable and fully useable.

ok, whatever. i will test the pII () on my board and think it will work well. if it does not, well i'll wait either for am3 or buy an sb750mobo.

what i surely wont do is buy an sb750 without checking whether the sb600 isnt enough to come around till am3 is available or even beyond.. well i think the new asus is a good board but i wont buy it because 3,2ghz compared to 3ghz of maximum oc isnt worth it. if acc would bring a 3,4ghz chip to 3,6ghz stable, that would be something else then 3ghz to 3,2ghz. but it does not from what i know from my readings.. but pII will reache these points easily, and i think the acc impact will be equal or even less. of course i dont know really

[Stuen4y]

htt oc IS working well on the phenoms. or my chip is the big exclusion from the rule.. i had it as high as 300mhz htt prime stable and fully useable.

ok, whatever. i will test the pII () on my board and think it will work well. if it does not, well i'll wait either for am3 or buy an sb750mobo.

I'll test it on SB600 too without doubt . Really your system is a kind of exception with this high HTT. Mine board does something like 230 HTT for stable .

ok of course you have to lower nb multiplier, raise nb voltage and such things.. and what is special with the m2r is you can adjust htt boot up clock and cpu htt clock sperately. dunno what is what exactly (ok sort of assumptions but never seen any real effect that points into that direction) but one has to set both the same for htt oc on phenom on the m2r. Kinda really funny thing...

I've done some tests when I changed the CPU but from then I don't have enough time for one serious bench session(I'm having sleep lack). I'm sure I could get 100MHz more for example, but don't have the passion when I know my CPU is a bad piece for overclocking.

[Oese]

ok of course you have to lower nb multiplier, raise nb voltage and such things.. and what is special with the m2r is you can adjust htt boot up clock and cpu htt clock sperately. dunno what is what exactly (ok sort of assumptions but never seen any real effect that points into that direction) but one has to set both the same for htt oc on phenom on the m2r. Kinda really funny thing...

[iDCT]

The MSI DKA790GX was a great OC'r, but the PWM section was a little weak. Mine died at 1.55 Vcore, and it was a little pricey.

Hi, I'm off topic here , sorry for that, but I can't get more than 1.45 Vcore from my DKA790GX with bios 1.2 and 1.3. Is there a special "trick" to go beyond 1.45 ?

Help would be greatly appreciate

[Daveburt714]

iDCT: You have PM

[Final8ty]

ok of course you have to lower nb multiplier, raise nb voltage and such things.. and what is special with the m2r is you can adjust htt boot up clock and cpu htt clock sperately. dunno what is what exactly (ok sort of assumptions but never seen any real effect that points into that direction) but one has to set both the same for htt oc on phenom on the m2r. Kinda really funny thing...

The last few days with your posts have been tempting me to get M2R again.