AMD PhenomII X3 710(720BE) 3core --> 4core PhenomII X4 920

[B-Unit]

Odds are that its the 4th core causing P95 to fail so fast. Im interested to see what it does when you revert to X3.

As a follow up, did you attempt P95 at stock speeds? The 4th core might only be flaky at higher speeds.

 

[misry]

Do you run OCCT?

 

[FrozenGpu]

Do you run OCCT?

Run occt plz, post results!!!

I am curious as to the same thing b-unit said, at stock speeds if it's stable enough.

 

[jcknouse]

Odds are that its the 4th core causing P95 to fail so fast. Im interested to see what it does when you revert to X3.

As a follow up, did you attempt P95 at stock speeds? The 4th core might only be flaky at higher speeds.

Yeah, I think it is too. I am thinking if I go back to X3, that the chip will clock up over 3.5. I got it to 3.5+, and it ran in 3DMark06 successfully with a benchmark of like 14,070 or something. I had it at 3465MHz and it ran through AMD Overdrive stress test with all 8 of CPU 0-3s test for 10 minutes.

I ran Prime95 for like 10 mins at 3.255GHz (5 or 6 iterations), before I got tired of sitting there.

The thing boots 4-core at that speed and runs everything fine.

I am planning on swapping back to X3 mode tonight and doing some benchmarking, especially with 3DMark06 to see if 3 cores is just as good for most gaming. The only time I'd wanna run X4 mode would be when I play Shadowbane, and that's because I run anywhere from 3-10 sessions of it and delegate individual processes to specific cores.

I'll let you know as soon as I do, and how that works out.

 

[jcknouse]

Run occt plz, post results!!!

I am curious as to the same thing b-unit said, at stock speeds if it's stable enough.

I will get OCCT tonight, or as soon as I can, and run it. I've never used it, so I will have to find it and install it.

Edit: Found the ocbase.com website. Will download OCCT tonight when I get home, and run a few tests, if I can. My mom is going to see a specialist today. I might have to go back over to see her and dad tonight.

 

[misry]

Update.
First the bad news. Core 2 is unstable with less than 1.4v. I think I've identified my bad core and why it failed. Second, I can't seem to go over x16 with 4 cores no matter what and to touch my FSB I had to drop the NB/HT multi to x9.

Now the good. I've got it at 3.33G/4 cores OCCT 1 hour stable. FSB at 208 which makes NB/HT 1881. Core is 1.45v, NB is 1.325v and temps are 30c/42c. L3 allocation is ALL CORES and ACC is ALL CORES/0%.

 

[yipsl]

It probably depends on which X3 720 anyone gets as to whether the 4th core is equal to the others. As jcknouse reports, he can get a decent overclock and get Prime95 to work, but he can't get the overclock he wants that can run Prime95. Seems like there are several variables deciding if a core is bad. At any rate, I don't think AMD disabled any good cores just to meet the X3 demand. All the X3's are probably classic "triple cripples" (using TC's amusing phrase that I see as almost a badge of honor; disabling cores or cache is standard for both AMD and Intel).

What I want is the rumored Phenom II X4 @ a native 3.6 that might be out later this year. What I'll probably settle for is whatever AM3 quad I can afford in April. I want the future option to move the CPU over to an "890GX" board with SB800 later this year if DDR3 and the chipset actually improves on my current setup.

I'd try overclocking this 8750, but I got it before the BE version arrived, so I wouldn't expect much from it. Wouldn't fool with voltages though. I'm too conservative for that.

 

[jcknouse]

Update.
First the bad news. Core 2 is unstable with less than 1.4v. I think I've identified my bad core and why it failed. Second, I can't seem to go over x16 with 4 cores no matter what and to touch my FSB I had to drop the NB/HT multi to x9.

Now the good. I've got it at 3.33G/4 cores OCCT 1 hour stable. FSB at 208 which makes NB/HT 1881. Core is 1.45v, NB is 1.325v and temps are 30c/42c. L3 allocation is ALL CORES and ACC is ALL CORES/0%.

I will try some of your settings tonight. I didn't mess with "ALL CORES" stuff, and I clocked the NB up with the Core volts (I thought I read somewhere that you were supposed to keep the NB at/near the same volts as the CPU.

I am getting about those temps too, except my idle is about 21-23c, and load is 39-40c, with the ambient temp in my house being about 69-72F/20-22C.

Thanks for the info on ur settings

 

[jcknouse]

It probably depends on which X3 720 anyone gets as to whether the 4th core is equal to the others. As jcknouse reports, he can get a decent overclock and get Prime95 to work, but he can't get the overclock he wants that can run Prime95. Seems like there are several variables deciding if a core is bad. At any rate, I don't think AMD disabled any good cores just to meet the X3 demand. All the X3's are probably classic "triple cripples" (using TC's amusing phrase that I see as almost a badge of honor; disabling cores or cache is standard for both AMD and Intel).

What I want is the rumored Phenom II X4 @ a native 3.6 that might be out later this year. What I'll probably settle for is whatever AM3 quad I can afford in April. I want the future option to move the CPU over to an "890GX" board with SB800 later this year if DDR3 and the chipset actually improves on my current setup.

I'd try overclocking this 8750, but I got it before the BE version arrived, so I wouldn't expect much from it. Wouldn't fool with voltages though. I'm too conservative for that.

I agree that AMD didn't just take x4s and turn a core off to have x3s to make money.

But, I still hold the belief (based on what friends in manufacturing have told me about QA and efficiency and cost-savings) that AMD is disqualifying production lots (groups of X4s) that a certain percentage of them fail in QA testing and sending them to be QAed and capped as x3s .

To me, that seems like the only reasonable explanation why some folks have gotten X3s that OC to 3.5GHz and higher on 4 cores. I don't think AMD could afford to spend (time- or money-wise) the resources to test every individual CPU that comes off their line.

It would take forever to get a tray of CPUs to market, or it would take armies of machines and people to verify each CPU and that would be a TON of money (and AMD doesn't have tons of money anymore).

I am waiting, like you are, to build my next PC to get the AM3 x4/890xx/SB8xx setup. I will probably wait til Q4 2009 though, as I just build a new rig in Sept/Oct 2008 and just did the 720BE/TA790GX upgrade.

 

[misry]

Oh no, I had no dream that I would have a core that had simply been turned off to fill an order. I disagree with that notion completely. In that respect my "bad" news is pretty good. I found the problem and have the fix. Nor am I lusting after anyone else's clock speed, as with any overclock YMMV. I got the chip assuming I could get 3.0G out of it, (my Opty won't even boot at that), and I did. Everything else is gravy.

As to the ALL CORES stuff, I'm just running it that way *now* to individualize performance of each core as much as possible, (I'm running unganged right now for pretty much the same reason). Later, probably after I get a "real" video card, I intend to go the other way to see what happens.

 

[jcknouse]

Alright, the X3 mode results:

Bus Freq: 203 MHz
HT Mult: 8x
CPU Mult: 17x
CPU Clock: 3451MHz
CPU Volt: 1.45v
NB Volt: 1.45v
DDR Volt: 1.85v

This was the best I could get, having it pass Prime95 Large FFT test, as well as OCCT.

Temps:

idle: 22-23c
load: 37-39c

I am about to run 3DMark06. If I get a higher benchmark in 3 core mode, I might leave it here. Otherwise, I'll go back to X4 mode.

Added: 3DMark06 - 13430

 

[Xenophage]

I just got the AM3 Gigabyte GA-MA790FXT-UD5P (the 790fx board) and a Phenom II X3 BE, and I see no ACC option in the bios.

 

[Xenophage]

Just read in some other boards that the Gigabyte board has some hidden BIOS menus that are accessible by pressing CTRL+F1 where you can enable the ACC. I'll try it in a few.

 

[Xenophage]

CTRL+F1 did indeed let me set the ACC to Auto. However, as soon as I chose to save and quit my computer turned off. Any attempt to boot the computer would result in an immediate shut-down. So, I cleared the CMOS and all is fine. Disappointing

 

[jimmysmitty]

Looks like either the mobo you have wont be able to boot with 4 cores or the CPU you have isn't able to because it is well one with a true bad core.

 

[fazers_on_stun]

Looks like maybe the X3 --> X4 trick is driving up P2 sales: Fuddie

A senior guy from AMD that we've met at Cebit has told us that AMD is gaining share with its Phenom II CPUs. Since the launch, the sale is doing really well and after the patch Phenom II X3 720 is selling like hot cakes.

However if true, this could be short-lived and actually have some negative consequences if too many people buy one, expecting a freebie upgrade to a 4-core at a steep discount, and then find out as the above posters that it doesn't quite work out that way.

 

[yipsl]

CTRL+F1 did indeed let me set the ACC to Auto. However, as soon as I chose to save and quit my computer turned off. Any attempt to boot the computer would result in an immediate shut-down. So, I cleared the CMOS and all is fine. Disappointing

The reports were it worked with certain Biostar boards:

http://www.theinquirer.net/inquirer/news/124/1051124/triple-core-phenoms-fourth-core-enabled

That's why I wouldn't expect it to work on mine.

Now, is this a bios screw up at Biostar? I can't see it being a a bios screw up by AMD or more boards would "support" the feature.

Plus, you might not have one of the lucky triples. Only by testing it in a biostar board with a bios reported to work would you know for sure.

 

[jcknouse]

I did some more tweaking. I am currently running stable on a 720BE AM3 in X4 mode at:

Bus speed: 206
HT Mult: 16.5x
Clock speed: 3.399GHz
CPU and NB Volts: 1.4625v

I ran the small FFT, Blend, and large FFT tests in Prime 95 for 10 mins+ each and no crashes. Ran the AOD stability test for 17 mins, no crashes.

I went over 1.45v, but i figure that is less than 1% over tolerance so it should not have any ill effects.

I could probably go higher (had it to 1.4875v earlier), but I don't want to take any other risk of damage.

I do have a question for those of you who have/know about OCCT:

I installed it while in 4 core mode, then went to 3 core, ran test, then went back and tried both Auto and All Cores setting. OCCT still only sees 3 cores.

Is there a config file that OCCT has that is only allowing to show the min cores ever? Do I need to uninstall and reinstall to get it to see 4 cores again?

I like OCCT. I would like to run it as a testing tool to confirm all my stability.

Thanks and I hope this all gives you guys hope of getting a higher-end CPU out of something affordable.

P.S.- Current Setup:

AMD 720BE AM3 proc in X4 setting through BIOS ACC Mode = "All Cores"
Biostar TA790GX 128M Mobo
4x2GB OCZ SLI-Ready Edition 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Desktop Memory OCZ2N800SR4GK

 

[misry]

High end for the 720 is 1.55v so you're still golden there. I installed OCCT under 4 cores and haven't "clocked back" to 3 but it shows 4 cores and that it's running 4 threads. Have you tried uninstall and reinstalling OCCT?

 

[amdfangirl]

My sister is getting a Phenom II 710, let's hope the 4th core works...

 

[jimmysmitty]

The reports were it worked with certain Biostar boards:

http://www.theinquirer.net/inquirer/news/124/1051124/triple-core-phenoms-fourth-core-enabled

That's why I wouldn't expect it to work on mine.

Now, is this a bios screw up at Biostar? I can't see it being a a bios screw up by AMD or more boards would "support" the feature.

Plus, you might not have one of the lucky triples. Only by testing it in a biostar board with a bios reported to work would you know for sure.

Unless these mobo makes (Biostar andthe otehrs) do a recall on AMDs behalf since they might not want to lose quad sales and they do. Then when the newer BIOS with the fix comes out this will all end.

So TBH, it will be a handful of ppl who will be able to actually do this.

 

[JAYDEEJOHN]

Theres already alot of bios hacks done, and the old bios has been saved for reflashing as well. If you really want to try this, you can

 

[jimmysmitty]

Thats a risk I wouldn't be willing to take. Its great that they can hack it but the stability and performance of the BIOS is not guranteed and chances are that the mobo vendor wont warranty it if anything goes to the point of no return wrong.

 

[JAYDEEJOHN]

True, but time will tell. If someone expects to have a quad and it fails to happen, and then is disappointed, well tough. Its a fine chip in its own regard, and having the fourth one active is only a plus, sometimes. It could cause instability, or make the quad run actually worse than the tri. To me, its a nice surprise, if it works

 

[misry]

Thats a risk I wouldn't be willing to take. Its great that they can hack it but the stability and performance of the BIOS is not guranteed and chances are that the mobo vendor wont warranty it if anything goes to the point of no return wrong.

Of course you realize you are replaying the exact arguments people have for overclocking in general. Please insert appropriate counter argument here,