Lame X2-3800 and X2-4200 OCs

[Rythan]

So, I've built two systems recently, one for work:

X2-4200 w/stock heatsink
Epox 9NPA+Ultra
2x512MB PC3200 Kingston ValueRam
GF 6600
Antec True380S
WD 250GB SATA

And one for home:

X2-3800 w/Scythe Ninja + Nexus RealSilent 120mm
Epox 9NPA+Ultra
2x512MB OCZ PC4200
GF7800GT
Antec TruePower 2.0 430
WD 74GB Raptor

The 4200 will run stable (according to 2x Prime95) at 2300 (230x10, HTT 3x, Memory at 133) but not at 2400 - This is after bumping CPU and chipset voltages up +0.1V, and I was keeping the FSB and memory speeds quite safe.

The 3800 will run stable (according to 2x Prime95) at 2200 (220x10, HTT 3x, Memory at 133) but not at 2300 - This is after bumping CPU and chipset voltages up +0.1V, and I was keeping the FSB and memory speeds quite safe.

Temps are in the 30-35 range idle, and 50-55 load for both systems.

So, a whopping 100 MHz OC on the 4200, and a 200 MHz OC on the 3800 :disgust:

I thought these 90nm A64s were supposed to OC to 2.4 easy, and 2.6 with some luck? I thought that between the two at least one would make a happy core for my home gaming machine, but I appear to have struck out unless I'm missing something.

I'd love any ideas!

P.S., I love the Nexus RealSilent fans - they actually are quiet with good airflow, unlike several Panaflos, Papsts, etc that I've bought for previous systems

 

[Uhtrinity]

Sounds like a PCI lock issue. Not familar with that board, but some of the boards (like my MSI) can't operate at fsb's higher than 230 - 240 without disabling SATA 1 & 2. I can however use SATA 3 & 4 all the way to 300 Mhz FSB.

 

[n7]

Originally posted by: Uhtrinity
Sounds like a PCI lock issue. Not familar with that board, but some of the boards (like my MSI) can't operate at fsb's higher than 230 - 240 without disabling SATA 1 & 2. I can however use SATA 3 & 4 all the way to 300 Mhz FSB.

It's not a PCI lock issue.
That mobo has them; it's excellent for OCing also.

Rythan, are you using a RAM divider?

I hate to say this, but i'd guess you are doing something wrong.

Tell us what settings you used, etc.

 

[Rythan]

I had thought that this Epox had the appropriate PCI locks, as I saw no mention of it being a limit in the various good reviews. I'll try switching SATA ports just for completeness, though.

I am using RAM dividers, as I wanted to isolate the highest CPU speed - I set the Epox memory speed to 133 (which controls the RAM divider), which means that I wouldn't be overclocking the memory until I exceeded an FSB of 300 (300 * 133/200 = 200). I verified this using cpuz.

I may crank the voltage up a bit more just to see, but it seems most others don't need to push the voltage hard to get to 2.4.

Addendum:

The settings are all at BIOS defaults, with the exception of the HTT, memory divider, FSB, and CPU/chipset voltages I mentioned in my first post. I'm running plain old WinXP SP2, with the latest Nvidia platform and video drivers.

 

[evilharp]

Hey Rythan,

It could be your power supplies.

X2-4200 w/stock heatsink
Epox 9NPA+Ultra
2x512MB PC3200 Kingston ValueRam
GF 6600
Antec True380S
WD 250GB SATA

And one for home:

X2-3800 w/Scythe Ninja + Nexus RealSilent 120mm
Epox 9NPA+Ultra
2x512MB OCZ PC4200
GF7800GT
Antec TruePower 2.0 430
WD 74GB Raptor

Both are cutting it close to the minimum specs for the systems in question.

 

[Markfw]

Well, I run my 4400+@2550. Only difference is I have a 9600pro, that 380 is probably fine with a 6600. And the 430 is probably OK on the 7800, since I hear they use less than the 6800!

What does CPU-Z say your vcore, regardless of what you bumped ? If it less than 1.45 at full load, crank it up some more !

 

[sixth]

Guys...it has nothing to do with power supplies ... sorry.

I have this board, and a X2 3800, i cant only run stable at 2300mhz with a +.125 vcore, i have ran this board through everything, and i think its the board, or its the chip. I have the manchester core and from what i have found out i cant get any higher than 2300 STABLE. i had her booting to 2450 but there was a ton of heat, and vcore was around 1.65-.68. I have messed with everything. (running the 4600 heatsink from thermaltake)

Rythan- if you want to see what i mean jack your vcore to +.2v set the board 10x 245 and she will boot (should and run prime, it will pass but vcore will jump to 1.65-68 and get way hot.

Best thing i could say is mess aroudn in the bios, but i have tried EVERYTHING and cant get this board past 2300 stable without a CRAZY! vcore..

Goodluck and let me know if you get passed 2300

 

[n7]

You can safely set vcore to 1.55V

I'd try that, & see what happens.

It does seems rather odd. Maybe there's an issue with X2s & that Epox?

It seems to OC well with single core A64s...maybe it's an X2 issue...have you guys tried different bioses?

Here's another thing:

Have you guys run with a lower multiplier & higher HTT to check if it's the mobo limiting things?
E.g., on the X2 3800+ system, try 9x250 (2250 MHz)) instead of 10x225 (2250 MHz)
Also, i believe it's possible for lower multipliers to work better in rare cases.

Try that too, & that will remove the mobo from being culprit if it works.

I'd try a low multi like 7x, & then see how high you can get the HTT.

 

[n7]

damnit dp

forums keep timing out

 

[Rythan]

After a night of playing with the X2 3800, I found the following:

- With the memory, HTT and CPU set to low multipliers (DDR200/2x/5x) I can get the FSB up to 320 and still be Prime95 stable at stock voltages. So, the motherboard seems good to go.

- I got the CPU to be Prime95 stable at 230x10, at 1.55V, but I can't get to 240x10 at that voltage. Temps get up to 57C under load at 2300.

- The CPU isn't stable at 2200 at stock voltages.

- I'll look around and see if I've got a bigger PS in one of my other computers, though it seems unlikely to be the cause - it's a high power CPU and video card, but that's about all that's in the system other than the HD and some fans.

So, now I am pondering a DFI M/B or an Opteron 165, but I still don't really know what the OC problem is.

 

[Leper Messiah]

Originally posted by: Rythan
After a night of playing with the X2 3800, I found the following:

- With the memory, HTT and CPU set to low multipliers (DDR200/2x/5x) I can get the FSB up to 320 and still be Prime95 stable at stock voltages. So, the motherboard seems good to go.

- I got the CPU to be Prime95 stable at 230x10, at 1.55V, but I can't get to 240x10 at that voltage. Temps get up to 57C under load at 2300.

- The CPU isn't stable at 2200 at stock voltages.

- I'll look around and see if I've got a bigger PS in one of my other computers, though it seems unlikely to be the cause - it's a high power CPU and video card, but that's about all that's in the system other than the HD and some fans.

So, now I am pondering a DFI M/B or an Opteron 165, but I still don't really know what the OC problem is.

Hm. Maybe you are just unlucky. My manchester 3800 X2 does 2.6GHz at 1.37vcore, and I'm running the NF3 Epox mobo.

 

[Gronich]

Do you have a good friend with a system which overclocks well you can perhaps try your chips in? (I know its a big ask...) OR

The first thing I would try is a bigger power supply 480W+ think of it as buying a spare for either system.

 

[sixth]

I have the OCZ Powerstream 650 + 8 hard drives, 10 total with the two raptors, plus one optical drive and a x800....my power is SOLID, this is NOT a power issue...i havent tried lowering the mulitplier to like 5 or 6 but i know i cant boot at 9....no go..i will try to lower the multiplier and vcore...

 

[Diogenes2]

When the OC fails is it rebooting?

I found out the 12v rail was cratering on my Antec TrueBlue 480 when I went to 1.55 and above with my X2...
It would do 1.7 no problem with a 3500+ Newcastle..

 

[Doctorweir]

Originally posted by: Rythan
After a night of playing with the X2 3800, I found the following:

- With the memory, HTT and CPU set to low multipliers (DDR200/2x/5x) I can get the FSB up to 320 and still be Prime95 stable at stock voltages. So, the motherboard seems good to go.

- I got the CPU to be Prime95 stable at 230x10, at 1.55V, but I can't get to 240x10 at that voltage. Temps get up to 57C under load at 2300.

- The CPU isn't stable at 2200 at stock voltages.

- I'll look around and see if I've got a bigger PS in one of my other computers, though it seems unlikely to be the cause - it's a high power CPU and video card, but that's about all that's in the system other than the HD and some fans.

So, now I am pondering a DFI M/B or an Opteron 165, but I still don't really know what the OC problem is.

The PSU shouldn't be a problem...430 is well enough with the 7800GT not even in 3D mode. For comparison: My rig below sucks 205 Watts peak at full 3D load...add for the second CPU core and the 7800, but IMHO no big deal and still <300 Watts total consumption, especially in 2D / Prime testing...

 

[Rythan]

No, no rebooting, just Prime95 failures. Though bumping up an extra 100MHz on the CPU usually leads to Windows crashes on both the 3800 and 4200.

 

[n7]

Originally posted by: Rythan
So, now I am pondering a DFI M/B or an Opteron 165, but I still don't really know what the OC problem is.

You just identified the problem.

It's not your mobo, since it can hit 320 HTT (very good BTW ).
It's your CPU.
Apparently you got a bum X2 or two.

Here's the thing though.

Not everyone is guaranteed 2.4 GHz. Granted, there are few people i've seen that haven't gotten that high, but it seems you got some CPUs that really do only perform to their rating.

I don't see why you'd want to get rid of what is already an extremely powerful CPU just because it doesn't OC well, but i guess each to their own.

 

[ethen]

you are not the only bud, one of my 3800 X2 wouldnt even go over 2.2ghz E4

So i took my wife's X2 3800 which is E6, and it was stable on 2.4Ghz, so yeah being a bad guy and swap my cpu with my wife's

 

[Rythan]

Yeah, at this point I'm guessing I just got really unlucky. I wouldn't be quite so annoyed at my overclocking failures if it weren't for the fact that I built two entire systems - I figured one of them ought to be decent. Obviously these X2's are quite happy CPUs even at stock speeds, I guess I just had high expectations after so many reports of high OCs at stock voltages.


Edit: I guess I should pull off the heatsinks and write down the batch numbers to warn people off whatever lot these CPUs were from. CPUZ reports that both are E4, though.

 

[Doctorweir]

Originally posted by: ethen
you are not the only bud, one of my 3800 X2 wouldnt even go over 2.2ghz E4

So i took my wife's X2 3800 which is E6, and it was stable on 2.4Ghz, so yeah being a bad guy and swap my cpu with my wife's

ROFL :laugh: As long as your wife isn't a hardcore overclocker herself and finds out... => vacation

 

[Zebo]

Originally posted by: ethen
you are not the only bud, one of my 3800 X2 wouldnt even go over 2.2ghz E4

So i took my wife's X2 3800 which is E6, and it was stable on 2.4Ghz, so yeah being a bad guy and swap my cpu with my wife's

LOL What's so bad about that? Mines lucky to have a working computer half the time.

 

[Zebo]

Could be memory. Test by setting multi to 9 (underclocking chips)

Could be PSU. Test by using two PSU's on one board. Short green and black on 24 pin on one. Connect that ones 4 pin and molex to board. Other connect to board as normal. Need heavy 12V rails with X2/s and high power video.

Could be chips. But I find it impossible to believe AMD does not have more tolerance for dusty office conditions which raises temps and thus chips run lower.

Could be temps- Try to keep less than 50C load. Stock HSF sould even do this is property mounted with a half rice drop of Arctic silver in center.

 

[GuitarDaddy]

Before you give up on those chips, I would try a more powerful PS with a strong 12v rail. IMO the first rule of thumb for overclocking is "overkill" the PS. Both of the PS's that you've listed have very low amps on the 12v rail, and I wouldn't even consider overclocking a single core with them. The dual rail 430 is a notoriously bad overclocker, I worked with a guy a few month's back that only had a 3000+ and a 6600gt and he couldn't get past 2.3ghz regardless of what he tried. He switched to a 480w single rail PS with 24a on the 12v and his OC went up to 2.65ghz. Can't hurt to try

 

[Zebo]

Antec not so good at PSU..Good at marketing though. My true 480 went up in flames..literally smoke from cheap china caps exploding. Not only that they have terrible HS inside and rate thier PSU's at like 0C...while inside case is 40C signifigantly lowering rating.

 

[Rythan]

Hmm... the True 430 is rated at 17A/17A on the 12V1/12V2 lines, while something like the OCZ PowerStream 600 is rated for 20A/18A, with nearly identical line and load regulation. Doesn't seem very significant. I will look around for something else to test with, though. I think I've got a nice server power supply at work I might borrow.