FX 60: Results and Advice

Indyxc1 · Feb 24, 2006

Well, I decided to splurge on an FX-60, mainly because I got a good deal on it, and the stepping/code I found was suppose/is to be excellent.

Stepping is CBB2E 0536 RPMW. Searched around the net, and found 3 others with this stepping, and they were all doing 3150 dual SuperPI 32m stable, one even at 3200. Another guy got his to 3800mhz on DICE. Solid stepping it seems.

My realistic goal was/is 3000 mhz prime stable all day long.

Cooling is:
Swiftech 6002MCW
Quietone 1200
77" boneville Heatercore

So off to the overclocking:

-2600mhz at 1.31v dual prime 95 stable for 9 hours, burn in time and check to make sure it was stable. (first attachment) Load temps =38C

-I bumped it up to 2700 on the same voltage (1.31v), and it passed 6 hours of prime 95 dual prime stable. I forgot to take a screenshot. Load Temps= 38C

-2800 mhz at 1.31v it failed 2hours 58mins into prime 95, so I bumped the voltage up to 1.344, and it was prime 95 stable 6 hours. Stock voltage is 1.30 to 1.35, so it's doing FX-57 speeds at stock voltage.

Load temps=38C

-2900 at 1.344 failed prime 95, so I bumped the voltage to 1.40, it failed after 5 hours, so I bumped it again, and in the end it was stable 2900 at 1.45volts. (no attachment) Load temps= 40C.

So at this point I'm thinking, prime95 stable at 2900 @ 1.45, this thing I rocking. 3000mhz should be no problem right...

Onto 3000.

Boot into windows at 15x200 at 1.45 volts. Run a dual Super PI 32M as a quick assesment. It passes, in something like 27 mins, full loading both cores. Run it two more times, passes both times. Then I decided to try 3dmark06 as it's good at multithreading. It does infact load both cores, passes that fine. (twice) Off to battlefield 2. Play battlefield 2 for about 3 hours, then I get a disconnect error. (dam dual core patch).

I'm feeling good at this point. So I try my old friend prime 95, and he makes me feel different.

Core 0 fails instanstly. No problem I bump up the voltage to 1.50volts. Prime 95 core Core 0 fails in 3 mins. Core 1 goes on and on. Ok, so I bump the voltage again to 1.55. Core 0 again fails prime 95 3 mins in. Bump the voltage again to 1.57, and then 1.60, but it fails core 0 again.

So at this point I'm thinking, lets lower the clocks a bit, and see what happens.

Well, it's prime stable up to 2950 at 1.49 volts, but after no matter how much voltage I do, it won't prime stabilize at 3000 mhz. That's promising in a way, because I think it could probably do 2950 on the stock cooler, because the volts are pretty low. It's Super PI 32M stable up to 3140 or so just like those other steppings.

So the whole time I'm running windows x64, and figure maybe I'll try x32. Read some threads that x32 is easier on the OC. Didn't make a difference.

Noticed my PWMIC temps skyrockets with this dual core, especially as voltages increase. With my opteron 148 max PWMIC was like 42C. Here it gets up to 60C. Oddly it's usually stable at 3000mhz in prime for about 3mins or so, until the PWMIC climb. Maybe it's a power issue?

Also thinking of trying a different bios. Right now I'm running the 704-2bta, but I just go FX60-BTA, and the "official" 11/14 DFI FX60 test bios. Any ideas?

I also found that people were having luck with Drive Strength, and Data strength with getting their Opteron 170/175 to the promise land.

Overall, even though I'm 50 mhz shy of 3000, this thing is decent. I looked around and there are some "star" Opteron 170s doing 2900 to 3000 mhz on similiar volts, but in general most dual cores top out in the 2700-2800 range.

Advice, thoughts welcome!

Know Nuttin · Feb 24, 2006

Indyxc1 said:
Boot into windows at 15x300 at 1.45 volts.

I'm guessing you mean 15x200?

You definitely will want to cool the PWM area.

Indyxc1 · Feb 24, 2006

Know Nuttin said:
I'm guessing you mean 15x200?

You definitely will want to cool the PWM area.

Yeah, sorry, meant 15x200. Fixed.

I'm thinking that it's part of the problem. Oddly I noticed when I disconntect 2 120 mm fans right now, it made it longer in prime 95. 8 mins. at 15x200 1.50volts. Yippe.

Likewise, like I mentioned it seems to last those couple mins until the PWMIC skyrocket to 50C+ then it fails. I'll look into that.

Spade · Feb 24, 2006

agreed. in most cases when you start to clock that area gets very very HOT.
you MIGHT get a touch more out of that chip watercooling the fets..
but you will NEED to at least air cool them.

that's still a mighty impressive OC

congrats!

OldOne · Feb 24, 2006

Indyxc1 said:
Well, I decided to splurge on an FX-60, mainly because I got a good deal on it, and the stepping/code I found was suppose/is to be excellent.

Stepping is CBB2E 0536 RPMW. Searched around the net, and found 3 others with this stepping, and they were all doing 3150 dual SuperPI 32m stable, one even at 3200. Another guy got his to 3800mhz on DICE. Solid stepping it seems.

My realistic goal was/is 3000 mhz prime stable all day long.

Cooling is:
Swiftech 6002MCW
Quietone 1200
77" boneville Heatercore

So off to the overclocking:

-2600mhz at 1.31v dual prime 95 stable for 9 hours, burn in time and check to make sure it was stable. (first attachment) Load temps =38C

-I bumped it up to 2700 on the same voltage (1.31v), and it passed 6 hours of prime 95 dual prime stable. I forgot to take a screenshot. Load Temps= 38C

-2800 mhz at 1.31v it failed 2hours 58mins into prime 95, so I bumped the voltage up to 1.344, and it was prime 95 stable 6 hours. Stock voltage is 1.30 to 1.35, so it's doing FX-57 speeds at stock voltage. Load temps=38C

-2900 at 1.344 failed prime 95, so I bumped the voltage to 1.40, it failed after 5 hours, so I bumped it again, and in the end it was stable 2900 at 1.45volts. (no attachment) Load temps= 40C.

So at this point I'm thinking, prime95 stable at 2900 @ 1.45, this thing I rocking. 3000mhz should be no problem right...

Onto 3000.

Boot into windows at 15x200 at 1.45 volts. Run a dual Super PI 32M as a quick assesment. It passes, in something like 27 mins, full loading both cores. Run it two more times, passes both times. Then I decided to try 3dmark06 as it's good at multithreading. It does infact load both cores, passes that fine. (twice) Off to battlefield 2. Play battlefield 2 for about 3 hours, then I get a disconnect error. (dam dual core patch).

I'm feeling good at this point. So I try my old friend prime 95, and he makes me feel different.

Core 0 fails instanstly. No problem I bump up the voltage to 1.50volts. Prime 95 core Core 0 fails in 3 mins. Core 1 goes on and on. Ok, so I bump the voltage again to 1.55. Core 0 again fails prime 95 3 mins in. Bump the voltage again to 1.57, and then 1.60, but it fails core 0 again.

So at this point I'm thinking, lets lower the clocks a bit, and see what happens.

Well, it's prime stable up to 2950 at 1.49 volts, but after no matter how much voltage I do, it won't prime stabilize at 3000 mhz. That's promising in a way, because I think it could probably do 2950 on the stock cooler, because the volts are pretty low. It's Super PI 32M stable up to 3140 or so just like those other steppings.

So the whole time I'm running windows x64, and figure maybe I'll try x32. Read some threads that x32 is easier on the OC. Didn't make a difference.

Noticed my PWMIC temps skyrockets with this dual core, especially as voltages increase. With my opteron 148 max PWMIC was like 42C. Here it gets up to 60C. Oddly it's usually stable at 3000mhz in prime for about 3mins or so, until the PWMIC climb. Maybe it's a power issue?

Also thinking of trying a different bios. Right now I'm running the 704-2bta, but I just go FX60-BTA, and the "official" 11/14 DFI FX60 test bios. Any ideas?

I also found that people were having luck with Drive Strength, and Data strength with getting their Opteron 170/175 to the promise land.

Overall, even though I'm 50 mhz shy of 3000, this thing is decent. I looked around and there are some "star" Opteron 170s doing 2900 to 3000 mhz on similiar volts, but in general most dual cores top out in the 2700-2800 range.

Advice, thoughts welcome!

Core 0 failed in prime also at 3.0 ,2.9 does just fine at 1.55 volts but got to hot for me.I am running 235X12 = 2820mhz at 1.33volts,thats good enough untill I get something to throw on the fire to cool it down.Right now I am running duel prime at 45c.

Did you plug in the ez plug on the motherboard?

LoneWolf121188 · Feb 25, 2006

I'd definetly throw at least some staples or pennies on those MOSFETS...see if you can get a small fan blowing on them too. Some place had uber-cheap ramsinks that you could probably use for MOSFET sinks...check in cyber deals...

Indyxc1 · Feb 25, 2006

Plug in the ez button? Lol.

Well, this chip is confusing me. Normally, from experience when a chip doesn't pass prime 95 for 4 hours + I find that eventually it will crash when you are playing a game, or whatever. So, I usually don't consider a chip stable, unless it can pass prime.. Especially, one that fails prime 95 in under 5 mins. That usually means programs will crash quickly, or at least sometimes.

Well last night, at 3000 mhz, 1.5v I loaded up Battlefield 2 to see if the chip would crash, set affinity to the "bad" core, and played the game for probably 4-5 hours, with the core being loaded 98% of the time, and the other core being loaded 30-40% the whole time.

Didn't crash on me once, no problems, nothing. Ran Super PI 32M 3 times-> No Problems. Then I ran 3dmark 06 5 times, and watched it fully load both cores on the CPU tests, and it gave me no problem.

So at this point I'm very confused. How is it failing prime 95 so fast, but no problems anywhere else?

On a side note, I noticed that x64 Windows gives you a decent boost in memory bandwidth. With DDR 600 in x64 I was getting roughly 8.0 GB/S, with x32 it's down to roughly 7.5 gb/s.

Indyxc1 · Feb 25, 2006

LoneWolf121188 said:
I'd definetly throw at least some staples or pennies on those MOSFETS...see if you can get a small fan blowing on them too. Some place had uber-cheap ramsinks that you could probably use for MOSFET sinks...check in cyber deals...

Well I have 2 fans on my side panel cut. 1 120 mm fan, and 1 80 fan, but niether was directly at the Mosfet area.

So last night I stuck a 80 mm fan right over the area, didn't help with prime 95.

At this point I'm not sure if I want to keep this thing. It's everyday stable at 3000 I guess, but it's a lot of money to me. :shrug:

It's already paid for, but what do you guys think? Worth keeping?

If I had a decent amount of money I'd surely keep it.

PhoenixOfChaos · Feb 25, 2006

I would keep testing it for a bit longer, see if you can try some alternative cooling solutions to really get that puppy flying. Also what hs do you have on it?

jmorgan · Feb 25, 2006

keep burning it in. Prime would fail on me 1 min in at [email protected] I used Burn-in for like an hour and after that it was 5+hour prime stable, it was weird but worked and now can get 2.7 at 1.423v also I would try a diffrent multi. I swear my chip did not like 10x at 270 (cound not really get it stable past 2700) so i was like what the hell let me try a 9 multi (it booted and was stable higher with the 9 multi) now im at (2750 or so and am burning it in) I say take small steps see if you can burn in a 3ghz for a while and see if that makes prime stable.

Indyxc1 · Feb 25, 2006

PhoenixOfChaos said:
I would keep testing it for a bit longer, see if you can try some alternative cooling solutions to really get that puppy flying. Also what hs do you have on it?

It's on water.

Well, I found a thread on a different forum, about dram strengh and drive settings, and it was suggested to try 5/1 instead of 7/2.

I did that, booted into Windows, prime it, makes it 10mins, and I was amazed, because it never made i that far. Made it 16 mins, and failed.

It made it to 16 mins once, and now it won't make it to 5 mins again. :shrug:

One other thing I noticed, is the "bad" core makes it a lot longer in prime, when I prime it alone, when I do them together if fails a lot faster.

I'll keep working at it.

Thanks for suggestions.

OldOne · Feb 25, 2006

Indyxc1 said:
Plug in the ez button? Lol.

Well, this chip is confusing me. Normally, from experience when a chip doesn't pass prime 95 for 4 hours + I find that eventually it will crash when you are playing a game, or whatever. So, I usually don't consider a chip stable, unless it can pass prime.. Especially, one that fails prime 95 in under 5 mins. That usually means programs will crash quickly, or at least sometimes.

Well last night, at 3000 mhz, 1.5v I loaded up Battlefield 2 to see if the chip would crash, set affinity to the "bad" core, and played the game for probably 4-5 hours, with the core being loaded 98% of the time, and the other core being loaded 30-40% the whole time.

Didn't crash on me once, no problems, nothing. Ran Super PI 32M 3 times-> No Problems. Then I ran 3dmark 06 5 times, and watched it fully load both cores on the CPU tests, and it gave me no problem.

So at this point I'm very confused. How is it failing prime 95 so fast, but no problems anywhere else?

On a side note, I noticed that x64 Windows gives you a decent boost in memory bandwidth. With DDR 600 in x64 I was getting roughly 8.0 GB/S, with x32 it's down to roughly 7.5 gb/s.

The 4 pin molex plug.I was thinking it might make the board more stable if you put a little more juice to it. Dont know,just thinking.My system in still running prime so when that is done I am going to plug it in.The manuel said plug in if running two video cards without aux power plugged in.Dont think it will hurt any.

PhoenixOfChaos · Feb 25, 2006

Well your on water so I'd say your volts could go a bit higher, maybe 1.60. I reccomend like someone else mentioned tweaking your dividers and multipliers. My Venice likes lower multipliers with higher FSB to get the maximum out of it. You might want to try tweaking that.

Indyxc1 · Feb 25, 2006

OldOne said:
The 4 pin molex plug.I was thinking it might make the board more stable if you put a little more juice to it. Dont know,just thinking.My system in still running prime so when that is done I am going to plug it in.The manuel said plug in if running two video cards without aux power plugged in.Dont think it will hurt any.

Yeah, the DFI boards have the 24 bin ATX, 4 PIN, Molex, and a plug from the floppy drive. So all are plugged in.

Thanks for the tip though.

Right now I'm gonna try to burn it to in. Never been a fan with that but I'll try.

I refused to believe it is prime stable at 2900 1.45 volts, but not prime stable at 3000 at any voltage.

Oddly it goes the longest prime stable at 1.50 volts, and anything over that it fails quicker.

Another thing I'm thinking is maybe my motherboard is bad. It's always given me inconsistent Overclocks, now with 3 cpus. An opteron 148, FX-57, and now the FX-60. The opteron 148 at first would fail prime 95 2 hours in, then 5, then after a month 10, by the end it never failed.

Likewise,
The FX-57 would barely do better on water, than the previous owner did on air with less voltage.

OldOne · Feb 25, 2006

Ben805 on xtremesystems has a thread on this board,have you seen it? That thread has helped me alot as with this one.

GEK · Feb 25, 2006

Go NAKED! Sounds like one core could be getting hotter than the other.

hawtrawkr · Feb 26, 2006

GEK said:
Go NAKED! Sounds like one core could be getting hotter than the other.

i wouldnt go that far yet. id play with the drive strength settings and go from there. removing the ihs on a 1k cpu is a bit diffrent than a 200$ one and usually a last resort

the memory controller revision you have isnt the best for phase but supposedly does well on air/water maybe the drive strength is causing you some troubles id look in that direction before permanantly altering the cpu itself.

Indyxc1 · Feb 26, 2006

hawtrawkr said:
i wouldnt go that far yet. id play with the drive strength settings and go from there. removing the ihs on a 1k cpu is a bit diffrent than a 200$ one and usually a last resort

the memory controller revision you have isnt the best for phase but supposedly does well on air/water maybe the drive strength is causing you some troubles id look in that direction before permanantly altering the cpu itself.

Yeah, I don't think I'll be taking the top off this FX-60. I'm fairly confident that it would still work, but it isn't woth the risk or the potential payoff.

On that note though. Last night I lowered the room temp, so the load CPU temp/ PWMIC, and Chipset temps were all in the mid 30s.

Well, guess what. It passed prime 95 on the "bad" core until this morning. A whole 6 hours 59 mins. CPU vcore was only 1.500 volts.

Ran it again this morning, and it went 5 hours or so.

So I raised the temps back up, and sure enough when the CPU/PWMIC/NF4 get into the 40s soon after it crashes.

At this point I'm not sure which of the components is the problem. I think if it were the cpu, then the boost in voltage from 1.50 to say 1.60 would cure that, as it usually does, but infact it errors faster at higher voltages. I wish I had a DFI expert to test the chip on, or another motherboard in general.

Another ODD thing I noticed was the prime 95 fatal errors. I've never seen there before. Normally, they are like .5 result, .49 expected. Well, I'm getting result was F96EE51C expected F6BD76. Some type of binary? Anyone see these before?

OldOne · Feb 27, 2006

From what I have seen,core 0 is the core that fails first and around 3 ghz running prime95.Is that the side that the mem controler is on?Not that we can do anything about it,but just a thought.

Indyxc1 · Feb 27, 2006

OldOne said:
From what I have seen,core 0 is the core that fails first and around 3 ghz running prime95.Is that the side that the mem controler is on?Not that we can do anything about it,but just a thought.

Oldone, whats the stepping on your FX-60?

CBB2E 0536 RPMW, SPMW, MPMW, or WPMW?

FX 60: Results and Advice

Member

Attachments

Senior Member

Member

Disabled

Member

Member

Member

Member

Member

Member

Member

Member

Attachments

Member

Member

Member

Registered

Member

Member

Member

Member

Similar threads