AMD "Thuban" Core (Phenom II X6) XS Overclocking Charts

[Polizei]

Locked multipliers can go down, but the "locked" means that they can't exceed their stock value.

For instance, the 1055T is locked at 14, meaning you can't go higher, but you can go lower, which is why you see the 13, 13.5, etc. You can use any multiplier up to and including 14, but not above 14.

[crazydiamond]

locked multiplier means its only locked higher than stock, you can still use all the multipliers that are lower than stock to help fine tune nb/ram/etc if you want

[biohead]

Here's one with a multiplier of 13: http://www.xtremesystems.org/forums/...&postcount=261
And another with 13.5: http://www.xtremesystems.org/forums/...&postcount=247

There's others, but those are what I found in a very quick search.

you must be new to overclocking locked@14x means max=14x.

[seiferoth10]

I am pretty new, and my last two procs (7750 kuma and x3 720) were black editions, so it's the first time I'm playing around with a locked multi.

[Particle]

At 1.52V load (1.55V BIOS) and 4004 MHz, this 1055T system is pulling 420W. 247W idle. Talk about guzzling power... I'm thinking we may need to revise our 1.55V max for 45nm recommendation. That would be suicide for a budget board.

[Polizei]

At 1.52V load (1.55V BIOS) and 4004 MHz, this 1055T system is pulling 420W. 247W idle. Talk about guzzling power... I'm thinking we may need to revise our 1.55V max for 45nm recommendation. That would be suicide for a budget board.

Define budget board.

[Particle]

In this case, I'd be most worried about most boards with only four power phases and those without a heatsink on the MOSFETs. This same system only pulls about 175W idle in a lower vCore configuration, so ALL of that differential is being dumped as heat into the heatsink (minus 25ish percent lost in the PSU and CPU power regulators). We're talking somewhere over 200 watts for the CPU (all heat). This particular board even has vCore options in the BIOS all the way into the 1.8V range, so it is probably designed for a little abuse.

If anything, I'm most impressed with this Tuniq Tower 120 Extreme.

[zhadoom]

Define budget board.

Maybe a non properly cooled VRM system or medium/low quality standard capacitors to power the processor...
The most common and low cost boards designed to support 95W processors ( being optimistic since I already have a Biostar AM2 board that recommended only 65W or lower processors ).

@Particle,
I'm very confused with the temperature reading from every software that I tried. All register 13-14ºC Idle but the environment ( ambient ) is 22ºC. Full load temps are 34-35ºC. That temps looks unrealistic to me. Any guess about this ? since the readings are directly from the core can I supose that the temperature curve ( or something like ) is changed ?

[slaveondope]

I nominate cpuz/hwmonitor for the standard on voltage(under full load) and cpu temp (not core temp)

[p4spooky]

Maybe a non properly cooled VRM system or medium/low quality standard capacitors to power the processor...
The most common and low cost boards designed to support 95W processors ( being optimistic since I already have a Biostar AM2 board that recommended only 65W or lower processors ).

@Particle,
I'm very confused with the temperature reading from every software that I tried. All register 13-14ºC Idle but the environment ( ambient ) is 22ºC. Full load temps are 34-35ºC. That temps looks unrealistic to me. Any guess about this ? since the readings are directly from the core can I supose that the temperature curve ( or something like ) is changed ?

On my Foxconn A79A-S that has 5 phases and the mosfets have a heatsink, I noticed running 4Ghz at 1.39V (1.42 under load), the caps and the ferrite cores were too hot to touch - measured at 62C which I think is not good long term.

The board was reporting temps at 56C (core temp was reporting 52). With vcore set to default - 1.28v (1.31 under load) and multiplier to 19, I now have it running prime95 at 47C for the last 9 hours. Huge difference in temps.

I think overclocking Thubans by increasing vcore beyond 1.45v is something I would not attempt unless you have watercooling and the board has 8+2 phases capable of handling at least 140W with active cooling on the VRM.

[zhadoom]

On my Foxconn A79A-S that has 5 phases and the mosfets have a heatsink, I noticed running 4Ghz at 1.39V (1.42 under load), the caps and the ferrite cores were too hot to touch - measured at 62C which I think is not good long term.

The board was reporting temps at 56C (core temp was reporting 52). With vcore set to default - 1.28v (1.31 under load) and multiplier to 19, I now have it running prime95 at 47C for the last 9 hours. Huge difference in temps.

I think overclocking Thubans by increasing vcore beyond 1.45v is something I would not attempt unless you have watercooling and the board has 8+2 phases capable of handling at least 140W with active cooling on the VRM.

I`m troubled about that readings. All softwares show the same.
This snapshot is about 2 hours of x264 conversion.
Anyone knows the default VID for each p-state ?

[Particle]

I'm very confused with the temperature reading from every software that I tried. All register 13-14ºC Idle but the environment ( ambient ) is 22ºC. Full load temps are 34-35ºC. That temps looks unrealistic to me. Any guess about this ? since the readings are directly from the core can I supose that the temperature curve ( or something like ) is changed ?

Your results are fairly consistent with what we're all seeing. Nothing appears to be calibrated for Thuban's temperature probe yet. +10-20C for real temps are what we're guessing.

[slaveondope]

On my Foxconn A79A-S that has 5 phases and the mosfets have a heatsink, I noticed running 4Ghz at 1.39V (1.42 under load), the caps and the ferrite cores were too hot to touch - measured at 62C which I think is not good long term.

The board was reporting temps at 56C (core temp was reporting 52). With vcore set to default - 1.28v (1.31 under load) and multiplier to 19, I now have it running prime95 at 47C for the last 9 hours. Huge difference in temps.

I think overclocking Thubans by increasing vcore beyond 1.45v is something I would not attempt unless you have watercooling and the board has 8+2 phases capable of handling at least 140W with active cooling on the VRM.

I agree with this basically
The x6 oc'ed is putting rather high demands on the pwm
I stick by cpuz/hwmonitor as my refrence of choice.
Any other temps/voltage readings I dismiss as invalid imo

[PaganII]

Read somewhere to use temp monitor that came with the board.
Ran Bionic for 8 days at stock speeds but 100% on all cores using the stock X6 cooler. Outside air temp was in the mid 90s F most of the week and I don't know what the inside temp was because the thermostat only goes up to 90F but it was warm. Asus probe never budged from 48C. Only the cpu cooler fan speed varied from ~2500 rpm at night to 4800 rpm in the afternoon. Computer never blinked. 250mm fan above the board on Antec Skeleton case.
It is in air conditioned room now and OC to 4.0G/1.4v still with stock cooler and can game all night without hiccups.
I'm really happy with this 1090T.

[nex_73]

Regarding temperature readings: as said earlier the CPU temperature is about 10-15C higher then core temperatures. (At full load)
To get the CPU temperature -
1. New version of CPUID Hardware Monitor (1.16)
2. Use software provided by motherboard manufacturer (example: Asus PC Probe II)
3. Try AMD Overdrive -> Status Monitor -> Board Status
4. Other software? Usually TMPIN0 (May differ between motherboard manufactures?)

[ReKcOlNu]

Regarding temperature readings: as said earlier the CPU temperature is about 10-15C higher then core temperatures.

I hope for BullDozer AMD just disables reading core temps all together, if not then they really need to include the max core temp ratings in their CPU specs and give programmers the correct code (formula) for reading them.

Both of my AMDs core temps are laughable,.. both on air coolers and core temps are 2+C lower then room temp at idle.

[Polizei]

In this case, I'd be most worried about most boards with only four power phases and those without a heatsink on the MOSFETs. This same system only pulls about 175W idle in a lower vCore configuration, so ALL of that differential is being dumped as heat into the heatsink (minus 25ish percent lost in the PSU and CPU power regulators). We're talking somewhere over 200 watts for the CPU (all heat). This particular board even has vCore options in the BIOS all the way into the 1.8V range, so it is probably designed for a little abuse.

If anything, I'm most impressed with this Tuniq Tower 120 Extreme.

Oh, I see. My MSI 890GXM-G65 only has four phases, but the heatsink over the MOSFETs has a "140W Ready" sticker. Budget? It's probably budget because it's a 890GX board but I don't know.

[Particle]

It's not to say four phase boards cannot do it reliably, but it's going to be a lot less common than say a typical six or eight phase board. Everyone exceeds design specs at their own risk.

[kms108]

CPU Model: AMD Phenom II X6 1055T
CPU Stepping: CCBBE CB 1011MPMW
CPU Frequency: 4197 MHz
CPU vCore: 1.55 idle 1.62v load
CPU Multiplier: 18x
CPU Turbo: Disabled
CPU NB Speed: 2796 MHz
HT Ref Speed: 233 MHz
RAM Speed: DDR1555
RAM Timings: 7-7-7-19 T1 unganged
RAM Configuration: 4 GB (2 x 2 GB)
RAM vDIMM: 1.55v
Motherboard: DFI LP Jr 790GX M3H5
Chipset/Socket: AM3, 790GX + SB750
Cooling: H50
Temps: 28ºC Idle / 55ºC Load
Operating System: Win7
32/64-Bit: 64
Stable/Suicide/Untested: Untested/Stable [your pick] 35 mins Prime blend and IntelBurn at the same time

isn't this suppost to be 1090T

[=SOC= Admiral]

CPU Model: AMD Phenom II X6 1055T
CPU Stepping: CCBBE CB 1011BPMW
CPU Frequency: 3710 MHz
CPU vCore: 1.48V
CPU Multiplier: 14x
CPU Turbo: Disabled
CPU NB Speed: 2120 MHz
HT Ref Speed: 265
RAM Speed: DDR3-1413
RAM Timings: 9-9-9-24-1T
RAM Configuration: 2 x 2GB
RAM vDIMM: 1.62v
Motherboard: MSI 790FX-GD70
Chipset/Socket: 790FX + SB750, AM3
Bios Version: 1.12 (1.C)
Cooling: TRUE 120 Push Pull
Temps:22C idle 42C load
Operating System: Windows 7 Professional
32/64-Bit: 64


Stable/Suicide/Untested: Unstable

Folding@Home: Failed EUEs
Prime 95: Untested
OCCT Burn: Passed 1hr Burn
OCCT Linpaq: Untested
Overdrive: Failed Calculation error
CineBench 11.5: Passed 6.28pts
CineBench 10: Passed 1X - CB 4280; 6X - CB 20391; Multi speed up - 4.76x
HyperPi: Passed 1M, 8M, 32M Time ~40min
Intel Burn: Untested
wPrime: Untested
Vantage Score: Passed P17970 http://service.futuremark.com/compare?3dmv=2194745

Verdict: Unstable Froze right after completion of OCCT.

[slaveondope]

isn't this suppost to be 1090T

Yep your right its a 1090T

Edited earlier post

[gooface]

4.1ghz 9 hours stable... this is as far as i think it will go until i have to push it to 1.6v and i dont want to do that, its weird though, only 7.7 in windows cpu score, what is a 7.9??? and who has a stable 24hour system like that??
btw its a CCBBE CB 1011MPMW

[FlanK3r]

CPU Model: AMD Phenom II X6 1090T
CPU Stepping: CCBBE CB 1015DPMW
CPU Frequency: 4300MHz
CPU vCore: 1.478V
CPU Multiplier: 20.5x
CPU Turbo: Disabled
CPU NB Speed: 2726 MHz
HT Ref Speed: 209.8
RAM Speed: DDR3-1398 MHz
RAM Timings: 9-9-9-24-1T
RAM Configuration: 2 x 2GB
RAM vDIMM: 1.55v
Motherboard: ASUS Crosshair IV Formula 0707
Chipset/Socket: 890FX + SB850, AM3
Bios Version: 0707
Cooling: aircooling Xigmateg 1283 DK
Temps:20C idle 45C load (coretemp)
Operating System: Windows XP Professional SP3
32/64-Bit: 32
stable: LInX about 20 minutes

stable for hard benchmarks: max to 4370 MHz
stable for easy benchmarks: max to 4600 MHz
validation: 1.55V 4800 MHz




next: testing Win7 64-bit stability

[dalearyous]

for those of you using an H50, did you replace the thermal paste that comes on it? i accidentally bought some mx-3 thinking i would need it. should i leave whatever is on there or replace it?

[crazydiamond]

replace it