***955 Unleashed***

[Slappa11]

Hey there chew. Slappa from OCN here.

Very incredible work. I too, had the privilege of getting my hands on one of these early. Here is my benchmark thread (just air results) http://www.overclock.net/amd-cpus/49...enchmarks.html

However, on both my MA790FXT-UD5P gigabyte board, and my A79A-S, I cannot get 4GHz 100% stable for the life of me. It seems that the further I raise the voltage past 1.488, the more unstable it gets. It is odd.

Like you, I could hit 4GHz on stock volts (which of course isn't stable). I believe my chip has the same potential as yours has, but there must be something I'm doing wrong. I tried everything I could think of. From raising the HT Ref clock vs just multi, loosening ram, upping different voltages, using odd combinations etc.

I'm running:

955
A79A-S or MA790FXT-UD5P
DDR2 800/DDR3 1333MHz
TRUE 120 w/ 2 fans in push pull. Temps are great.
Corsair HX520W PSU
Tried in 32bit XP, 32 bit vista, 64bit win 7, no luck

If you have any tips for me, I'd really appreciate it.

Thanks, Slappa

[FlanK3r]

The best for u maybe screens BIOS from *Chew (for me too ideal )

[ecat]

Hmm you super pi1m time is...slow hmmm maybe chang to some higher nb clocks?, are you runing 1600 nb?

Thanks for that
I'm not worried about the actual time. I'd spent all day chasing issues that in the end came down to the psu, I just wanted to see if I could run something at 4.2.

A side note on SuperPi 1m, tweaking the 955 3.8 settings as close as I can to match my previous 720 3.8 settings the 720 puts in better times. Much to learn about this new memory controller.

[Oese]

what you guys think is better for performance.. 1333 cl5 or 1600 cl7??

everest is reading about the same, actually 1333 cl5 is slower in bandwidth and about the same in latency curiously... but i would think 1333 cl5 is better...

[ecat]

Hmm you super pi1m time is...slow hmmm maybe chang to some higher nb clocks?, are you runing 1600 nb?

Getting there, @4.1 btw. ddr2 of course

[Oese]

nice... i cannot boot into xp.. @ same speeds that i can prime in vista 64 and i dont get it.. someone any idea why?

probably its the hdd.. incompatibility or something...

[chew*]

Hey there chew. Slappa from OCN here.

Very incredible work. I too, had the privilege of getting my hands on one of these early. Here is my benchmark thread (just air results) http://www.overclock.net/amd-cpus/49...enchmarks.html

However, on both my MA790FXT-UD5P gigabyte board, and my A79A-S, I cannot get 4GHz 100% stable for the life of me. It seems that the further I raise the voltage past 1.488, the more unstable it gets. It is odd.

Like you, I could hit 4GHz on stock volts (which of course isn't stable). I believe my chip has the same potential as yours has, but there must be something I'm doing wrong. I tried everything I could think of. From raising the HT Ref clock vs just multi, loosening ram, upping different voltages, using odd combinations etc.

I'm running:

955
A79A-S or MA790FXT-UD5P
DDR2 800/DDR3 1333MHz
TRUE 120 w/ 2 fans in push pull. Temps are great.
Corsair HX520W PSU
Tried in 32bit XP, 32 bit vista, 64bit win 7, no luck

If you have any tips for me, I'd really appreciate it.

Thanks, Slappa

Slappa, try dropping your NB (IMC) speed a tad and NB vid voltage, some chips just add way to much heat when you try to give it NB, Also you might want to try backing off on ddr 3 clock speeds/timings.......

I have found it necessary to increase NB not NB vid voltage to gain stability at 1600 and up ddr 3 clocks, especiallly when trying to run tight timings ( 1.28-1.30 )

[chew*]

nice... i cannot boot into xp.. @ same speeds that i can prime in vista 64 and i dont get it.. someone any idea why?

probably its the hdd.. incompatibility or something...

XP boot is harsher and creates more heat than vista/win 7 if you can believe that.......we have noted this quite a few times when benching on LN2, XP load screen is the equivalent of like running cinebench/Wprime 1024 M, I have no explanation why but it definitely puts a strain on the CPU.

Try booting in cold, etc with the PC off for a while.


I found 1333 5-5-5 to be less strain and easier on the IMC.......in my comparsion I think it was 1 sec slower in 32m and the 1600 ram clocks had a 30mhz advantage in NB speeds.........so the 1 second could be contributed to that.....

[xPliziT]

I am having issues getting a higher NB aswell.
As soon as i raise the NB mulitplier just one step the board refuses to boot.
Even raising NB vid and stuff causes no boot, but still working on it.
Running bios version 1.2.


So far got my system memory stable very easy which i did not expect for 8 GB .

running mem 8GB at 7-7-7-24 all 4 slots populated.

[Oese]

XP boot is harsher and creates more heat than vista/win 7 if you can believe that.......we have noted this quite a few times when benching on LN2, XP load screen is the equivalent of like running cinebench/Wprime 1024 M, I have no explanation why but it definitely puts a strain on the CPU.

Try booting in cold, etc with the PC off for a while.


I found 1333 5-5-5 to be less strain and easier on the IMC.......in my comparsion I think it was 1 sec slower in 32m and the 1600 ram clocks had a 30mhz advantage in NB speeds.........so the 1 second could be contributed to that.....

1 sec lower in 32m is more or less nothing. whatever my ram needs > 2.0v for 1333 cl5 so no option. try 1333 cl6 now with 6gb..

i really never experience such boot problems with xp. it was always easier compared to vista 64. but perhaps you are right and the system is better on 64bit and so the harsher boot sequence comes into play.. i suppose you compared with vista 32bit?

whatever its for benches only so i do not care that much.. but nice to know..

[chew*]

1 sec lower in 32m is more or less nothing. whatever my ram needs > 2.0v for 1333 cl5 so no option. try 1333 cl6 now with 6gb..

i really never experience such boot problems with xp. it was always easier compared to vista 64. but perhaps you are right and the system is better on 64bit and so the harsher boot sequence comes into play.. i suppose you compared with vista 32bit?

whatever its for benches only so i do not care that much.. but nice to know..

Compared to win 7 32bit, should be same as vista.......win 7 32 burns more ln2 altogether but winx xp32 literally boils over the ln2 pot at load screen........after boot we use less LN2 in Xp....temps diff between both OS is about 2C within the same benching session, just swapping HD's....... Mind you this is with a Kingpin pot we have noticed this with, and the Kingpin pot handles loads very well........so to actually see this temp increase with that the load has to be very intense.....

It would be safe to say win vista32 and win 732 run the cpu hotter than xp32 for 24/7 usage......but cooler on bootup.....


I've noted this on air as well with a temp probe attached to side of IHS....I have temps on these down to a science now, I know what temps benches will fail at, prime will fail at and how bad temp monitoring software via aod and or bios is off ( etc its innacurate ) I tend to be very very thorough, some would even say that I'm some what anal.........

If the side of IHS is hotter than software I can imagine how warm the core temp actually is........

Here's an example of how "anal" and thorough I am...... Im sure some of you may have experienced a crash with cinebench.........sometime you will blue screen, somtimes you will just black screen and sometimes the bench will just crash ( dissapear, etc just shut down ) and windows will still be up...........the blue screen is NB vid/IMC memory related, the black screen is core clocks/cpu voltage related and the just crash/dissapear from desktop is temp related.............

[Oese]

probably i try with aod one time.. boot into xp at lower speeds and then push it afterwards..

[Slappa]

Slappa, try dropping your NB (IMC) speed a tad and NB vid voltage, some chips just add way to much heat when you try to give it NB, Also you might want to try backing off on ddr 3 clock speeds/timings.......

I have found it necessary to increase NB not NB vid voltage to gain stability at 1600 and up ddr 3 clocks, especiallly when trying to run tight timings ( 1.28-1.30 )

Thank you for your response.

I tried this, and still no dice. I'm just stumped. Tried with both boards again as well.

It is not my ram causing instability, it is just the CPU.

Even raising the voltage on the cpu to 1.55 is more unstable than 1.4V. It's weird.

[Oese]

Slappa, try dropping your NB (IMC) speed a tad and NB vid voltage, some chips just add way to much heat when you try to give it NB, Also you might want to try backing off on ddr 3 clock speeds/timings.......

I have found it necessary to increase NB not NB vid voltage to gain stability at 1600 and up ddr 3 clocks, especiallly when trying to run tight timings ( 1.28-1.30 )

lol nb? really?? how is that...

[chew*]

lol nb? really?? how is that...

Ancient chinese secret , seems to be something that come up with the revision of the IMC.....

Like I said i'm extremely thorough.....I test all aspects even the more obscure, less obvious and unimaginable settings that shouldn't work....

You will see an example of this in a few days as I am documenting how thorough I am just with testing 2 sticks of memory.....day 3 and still going.......

Heres page 1 of 5 so far day 1...........Keep in mind voltage is always and always will be a last resort for me.........

[Oese]

^^ thats nice..

[cdawall]

Ancient chinese secret , seems to be something that come up with the revision of the IMC.....

Like I said i'm extremely thorough.....I test all aspects even the more obscure, less obvious and unimaginable settings that shouldn't work....

You will see an example of this in a few days as I am documenting how thorough I am just with testing 2 sticks of memory.....day 3 and still going.......

Heres page 1 of 5 so far day 1...........Keep in mind voltage is always and always will be a last resort for me.........

looks like my desk i always wondered if i was the only person who had to write all that stuff down

[[XC] gomeler]

I see you missed a number with the CAS settings, it should be CAS6 there

looks like my desk i always wondered if i was the only person who had to write all that stuff down

Nope, I have notebooks full of settings for hardware I no longer own. It is the only way to be consistent when tweaking memory

[FlawleZ]

XP boot is harsher and creates more heat than vista/win 7 if you can believe that.......we have noted this quite a few times when benching on LN2, XP load screen is the equivalent of like running cinebench/Wprime 1024 M, I have no explanation why but it definitely puts a strain on the CPU.

Does this hold true for XP x64 as well?

[chew*]

I see you missed a number with the CAS settings, it should be CAS6 there



Nope, I have notebooks full of settings for hardware I no longer own. It is the only way to be consistent when tweaking memory

Thats scheduled once cas 7 is done lol, probably be on pages 7-14

Wouldn't know flawlez....I don't have 64 bit software thus no 64bit OS........

[XRogerX]

Damn Chew been busy i see lol, check this out lol

http://www.ripping.org/database.php?cpuid=961

[chew*]

Hmmm somehow I managed to fit in a skateboard shred session today as well well Maybe not shred, first time back on a board since september when I broke my ribs skateboarding.

[Slappa]

Hmmm somehow I managed to fit in a skateboard shred session today as well well Maybe not shred, first time back on a board since september when I broke my ribs skateboarding.

You skateboard?

Thats awesome. I skateboard too.

Sprained both my ankles doing it last year, and broke a bone in my foot.

Never met another OCer that skateboarded

[ecat]

Just a quick update on my chip:
3.857 appears optimal but even this requires 1.4375 v for stability.
2.66 is looking to be good for NB speed, I'm working on the voltage, at 1.2875 for now. 2.394 was fine at 1.1125 v, 2.926 failed at 1.3v.

I can make 3.99 stable but I need 1.5125v to do it. A big jump. 4.0 fails at this voltage.

I'm messing with 266 ref speed due to the 3.99 testing. I'll play with lower ref speeds when/if I get around to it.

Of note perhaps, and I'll have to retest my results. At same cpu speed, memory settings, NB settings my Pi1M results do appear slower than those of the 720be

[FlanK3r]

ecat:which is your MB?