• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Phenom II X4 955 overclocking adventure

Sylvanas

Sylvanas

Diamond Member
Lets get to it. I have had it installed a few hours and already I have dabbled in some overclocking 😀. It is noticably faster than my 'old' 9950 in games but more on that later. Keep in mind this is preliminary results and in no way what I may find to be the max or 24/7 stable. So, with stock voltage on everything and the CPU core bumped up a tiny bit to 1.36v from 1.35v- I dialled in 3.6ghz and 2.4ghz NB and here it is 1 hr AOD stable (to which I find heats up the CPU the same as Prime). Temps were 39c across all cores at 100% load the whole time. I'm using the latest Beta BIOS (21/4/09) for my board.

This is a quick post and I will update this post as go further.

Update 1 : After that screenshot was taken I got some random BSOD's after an hour or so of it sitting Idle, It appears it doesn't like the higher NB. I am now at 3.8ghz / NB 2.2ghz / HT 2.2ghz @ 1.4v and which appears stable (40min Prime and counting) but I still need to run an extended prime.

Update 2: After some further testing 3.8ghz / 2.2ghz NB and 2.2ghz HT is Fully stable at 1.44vcore and 1.33v NB (higher voltages can raise temps quite a bit as you can see in the pic). It seems I may have hit a wall, it does not seem to go above 3.8 with any amount of CPU VID and the same can be said for the NB- it'll boot into windows at 2.4ghz but won't pass any tests, 2.6 is a no go. It could be a conbination of things, 1) the CPUs limit has been reached 2) Vista 64 3) the Bios. I say the Bios because I have noticed a few oddities like that if I raise the multipliers within AOD and then run Prime it'll run for a while at 3.9 but then stop, but if I raise it in the Bios it will hardly even get past POST. So I will do a little more testing but I am thinking I may be at the limit for this CPU- not to worry though, I am very pleased with 3.8ghz and 2.2NB/HT as it is vastly higher than what I got stable with my 9950. I will post some everest scores soon.

Update 3:
Breakthrough!

I'm glad you mentioned it Geokilla as I enabled ACC across all cores +2% and I am now benching at 3.9ghz no worries, OCCT is half an hour through and still going no worries (ill update with a pic once it's done) when before it didn't even boot into windows! If early indications are anything to go by, I think ACC does infact do something for 45nm CPUs. I remember before with my 9950 the sweet spot was +4% all cores with higher voltage so I still may have some headroom left. Generally higher positive percentages 2-10% increase max clocks with high voltage lower negative pencentages -2 -10% increase stability with lower voltages- this is how it worked with the 65nm Phenoms.

Here is 1hr OCCT 3.9ghz

Update 4: Breakthrough for NB speed!

Wow it appears I missed a major option in my Bios. When entering the BIOS in the lower menu there is an option of 'change setup mode: press F9' by pressing F9 turns the BIOS to 'Advanced mode' just what I need! So after entering advanced mode a whole slew of settings are revealed which were otherwise not visible in 'Basic' mode- this applies especially to memory as there are many more subtimings than were available beforehand. Most interesting of which was when I was in the 'CPU feature' menu, an option appeared called 'CPU-NB VID' with a hex value of 24. I thought to myself "wtf? I already have that setting in the voltage menu" it appears that in the voltage menu there is a setting called 'NB-CORE' (which I have been adjusting all this time) and then this new setting in CPU Feature called 'NB VID' these are most definitely separate settings. So in fact all this time I have been adjusting the wrong setting hoping that it'll improve my NB speed. I think this 'NB CORE' may in fact be in relation to the 790FX chip on the motherboard itself and thus have no actual impact on NB at all, even though they share a similar name.

So, after finding a very helpful thread on this topic through 10seconds of google searching it appears that the default NB VID of a 955 CPU (at least on this board) is 1.1v! After identifying the HEX value that corresponds to a .1v increase in NB VID (1C) I dialled it in an hew presto I am now priming at 2.4ghz no problem- not quite done yet 🙂
 
Very good. Is your DFI - SB600 version or SB750 version? I hear the SB600 version of DFI board is rocking on 955BE just as well as on SB750 version.
 
Update 2: After some further testing 3.8ghz / 2.2ghz NB and 2.2ghz HT is Fully stable at 1.44vcore and 1.33v NB (higher voltages can raise temps quite a bit as you can see in the pic). It seems I may have hit a wall, it does not seem to go above 3.8 with any amount of CPU VID and the same can be said for the NB- it'll boot into windows at 2.4ghz but won't pass any tests, 2.6 is a no go. It could be a conbination of things, 1) the CPUs limit has been reached 2) Vista 64 3) the Bios. I say the Bios because I have noticed a few oddities like that if I raise the multipliers within AOD and then run Prime it'll run for a while at 3.9 but then stop, but if I raise it in the Bios it will hardly even get past POST. So I will do a little more testing but I am thinking I may be at the limit for this CPU- not to worry though, I am very pleased with 3.8ghz and 2.2NB/HT as it is vastly higher than what I got stable with my 9950. I will post some everest scores soon.

Very good. Is your DFI - SB600 version or SB750 version? I hear the SB600 version of DFI board is rocking on 955BE just as well as on SB750 version.
Click to expand...

It is the SB 750 version. Overall I have found it an agreeable board and the fact that the latest Bios makes detecting minor voltages and temps though AOD a breeze (with the exception of CPU Vid) is a bonus as usually those things are detected incorrectly and it just becomes an inconvenience. The current Beta BIOS I am on seems fine with the exception of what I've noted above but I am still hanging out for any future releases DFI will bring out. Maybe next stop is memory timings and subtimings adjustments.
 
:thumbsup:

Outstanding result. I heard that 200MHz gain just by going from 64-bit OS to 32-bit OS, so I'd consider it as 4.0 GHz. 🙂

How is AOD? I didn't really use it although it was because of AOD that I found out my initial problem (4.0 GHz one core, and the rest at 0.8 GHz) I haven't really used it other than for monitoring purposes.
 
Originally posted by: lopri
:thumbsup:

Outstanding result. I heard that 200MHz gain just by going from 64-bit OS to 32-bit OS, so I'd consider it as 4.0 GHz. 🙂

How is AOD? I didn't really use it although it was because of AOD that I found out my initial problem (4.0 GHz one core, and the rest at 0.8 GHz) I haven't really used it other than for monitoring purposes.
Click to expand...

Thanks 😀 4ghz is more a psychological barrier than anything else, its so close yet so far 😛.

I mainly use AOD for 1) Individual core multiplier adjustments b) monitoring everything especially though the 'Board status' menu of the Status monitor, here it keeps tabs on voltages aswell as case temps and motherboard temps. The bench and stability tests are useful as is the memory timing manager but other users around the net are using the tried and tested apps so I figure I may aswell use them for comparison purposes.

Here is my Everest run, I am going to see if I can dig up some other results for everest (are those numbers similar to yours Lopri?) to see if everything is in proper order.
 
I have to look my screenshots as well, but I don't think my Everest numbers were similar to yours. I would always get high write and low copy. Don't remember I've ever seen copy over 9000.

Found one. Mine goes like 7287-8687-7090-58.7. This one was with NB clocked @2.60GHz, though. DDR2-800/5-6-6-18. I'll re-run it later on.
 
Originally posted by: lopri
I have to look my screenshots as well, but I don't think my Everest numbers were similar to yours. I would always get high write and low copy. Don't remember I've ever seen copy over 9000.

Found one. Mine goes like 7287-8687-7090-58.7. This one was with NB clocked @2.60GHz, though. DDR2-800/5-6-6-18. I'll re-run it later on.
Click to expand...

Interesting. That is quite a large gap in numbers there, I wonder why the discrepancy between writes and copy- could come down to Bios / subtimings.

I was a little curious and I changed my ram from 2t to 1t and there was a marked increase in Everest readout- I am now at 7282 / 7481 / 9185 58.9ns same frequency as before.
 
If I venture to guess it's the difference BIOS. Isn't there some genious in DFI Bios team? Forgot his name..
 
I took a look at your Everest shot again. Where was your NB at? It's interesting, indeed. The memory read bandwidth looks as if limited by L3 read bandwidth. That is kinda typical until NB is clocked higher. So you clock up the NB until your memory read bandwidth starts lagging, and I'm guessing that's a kind of a first sweet spot, memory-wise. From there, you can start overclocking memory or use 3:8 divider (or tighten timings), so that memory can catch up with NB again. Rinse and repeat until you reach the limit of NB.

Oh, and I noticed your memory is 'ganged'. My board doesn't seem fond of 'ganged' mode so I use 'unganged'.

Edit: For example, if:

Memory -------- NB/L3
6000 ---------- 6000

then the NB is likely the bottleneck. So overclock the NB. If you get a result like:

Memory -------- NB/L3
7000 ---------- 8000

then memory is not keeping up with the NB. So you overclock memory. (or tighten timings, different dividers, etc.)

It's just my theory.. 😱
 
It seems that all the 955 are topping out at around the 4Ghz range. Didn't the Phenom II 940 top out at more or less the same clock speeds as well? Maybe a new stepping will let overclockers push it to over 4Ghz.

@lopri and OP. Did you guys enable ACC? I'm not good with Phenom II overclocking, but I believe ACC makes a difference.
 
Originally posted by: geokilla
It seems that all the 955 are topping out at around the 4Ghz range. Didn't the Phenom II 940 top out at more or less the same clock speeds as well? Maybe a new stepping will let overclockers push it to over 4Ghz.

@lopri and OP. Did you guys enable ACC? I'm not good with Phenom II overclocking, but I believe ACC makes a difference.
Click to expand...

Average OC with 940 seemed to be around 3.6, lucky if got 3.8. This is 3.8 and it seems many people can get 4.0Ghz on 32-bit.
 
@ Lopri: NB was at 2.2ghz (1.33v) in that screenshot and also unganged. Your theory looks sound but I don't think I can really test it out on this OS 🙁 as it seems my NB will not budge for anything over 2.2- sometimes it'll load windows but fail benches and others it'll just plain not boot with any NB 2.4+ so it looks like I am to make the most of 2.2ghz NB (thats why I have changed the memory to 1t, trc= 22 TRRD= 2). This is not a fresh OS format aswell I just put the new CPU in over the 'old' 9950 installation but I don't think it should make a difference, I hope Windows has altered the appropriate settings under the hood and there is not a software/hardware conflict holding me back.

@ Geokilla: I have not enabled ACC yet as I was under the impression it does nothing for 45nm Denebs as I thought the overclocking benefit is built into the chip this time unlike the 65nm predecessors. Correct me if I'm wrong and I'll definitely give it a go.

I will see how far my bargain bin memory will go at 1t but I wouldn't think too much further, I am amazed they work in 1t as some higher end memory has problems with it- gogo Gskill 😎 .
 
geokilla: I guess both 940 and 955 have advantages and disadvantages. On average, 955 will OC slightly better but what's more notable is 955 needs less voltages. Good 940s will match 955s clock-wise. Oh and ACC.. I haven't figured it out what it does or if it does anything.

nyker96: Not true!! 😛 The wall is 4.0 GHz to be correct. It's more to do with the fact that running 3.9 GHz core with 2.0 GHz NB isn't worthwhile compared to lower core frequency + higher NB frequency.

Sylvanas: I'm going to take some Everest shots. Can't believe I took only a few Everest shots.. My 'theory' could be wrong or right but I think I did see the trend of memory read bandwidth =< L3 read bandwidth (by like 5%).

And you've had the chip for only a day or two.. I had the same experience. A week or two later your NB will be at 2.6 GHz+.
 
Breakthrough!

I'm glad you mentioned it Geokilla as I enabled ACC across all cores +2% and I am now benching at 3.9ghz no worries, OCCT is half an hour through and still going no worries (ill update with a pic once it's done) when before it didn't even boot into windows! If early indications are anything to go by, I think ACC does infact do something for 45nm CPUs. I remember before with my 9950 the sweet spot was +4% all cores with higher voltage so I still may have some headroom left. Generally higher positive percentages 2-10% increase max clocks with high voltage lower negative pencentages -2 -10% increase stability with lower voltages- this is how it worked with the 65nm Phenoms.
 
Originally posted by: Dadofamunky
What are your temps now at those settings? It's sure great to see the AMD side of the fence get some love.
Click to expand...

Max temps in OCCT is around 53c- this is at 1.5v. I decided to bump the voltage for these tests to see what the physical max of the chip is then after that is established I can wind it back if I want to a cooler point, but really, 53c is fine for a quad.

Here is 1hr OCCT 3.9ghz this is with +2% ACC all cores. Next stop the big 4.0 😀

Validation 4.0, still testing
 
Originally posted by: Stoneburner
.... And I thought ACC was a waste of time. Maybe I can hit 3.8 with my 920!
Click to expand...

Definitely give it a go, +2% ACC gave me 200mhz no question about it.

After testing 1.5-1.52v with 2%-6% ACC @ 4ghz it reboots after about 10minutes of OCCT so at the moment it is a no go, but it appears to be near the tipping point. I have still yet to try different voltages, HTT overclocking and maybe 1 more notch of ACC to try and get 4ghz stable on Vista 64 but I am nearing the upper limit of what voltage I am happy with, even on water. I am pleased with 3.9ghz stable and of course I am yet to see how ACC affects NB overclocking- so still much to do.
 
Originally posted by: Sylvanas
Breakthrough!

I'm glad you mentioned it Geokilla as I enabled ACC across all cores +2% and I am now benching at 3.9ghz no worries, OCCT is half an hour through and still going no worries (ill update with a pic once it's done) when before it didn't even boot into windows! If early indications are anything to go by, I think ACC does infact do something for 45nm CPUs. I remember before with my 9950 the sweet spot was +4% all cores with higher voltage so I still may have some headroom left. Generally higher positive percentages 2-10% increase max clocks with high voltage lower negative pencentages -2 -10% increase stability with lower voltages- this is how it worked with the 65nm Phenoms.
Click to expand...

Hooray! Good luck on your 4Ghz. 1.5V seems to be a bit high though for a 45nm chip. Have you tried dropping the HT speeds? Does it scale well with the additional voltages?

@lopri. Well the 940 is cheaper so.......940 for me lol. What's the max that 920 is hitting? I haven't been keeping up with Phenom II overclocking since people seem to be purchasing the Black Editions instead. And what's this ACC 2%-10% that you're talking about....

Does the NB have a big impact on CPU speed? I have no idea what NB does now on AMD CPUs since the Athlon X2 didn't have a NB and Intel has a NB for some other stuff that I do not know about.

Man Phenom II overclocking is as complicated as Core i7 overclocking. Heck I even find C2D difficult to overclock, with all these FSB, NB, etc. settings to set. With the old Athlon X2, I had to change only like 3 settings on my M2N-E and I easily got 2.4Ghz.
 
Small update: I have been able to reduce the CPU VID to 1.48 for complete stability at 3.9ghz which is what I am running now. I have tested up to +8% ACC with 1.52v and it still will not stay stable at 4ghz so I may be at the wall on Vista 64.

@ geokilla: The NB is valuable part of OCing these quads, it reduces the L3 cache latency and can have a serious impct on performance, sometimes moreso than clock speed. I am still yet to thoroughly test a higher NB but in a few small tests it didn't like anything above 2.2ghz.
 
Originally posted by: Sylvanas
I'm glad you mentioned it Geokilla as I enabled ACC across all cores +2% and I am now benching at 3.9ghz no worries, OCCT is half an hour through and still going no worries (ill update with a pic once it's done) when before it didn't even boot into windows! If early indications are anything to go by, I think ACC does infact do something for 45nm CPUs. I remember before with my 9950 the sweet spot was +4% all cores with higher voltage so I still may have some headroom left. Generally higher positive percentages 2-10% increase max clocks with high voltage lower negative pencentages -2 -10% increase stability with lower voltages- this is how it worked with the 65nm Phenoms.
Click to expand...
I read a little more about ACC.

AMD's Overclocking Guru, Sami Mäkinen, has a specific technique for testing the overclockability of Phenom II CPUs. By setting the Advanced Clock Calibration (ACC) to zero percent initially, then overclocking to the maximum stable all cores will accept, before checking each core individually with the latest AMD OverDrive.

Unlike the original Pheno's which benefited from a negative two to six percent ACC, Sami explained that with 45nm Phenom IIs it can be used to boost "weaker" CPU cores to higher speeds by using positive ACC. If there is one (or several), applying two percent ACC to that core(s), while leaving the rest at zero percent can eke out a few extra MHz. However he was also keen to point out it may or may not provide additional MHz over leaving ACC disabled - it's entirely down to the quality of the CPU and its synergy with the motherboard and BIOS.

Sami explained that a common overclock was around 3.7GHz with 3.9GHz the best case with high-end air cooling. CPU core voltages from 1.45-1.55V should be OK, but it's not uniform that "more voltage = better" as some prefer closer to 1.45-1.5V rather than 1.55V. Since AMD is still using Silicon on Insulator (SOI) technology for its 45nm products, it's less sensitive to voltage increments unlike Intel's 45nm with High-K MG.
Click to expand...

http://www.bit-tech.net/hardwa...-phenom-ii-x3-720-be/1
 
I thought liquid cooling was supposed to take these chips well into the >4GHz territory?

http://i272.photobucket.com/al...PhenomIIHeadroom-1.jpg

I understand the air-cooled chips hitting sub-4GHz, and the LN2 chips hitting >6GHz...but shouldn't water-cooled be somewhere in the 4-5GHz territory? This isn't so much a question to the OP, but more generally are there LC OC'ing reports out there to support AMD's initial marketing assertions for liquid cooling OC's?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top