• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Look here: My top 10 "quick facts" about AM2/AM2+/AM3/AM3+ platforms.

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

BeepBeep2

Member
Joined
Apr 3, 2012
Location
Barberton, OH
Howdy! :) I have been reading some really weird stuff in this section, wondering how people came up to such conclusions, and some information passed on here that are taken as fact but not proven or backed up. The layout of this post is simple, starting from oldest to newest. Simply find the section that applies to you and read up! :D

Please note that this is not an overclocking guide, but a collection of some interesting quick facts that can definitely help you get to know the platform you are dealing with a little better after you get the basics down.

I hope this is a good read for you all, and by the way...if you feel something I say here is inaccurate or incorrect, I encourage you to try it yourself! :D I promise, nothing I suggest will harm your hardware outside of anything you were already doing. These are common-ground facts, gained through extensive testing and validation through many people!

However, I have a disclaimer: Do not take something for fact, or accuse it of being untrue unless you have experienced it yourself. Just because you read something on the internet or make an educated guess at something doesn't automatically mean it's true!

10 quick facts about AM2 (Athlon 64 X2 CPUs)
1. HT Link is usually limited to 1300-1400 MHz. There is no reason to overclock this, and is best left close to 1000. In some cases though, underclocking this can help with coldbugs. Please note that on some motherboards, it is not uncommon to find an HT multiplier that does not work. Some boards have bugged HT multis.

2. RAM speeds are based on CPU speed, they are divided from core speeds, not HT Ref. Clock. (often misconceived as FSB) For some models, RAM speeds are not exact to what the multplier in BIOS suggests.

3. Chances are, if you are running an AM2 setup, you are not running a Sempron or Athlon 64 single core. The X2's came in 2 variants, a 90nm design codenamed "Windsor", and another which is a 65nm design named "Brisbane.

4. "Windsor" based CPUs are faster. They overclock better, and run hotter. They are also more tolerant to voltage. I would suggest 1.6v max for Windsors and 1.55v max for Brisbanes, however 1.7/1.65v for those eager benchmarkers shouldn't hurt for short stints. Keep in mind they get hot and usually don't scale well at high voltages.

5. Some CPUs have weak IMCs, which will limit DRAM overclocks. Others can fly up past DDR2-1300 provided good ram.

6. These CPUs coldbug, BAD. If you want to try LN2 for the first time, this is the wrong platform. Coldbugs are HT Ref. Clock related, and will limit your overclocks. Windsors usually coldbug between -20 and 0c. Brisbanes between -10 and 5c.

7. Some ASUS boards support CPUs all the way up to Phenom II! I think that is pretty cool. :D

8. Max 24/7 clocks for Windsor-based CPUs are in the 3000-3500 range. Max for Brisbanes are in the 3000-3400 range. They don't usually do as well as the Windsors.

9. Windsors are faster when compared clock per clock too! They have 2MB L2 cache vs 1MB L2 on Brisbanes. However, the difference equates to only 3-5%.

10. CAS 4 is the way to go on this platform. With Micron D9 sticks, you can be rocking all the way up to 1066-1200 4-4-4-12! Compared to CL3 and CL5, especially for weaker IMCs, you will get the best results with CL4 every time.

5 quick facts (due to inpopularity) about AM2+(Original "Phenom")

1. These CPUs have varying default HyperTransport speeds depending on the model. Fear not, HT speed is usually unrelated to performance unless overclocking an integrated motherboard GPU.

2. There are two retail design steppings for these CPUs. B2 and B3. B2 stepping suffered from a bug that could cause BSOD's in some situations. For more information, google "Phenom TLB bug" :D

3. These were touted to work in existing AM2 socket boards. While they physically fit, the amount of compatible boards were rather slim. This will cause a no-boot situation, if your AM2 board is not compatible. However, if you upgrade to an AM2+ board, chances are it supports Phenom II CPUs too!

4. To check compatibility, reference your board manufacturer website or see here: http://products.amd.com/en-us/RecommendedMBFilter.aspx

5. Maximum 24/7 overclocks on this platform are between 2800 and 3200 MHz. In reality, you are limited by heat...these chips heat up pretty fast when you pile on the voltage.



10 quick facts about AM2+/AM3 (Sempron 1xx/Athlon II/Phenom II X2/X3/X4)

1. All AM2+/AM3 CPUs except Phenom II X6 and X4 models denoted with a "T" come in two retail design steppings. The first is "C2", which was the release stepping for all AM2+ CPUs. The second is C3, which came in later releases. It is a better overclocking CPU with better power efficiency. It is not faster per clock than C2 CPUs. All of these CPUs have the potential to unlock hidden cores too, in case of X3/X2/Sempron models. Many boards have built-in unlocking features.

2. C2 CPUs often overclock and retain stability better on 32-bit operating systems vs 64-bit. The reason for this is unknown. This phenomenon does not occur with C3 revision CPUs. In any case, good 24/7 overclocks for C2 CPUs are between 3.6 and 3.8 GHz. 4 GHz is usually a stretch.

3. C2 CPUs usually have weaker IMCs and CPU-NB's too. On AM2+ platforms, this is insignificant. On AM3, maximum CPU-NB overclocks are good to maximize DRAM bandwidth with higher DRAM clock frequencies. C2 CPUs usually can only overclock DDR3 to 1600-1700, however I have heard of CPUs that only do 1333-1600.

4. C3 CPUs are preferred for overclocking. These can usually reach higher overclocks all around compared to C2 stepping CPUs. Most of these can get close to 4 GHz and above for 24/7 stable overclocks using standard cooling methods. Most every one of these C3 stepping CPUs can hit 1700-1866 MHz DDR3 and CPU-NB above 2.6 GHz. In the hands over experienced overclockers, select CPUs can even be pushed upwards of 1900 MHz. Maximum recommended voltages for C2 and C3 CPUs are 1.5-1.55v vcore and 1.35v CPU-NB, these CPUs typically do not scale much with higher voltages, but add heat.

5. Some earlier AM3 boards have DRAM overclocking limits! (Gigabyte being notorious for this) - don't be surprised if you have a C3 stepping CPU that did 1900 MHz on someone elses board and only does 1700 on yours. Some boards lose stability at a certain frequency and then subsequently have a hard frequency wall that will not change no matter what CPU you put in it.

6. HT Link is not useful to overclock in most cases. Some guys here say to keep it at 2000 here all the time, however most boards are good up to 3000, which is usually about the limit of CPU-NB (whose minimum multiplier is the same as HT Link) on standard cooling. In other words, you will never have to worry about pushing this frequency too far...but you have no reason to do so anyway.

7. Contrary to popular belief, CPU-NB voltage does not help you reach a higher core overclock! (This is an internet myth, started way back in 2009!) ie. Your CPU will not be able to magically reach 4 GHz instead of 3.95 GHz if you increase CPU-NB voltage. All this does is add heat. If you want to raise your CPU-NB higher, that is a different story.

To address another myth, only of which I have heard on Overclockers.com, that much over 1.5v DRAM Voltage can kill your CPU? Nah. C2 and C3 (before 2011 weeks) non X6 or "T" model CPUs can take up to 2.4v without issue. However, there is a great chance that your DDR3 can not. Most DDR2 can, though! :D For later X4 and X6 models, however, stick to 1.8v or less.

8. Athlon II CPUs often can overclock higher CPU-NB frequency than their Phenom II counterparts. (Assuming same manufacturing time and stepping)...this is because Athlon II CPUs have no L3 cache, which is often a limiting factor in CPU-NB overclocking. (L3 is overclocked)

9. Overclocking your CPU can not damage it, all you will do is reach the edge of stability and have to pull back. You can not degrade the DRAM controller by overclocking your DRAM! The main keys to degradation are A. VOLTAGE and B. TEMPERATURE. That is absolutely all. Telling your CPU to attempt to run faster does not kill it! Okay, maybe it will reduce the lifespan, but by an infinitesimal amount...

10. Maybe you have heard about these CPU's "heat wall". This "wall" is different for every CPU...but these CPUs do a lot better when cooled properly. Usually 55-60c is acceptable with normal cooling methods, some CPUs will take 65+ without issue or losing stability/OC headroom however.

3 quick facts about AM3 (Phenom II X4 "T", Phenom II X6, cont. from above, please read "AM2+/AM3")

1. These CPUs are equipped with a Turbo p-state, which means they can overclock past base frequency when under light loads. This is rather broken and extremely useless on Phenom II platforms. To avoid instability, turbo is best left OFF.

2. These CPUs overclock similarly to the other Phenom II CPUs, and the X4 "T" CPUs are Phenom II X6 based, meaning you can possibly unlock cores. Oh, and by the way, special to these CPUs, you can hit much higher CPU-NB frequencies when you disable some cores vs all cores enabled! I'm talking huge gains, like 3.1 GHz to 3.4! Try it sometime. :)

3. The temperature sensors don't work right on these, the rule of thumb is to take the CPU sensor readout and add 15c for a semi-accurate estimate of CPU temperature. :)

10 quick facts about AM3+ (AMD "Bulldozer" "FX" 4xx0/6xx0/8xx0 series)

1. These CPUs run HOT HOT HOT. Acquire proper cooling before thinking about thinking about thinking about overclocking. :D

2. These are slower CPUs per core than Phenom II in most cases, be advised. They are much slower when compared -per clock to older CPUs, however can be overclocked to higher frequencies.

3. CPU-NB is a rather unimportant component for these CPUs. You are fine at stock, however those "power users" will want to push anything and everything :bday: but these CPUs are not bandwidth strapped, as long as you overclock the DRAM. If you are running into trouble with heat while overclocking, the first thing to suggest is to undervolt/run the CPU-NB at stock voltage and make sure it is at stock speeds. Adding voltage adds heat, and core clocks matter most with FX.

4. For reasons related to chip harvesting (defective parts), 4 core and 6 core FX series CPUs usually overclock worse than their 8 core sisters. Think about this before you buy. In almost every case, a Phenom II part can beat out a 4 or 6 core FX part. The 8 cores have just enough oomph usually in single threads with the extra headroom to match or edge past Phenom II counterparts, however you are really playing lottery with the others. :popcorn:

5. FX is extremely power-inefficient. Don't worry if power consumption spikes after moving from a Phenom II part or intel part...even up to twice as much at the wall compared to competing intel CPUs.

6. Normal overclocks for these CPUs are in between 4.2 and 4.8 GHz, with CPU-NB usually hitting 2400-2600 MHz with ram frequencies through the roof. (2133-2600 maximum 24/7 depending on CPU and if you tweak hard enough) and clock per clock, they are overall 10-20% slower than Phenom II at similar clocks... You do the math, and YMMV. :salute:

7. HT Link has absolutely no benefit at all since there are no IGP chipsets anymore, and all the CPUs are multiplier unlocked. Just leave it on AUTO :bday: or keep it at stock, no reason at all in any case to OC it.

8. These CPUs are a blast to play with, but not for the average joe. Intel has blistering fast single thread speed, and excels in gaming/3D compared to Bulldozer, sometimes up to 50%. If you encode media all day using all 8 cores, that's great...but please note that your power bill will add up compared to competing intel 4 core/8 thread CPU.

9. Just like every other recent AMD SOI CPU known to man, they love voltage. However, the heat is a huge issue, despite frequency scaling. To keep heat down on air/water, 1.45v / 1.3v CPU and CPU-NB voltages are recommended...with a hot-running CPU (not all are the same), you could easily be seeing 75c at load on water cooling with those voltages.

10. The temperature sensors don't work on these! Rule of thumb is to add 20c to the temperature readout. :)


...
Anyway, hope this helps someone... I wrote it for a reason :D

-Sam
 
Last edited:
THANKS! beepbeep2... all looks spot on and seems like you have used my fx and phenom.
 
Now it's add 20C to the Bulldozer temps!?! Jeez, was my CPU really running at 70C at full load, undervolted at stock speeds on the stock cooler? .___.

Also, I think this should be a sticky as it's very informative. Most of the information people need to know in a quick, easy to read post.
 
. . .
10 quick facts about AM2+/AM3 (Sempron 1xx/Athlon II/Phenom II X2/X3/X4)


7. Contrary to popular belief, CPU-NB voltage does not help you reach a higher core overclock! (This is an internet myth, started way back in 2009!) ie. Your CPU will not be able to magically reach 4 GHz instead of 3.95 GHz if you increase CPU-NB voltage. All this does is add heat. If you want to raise your CPU-NB higher, that is a different story.

I go on record as disagreeing with this. This runs contrary to my experience with overclocking my own equipment and with others on the forum I have helped. Just last week I helped someone who ran into instability and when I had him increase his CPUNB voltage from stock to 1.25 as I recall, the computer became stable. He made no other changes and his CPUNB was at stock frequency.

To address another myth, only of which I have heard on Overclockers.com, that much over 1.5v can kill your CPU? Nah. C2 and C3 (before 2011 weeks) non X6 or "T" model CPUs can take up to 2.4v without issue. However, there is a great chance that your DDR3 can not. Most DDR2 can, though! :D For later X4 and X6 models, however, stick to 1.8v or less.

I disagree with this as well. To be honest I have never killed a CPU with too much voltage but I have also been very careful not to exceed 1.55 volts because many knowledgeable and experienced overclockers on this forum, people whose opinions I respect, recommend not exceeding that amount for 24/7 computing use. Beepy, if you want to shoot 2.4v to your CPU, be my guest. Besides, the CPU is not the only component affected by high voltage. Motherboard components can take a beating.


10 quick facts about AM3+ (AMD "Bulldozer" "FX" 4xx0/6xx0/8xx0 series)

10. The temperature sensors don't work on these! Rule of thumb is to add 20c to the temperature readout. :)

I agree their temp sensors read quite a bit on the low side but I'm not sure it's that much off.


-Sam


Some valuable info here but I disagree with you at some points, Sam, as you see in the lavender-colored comments.
 
Last edited by a moderator:
I have to agree with both of you.
my cpu/nb at all clocks is 1.175 my fx is happy at that.
my fx's can all stand over 1.55v, they regularly see 1.66v for extended periods of time and are still alive, my motherboard is another story, the "caps" are rising and i am told that is a sign of the coming of mr death from overheating.

the other "myth" is being unstable above 1.5v.

note, this applys to my 4 fx 8120 processors.

i was having issues with my 955be machine, trents had me raise my cpu/nb voltage and i cleared 4.0ghz right away.

your both right.
 
caddi, when I think "extended periods of time" I'm thinking in terms of 2-3 years though I realize many of us will upgrade before then.
 
Here's a question that I would like to get some kind of definitive answer on, and maybe one of you guys know as I'm not too sure myself. If you pump a high voltage (2.7v+) into ddr2 on a AM3 chip will it likely damage that chips IMC ? And what is the rated speed Imc for ddr2 on AM3 (ie it's 1333 ddr3) is it the same for ddr2.
Thanks
Mark
 
I think that beepbeep2's point is that just upping your vcore a click or two is not going to kill your fun, i watch plenty of noobs, as i am, sit frustrated that they are stuck at whatever ghz, afraid to, or not knowing how to, raise that vcore and reach the next ghz. look at how much i have in this low, mi range box. a poor guy or gal has just spent a grand or a grand and a half and durn well should be scared.
 
Here's a question that I would like to get some kind of definitive answer on, and maybe one of you guys know as I'm not too sure myself. If you pump a high voltage (2.7v+) into ddr2 on a AM3 chip will it likely damage that chips IMC ? And what is the rated speed Imc for ddr2 on AM3 (ie it's 1333 ddr3) is it the same for ddr2.
Thanks
Mark

I haven't seen this in awhile, but when the AMD IMC first came into play, the high Ram voltage could and did damage some IMC's. Remember the IMC was made up of part of the NB that used to reside on the mobo. Way back then we said the IMC was damaged since the ram worked fine in another motherboard but the ram gave issues in the AMD board with IMC that had had high Ram voltage given to it. I think the term used was back EMF causing the damage.

Scratching my head trying to remember all the surrounding circumstances. Mainly trying to remember when the AMD started to use or call out a dual power plane as in use. What we found to do was to use more Vcore than necessary to run the cpu mhz but the extra voltage saved the IMC. You got to remember this was a few years ago now.

When you look at most AM3 and AM3+ boards, you will see a memory spec of 1.5V in most I have read lately. NO I have not read every AMD motherboard spec for memory manufactured in the world. Just saying 1.5V ram volts seems the standard today in AMD world.

I am looking toward PD and had to make a NewEgg order the other day and ordered me some 1.5V ram because " I " did not want to take a chance on hurting "my" IMC. I still have quite a bit of older DDR3 but it takes 1.65V in general.

So when we type into a forum whee gawd himself only knows who is reading, we need to almost always err on the side of caution. IMO only. YMMV.

RGone...ster. :chair:
 
Last edited:
I agree with RGone. I've always said the 1.65v ram makes some AM3/AM3+ systems unstable when the ram is run in that upper frequency/voltage range. Somehow it seems to spill over into the CPU.
 
Cheers guys . I'm just messing with a AM2+ board with a AM3 chip in and busting out some old ddr2 to bench on ,and if high ram voltage was going to damage the Imc (or potentially damage) then a sempy 145/140 was going in and not a thuban, even though thuban Imc is better I'm not going to risk it, I've ran thuban at 2.1v dram voltage up to now with no obvious deteriation in the Imc, i know AM3 have a mem controller for ddr2/3 and what with jedec for ddr2 being 1.8v it should be able to run that at a minimum on ddr2 surely ?
 
First of all, well composed and very informative post. Only one point that I can personally attest against.

. . .
10 quick facts about AM2+/AM3 (Sempron 1xx/Athlon II/Phenom II X2/X3/X4)


7. Contrary to popular belief, CPU-NB voltage does not help you reach a higher core overclock! (This is an internet myth, started way back in 2009!) ie. Your CPU will not be able to magically reach 4 GHz instead of 3.95 GHz if you increase CPU-NB voltage. All this does is add heat. If you want to raise your CPU-NB higher, that is a different story.

trents said:
I go on record as disagreeing with this. This runs contrary to my experience with overclocking my own equipment and with others on the forum I have helped. Just last week I helped someone who ran into instability and when I had him increase his CPUNB voltage from stock to 1.25 as I recall, the computer became stable. He made no other changes and his CPUNB was at stock frequency.

More CPU-NB voltage may not help under air... But I can, without a shadow of a doubt through extensive personal testing, confirm that increasing CPU-NB voltage helps under LN2 after vcore stops scaling frequency. I have tested this extensively on Phenom II as well as Bulldozer. Each time I broke 7GHz (3 times), and when I broke 8GHz (once), I got the last 100MHz or so by stopping vcore where it no longer scaled and increasing CPU-NB voltage.

I cannot attest to if this always works, and I wouldn't increase CPU-NB voltage for core frequency on air for benching or for 24/7 overclocking as its unlikely to make an important difference in those situations... But for extreme benchmarking for max frequency, increasing CPU-NB voltage with low CPU-NB frequency helped increase core frequency when more vCore did not.
 
Last edited:
BeepBeep2 said:
Look here: My top 10 "quick facts" about AM2/AM2+/AM3/AM3+ platforms.

Well it did not turn out to be only 10 quick facts anyway. More like 10, 5, 10, 3 and 10 facts. I don't know if they are all facts by the definition of a fact. A fact could be defined as something that is existing or has existed or is known to be true.

The problem with the idea of being true is the qualifier, stating in my experience or in most cases or words to that effect. Oddly in life there are many things that are true up until a thing does not appear always true any longer. I am sure every culture does not assume that true means always so, but the culture I am from takes true to mean, always a certain outcome will result from doing a certain thing or following a certain set procedure.

Some have experienced over time that raising CPU/NB voltage can in 'fact' have a good effect on their overclock.

Something can certainly be a fact in one's own personal experience but it is a little too far reaching in computer land to make great broad sweeps with the brush of fact since that brushing cannot really brush away others experiences either.
 
I enjoyed reading it, and it must have taken a while to compose :thup: There were some nice points brought out and I definately learned something from reading it (I've probably forgot it already as my minds like a sieve) I don't think I would have the cahooners to pump 2.4v into a phenom ii :eek: man I start sweating putting 1.8v into them when i am sub zero.
 
@ "keny". I keep folders with txt files in them have the URLs to threads like this. So I can find them again. My brain takes a faht much now as I age. Hehehe.

I have foldes like OCF Posting. AMD Stuff which is where this one would go. After I go an dig anything up that is strking, I don't want to have to find it again. So I keep the URL by 'what it is' in the txt title. Some of this stuff is not a RE-type but a copy and paste if the issue has been covered time and again.

I don't have to believe an entire post is fact, in order to get some relevant info from them. In journey across the net, most things I search for have one poster on one side of an issue and in most situations there is someone else on another side. Those two do not compute. You would almost have to believe one is fact and the other not, but often it two people coming at something from either direction. Odd stuff for sure.
 
I think he meant ram voltage over 1.5v rather than core voltage over 1.5v.
I'd be rather surprised if a PhII could live at 2.4vcore for long. Now powering 2.4v ram, quite possibly.
 
Agreed, from my experience benching, he was obviously talking about ram voltage.

Most everyone considers 2V vcore to be lethal in the short term, even on LN2 for PHII... It can be done, and has been done, without death. But very few people would run that voltage on the core without recognizing there is a good chance for death.

suggesting 2.4V makes it clearly a comment concerning RAM to me.
 
Did you put information in there regarding killing the IMC at speeds over 1600Mhz (in that it DOESNT do that?). Sorry if I missed it.

Great post!
 
Back