Sign in with
Sign up | Sign in
Your question

Reached the limit at 2.8??

Tags:
  • AMD
  • Overclocking
  • Product
Last response: in Overclocking
Share
April 17, 2009 7:14:40 PM

I have a AMD 5000+ Black edition running on a Asus M2N-SLI Deluxe motherboard, with 2 GB 1x1GB OCZ DDR2 800 PC6400 RAM. I am trying to overclock to a stable 3.0 Ghz speed, however I can’t seem to do so. I have tried to simply up the multiplier and leave everything on auto, Orthos gets 2 seconds in and fails. I have increased the Vcore voltage up to 1.4125, and the same thing happens. Changed my RAM voltage and timings, still the same. Have I reached the limit of stability at just 2.8 Ghz??

I know the articles say you can OC this proc to 3.1 on stock voltage, and i realize that each setup and chip is different. Stock settings seem to be much lower for me as compared to everyone else to has this chip.

Thanks!
Zach

More about : reached limit

a b K Overclocking
April 21, 2009 5:11:22 AM

Hey, ZM
You may try going a little different route. When I had the 5400 BE in this board I had more stable over-clocks by dropping the multiplier and raising the FSB. My old settings looked like this: 13.5x 238 = @3224. Of course yours will be different than that, but by trying it this way you may just get over the @2.8 hump.
April 21, 2009 2:38:12 PM

Z have you read the stickies on this topic, and or browsed around the web? Set all the timinings and voltages manually. Leave the freq at 200mhz. Back the vcore down. A high vcore can damage the cpu, try the OC at stock range voltages with small increases as needed. You must have good cooling. Multi to 16 freq to 200 = 3.2GHZ. If the system (with the latest bios) doesn't run right and it should, try 15.5 and push the freq up a bit at a time.
Related resources
April 21, 2009 2:41:15 PM

Thanks guys i will try those things and give you an update! At one time i had the Vcore up to 1.45, i have since backed it down to stock. I only tried to boot it once at that level of voltage, that shouldn't have hurt the chip right? Will it boot if you over volt components??
April 23, 2009 10:29:11 AM

I currently have my 5000+ BE at 3.2GHz stable with 16 x 200 @ 1.42V on an MSI K9A2 CF v1.X mobo.

Each chip is a little different, and the only way I could get beyond 3.1 stable (15.5 x 200 @ 1.35V) was to increase my voltage. Mind you, I'm also using an aftermarket cooler - ZEROTherm BTF80 - and my 2x2GB OCZ Reaper HPC DDR2-800 timings are 4-4-4-15 at 2.1V. My RAM, and the Crucial Ballistix it replaced, has never worked well with FSB modifications that take it over 400MHz, even at 13-15x multipliers when I had the timings rather tight and tried the bumping up the FSB method.

That said, with the multiplier directly affecting the RAM speed, I decided to leave the FSB at 200 and went to work getting to 16x multiplier to get the memory to 400Mhz. Once I got it there, that's where I stopped. When I tried loosening the memory to 5-5-5-18 timings, I was able to go over 3.4 GHz by increasing the FSB to I think it was 214, but that also required a little more voltage. I decided against that setting and went back to my 3.2GHz setting. I've never messed with RAM or northbridge voltage, either. I dunno... it gets complicated, as you can plainly see.

It really is trial and error with each chip, especially when trying to find which method (multiplier only, FSB only, combo of each) works best for yours. Voltages under 1.5V should not damage these chips with a good cooling solution, so you're probably fine. I don't like the voltage game much, so I'm trying to stay under 1.45V myself. IF I could lower the voltage and see the same speed, as well as keep my memory at ~400MHz with tight timings, I'd do it. But I just can't get it there without a hassle... Maybe I should stop changing this stuff in the BIOS and give AMD Overdrive a shot instead. At least then I wouldn't have to reboot constantly to see if Windows will even start with a particular setting :) 

I really want to max mine out because it is my bottleneck in many of the games I play. At the same time, I don't want to burn it or anything else in the process. I've had mine well over a year now, and I know I'd be better off replacing it with a Phenom II 720, 940, or even the brand new 955, but I can't help but want to see how much I can push this thing. If I decide to replace it, I'll probably go for it, cause burning it out then wouldn't be too much of a loss, so long as it doesn't take anything else with it.
April 23, 2009 5:18:13 PM

Wow, thanks for the awesome response RazberyBandit!

I know my cooling is not that great, so that is a hindrance to me right now. The other thing i am concerned about is the longevity of the processor at higher voltage. Right now it's at stock, BIOS reports at 1.38. I am definitely a noob when it comes to overclocking, so i dont have any experience with long term effects. If i invest in a better cooler, will upping the voltage to get higher speeds have a long term effect on the processor? It's currently about a year old, always been run on stock voltage. I dont plan on buying buying for at least a year or 2.

Have you had any experience with higher voltages affecting the processor life??
a b K Overclocking
April 23, 2009 8:12:47 PM

Small increases in voltages should not have much of a long term effect on CPU longevity. Of course going to extremes, especially with high voltages are going to shorten it's life expectancy. With my 5400, increases in voltages didn't seem to help much. I could get to @3.4 but was nowhere close to stable. That's why I went increasing the FSB.

Keep in mind that if you do decide to OC using the FSB method that you are also increasing speeds to other components as well (HDDs, Memory, etc...)

This is a CPU-Z screeshot of my 24/7 OC, which I ran for nearly 5 months. Note the voltage, Stock!!

April 24, 2009 8:18:28 AM

The Brisbane core's safe zone for voltage goes up as high as 1.5 volts. Once you break that 1.5V mark, you enter into the danger zone where severe damage can occur rather easily. Why? Because more voltage directly translates into more heat, the primary enemy of such sensitive, micro-electrical devices. (Secondary enemy is of course, water/moisture while powered.) My 5000+ BE is just over 15 months old now, and has been running at 3.2 for about half that time at the aforementioned 1.42V. It seems every bit as rock solid as it was day one. It passes through run after run of SiSoft Sandra benches, and has repeatedly passed occasional overnight runs of Orthos Prime 2004 and Prime95.

The heat that overclocking and voltage increases create can be countered by using high quality heatsink/fan units, as well as liquid cooling options. Heatsink/fan units which have a solid copper base, copper heatpipes, copper or aluminum fins, and a variable speed fan are ideal. Nickel plated copper is another excellent material that's starting to become more common lately. Solid aluminum heatsinks, while less-efficient than copper, can cool rather well provided they're used on CPUs run at stock settings, or overclocked low-wattage CPUs. My ZEROTherm BTF80 cooler consists of a solid copper base, four copper heatpipes, and a large stack of aluminum fins. My CPU has never once broken 46C with my cooler. For the longest time I left the fan speed at 100% constantly. While it was at 100%, my CPU wouldn't even get out of the 30's under load.

Now, when it comes to the heatpipes in coolers, "the more the merrier" rule applies in most cases. The use of four is most common, but they are sometimes used differently. Some pipes come off the sides of the base, bend slightly, then run straight up through the fins and end. It works, but there's another variation. The other type will follow the same path up through the fins, but once they're through them, they bend 180 degrees and continue back down through a different section of the fins. Of the few I've used, the latter, 180-degree bending variety gave me better results than the single-direction type.

I hope this info brings a little more light to the subject, and perhaps even a little more life and performance out of your CPU.
May 11, 2009 9:37:13 AM

zm15 said:
I have a AMD 5000+ Black edition running on a Asus M2N-SLI Deluxe motherboard, with 2 GB 1x1GB OCZ DDR2 800 PC6400 RAM. I am trying to overclock to a stable 3.0 Ghz speed, however I can’t seem to do so. I have tried to simply up the multiplier and leave everything on auto, Orthos gets 2 seconds in and fails. I have increased the Vcore voltage up to 1.4125, and the same thing happens. Changed my RAM voltage and timings, still the same. Have I reached the limit of stability at just 2.8 Ghz??

I know the articles say you can OC this proc to 3.1 on stock voltage, and i realize that each setup and chip is different. Stock settings seem to be much lower for me as compared to everyone else to has this chip.

Thanks!
Zach


Hey zm15,
If you use this system for gaming mainly... I have the EXACT same mobo and cpu as you do and OC to 2.8 and i do not have any problems running any games. A 2.8ghz dual core is good enough to run just about any game out right now. I would invest your time into getting more ram, and sli a nice pair of cards. 9800gtx+ based chipset or higher. You will get more gaming out of your money doing this than buying a better heatsink to squeeze out an extra .2 ghz from your processor. Personally, i can run mine at 3.2ghz 70C+ full load (air cooled) but there is no noticeable difference to my games at this setting over my 2.8ghz profile, 3.2ghz didnt let me bump up my screen resolution, my video cards did. I run at 2.8 at 58C full load (air cooled) for the stability and to elongate the life of my CPU. In a year or so when the game developers start programming using more threads and the programming starts to catch up to where you will need a quad+ core processor to even run a game, you will still have the video cards to sli. Its a shame our systems are so out of date already, but thankfully for gaming, video cards and ram are more important as of right now.

Pain
May 11, 2009 1:10:59 PM

Hi Pain, thanks for your insight. It's great to hear someone else with the same issues. I only do minimal gaming, mostly video production using After Effects. So it's getting to be an issue with the dual core more than anything i think. I am running windows 7 RC 64 and 8 gigs of ram and it is beautiful! Now i just need to buck up and get a quad core to stay at a decent speed. Thanks for filling me in on the graphics card sitch. I'm not too well versed in that area. I was looking at a 9600gt, is that a decent card? I don't want to speed a ton on it as i can't take advantage of it too much right now.
May 11, 2009 3:01:47 PM

zm15,
I personally wouldnt get anything less then a 9800gtx+ card. The 9600gt is $77 on newegg and the 9800gtx+ is $125. The 9800gtx+ has double the stream processors then a 9600gt. Dunno if you plan on using sli or not, but at these prices the 1 9800gtx+ would probably perform just as well as the two 9600gt's in SLI, but if you can get the same performance for the same price, i would go with the single card. Less things to power in your system, possibly saving you a power supply upgrade. Also the less parts in a computer, the less chance for things to break. And furthermore, one card would be cooler and quieter inside the case then two cards with fans running. The 9800gtx+ is not a new cutting edge tech card, but 2 of these in SLI should still perform AT LEAST as well as a single GTX 280, which is a very top of the line video card priced at $265. If you follow the tips from the earlier posts, you should be fine as far as speeding up the processor, but if you cant get it to work, then i would save your money for a whole system upgrade instead of dropping more cash on a new heatsink. Unless you plan on getting something like a TRUE that will keep even the newest CPU's cool. Oh, just to let you know, our board comes with a "asus cool n quiet" feature installed on it. If you plan on overclocking, make sure you disable this feature from inside the bios and keep that fan blowing full speed all the time. In the worst case, your fan burns out and you get to buy a new better one that lets you overclock even further. In fact my, all the case fan circuits on the MOBO fried before any of my fans did. Luckily the circuit for the CPU fan is separate then the case fans or i would have been really SOL. I had to wire all my case fans directly to the power supply, not a big deal as i didnt want to use the built in mobo fan speed controls anyhow. Also try not to overclock with that stupid Ai booster windows based program that comes with this board. Always change these settings through your bios. About the only good function of this program is that it will let you save overclock settings profiles an switch them from windows without having to restart your rig, which comes in handy if you want to move from doing that B.S. term paper to a quick Crysis frag fest. Thats about the only other advice that i can give you that hasnt been covered in previous posts. Good luck, i know you can get it up to 3.0ghz, 3.2 ghz is where it starts to get dicey for me with good air cooling. Oh, ill throw in my OC settings that run stable with XP sp3. I dunno what kinda cooler or power supply you have, but i have the same type of RAM (only 4 gigs)/ MOBO / CPU / 650watt powersupply, an aftermarket zalman fan, not top of the line, but a $40 job, dont remember the model off hand and im too lazy to get up an look through the case window.
First i set the external CPU freq. to 205. (this OC the FSB to 1025, pretty mild, push this further to get to 3.1ghz)
Second i set the CPU mult. to 15x
Third i set the CPU voltage to 1.3625 (seems i got one of the good CPU's as most people have this set higher to run stable, but this works for me and keeps the heat low)
I didnt try to OC my ram that much, settings are 5,5,5,15, 2T at 2.1v
And yipee skippy, 3.0ghz on my system, which is very similar to yours :) 
Good luck, i know you can make it happen, this processor was a beast in the day, and it can take some punishment.

Pain
May 11, 2009 8:51:25 PM

Pain, zm. I just wanted to point out that when you use an odd-numbered (or half-step) multiplier with the FSB set to 200MHz, your RAM frequency will drop below 400MHz due to the multiplier's control of the memory divider. Pain's on the right track boosting the FSB a bit to compensate, but in order to actually get the memory up to 400MHz in his machine, he'd have to take the FSB to 213MHz. Any multiplier between 14.5 and 16 uses a divider of 8. I'll show you the math, illustrating the relationship between CPU frequency, memory divider, and RAM frequency.
15 x 200 = 3000 MHz (CPU)
3000 / 8 = 375 MHz (RAM)
15 x 205 = 3075 MHz (CPU)
3075 / 8 = 384.375 MHz (RAM)
15 x 213 = 3195 MHz (CPU)
3195 / 8 = 399.375 MHz (RAM) - almost perfect!
15.5 x 205 = 3177.5 MHz (CPU)
3177.5 / 8 = 397.1875 (RAM) - pretty close!

Now, if we move to either 14x or 16x multiplier, and keep the FSB at a stock 200MHz, we'll see the RAM frequency hitting the target of 400MHz. 14x multiplier uses a divider of 7, while 16x, like 14.5, 15, and 15.5, uses a divider of 8.
14 x 200 = 2800 MHz (CPU)
2800 / 7 = 400 MHz (RAM)
16 x 200 = 3200 MHz (CPU)
3200 / 8 = 400 MHz (RAM)

So, if you're using any half-step or odd-numbered multiplier, it's likely your RAM is not running at it's rated speed. You can make adjustments to the FSB to compensate for this, but you must be careful. FSB increases will speed up the CPU and affect overall stability, and likely add more heat. More heat of course requires more cooling. More cooling requires... Yeah... it goes on and on... :) 

See this article for more detail.

Oh, and I forgot to mention... When increasing the FSB, the HT link also goes up in frequency. The HT link for these CPUs does NOT like much overclocking. It's default multiplier is 5x, so I strongly recommend that once you get to any FSB setting above 215 (if you dare go that high) you drop the HT multiplier to 4x. Yeah, it'll drop to 900-something, but it'll be stable there, whereas it likely won't be stable at 1100+.
May 11, 2009 10:44:23 PM

RazberyBandit said:
Pain, zm. I just wanted to point out that when you use an odd-numbered (or half-step) multiplier with the FSB set to 200MHz, your RAM frequency will drop below 400MHz due to the multiplier's control of the memory divider. Pain's on the right track boosting the FSB a bit to compensate, but in order to actually get the memory up to 400MHz in his machine, he'd have to take the FSB to 213MHz. Any multiplier between 14.5 and 16 uses a divider of 8. I'll show you the math, illustrating the relationship between CPU frequency, memory divider, and RAM frequency.
15 x 200 = 3000 MHz (CPU)
3000 / 8 = 375 MHz (RAM)
15 x 205 = 3075 MHz (CPU)
3075 / 8 = 384.375 MHz (RAM)
15 x 213 = 3195 MHz (CPU)
3195 / 8 = 399.375 MHz (RAM) - almost perfect!
15.5 x 205 = 3177.5 MHz (CPU)
3177.5 / 8 = 397.1875 (RAM) - pretty close!

Now, if we move to either 14x or 16x multiplier, and keep the FSB at a stock 200MHz, we'll see the RAM frequency hitting the target of 400MHz. 14x multiplier uses a divider of 7, while 16x, like 14.5, 15, and 15.5, uses a divider of 8.
14 x 200 = 2800 MHz (CPU)
2800 / 7 = 400 MHz (RAM)
16 x 200 = 3200 MHz (CPU)
3200 / 8 = 400 MHz (RAM)

So, if you're using any half-step or odd-numbered multiplier, it's likely your RAM is not running at it's rated speed. You can make adjustments to the FSB to compensate for this, but you must be careful. FSB increases will speed up the CPU and affect overall stability, and likely add more heat. More heat of course requires more cooling. More cooling requires... Yeah... it goes on and on... :) 

See this article for more detail.

Oh, and I forgot to mention... When increasing the FSB, the HT link also goes up in frequency. The HT link for these CPUs does NOT like much overclocking. It's default multiplier is 5x, so I strongly recommend that once you get to any FSB setting above 215 (if you dare go that high) you drop the HT multiplier to 4x. Yeah, it'll drop to 900-something, but it'll be stable there, whereas it likely won't be stable at 1100+.


Thanks Raz.
Very nice explanation. Like i said, i dont usually run my system at 3.2 or 3.0, im very happy with the what i get out of the 2.8 setting. My ram is actually rated at 5,5,5,18 1.9v, i put it up to 2.1v as i thought this would compensate for the odd multiplier and keep my ram where it should be, at 400mhz. Am i wrong with this train of thought? Guess ill check it with CPU-Z and see what it says. If i run my system OC ill set it to 3.2ghz anyhow, but was just as i said, i dont like to run hot all the time.
Thanks again Raz
May 12, 2009 6:27:28 AM

Unfortunately, the voltage setting for your RAM won't affect it's operating frequency at all. I don't know who told you that, but they were sadly mistaken. However, raising the voltage slightly will often make DDR2-800/PC2-6400 (400MHz) stable at frequencies beyond 400MHz, just like increases in voltage often make overclocked settings for CPUs stable.

Now, there's one last twist when it comes to RAM operating frequency. In most BIOS versions, there's a setting for the FSB/Memory ratio, which is usually defaulted to 1:2 for X2 processors. If you should change that to 1:1.66, the memory will frequency will drop significantly. This is exactly what I believe Unclefester did in the setup of which he displayed the CPU-Z results. It's important to remember this little tidbit when using a high FSB overclock with smaller multipliers. I'll again illustrate mathematically using his settings. Keep in mind, his multiplier was 13.5, which has a memory divider of 7.

13.5 x 239 = 3226 MHz (CPU)
3226 / 7 = 460.857 MHz (RAM)

Now, with the default 1:2 ratio, his memory should be running at a heavily overclocked speed of 460.86 MHz. Not much DDR2-800/PC2-6400 memory out there can be run at such frequencies. I think it's a pretty safe bet he used a 1:1.66 ratio. So, if we take that figure (460.86) and multiply it by 1.66, we get 765.02 MHz. Since this is DDR2, we have to divide that figure by 2 to get the final frequency, which would be 382.51 MHz - his final memory frequency.

Using the 1:1.66 ratio method is incredibly useful when you want to use very high FSB frequencies. I'll illustrate one method I spent nearly 3 days trying to iron the kinks out of, but was unable to get stable without taking my CPU voltage over 1.47V.

14 x 240 = 3360 (14x multiplier has a divider of 7)
3360 / 7 = 480 (RAM frequency with 1:2 ratio)
480 x 1.66 = 796.8 / 2 = 398.4 MHz (final RAM frequency with 1:1.66 ratio)

The final memory speed was damn near 400MHz, and should have worked great with tight memory timings (4-4-4-15-2T). The problem was getting the CPU to be stable at that speed, and countless tests with Prime95 with any voltage under 1.46V would give warnings, and sometimes the occasional error. Even 1.46V would pop the occasional warning after 15-30mins of testing. It also made the CPU hotter than usual (45-46C under full Prime95 load with my 16 x 200 @ 1.42V settings, vs 49-50C with these settings). So, I abandoned that setup as it just wasn't as efficient or cool as the one I'm using now, and I was not comfortable running my CPU at just under it's voltage danger zone.

I should also mention that with no change to the HT link multiplier (5x), my HT frequency jumped to 1200 MHz. In one final attempt to get that setting stable, I changed the HT multiplier to 4x and it dropped to 960 MHz, just a bit under the stock 1000. However, these settings were still unstable at 1.47V.

Maybe I should try this setup at 13.5 like Unclefester did... Possibly push the frequency up to 245 or 250, too. That would be a nice final speed 13.5 x 250 = 3375 Mhz and would leave me with a memory frequency of 402 MHz using the 1:1.66 ratio method.
a b K Overclocking
May 13, 2009 5:36:32 AM

@RazberyBandit, your giving Pain a lot of very good information. But I was running at 1:2 memory divider



It does take some of the higher quality RAM though. Plus I've been told that 1 Gig sticks will do this easier than 2 Gig sticks.

Here's a screen-shot just because somebody once told me I wouldn't be able to run at commandrate 1 at 800Mhz



Now I haven't tried this at @1066, as it's a lot of trouble to clear my CMOS ( I have to pull the soundcard videocard bla.. bla.. bla..)LOL
May 13, 2009 6:05:51 AM

I should have known better than to think you didn't go all out :) 

Hard to believe that setup was 100% stable. What RAM were you using and where can I get some? LOL
a b K Overclocking
May 13, 2009 6:26:33 AM

I'm using OCZ Reapers they are the 1 Gig 8500's with the D-9 chip set. I don't think you can get em anymore with the Micron technology. I bought the first pair in September for $53 after rebate a the Egg. Then never fails they had em for $26 after rebate in October, so I thought i better grab a second pair. 1 of the smarter things I've done when it comes to hardware LOL. If you see OCZ Reapers that are warrantied to 2.35v that be the ones.

Forgot to mention I ran at those settings for almost 5 months. I put this unit together at the beginning of September and added the PII 940 in the middle of January.
July 25, 2009 9:37:49 AM

Hey Z, definately make sure your cool and quiet feature is off, if you are going to try and get to 3.0ghz. When this feature is enabled i believe it will not allow the motherboard to unlock the higher multipliers. Razzy thanks for the good info bro. I also have a 5000+ Black at 3.2ghz on an M2N32-sli deluxe board. When i first started messing around with this board, i noticed unless the cool and quiet feature was off i couldnt get a stable sytem nor could i get the higher multipliers. Z you may try to disable that feature and set your vcore to 1.4v and miltiplier at 16x. Again guys thanks for all the great info, you guys rock :) 
!