SOLVED X2 550 unlocking issues

[BoundByBlood]

Thanks...so far so good.

This is what is going on atm:







So, it seems to be stable and running fine. I've been wondering something though. When I unlocked the additional cores with ACC in bios I had to set the HT link speed to 1.6 GHz or I wouldn't get a video signal otherwise. To compensate I put the HT multiplier to 10x for the HT link to be 2 GHz, but I've been wondering if I should lower it back down to 8x.

 

[coonmanx]

Someone had mentioned to do exactly that by using AMD Overdrive, which is what I believe you did. I tried it, but Overdrive was causing my system to not reboot properly so I uninstalled it. 1600 is plenty of speed and I believe that while running Windows Experience Index that I noticed absolutely no difference in performance. However, if you get it to work then so be it.

 

[stowenrat]

Glad you got it working, Blackheart.

Did you find much difference in load temps when using 4 cores rather than 2? I get a lot more heat when using 4. I am currently playing about trying to see if mine will be stable but I can't give it more than 1.35 vcore until I get a new cooler.

 

[BoundByBlood]

@coonmanx: Did I forget to mention my computer doesn't reboot properly either? I don't think it's just you. When I click on restart the computer shuts down and then nothing. My monitor light goes from blue (operational) to green (inactive). I have to manually power the computer down (pressing and holding the power switch) then turn it back on and it boots right up. It's one of the two oddities I think OverDrive has caused, the other being how the ACC settings are always in red saying the system needs to be restarted to apply the settings. The reboot thing doesn't bother me though since OverDrive hasn't caused any other issues and the system runs perfectly fine.

@stowenrat: No, i really haven't noticed a significant temperature difference, but it could be due to a number of factors such as my custom cooler or the fact I'm using a much better thermal compound than the crap you get with your stock heatsink and fan. With my Cooler Master GeminII and Tuniq TX-2 thermal compound my cpu idle temps as a dual core were 27-29*C and my load temps were around 37-39*C. As a quad core my idle temps are 32-34*C and my load temps are [dependent on the application and cpu usage] 36-38*C or 40-42*C. In the pics I posted if you look in the lower right hand corner you will see my hardware monitor in simple mode. Looking at the taskbar with Google Chrome, AMD OverDrive, CPU-Z, Photoshop, Peerblock, and uTorrent (peerblock and utorrent running as a background process) all open and running my load temp was only 36*C with 1% cpu usage. There is a difference in idle and load temps with the dual vs. quad, but it isn't anything drastic. Even with my stock heatsink and fan the temps weren't that dramatically different between the dual and quad.

Now, someone on Tom's Hardware forum told me that the core temps for the third and fourth cores are disabled so the temp reading I'm getting is probably from the motherboard 10*C lower than what it actually is. I'm not entirely convinced of this becuase it creates an inconsistency and doesn't make much sense.

First, the core temps for the third and fourth cores may be disabled in bios, but I don't think my readings are coming solely from the motherboard. I'm using a third party program, HW Monitor, for my readings. I believe it is reading and factoring in the temps of the last two cores when giving me a reading.

Second, if my idle temps as a dual core are on average 28*C then I'm roughly averaging 14*C per core. So, if as a quad my idles temps still read as 28*C and I add 10*C then I'm only averaging around 5*C per third and fourth cores. What happened to the other 18*C? As it turns out my idle temp as a quad is roughly 34*C and with that person's equation I should be reading it as 44*C. I would be scared if my processor really did idle that hot. No processor, not even with a stock heatsink and fan, should idle that hot because of the low amount of heat generated when the processor is idle.

So how much heat are you generating as a quad? You say you get a lot more, give me a roundabout figure.

 

[coonmanx]

I was told that socket temp (which you will still see with Speedfan) is actually higher than core temps by about 2 to 5 degrees. As for the AMD Overdrive, I do all of my OC'ing through BIOS so I don't have much use for it. I leave HT Link at 1600 since that doesn't really affect performance and with the OC it gets up to about 1800. You could try setting the HT Link speed back to 1600 before trying a restart and see if that works or not. It is definitely an odd thing (the failed restart).

Yes, I don't see core temps any more but that doesn't bother me one bit.

 

[stowenrat]

Not much difference at idle (maybe 3c) but at load, about 8-10c. Not a huge amount but enough for me not to want to test the quad for stability without replacing the stock HSF. That is at about 1.35v.

 

[coonmanx]

Probably a good idea to get a better HSF. These things definitely go hotter as quads under load.

 

[QuietIce]

Now, someone on Tom's Hardware forum told me that the core temps for the third and fourth cores are disabled so the temp reading I'm getting is probably from the motherboard 10*C lower than what it actually is. I'm not entirely convinced of this becuase it creates an inconsistency and doesn't make much sense.

First, the core temps for the third and fourth cores may be disabled in bios, but I don't think my readings are coming solely from the motherboard. I'm using a third party program, HW Monitor, for my readings. I believe it is reading and factoring in the temps of the last two cores when giving me a reading.

Second, if my idle temps as a dual core are on average 28*C then I'm roughly averaging 14*C per core. So, if as a quad my idles temps still read as 28*C and I add 10*C then I'm only averaging around 5*C per third and fourth cores. What happened to the other 18*C? As it turns out my idle temp as a quad is roughly 34*C and with that person's equation I should be reading it as 44*C. I would be scared if my processor really did idle that hot. No processor, not even with a stock heatsink and fan, should idle that hot because of the low amount of heat generated when the processor is idle.

So how much heat are you generating as a quad? You say you get a lot more, give me a roundabout figure.

I don't want to know who was feeding you that hogwash but ALL Phenom II CPUs (X2, X3, X4, and X6) have one and only one sensor on the CPU die. When an X2 or X3 has it's extra cores unlocked that sensor cannot be read by any program.

The best way to monitor the core temp is to drop back to the original configuration, an X2 in this case, and compare the load core temp ((Prime95 or OCCT at an OC similar to what you will be running unlocked)) with the CPU temp given by the motherboard's CPU socket sensor. Often there is a difference between these two readings but that can be used as an index to get rough core temps once the extra cores are unlocked again. For example, if your load core temp at 3.6 GHz is 43°C and your CPU temp is 47°C then the index is -4°C. When you unlock the cores again you would apply the index (subtract 4°C) from whatever reading you get from the motherboard CPU socket sensor. It's not perfect but it's better than just guessing ...

 

[coonmanx]

Exactly.

 

[BoundByBlood]

coonmanx said:

As for the AMD Overdrive, I do all of my OC'ing through BIOS so I don't have much use for it.

I do what I can in bios since it is more preferable, but for some reason I don't have much versatility with the vCore in bios. I can only ramp it up at +.05, +.10, or +.15V at a time to where I can do +.01V increments in OverDrive. That is the main reason I use the program.

You could try setting the HT Link speed back to 1600 before trying a restart and see if that works or not. It is definitely an odd thing (the failed restart).

A minor yet tolerable inconvenience. If I can fix it, great, if not then oh well. I could care less about the failed restart as long as the system runs as a stable quad and doesn't give me any other quirk. As long as all my apps run they way they're supposed to, they don't lock up or freeze, and my system runs as a stable quad then I'm good.

stowenrat said:

Not much difference at idle (maybe 3c) but at load, about 8-10c.

I'm only seeing a 4-6*C flucuation in load temps as reported by HW Monitor using a custom cooler [which performs up there within a 2-3*C striking range against TRUE and Cooler Master V8] and a highly rated thermal compound, so you seem to be still within a tolerable range. I wouldn't really stress it unless the temp displacement was greater. If it was 10-15+*C then I would say you might have a problem.

Not a huge amount but enough for me not to want to test the quad for stability without replacing the stock HSF. That is at about 1.35v.

What I would suggest you do as far as stability goes for the quad is start off with something such as 3DMark06 with all tests enabled and the optimal settings in accordance to your hardware. Once that is done load up prime95 and do a six hour torture test which tests the processor and memory. Finally, do a few hours in Furmark which is "a very intensive OpenGL benchmark that uses fur rendering algorithms to measure the performance of the graphics card. Fur rendering is especially adapted to overheat the GPU and that's why FurMark is also a perfect stability and stress test tool (also called GPU burner) for the graphics card" [as stated on their website]. Furmurk is also good because it taxes the HT link, the ref clock, and the system bus stressing communication between the processor and other components. If your processor can survive all of that without errors then you have a perfectly stable quad. NOTE: When using Furmark turn your gpu fans to 100% since the program is so much more demanding on it than any game on the market.

coonmanx said:

Probably a good idea to get a better HSF. These things definitely go hotter as quads under load.

Indeed. Getting a better cooling solution stowenrat is always a good idea if you can afford it. I'm going water with my next build. I was thinking about phase change [keeping the processor at sub-zero temps does have its appeal], but it's still a new technology and because of the massive AC fans phase units have they are the loudest cooling solutions on the planet.

@stowenrat: If you do upgrade your cooling solution I recommend getting a better thermal compound or some other highly rated TIM (thermal insulated material) as well such as Innovation Cooling's Diamond 24 carat or Tuniq's TX-4 which are supposed to be two of the best. I'm pretty pleased with the TX-2 and I did want to try the TX-4, but I couldn't find any of the -4 for sale. It did get good reviews though. I like Tuniq's compound because it doesn't require any curing time and performs well. From what I've read on the TX-4 on other overclocking sites it performs really well.

 

[I.M.O.G.]

Blackheart - if its a daily driver, what is the point of spending an entire days worth of time running benchmarks to determine stability? If you can run it for a few days and not crash, that's stable for you. If you have an occasional crash, tweak the settings.

I think overly rigorous stress testing is over-rated. If you are doing anything with your pc where an occasional crash is unacceptable, then I don't see any business running an aggressive overclock where stability is critical.

I'm not saying you are wrong, and I'm not saying it to criticize you - I'm just saying it because overclocking is a hobby and its supposed to be fun. Sometimes group think gets people thinking exaggerated things about "stable" and it loses focus of tweaking being fun and playing with settings as you go along! It doesn't have to be a boring stress testing process, unless you are struggling with problems you can't put your finger on - then its a good idea to stress test to identify where the problem resides more exactly.

 

[stowenrat]

Blackheart, I will be getting a decent thermal compound.

I am after a budget cooler like the AS Freezer 7 or Hyper 212 but have concerns about whether they will face towards the rear of case, so I might go for the hyper TX3 or maybe the old AS Freezer 64.

 

[trents]

Blackheart - if its a daily driver, what is the point of spending an entire days worth of time running benchmarks to determine stability? If you can run it for a few days and not crash, that's stable for you. If you have an occasional crash, tweak the settings.

I think overly rigorous stress testing is over-rated. If you are doing anything with your pc where an occasional crash is unacceptable, then I don't see any business running an aggressive overclock where stability is critical.

I'm not saying you are wrong, and I'm not saying it to criticize you - I'm just saying it because overclocking is a hobby and its supposed to be fun. Sometimes group think gets people thinking exaggerated things about "stable" and it loses focus of tweaking being fun and playing with settings as you go along! It doesn't have to be a boring stress testing process, unless you are struggling with problems you can't put your finger on - then its a good idea to stress test to identify where the problem resides more exactly.

Amen!

 

[coonmanx]

From what I've seen from YouTube videos, the Hyper 212 Plus points to the back of the case. Although the guy was doing it on an Intel board. I don't believe that should matter.

 

[stowenrat]

From what I've seen from YouTube videos, the Hyper 212 Plus points to the back of the case. Although the guy was doing it on an Intel board. I don't believe that should matter.

I think it can make a difference, it is specifically an AM3 issue regarding the way the AC Freezer 7 faces for example. However I do think the Hyper 212 will be OK.

 

[QuietIce]

I stand corrected, the Hyper 212 does mount side-to-side - and it's about the only one except for the Sunbeam (which uses the clips). Note that the Hyper 212 Plus is a core contact heatsink, which are better sinks than the normal copper block base type so if you have the option to get the Plus then do so.

From what I've seen from YouTube videos, the Hyper 212 Plus points to the back of the case. Although the guy was doing it on an Intel board. I don't believe that should matter.

Of course it matters, which is why so many coolers do NOT point toward the back. The Intel socket is square, the AMD socket is rectangular. If the long side of the rectangle on the mount does not go the same direction as the heatpipes then it'll be up-down, like my TRUE, instead of side-to-side ...

 

[BoundByBlood]

Originally posted by I.M.G.O.G.:
Blackheart - if its a daily driver, what is the point of spending an entire days worth of time running benchmarks to determine stability? If you can run it for a few days and not crash, that's stable for you. If you have an occasional crash, tweak the settings.

I think overly rigorous stress testing is over-rated. If you are doing anything with your pc where an occasional crash is unacceptable, then I don't see any business running an aggressive overclock where stability is critical.

I'm not saying you are wrong, and I'm not saying it to criticize you - I'm just saying it because overclocking is a hobby and its supposed to be fun. Sometimes group think gets people thinking exaggerated things about "stable" and it loses focus of tweaking being fun and playing with settings as you go along! It doesn't have to be a boring stress testing process, unless you are struggling with problems you can't put your finger on - then its a good idea to stress test to identify where the problem resides more exactly.

Please, allow me to clarify. All throughout my life two interests of mine have been science and math, I apply the empirical method to test and corroborate the legitimacy of every thought process or idea that crosses my mind. A large part of science is validation. When we come up with new ideas and apply them in a practical environment as we test their premise we need to substantiate the results which can also help us discover problems with our methods we may not have been aware of before testing.

You're absolutely right when you saw tweaking is supposed to be a fun hobby, but when you think about it overclocking isn't much different than scientifically testing a hypothesis. You come up with an idea [your oc goal], you create the parameters to test the data [the general reference criteria for your oc settings], you test the data [experiment with different settings], and then you validate the results [ensure your system is stable]. While stress testing may be boring and tedious to you I encompass it as part of the fun process which is overclocking.

Even though it may not always be necessary it can also bring to light hidden problems within our systems which escape our attention. Any pc modder knows because computers comrpise of many different parts operating within various parameters that sometimes ***** just goes wrong, but the evidence of a problem may not always be apparent or obvious. A good example of this would be a while back when I was getting BSODs and system lockups while gaming. To investigate the issue I tried to load up event viewer so I could read the system error dump files and minidumps, but the service wasn't running. I tried to start up the windows event log service, but it wasn't available. Confused as to why I couldn't get a running event viewer I went into a command prompt and decided to check the integrity of the windows installation with system file checker. I did a "sfc /scannow" and sure enough some files, particularly iassdo.mui.dll, was corrupt. I can't tell you the number of times I've taken ownership of the corrupt files, replaced them with the originals from the Win7 dvd and rebooted, but every time I do a sfc scan the same files keep being reported as corrupt. Well, the issue with the system crashes turned out to be something different (the video card), but I wouldn't have become aware of the issue with my windows installation if I hadn't done a sfc scan. It's a peculair problem which doesn't impact system performance or my apps, but at least I'm aware of it as opposed to being completely oblivious and can keep the information of the issue as a refernce within a realm of possibilities when investigating future problems should they occur.

Now, nothing I said was meant to convey these were things stowenrat had to do in order to test stability. He expressed a concern for temps and a desire to test for stability so I simply informed him of what I would do in his predicament. What he takes away from it and how much or how little of my suggestions he follows is on him. I wouldn't recommend anything to anyone I wouldn't do myself.


On a different note the issue with the random reboots, which I'm not currently experiencing, turned out to be with the power supply. It just isn't adequate enough to provide enough power throughput to allow me to stably unlock the quad and overclock it. So, after starting a thread in th psu section and having a discussion with Super Nade amongst others, I've decided to go with an Antec 650W. Antec seems to be a pretty solid manufacturer so it should cover my needs.