Overclocking FX-6300

[RGone]

Well let us look at the CPUz SPD tab.
1. XMP programmed by SPD says DDR2133 is 10, 11, 11, 30, 52 and 3T at 1.65Volts applied to the ram. That is the Intel SPD.

2. The AMP preprogrammed SPD shows 10, 11, 11, 30 and 51 and it is likely 2T and 1.65Volts applied to the ram.

3. You should manually adjust tRC to 51 and not the 44 it is currently showing it is set to in CPUz Memory tab. Set CR to 2T.

4. Then I would ensure that the ram voltage is set to 1.65V if you plan to run the ram at DDR2133.

5. Now you are using 210 FSB and the ram is being overclocked from its rating of DDR2133 to DDR2236. Who knows if that ram will run at DDR2236? It is rated to run at DDR2133 and you are not validating the ram at its' rated speed but with it overspeeded. Drop the FSB to 200 which is stock/default and see then if Prime drops any workers with the ram running at its' actual rated speed.

6. That is what I would do first. Not overclock the ram and hope.

 

[Fortunesfool]

RGone Thanks again man Ill let you know how it goes at 200fsb and ill double check the tRC & CR

and Im dumb because CINEBENCH has nothing to do with Ram?

 

[Fortunesfool]

What's a good program to bench overall system performance? I thought CINEBENCH was the one but that's just CPU or GPU I think?

 

[RGone]

Cinebench will show some increase due to ram speed increase. Since it shows values in X.xx format the result number is pretty small and if you are not completely familiar with the previous scores it maybe hard to know there has been an increase in peformance. I had so many of my results recorded, that I can see ram speed changes in the results from the Cinebench results. Like I said the X.xx results format does not allow for showing large amounts of increase.

Okay you might try this but then again without previous data or results from your previous ram, you will not know how much of an increase you have accomplished. Baseline or previous data is a must to really know if our money spent has brought us much greater performance.

http://www.maxxpi.net/pages/downloads/maxxmemsup2---preview.php
MaxxMEM² is freeware. Aida64 is for pay program that has a similar ram speed testing and reporting. The only problem I have seen with any ram performance testing is to know what I was seeing. What is good? I have looked at many types and sticks of ram and the results mean at least a little to me. Just jumping in and gettting a resulting answer may mean little.

If you game try your games out. Considerably faster ram with good tight timings should result in better FPS. Maybe not a lot but some. Ram is only going to mean just so much in a game anyway. But often tighter ram timings over just a couple of 100Mhz faster ram will actually perform better. Like I said test out your gaming and see if you get a better gaming experience. If you really do then you have made progress for sure.
RGone...

 

[Fortunesfool]

Rgone,

I set my FSB to 208 down from the 210 I had set for my previous clock. I passed 8hour P95 test and im settin about 4.48Ghz (I think). My CINEBENCH results are down from 5.80 to 5.78, insignificant drop off compared to the huge difference I noticed in game...

I pulled up the game yesterday to see if I could notice any difference and unless im fooling my self, it looked like it was running way smoother!! Im happy with where I am at now, I dont think I can squeeze anymore volts out of it rite now because its about 70* warmer in my house then it was when I was testing with the other RAM kit. Ill probably leave it alone for now and play around with it again next winter, unless theres anymore sugestions??

Ill post Screenshots when I get home from work.

Thanks again RGone and everyone else who has been helping me from the start with this learning process.

 

[ssjwizard]

Its kind of a juggling act dealing with RAM. Faster speed results in a higher potential throughput, however, tighter timings decrease command latency, but what will give the biggest performance boost? TBH its application dependent. I will say though that with modern CPUs and typical workloads your not going to exceed the total BW available to DDR2133 and timings are going to be your best angle of approach. I am running my 2133 @ 2052 because the HTT speed I run wont allow me to hit 2133(going to that multi puts me in the 2400Mhz range) but as a result I have pulled my timings down from the 10-11-10 rating to 8-9-8 with a touch of overvoltage and the performance difference is pretty drastic. Since most workloads wont push the limits of the speed you really want to persue better timings.

 

[Fortunesfool]

ssjwizard,

That makes alot of sense. Maybe Ill try to drop my RAM Ghz down a notch (not sure exactly what that will be at, then ill try and lower the timings. Do you have any suggestions on jerking the timings around? Should I try to lower each timing by 1 then retesT>?? or should I lower one timeing more than the other? Im assuming that you want the numers to be around the same intervals like if im at 10-11-11 I want to try 9-10-10 eh?

 

[ssjwizard]

http://www.hardwaresecrets.com/article/Understanding-RAM-Timings/26
http://www.overclock.net/t/381699/ram-timings-explained
http://www.tomshardware.com/reviews/pc-memory,1698-2.html

Alot of it is guess work / trial and error. There is a bit of logic about it though. There are a few timings that must be set to higher values as they are limiters. The links above here are all good reads and taking some time to try and understand the way various timings interact with each other will lead you to more successful overclocks. RAM unfortunately is one of those things that most people dont know too much about but we all have to deal with, and its peculiarities can be very frustrating at times. Fortunately these new motherboards provide alot of assistance in dealing with them. Best of luck friend!

 

[RGone]

http://www.hardwaresecrets.com/article/Understanding-RAM-Timings/26
http://www.overclock.net/t/381699/ram-timings-explained
http://www.tomshardware.com/reviews/pc-memory,1698-2.html

Alot of it is guess work / trial and error. There is a bit of logic about it though. There are a few timings that must be set to higher values as they are limiters. The links above here are all good reads and taking some time to try and understand the way various timings interact with each other will lead you to more successful overclocks. RAM unfortunately is one of those things that most people dont know too much about but we all have to deal with, and its peculiarities can be very frustrating at times. Fortunately these new motherboards provide alot of assistance in dealing with them. Best of luck friend!

"ssjwizard" is showing the 'hardware secrets' link and is about as good a read to get off the porch as there is for newer users.

Then that increasing a few of the timings since they exact a limit to other settings is a sure thing. It is just about impossible to tutor ram settings by typing into the forum over the net.

And even by voice it is a job. I used to bench everyday and using AMD; the ram could make a difference. Problem was that there are about three of the timings that were used with DDR2 ram that all three had to be changed at the same time to 'see' a performance increase. So just changing a single timing and testing might be showing a door when it really was a window we were looking thru. Ram tweaking is sit at the keyboard and test. Of course it is helpful to know how the timings can interact as "ssjwizard" says.
RGone...

 

[Fortunesfool]

Rgone,

When u say RAM tweaking is sit at the keyboard and test, you reccomend using http://www.maxxpi.net/pages/download...---preview.php to test as you a previously posted?

SSJwizard,

Thanks for the links bro im going to read up rite now so I can try and have a little better grasp on what each timing does. Im thinking about just going home changing from 10-11-11 to 9-10-10 and dropping the Mhz down a notch that will put me in the high 1900mhz I believe. Ill test and see what happens lol

 

[Fortunesfool]

Its kind of a juggling act dealing with RAM. Faster speed results in a higher potential throughput, however, tighter timings decrease command latency, but what will give the biggest performance boost? TBH its application dependent. I will say though that with modern CPUs and typical workloads your not going to exceed the total BW available to DDR2133 and timings are going to be your best angle of approach. I am running my 2133 @ 2052 because the HTT speed I run wont allow me to hit 2133(going to that multi puts me in the 2400Mhz range) but as a result I have pulled my timings down from the 10-11-10 rating to 8-9-8 with a touch of overvoltage and the performance difference is pretty drastic. Since most workloads wont push the limits of the speed you really want to persue better timings.

What program are you using to see the performace difference when messing with your RAM timings?

 

[ssjwizard]

SuperPi is a great bench for showing memory effects, maxmem, aida64, and Sandra Memory tests are all good ones also. Basically your goal is to tune them and then throw a variety of tests at it and see what happens. You do however see the biggest changes from adjusting the main 5 values that you will see in CPU-z memory tab, but the sub timings can sometimes make a huge difference.

 

[Fortunesfool]

Temps seem to be 2-3*c better with the RAM set like this... I wonder how much of a performance gain I got? Haven't DL any benchmark programs yet for my ram so I guess Ill run the game and see..If my P95 test keeps goin strong..

 

[RGone]

No way to know what performance increase you got from changing the ram. You did not get read/wirte/combined and memory latency of earlier ram. So actually you might know where you are today, but not if you really gained any ram performance without knowing how well the older ram performed.
RGone...

 

[Fortunesfool]

Think it would be worth it to go back and bench my profile with the 1600 ram to compare the difference?
Wouldn't be to hard to do. Gonna run this for a few more hours and see what happens.
Then Ill bench it, save profile, and compare to the old VENGEANCE ram I had.

 

[ssjwizard]

Nothing wasted if it helps you learn something. I spent over a month torturing two different FX 8320s in my rig to see how they compared just to get a feel for them.

 

[Fortunesfool]

I just failed a worker after about 1 hour 30min of P95 blend test.
Maybe I should just set my ram timings back to manufacture specs and bench. I dont understand this SuperPI program..Its all Japaneese...

lol Show me a SS of what its supposed to look like if someone would please?

And this MaxMem program I dont understand either? How am I able to see statistics and compare them?

 

[RGone]

Run Superpi.exe.

Upper left of menu choose or click on Calculate.

Most then use 1M or 32M for the test to run. At least years ago that is what most used to run. I hardly every use SuperPi anymore though.

Click run.

It is pretty simple.

 

[Jscot123]

I have this chip(amd 6300). I would get those same errors in prime 95. What I did was up the cpu voltage +.2 volts in the bios, and it would clock to 4.6ghz stable. The computer shut down at 4.8. I run it at 4.4ghz. Im water cooled btw. Running at 25 degrees idle and 51 degrees under load with an H60. At stock clocks my temps are 21/44. The hyper 212 looked tacky imo. I used it, it was so big and chrome looking(I know its aluminum). It had to go. Seemed to cool well.