Is this the wrong CPU

[t1nm4n]

welp, apparently 4.8 isn't as stable as 4.7 or my voltage was too low, cause in the middle of posting this with p95 still running windows locked up, so I upped the voltage one tick and will start over again. but for an hour it was great, and there could of been other things seeing how I had several tabs opened in IE and all the background programs still running which I figure are a good thing to test stability.

 

[Theocnoob]

What Vcore are you at? You've gotten into the high range OC's for that chip. Keep in mind that the higher the voltage, the faster the chip will degrade. What temperatures are you reaching?

 

[bassnut]

OK you have some stability there but as trents said you need to have HWMonitor running in the background to show all your temps and voltages as your system is being stressed.

@Theocnoob ...... his VCore is shown just 4 posts back ...... 1.416v

 

[bassnut]

from here it is just a matter of getting it stable ...... at least 2 hrs of Prime blend ...... some may say more and if you choose to it is all good

 

[Theocnoob]

OK you have some stability there but as trents said you need to have HWMonitor running in the background to show all your temps and voltages as your system is being stressed.

@Theocnoob ...... his VCore is shown just 4 posts back ...... 1.416v

Was that for 4.8Ghz though? I'm still groggy from the Propofol yesterday so I don't trust my eyes.

 

[t1nm4n]

@trents, ok, I don't know what it is you all don't see, I see 100% cpu utilization under AMD FX-4350 package temp, so that under load, yes I cleared min/max so there was no confusion that this was a picture taken under stress. Min temp sitting at idle is 32c cpu mainboard 31c package 4c something is off on my temps I think.

@theocnood yes the 1.416 is from the 4.8 OC, it did end up failing after posting the pic of proof of runtime, so I upped the voltage one more tick and haven't run a blend as of yet, just been ding normal pc stuffs.

Edit: it just dawned on me, when you say have it running (I don't start the blend until everything else is set up) you want to see min max and do you want it in the same shot as the cpuid tabs, so that it was cut from the same shot, or is it ok that I'm cutting them separately? that might be what I'm not understanding

 

[RGone]

IF you use HWMonitor Free from CPUID com, then you open and have it running and then open P95 Blend mode or Even OCCT P95 Mode. And run it until you reach too high temps, some sort of failure or until you reach the limit of the length of time you want to run. Then P95 OFF and now while never stopping HWMonitor from running, capture the pic of HWMonitor from the Volts down 'thru' Package Temp for FX-4350. That then will show a Min Temp and Voltage with nothing happening but system booted. Then when you begin to run P95 Blend, the Volts Temps will Max out and that is what we want to see. Unloaded Min and loaded Max and they will be in the same Capture of P95 if you start HWMonitor before you begin running P95. AND HWMonitor is still running after p95 is turned off or it fails.

AND I will be just flat honest, I don't care if I ever see a capture of P95 Blend, since if you tell a fib about how long it ran before failing or passing...you ain't hurt me, but are fooling your ownself. BUT I do always wish and really need to see HWMonitor volts thru CPU Package temps. And CPU-z > Cpu Tab > Memory tab and fairly regular post an SPD capture as well so we don't have to go back so far to see what ram and timings are suggested by ram manufacturer. This you do ot help us help you. If you don't need help, then we don't regularly need to see the captures. Simple.
RGone...ster.

 

[t1nm4n]

@RGone so the Min is important for a relative comparison as to how much a system rises and how much voltages may increase or decrease, I can see their relevance. Attached is a 3 hour test of p95 blend, I still think there might be a something wrong with the temp sensors, they do seem a tad low. I will run OCCT and see if they increase more.

Now where to go from here, according to the article jumping to 230-250 fsb seems a bit excessive to me, ill go with 210 and start the increase there, I wont make the same mistake I did the other day with leaving the mem settings too high either. But I will do more testing with this setting I have right now before attempting anything else.

Side note, does viewing streaming movies while your test is going on hurt the test, and are my voltages supposed to fluctuate some while the test is ongoing?

 

[RGone]

1. Your CPUz captures are ALL of the same screen and not CpuZ > Cpu tab > Memory Tab and SPD tab, so not as much info as would have been nice.

2. Okay something about guides/tutorials/how-to's, they are just pointers of a direction that one might take. The problem is that many think that the guide means do A and B and C is the result. BS is what I say. There are almost too many variables to take a guide done on one mobo and with one cpu and expect even 10% of the one's who use the guide as a must do, to actually do the exact same.

3. I can take say one of three 8 core FX processors that i have and the first time I ran either of them, I expected to rather quickly get to 4.5Ghz using a voltage that I had seen used on the first one I ever tested. AFTER that all comparison to my previous attempts was forgotten, since this was a new and different cpu. Imagine how I would flounder and flail around if I tested the second cpu and expected to do the very same things and get the same results? If it were that way, then why would we have people write in asking how to overclock? Just read one tutorial and job done. it is not so. Why do my buddies send me private communication saying they just bought ANOTHER cpu to try and go faster? Why? They do it because they are looking for that better cpu that goes faster. They spend $100s, looking for that 'other' cpu. That one that is really different and better.

4. So we write > test at known stable. Know where that is. Raise multiplier and test until voltage or temps are just insanely silly. Then and most of the time, only then, do we begin to adjust for CPU_NB speed, HT Link speed and ram adjustment toward higher performance if there is any in the ram for real.

5. Key to any real performance of any rig by any maker is raw cpu speed. Period. Has been and as long as a cpu is made to work as they do work...this raw speed will produce the most performance. That is why it is find as much speed of cpu as possible in the realms of 24/7 sanity and then adjust other items for fine tuning. Speed speed speed is the name of the game. That is why the S4 Galaxy was 1.9Ghz but the S5 is 2.4Ghz I believe it is. Speed came into play so they could tout the new phone as faster. Hehehe

Get that cpu to run as fast as it will, mindfull of 24/7 use and then and only then worry about the other items. I have written what may be a number of tutorials about overclocking. I only read such for a very general idea about how a brand new cpu may respond to overclocking in general. But I never read a tutorial with the idea of learning how to overclock. I know how to do that. Raise cpu speed. Test new cpu speed. If new cpu speed fails, I add voltage to 'try' and stabilize and then rinse and repeat as I move to the max stable speed I can attain. It never changes. I monitor the other speeds that 'may' change as I adjust cpu speed faster and I adjust any other speed to keep it inside the realms of sensible and clock her until she hollers >> NO more please. Then my daily speed will be maybe 100 to 200Mhz slower just so I have what I think is a 'margin' for stability that will not be a problem in daily use.
RGone...ster.

 

[t1nm4n]

I feel like such a moron forgetting to change the tabs on cpu id. And all the trouble I went to get that thing uploaded, lol.

I get what your sayin about speed speed speed, its a great selling point to just about everything in this world, why you think so many people buy motorcycles, to go fast! I might also be the guy who get this nice clock and just decides to go back to stock, but I notice a difference in several applications with the OC, it is kinda of nice nwoign you are getting the most out of something, but to what extent (another personal question, I know)?

That sorta makes things easier, glad they are just a guideline of sorts, although some more ideas of min/max. never seen my cpu this high in temp, but it seems to have stabilized here I guessed based on ambient and potential of the cooler. but I am happy with this, just not sure where to go from here, seems stable, this is only a 45min. run so far, but I ran a couple of minor tests before hand to make sure other software issues wouldn't play a part, like Cinebench, all Aida64 benchmarks, 3Dmark physics tests ( only cause I like something to look at) and ran through all the Spi blocks just for something to do I guess, the more tests I throw at it the more comfortable I feel that this is a stable clock.

I will say this I never though this lil H100i AiO cooler would be this good, granted I keep it cold at home, but that's cause I'm cheap and don't want to heat in the winter time. but ambient right now is 77 around the cooler intake.

Edit: I really hope I got all the required info in there for you to see what is happening, or as much as I can get in there at least.

 

[Johan45]

You're doing great t1nm4n, Temps are good the voltage is still reasonable. You have room for more if you want it. One more tick on the multi would put you really close to 5.0

 

[t1nm4n]

you think so? I might try one more tick, I'm gonna let this test run some more, its close to 2 hours no problems while surfing the interwebz, been up all night getting to were I'm at, not sure I can hold out much longer. temps still holding nice at 48-50 still, has anyone noticed that the longer the FFT length of a particular test the higher the temps are? my cpu temps seem to bob up and down along with the longer FFT length tests, might just be my being tired.

 

[RGone]

Looks good man. Yes temps can move based on 'where' in the stress test the testing has reached. Not unknown at all.
RGone...ster.

 

[Theocnoob]

You're on the right track dude. Great work. Merry Christmas!