SOLVED Can I get more out of my 4170?

[Osnaps]

I will definitely make sure to keep it < or = from here on out. I'm still surprised it was P95 stable for 2 hours with the settings being mismatched like that. It also makes me wonder if thats what caused instability issues in the first place when trying for 4.7

I think I need a reason to go home early....

edit: After seeing keny's screenshot up there, I'm also going to reload the optimized defaults to see how it was setup originally. Its kind of sad I can't remember.

 

[RGone]

I will definitely make sure to keep it < or = from here on out. I'm still surprised it was P95 stable for 2 hours with the settings being mismatched like that. It also makes me wonder if thats what caused instability issues in the first place when trying for 4.7

I think I need a reason to go home early....

edit: After seeing keny's screenshot up there, I'm also going to reload the optimized defaults to see how it was setup originally. Its kind of sad I can't remember.

I glimpse what my bios showed before I changed it on CHV and it was 2200 HTL and 2400 CPU/NB. I usually take digital pics of my bios but did not this time since I knew about where I was going from helping so many set up FX and not even owning one. Hehehe.

Good luck on early going home.

RGone...ster.

 

[keny]

I just done a bit of super quick testing to get to the bottom of why the HT is set so high stock and i still dont 100% know, i took a few screens for you to have a look at and as you can see the higher HT hinders the memory benches (SuperPI) but it helps the CPU based benches (UCBench), i ran these test a few times and the results were almost the same each time.Once the cpu was clocked above 5GHZ the HT almost always needed to match the CPU/NB to finish a bench hence the 5ghz cap.Hope this helps the OP to make the cpu/nb match the ht for stabilitys sake.

High HT


Matched HT = Lower score



High HT



Matched HT = Higher score

 

[Osnaps]

OK, almost to the end of the work day. Gonna test out the 2600/2600 first and see what happens. If unstable going to drop it down to 2400 cpunb and 2200 ht link. Also going to check and see what the default values were.

Anything else to check or suggestions from you guys?

 

[Osnaps]

@keny - I'll be honest, I think most of that just went over my head ,not quite sure I'm reading that right. It hinders the memory bench, but helps the cpu bench. So, possibly being higher it would help in more cpu intensive things? (forgive me here for ignorance)

 

[keny]

@keny - I'll be honest, I think most of that just went over my head ,not quite sure I'm reading that right. It hinders the memory bench, but helps the cpu bench. So, possibly being higher it would help in more cpu intensive things? (forgive me here for ignorance)

Yeah that seems to be the outcome! It's surprised Neva bit as all other amd platforms I used up until bulldozer the HT has done nothing but induce instability. In these quick tests the best stable standpoint seems to be matched nb&ht

 

[Osnaps]

Yeah that seems to be the outcome! It's surprised Neva bit as all other amd platforms I used up until bulldozer the HT has done nothing but induce instability. In these quick tests the best stable standpoint seems to be matched nb&ht

Neat! Ive got mine both matched at 2600 and about to start a prime 95 run on it and then go from there.

UPDATE **After resetting bios values back to normal, BIOS shows my cpunb @ 2200mhz and the HT Link Speed @ 2400mhz**

time to burn some prime will keep this updated

 

[Osnaps]

***UPDATE***

After 33 mins core 3 dropped out on me w/ illegal sumout error. Going to try one tick on the vcore before looking to drop the 2600/2600 down

 

[keny]

2600 on the NB is maybe pushing you're NB a bit too much

 

[Osnaps]

2600 on the NB is maybe pushing you're NB a bit too much

I'll give her a kick down then, I bumped my vcore to 1.43 in bios from the 1.425 just to see what would happen, but if 2600 is to much I definitely don't wanna mess something up.

 

[keny]

It seems from the few tests I have done, (which by no means is extensive), that anything over 2500 is a big ask for the NB on these chips, I don't know why as the thubans were 3000+ capable, maybe it was a trade of to achieve a good clock from the IMC

 

[Osnaps]

I'm almost 30 into another test @ the 2600 mark, last crashed at 33. I think I'm going to stop it. If 2600 is to much, I'm worried about damaging something.

 

[Osnaps]

Just curious now though, is it an issue of damage or just stability?

*The reason I ask is because temps are good and everything is rolling nicely

 

[keny]

If you can get away with running 2600 then great, you won't break anything by running it at those speeds , it's only a stability issue, and if you can run it faster it will help with you're memory performance

 

[Osnaps]

Suh-weet I'll definitely keep it running then to see what happens. Its got about 10 mins to make it to an hour.


*Update*
1 hr 15 min in, so far so good. cpu - 37c , mb - 37c , all cores @ 50c (max 52c)

 

[Osnaps]

worker 3 down for the count after 1hr 20, dropping cpunb down and going again.

I could probably tap a little more into the vcore, but idt it would be worth it and i'm pretty sure id start getting into the 55c range

 

[keny]

If a worker is stopping and it's not crashing, I would say its probably a vcore bump that's needed, but you're temps are pretty high now, also I'm not a big fan of long term stability testing, and 1hr20 of prime would be good enough for me (personally) some say 24h some say 8h, I say 1h

 

[Osnaps]

I usually go by 2hr and call it good there... i'm way to paranoid to let it run 24 hrs with no supervision lol.

 

[trents]

I usually go by 2hr and call it good there... i'm way to paranoid to let it run 24 hrs with no supervision lol.

I'm with you man! I've heard guys say to run those super long Prime stability tests doesn't put your equipment at any more risk but that just doesn't make sense to me. Would you run your automobile at full throttle for hours on end? If you're on the Folding team maybe it would be necessary but not for normal stuff like gaming and surfing the net.

 

[Osnaps]

I'm with you man! I've heard guys say to run those super long Prime stability tests doesn't put your equipment at any more risk but that just doesn't make sense to me. Would you run your automobile at full throttle for hours on end? If you're on the Folding team maybe it would be necessary but not for normal stuff like gaming and surfing the net.

That pretty much sums up how I look at it too lol

* So I'm stable @ 4.6 2400/2400 and 1.425(voltage from op) *

I have noticed though that when I up my cpunb, either manually or via fsb, that sometimes my comp will boot up and windows looks funny. Like the task bar gets changed around from its normal look to some kinda weird chromed out thing and if I go into the net, the pages are slow to load and sometimes it stalls out. I set my ram voltage manually in bios to 1.507 and that seemed to help.

I have 2 questions : 1 - Should I mess with the cpunb voltage any and 2 - Where can I find what my default cpunb voltage is set at? I am having a hard time for some reason finding it in the bios.

Thanks for all ya'lls help so far, this has been a freaking blast!