Hi guys I've just read through the rog guide and this thread as I have an fx8350. In the rog guide they had it running at 76 deg c and said it was within limits, yet I've seen literature from amd saying its limits are 61.1 deg c on the core and 70 deg c at the socket before it throttles, so what are it's temperature limits and if 61/70 limit is true how did rog get it to run at 76?
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
FX8350 Overclocking FSB vs Multiplier
- Thread starter DMF1986
- Start date
Soldato
It wont throttle till around 70-75 (last time i checked, been on water for a long time) but you should aim to keep it below 61 if you want to prolong life, this is why AMD recommends that limit.
If it actually throttles at the recommended limit (rather than the upper limit) the stock cooler would have to be replaced, as it is utter gash.
If it actually throttles at the recommended limit (rather than the upper limit) the stock cooler would have to be replaced, as it is utter gash.
I have mine under a koolance cpu 380a. At 4.7ghz with a corsair hydro h60 v2 it was at 52 deg c core and 64 deg c socket. This was using amd overdrive stability test for an hour. (I don't render videos or anything just play games so I didn't feel the need to throw prime 95 at it). With the custom loop it's quite strange same oc settings I'm getting 41 deg c at the core and 61 at the socket. So the core is 11 deg c cooler, the socket is only 3 deg c cooler. What could account for this? I wanted to see if I could get 5 ghz out of it. I have 2 x 120mm rads cooling it, 1 is 40mm thick, the other 50mm thick. No other components on water yet (thinking of a gtx770 on water but not sure how it would cope). I play fsx which is more cpu intensive than most games hence wanting to oc it. 
Associate
Presumably because the socket isn't in direct contact with the cooler, so worse direct cooling of it?
Soldato
Simply for the fact he would be a great opinion on the intel/amd debate owning both at the same time and could do all kinds of testing.
Soldato
Anyway, anyone get some interesting findings in the BIOS with regard to overclocking on the FSB?
I didn't get much time but seem to have found the spot my CPU starts requiring lots of voltage for stability - 4.7Ghz
I didn't get much time but seem to have found the spot my CPU starts requiring lots of voltage for stability - 4.7Ghz
Same here not much difference between multi & fsb but it likes voltage after 4.7ghz
I also found voltage jumps going up to 4.6ghz were relatively small then to get 4.7ghz stable needed a much bigger jump.
I'm using a asus m5a99fx pro r2.0 motherboard
I have my multiplier at 20 still just using the fsb it's on 235. Nb and ht 2375 I believe and ram 1545 or something similar. Next notch on the ram is 1880mhz I've got 1866 rated ram so I presume it will be ok at that speed?
I'm using a asus m5a99fx pro r2.0 motherboard
I have my multiplier at 20 still just using the fsb it's on 235. Nb and ht 2375 I believe and ram 1545 or something similar. Next notch on the ram is 1880mhz I've got 1866 rated ram so I presume it will be ok at that speed?
Associate
I'm running mine between 4.8-4.9, though seems to want to go from 1.42v to 1.48v for anything higher. V tempted for 24/7 to have the holy grail of 5Ghz, but am running out of headroom temperature wise - plays bf4 at 62C but can't quite finish asus realbench (locks up due to high temps). Now working out if I can afford a full wc loop... 
5Ghz does give much better cinebench results though. Should have tested it on single core programmes like WoT... May try that tomorrow on my day off
5Ghz does give much better cinebench results though. Should have tested it on single core programmes like WoT... May try that tomorrow on my day off
Soldato
OK what voltages?
I found that 1.44v gets it into windows and doing simple tasks but stressing returns the odd error when using realbench 2.
I think it needs more voltage, maybe we should create a settings list for successful clocks? The asus boards will have similar settings, i.e. if you set a voltage number and not use auto.
Last edited:
Soldato
Would keep the RAM multiplier low as possible while exploring the clocks, but should be fine to up it to near your original RAM speed. It wont always like it (especially if you have an unco-operative IMC) but tweaking timings and upping voltages (on RAM and nb/cpu) could fix that
Associate
Ram does seem to be a bit touchy on these, for instance my vengeance 2133 doesnt run at its XMP settings, it requires 1.625v rather than the 1.5v stated. Maybe an idea for me to check out the new AMD stuff designed for us...
Ok this is my results using amd over drives stability test. My experiences have shown that running small fft in prime 95 increases temperature but doesn't seem to affect stability.
4.5ghz @ 1.4000 vcore 42deg c core and 52deg c socket stable after 30m
4.6ghz @ 1.4125 core 5 failed after 2m
4.6ghz @ 1.425 core 5 failed after 4m
4.6ghz @ 1.4375 47 core 57 socket stable after 30m
4.7ghz @ 1.4625 fail after 5m
4.7ghz @ 1.475 fail after 25m
4.7ghz @ 1.4875 55 core 64 socket stable after an hour
Nb and ht- 2350mhz
Ram - 1566mhz
This was all done under a corsair hydro h60 v2. As those temps were approaching amd limits and I had to bump up the voltage so much I didn't think I'd get 4.8 and not be too hot so I stopped there.
4.5ghz @ 1.4000 vcore 42deg c core and 52deg c socket stable after 30m
4.6ghz @ 1.4125 core 5 failed after 2m
4.6ghz @ 1.425 core 5 failed after 4m
4.6ghz @ 1.4375 47 core 57 socket stable after 30m
4.7ghz @ 1.4625 fail after 5m
4.7ghz @ 1.475 fail after 25m
4.7ghz @ 1.4875 55 core 64 socket stable after an hour
Nb and ht- 2350mhz
Ram - 1566mhz
This was all done under a corsair hydro h60 v2. As those temps were approaching amd limits and I had to bump up the voltage so much I didn't think I'd get 4.8 and not be too hot so I stopped there.
Soldato
I dont think it's the memory as the errors are not typical blue screens or lock ups - more likely a single core that it weaker than it's brothers.
I did get some WHEA errors booting up on 4.8Ghz before I upped the voltage so it's probably that.
When you say the RAM multiplier A7F do you mean selecting it's mhz on the drop downs to a lower than 1600 value?
That seems consistent Damo with what I am finding. I got to 4.7Ghz with 260 fsb but had errors when benching, this was only on 1.44v though. Temps were always good though as I have a H100.
I did get some WHEA errors booting up on 4.8Ghz before I upped the voltage so it's probably that.
When you say the RAM multiplier A7F do you mean selecting it's mhz on the drop downs to a lower than 1600 value?
That seems consistent Damo with what I am finding. I got to 4.7Ghz with 260 fsb but had errors when benching, this was only on 1.44v though. Temps were always good though as I have a H100.
Last edited:
I've now got it under a koolance cpu 380a in a loop on its own with two thicker 120mm rads. I've fitted sp120 high performance fans on either side of each one too. I'm just trying to sort out my flow issues I bought a pump without realising was way too powerful and the water is re entering the bay res with such force it's turning the coolant half into froth. When I sort that out I'm going to see how much more she has to give.
Soldato
Thats the one!
Associate
help o/cing a 8350
...
...
Soldato
Eh? 
Associate
Now I'm confused
With this chip it's not the amount of rad that counts past a point - it's the waterblocks ability to remove all that heat fast enough. I've got a 240 with 4 fans on a lapped chip with liquid metal Tim and 4.8 on all 8 cores is all I can manage without throttling in prime.
If I didnt multitask I'm sure I could get a few hundred more MHz with turbo core enabled but blah!