-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
amd fx 8350 vs 8370 any differences that matter
- Thread starter shawnmcc
- Start date
- Joined
- Jul 14, 2013
- Location
- Holed up in Branford, CT
I agree with everyone on the awesomeness of SSD's. It is actually painful to use a mechanical drive as anything more than just a place to dump large files.
- Joined
- Jan 12, 2012
to that chris! I don't think I can use a pc anymore, without a SSD!
Well then do that then... but also get a $50 128GB SSD to boot the OS from!
Seriously, Just Do It (tm)
<peer pressure>Another "kinda alternative" is the old Crucial M500 960GB ... I'm pretty sure they're closing them out, so you'll see it flux from $290 ~ $310 places. They're not the fastest SSD for sure, but they're hella better than any HDD. I've bought 2 over the last year.
Though I keep hearing that "SSD prices are gonna drop this summer" but... you always hear that every year, for everything so meh - you can only buy what's here now.
Regardless what you get, keep us in the loop
Last edited:
- Joined
- Aug 11, 2015
Hello, new to the forum. First post in an overclocking thread. Background is with a 3 year old FX-8350 chip that has been overclocked to 4.4 GHZ. First two years at 4.2GHZ on air with Hyper 212. Last year at 4.4 Ghz on H105 AIO watercooling. Will upgrade to a FX-8370 tonight that I just received. I believe that the FX-8350 was damaged during the first two years on air cooling. Never had an issue till the last couple of months with errors on SETI CPU tasks with what appears as a bit-flip. Finally got around to trying Prime95 tests and found it failed immediately on Core#5. Have tested the RAM many times now with MemTest86+ with no errors so deduced that the CPU was causing my CPU errors on my SETI tasks. Downclocked the chip back to stock 4.0 Ghz and can pass Prime95 Blend and Small FFT tests for several hours with no errors. Will start stock clocks and Prime95 to verify the new FX-8370. Then attempt to move on to an overclock at least equal to what the 8350 was doing before failing. Should I attempt to try the 4.6-4.7 Ghz speeds that I see have been reached on watercooling in the many threads I have read today? I have also stuck a 50mm fan on the backside of the CPU socket to cool the socket temps. Made a big difference on temps but unfortunately did not help at all with the overclocked 8350, I guess the damage has been done. I will keep that configuration with the 8370 since I will try to overclock that eventually. I need stable running producing no errors on distributed computing projects SETI@Home, MilkyWay@Home and Einstein@Home. The system is loaded 100% for 8-10 hours during the summer and 24/7 during the winter. My other system is basically a twin of this one except a year newer 8350 that runs 4.4 Ghz at stock VID. A better piece of silicon than the failed 8350 which runs at .08V less VID. I'm hoping I won the silicon lottery with this new FX-8370. Comments about my aspirations?? Thanks in advance.
- Joined
- Sep 7, 2013
I'd start a new thread to get more answers
Also, sounds like you needed just a bit more Vcore on the 8350, and overclock in steps with the 8370 (or any CPU for that matter) raise multi, prime95 if pass, raise multi and test again if fail, raise Vcore and test again.
Wash, rinse, repeat.
to OCF Keith! 
I figured i would add this to the thread in case anybody else had a similar question. Rumor has it that the 8370 is binned slightly higher and is more of an "enthusiast grade" chip. Not stating that it will perform better than an 8350 but that it was selected as a more capable chip when it comes to overclocking. In my small experience i find it to be true. I am by no means an expert or an extreme clocker but it seems my 8370 requires less voltage than my 8350 in comparison. It seems more willing to reach a higher clock. That is all.
- Joined
- Aug 11, 2015
Failed 8350 confirmed IMHO, new 8370 appears good with modest overclock
I tried increasing the Vcore quite a bit already on the 8350, made no difference in passing Prime95 at my original 4.4 Ghz. If anything, caused a failure much faster. I have read through all the overclocking threads for the Bulldozer and ASUS overclocking threads. Think I have a fairly good handle on the concepts and practices. The 8350 chip ran for 3 years with no issues. It was only in the last few months where my invalids were called to my attention and the issue was heavily discussed in a Number Crunching forum thread. Got down to the realization that my CPU was not doing math correctly anymore. Never had any kind of other issues with the computer. Never had freezes, lockups or BSODs of Windows or Boinc Manager. I just think that I had a marginal chip in the first place, one which already had a poor VID voltage compared to my 1 year newer 8350 which runs the same clock speed at stock VID. The only way I could get the failed 8350 to pass Prime95 was to reduce clock speed to stock 4 Ghz. That, with my LLC settings in the BIOS allowed the chip to run at 1.30V VID rather than the stock 1.3875V that was causing the Prime95 failure. Anything I did that caused the Vcore to ramp up toward the stock VID caused Prime95 to fail. That was 99% of the time on Core#5. Thus my deduction that I had hurt that core in overclocking. When I was on air, I was almost always up at the 62 C. limit when fully loaded. Only when I put in the water cooling did I get the chip temps down into the 52-55 C. range. My goals are not to achieve the maximum overclock but to produce valid science results. Any kind of science result failure is to be avoided even if it means downclocking the chip to get valid results.
I have the 8370 at 4.6 Ghz at 1.34V with a multiplier bump and LLC settings. That is only .04V above the stock VID for the chip. The Vcore stays at 1.34V regardless of loading, so I think I have Vdroop dialed in. That puts the core temps at 50-55 C. at 100% load. I am happy with the overclock as it at least matches and betters slightly what I achieved with the 8350. Main thing is that I have produced valid science results again. I don't think I will go any higher on the overclock. Don't think I have enough thermal overhead to push the chip further, at least with my current AIO watercooling. A custom loop would do better of course, but that would entail a complete new build with newer case to fit the watercooling hardware and a lot of new costs. What would entice me to move to a custom loop is to drop the noise level of the dual GTX970s that run all the time. They produce 95% of the noise outputs from my computers. The AIO watercoolers produce only a very little amount of background noise even at 100% load. Watercooling the dual graphics cards with a custom loop would significantly reduce the systems noise level. Some day after shutting down the systems for the day's run at crunching, I might play around with seeing what this new 8370 can do with higher overclocks. Just for grins, I won't be running that way for normal usage. This 8370 already consumes 100W more than the old 8350 at full load. Thanks for the comment and the welcome.
I'd start a new thread to get more answers
Also, sounds like you needed just a bit more Vcore on the 8350, and overclock in steps with the 8370 (or any CPU for that matter) raise multi, prime95 if pass, raise multi and test again if fail, raise Vcore and test again.
Wash, rinse, repeat.
I tried increasing the Vcore quite a bit already on the 8350, made no difference in passing Prime95 at my original 4.4 Ghz. If anything, caused a failure much faster. I have read through all the overclocking threads for the Bulldozer and ASUS overclocking threads. Think I have a fairly good handle on the concepts and practices. The 8350 chip ran for 3 years with no issues. It was only in the last few months where my invalids were called to my attention and the issue was heavily discussed in a Number Crunching forum thread. Got down to the realization that my CPU was not doing math correctly anymore. Never had any kind of other issues with the computer. Never had freezes, lockups or BSODs of Windows or Boinc Manager. I just think that I had a marginal chip in the first place, one which already had a poor VID voltage compared to my 1 year newer 8350 which runs the same clock speed at stock VID. The only way I could get the failed 8350 to pass Prime95 was to reduce clock speed to stock 4 Ghz. That, with my LLC settings in the BIOS allowed the chip to run at 1.30V VID rather than the stock 1.3875V that was causing the Prime95 failure. Anything I did that caused the Vcore to ramp up toward the stock VID caused Prime95 to fail. That was 99% of the time on Core#5. Thus my deduction that I had hurt that core in overclocking. When I was on air, I was almost always up at the 62 C. limit when fully loaded. Only when I put in the water cooling did I get the chip temps down into the 52-55 C. range. My goals are not to achieve the maximum overclock but to produce valid science results. Any kind of science result failure is to be avoided even if it means downclocking the chip to get valid results.
I have the 8370 at 4.6 Ghz at 1.34V with a multiplier bump and LLC settings. That is only .04V above the stock VID for the chip. The Vcore stays at 1.34V regardless of loading, so I think I have Vdroop dialed in. That puts the core temps at 50-55 C. at 100% load. I am happy with the overclock as it at least matches and betters slightly what I achieved with the 8350. Main thing is that I have produced valid science results again. I don't think I will go any higher on the overclock. Don't think I have enough thermal overhead to push the chip further, at least with my current AIO watercooling. A custom loop would do better of course, but that would entail a complete new build with newer case to fit the watercooling hardware and a lot of new costs. What would entice me to move to a custom loop is to drop the noise level of the dual GTX970s that run all the time. They produce 95% of the noise outputs from my computers. The AIO watercoolers produce only a very little amount of background noise even at 100% load. Watercooling the dual graphics cards with a custom loop would significantly reduce the systems noise level. Some day after shutting down the systems for the day's run at crunching, I might play around with seeing what this new 8370 can do with higher overclocks. Just for grins, I won't be running that way for normal usage. This 8370 already consumes 100W more than the old 8350 at full load. Thanks for the comment and the welcome.
- Joined
- Aug 11, 2015
Won the silicon lottery with new FX-8370 it appears ....
That is the reason I went with the 8370 instead of the same 8350, even though it was $25 more. I had read also that the 8370 and 8370E benefited from improved manufacturing, or at least better binning. My 1 year newer 8350 seems to have benefited from better quality or better binning. It runs at stock 1.30 VID. My failed 8350 ran at stock 1.3875 VID and needed 1.42V to run at 4.4 Ghz. The new 8370 has 1.30V stock VID and runs at 1.34V Vcore at current 4.6 Ghz. I think I won the silicon lottery with this new chip.
That is the reason I went with the 8370 instead of the same 8350, even though it was $25 more. I had read also that the 8370 and 8370E benefited from improved manufacturing, or at least better binning. My 1 year newer 8350 seems to have benefited from better quality or better binning. It runs at stock 1.30 VID. My failed 8350 ran at stock 1.3875 VID and needed 1.42V to run at 4.4 Ghz. The new 8370 has 1.30V stock VID and runs at 1.34V Vcore at current 4.6 Ghz. I think I won the silicon lottery with this new chip.
- Thread Starter
- #50
I would reseat the cpu and put new paste on the cpu as well. Mine did the exact same thing until I did this and used far less cpu paste then what I used before. One more thing you need is sand paper the bottom of your heat sink sometimes when this happens it can get damaged doing so cleans and makes it like new.
- - - Updated - - -
I forgot to add the reason sanding the heat sink works is because of cooling and heating up the heat sink can warp and bend. By sanding it your flattening out the bottom again.
- Joined
- Aug 11, 2015
Well, that test will have to wait to see whether I want to build another crunching PC. I have the scrapped parts on the shelf already I think to rebuild with the old 8350. Might need some memory. Question is whether I want to add to my power bill. I would have to put it into another room, no more room now with the existing systems. Would have to go with a wireless card for Internet too since I won't be crawling under the house to run cable. I'm all out of ports on the router anyway. I could just run the failed 8350 at stock 4 Ghz since I already proved to myself it will do correct math at that speed. I would be using my old Hyper 212 air cooler on it for cooling. Good advice on the amount of TIM. I'm pretty sure I overapplied on the 8370 since I used Arctic MX-4 for the first time instead of my usual Artic AS5. The MX-4 turned out to have a lot less viscosity and more than enough came out of the tube when I pushed the plunger. Had to scrape off probably 50% of what came out. Not sure if I need to sand my AIO coldplate. It seems pretty stout with a good polished surface. Probably wouldn't be able to match the polished finish even with my supply of 1200, 1500 and 2000 grit sandpaper.
- Joined
- Jun 28, 2012
I would try this as a last resort.
