Testing does FX-83xx seem to flat line after 4.3GHz

[tym64]

While I am still waiting on my fx 8350, I am gonna push that bad boy and see how far I can get it. I Recall being able to get my fx 6100 on the same motherboard to a stable 4.5.

Now I may be wrong but I honestly think that part of the trick of getting the most out of the fx series is TRYING the following steps. I am not gonna say it will work. Its just a idea to try to see if it helps. Its how I got the fx 6100 to 4.5 when I hit the wall at 4.3.


Now Open AI Suite 2 You want to find DIGI PLUS VRM Tool.

Set CPU Load Line Calibration to Extreem, Full Blast.

Set CPU Current Capability to Full Blast 140 Percent.

Set the CPU Power Phase Control to Optimized.

Set the CPU/NB Current Capability to 120 Percent.

CPU/NB Load Line Calibration was left alone.

With the above settings, I got 4.5 stable, When I set the above back to normal settings, my 4.5 setting fail. I am not saying it will help. Just a idea you can try to get beyond your own brick wall.

 

[RGone]

Honestly I have already had my FX-8350 stable at 5.2Ghz and benched at 5.4Ghz. Don't think I will have to resort to just brute force.

I mean you got to realize who C_D is. He is the Vcore molester. If it don't go just force it. Hehehe. Where most would recoil in horror he just keeps upping the Vcore. I ran a cpu that he had had and know it had over 1.675Vcore on it. And that was not on DICE or LN2, that was on water. So we pretty much know how to run into a wall, but the end of the world on water is real. I told him I would not burn up his FX-6300 but he keeps telling me to NAIL IT. He might be sorry.
RGone...

 

[caddi daddi]

I think your just not wanting to stress out that scrawny chv-f, chuck that thing in that chubby chick you got laying around and shovel a few more scoops of coal under it......

 

[tym64]

Honestly I have already had my FX-8350 stable at 5.2Ghz and benched at 5.4Ghz. Don't think I will have to resort to just brute force.

I mean you got to realize who C_D is. He is the Vcore molester. If it don't go just force it. Hehehe. Where most would recoil in horror he just keeps upping the Vcore. I ran a cpu that he had had and know it had over 1.675Vcore on it. And that was not on DICE or LN2, that was on water. So we pretty much know how to run into a wall, but the end of the world on water is real. I told him I would not burn up his FX-6300 but he keeps telling me to NAIL IT. He might be sorry.
RGone...

Could you please give me a bit of info on your setting to run this fast. I thank you kindly in advance.

 

[RGone]

Could you please give me a bit of info on your setting to run this fast. I thank you kindly in advance.

Yeah. Go back to post #7 where it says:
Other Benches and BIOS Screenshots are in the Spoiler.
Then do the same thing for the next 4 or 5 posts and there are bios shots in each of the posts at the spoiler. If you don't have heavy duty water cooling; as I have, then I doubt you reach those speeds. But the Vcore and ram speeds and all are shown at the spoiler for speeds up to 5.0Ghz I think. I don't really remember how far up the bios shots went right now.

 

[RGone]

Could you please give me a bit of info on your setting to run this fast. I thank you kindly in advance.

Yeah. Go back to post #7 where it says:
Other Benches and BIOS Screenshots are in the Spoiler.
Then do the same thing for the next 4 or 5 posts and there are bios shots in each of the posts at the spoiler. If you don't have heavy duty water cooling as I have then I doubt you reach those speeds. But the Vcore and ram speeds and all are shown at the spoiler for speeds up to 5.0Ghz I think. I don't really remember how far up the bios shots went right now.

 

[tym64]

Well done RGone. That will teach me for being to lazy to read through the posts first Sorry.

Yea I will have liquid cooling with a decent size radiator and push pull fans. Using the same system, I overclocked the fx 6100 to stable 4.5 and it never got over 47c with all the stress tests. Be interesting to see how well it does with this new chip. I am NOT gonna push it much past 50c on the stress test. Gets much more than 50c, Im done.

Myself, won't leave it overclocked at the max I can reach more than a couple of weeks of normal use, Then if it passes that test with no issues, I have achieved what I wanted, and the cpu goes back to normal settings.

I only overclock as a learning experience, With the way I use my computer and things I do, Overclocking the chip and leaving will offer no benefits.

 

[RGone]

I think your just not wanting to stress out that scrawny chv-f, chuck that thing in that chubby chick you got laying around and shovel a few more scoops of coal under it......

I toilet yu at yu mat be sorrowful-up. Faht Chick putted en. Now we c what happins. Uhready at 4.8Ghzskis. Nuttin 2 et with a pretty goot mobo.

I am thiniking about doing some sort of mini-test of the Asrock Fatal1ty 990FX mobo and maybe just put it in AMD Mobo's. Not sure yet.
RGone...ster.

 

[ch0b1ts]

Thanks for all the useful info. It helped me a great deal.

I wish I could get mine stable at your low voltages. I was P95 stable (12 hours) @4.5 ~1.39v with core temps in the low 50s but it took ~1.46v to get stable (14 hours) @4.6.
I don't think I'm going higher than 4.6Ghz with this chip since my temps at 4.6 are already mid to upper 50s in P95. BTW This was my 2nd 8320. The first one I excahnged after I couldn't get any sort of stable overclock above 4.2.

 

[RGone]

Yeah so many of these chips are just all over the map for Vcore and its' result of HEAT. I have quit hammering on my FX-8350 after I found out it appears better than a lot of them. I have been running 4.6 for a daily speed and I think I am going to drop that further to 4.2Ghz and then do my home video editting at 4.8Ghz since the quickness of editting at that particular speed is just awesome for me.

4.2, 4.4, 4.6 and 4.8Ghz seem to be on the very top of high-points in my previous testing; so 4.6Gh would likely be a very sweet spot indeed. Good luck man.
RGone...

 

[caddi daddi]

as we have found, pushing into the fives is just a waiste of time, these things work best in the low to mid 4's.
sure i push this cpu up to the 5's, but that's just to impress myself.
with the better, faster ram i have found that the 8120 now works best at 4.4 instead of the 4.2 i found with the 1600 ram.

 

[RGone]

with the better, faster ram i have found that the 8120 now works best at 4.4 instead of the 4.2 i found with the 1600 ram.

There are a number of parameters that have likely been overlooked when speccing up to run the FX processors. I expect that DDR1866 is the better speed of ram to run with an FX processor since that can be made to run up t about DDR2000 or so and is why my performance numbers have seemed to stay up at speeds that normally do not show much increased performance in benches.

I have run faster ram than C_D all along so his increase in workable cpu mhz are more noticeable to him than to myself. I did not ever run slow ram speed since I had fast ram. Such running caused ram speed to seem not so significant as he now does see.

Now I just happened to find another strange occurance when testing the Fatal1ty board. I did n0t want to risk my good Gskil ram at the outset of testing the board. So I put a 4gig kit of Gskil DDR1333 ram in and ran it at DDR1443 I think it was and began a few benches. Felt everything was in hand so I swapped in my DDR1866 running the same timings I had used and the same speed of the ram as I had run with the 4 gig kit of ram. Now the ONLY difference was a move from 4gig in two sticks to 8gigs in two sticks and my bench marks dr0pped dramatically. Then I increase the speed of the 8gig kit of ram to closer to its' DDR1866 ram speed and the benchmark results took off again. This leads me to believe that 8gigs of slow ram produces less performance by far than 4gigs of slow ram. I can only image that 16gigs of slow ram would cost performance of an even greater amount.

So there are many items that have probably been greatly overlooked when buying parts for even the BD processor. Fast ram likely would have shown the 4.3Ghz peak point and then a slowing and taper off; not to be at 4.3Ghz but at 4.4Ghz.

Now none of this is etched in stone for sure. And like so much of what we are seeing will show ebbs and flows. I doubt DDR2400 would be an end all be all since the break-thru point is likely greater than dead on DDR2400 and also the point of diminishing returns may have already been passed at DDR2100 to DDR2200 already. I think I wrote back a few months ago that any speed ram slower than DDR1866 was a waste and now am more sure of that than ever with the results we are seeing in testing today. Thanks for the heads-up C_D.
RGone...

 

[NiteSmoker17]

RGone do you happen to know the max safe die temp for the FX-83xx?

I think its 70c but wanted another opinion. i haven't seen AMD publish thermal data yet.

 

[Mandrake4565]

Nightsmoker, I haven't seen anything posted yet either. In general, 70 on the socket and 60 on the cores is about where we like to stay. I've had at mine at 65 on the cores while benching it and it starts becoming unstable. It's performance clock for clock drops a bit at that point. It definitely likes cooler temps.

Now I just happened to find another strange occurance when testing the Fatal1ty board. I did n0t want to risk my good Gskil ram at the outset of testing the board. So I put a 4gig kit of Gskil DDR1333 ram in and ran it at DDR1443 I think it was and began a few benches. Felt everything was in hand so I swapped in my DDR1866 running the same timings I had used and the same speed of the ram as I had run with the 4 gig kit of ram. Now the ONLY difference was a move from 4gig in two sticks to 8gigs in two sticks and my bench marks dr0pped dramatically. Then I increase the speed of the 8gig kit of ram to closer to its' DDR1866 ram speed and the benchmark results took off again. This leads me to believe that 8gigs of slow ram produces less performance by far than 4gigs of slow ram. I can only image that 16gigs of slow ram would cost performance of an even greater amount.

Is that because more ram puts more strain on the IMC or could it just be a difference in the sticks? I actually have a set of Corsair 1600 2x2 that run the same timings as my G Skills 1600 "2x4" maybe I'll give them both some runs and see the difference.

 

[ivanlabrie]

Might have to do with the sub timings too...set them manually to be the exact same if you want a fair comparison.

 

[Mandrake4565]

Might have to do with the sub timings too...set them manually to be the exact same if you want a fair comparison.

Now i'm in trouble I haven't started playing with those yet I'm just getting the hang of messing with the, for lack of better word 1st tier timings?

 

[caddi daddi]

don't worry mandrake it'll just no boot, I'm just finding that out now!!

 

[ivanlabrie]

Jedec profiles tend to have similar timings...check those out and try to use a generic 1333mhz cl7 one or something like that.

 

[Mandrake4565]

don't worry mandrake it'll just no boot, I'm just finding that out now!!

Glad we're on the same page Caddi.

Jedec profiles tend to have similar timings...check those out and try to use a generic 1333mhz cl7 one or something like that.

will do Ivanlabrie thanks

 

[RGone]

Going to try my FX-8530 on an Asrock Fatal1ty 990FX Professional mobo and see if all is the same or pretty close to the same as on my CHV.
RGone...