Observations with an FX-8350

[grimpr]

AMDs stock is good but man, at 4.5ghz and running 5 rounds of LinX, it would sound like a jet airplane. Still its amazing that it gets the job done.

 

[grimpr]

Too bad AMD didn't put more into the CPU and less into the stock HSF!

Why? Its a very good cpu for the price and at 4.4-4.6ghz is a very strong multitasker and performer. Did you try running Linux on it? it flies on the FX. Try phoronix test suit on it, its much fun.

http://www.phoronix-test-suite.com/
http://openbenchmarking.org/

 

[itsmydamnation]

The extrapolation to 4.5GHz reads out at 1.49V, and 1.68V for 5GHz D: Are those considered silly stupid voltage for a 32nm AMD chip, or is that "safe and reasonable"?

(by comparison I had no qualms over shoveling 1.5V into my 32nm 2600K and running it at 93C with LinX...perhaps arguably "silly stupid" of me, but that is my threshold as a baseline in answering my question above)

might be your chip but i can get mine IBT for 30mins stable @ 1.275 @ 4600. I find setting Vdroop to high level's has a big impact on clock stability. I can easily go higher then 4600 but i lack the cooling to remove heat and i can see throttling in IBT but results remain accurate.

There is about a 30-35 degree temp differential where i live, summer max is generally low 40C winter max is around 8-10. i'll be able to ramp it up heaps more in winter so i will see how it goes then. In winter i can clock my 920-C0 at 4.0ghz in summer its 3.4 at best D: .

 

[CTho9305]

I know I asked earlier, but I cannot seem to find the response (thread is unwieldly) but what is the safe voltage for these pilderiver chips? We know AMD knows but isn't telling, but what is the general concensus?

Is 1.5V safe? What about 1.6V?

To completely ignore your actual question and post something only tangentially related because I think you'll find it interesting, have you seen this article (based on this paper, if you have access to it)?

 

[Idontcare]

might be your chip but i can get mine IBT for 30mins stable @ 1.275 @ 4600.

Is that with the newest (patched) version of LinX for Piledriver?

That is crazy low Vcore for the clockspeed. At stock (4GHz) mine needs 1.319V to be LinX stable.

You must have a golden sample :thumbsup:

To completely ignore your actual question and post something only tangentially related because I think you'll find it interesting, have you seen this article (based on this paper, if you have access to it)?

That is really cool, in many ways. Cool that AMD approved the release of that engineering data into the public domain, and the data itself is cool

From that article (I don't have access to the paper) can you tell me what the difference is between "worst case" and "best case"?

Are they showing the extremes of different chips, the bounds of the binning distribution that was coming from the wafers? Or is all the data for the same chip but running different application? Or is it a difference between a hot chip and a cold chip?

 

[itsmydamnation]

Is that with the newest (patched) version of LinX for Piledriver?

That is crazy low Vcore for the clockspeed. At stock (4GHz) mine needs 1.319V to be LinX stable.

You must have a golden sample :thumbsup:

i dont think it is, i just downloaded the lastest version of IBT. as i said the the high load line callibration seems to make all the difference. Also im running ESXI so even @ 100% im unlikely to be hitting the chip quite as hard as you. That said ESXI is pretty low CPU overload, its I/O where it gets bogged down.

 

[Idontcare]

i dont think it is, i just downloaded the lastest version of IBT. as i said the the high load line callibration seems to make all the difference. Also im running ESXI so even @ 100% im unlikely to be hitting the chip quite as hard as you. That said ESXI is pretty low CPU overload, its I/O where it gets bogged down.

I have my LLC set to be basically no Vdroop as well, and I confirmed it with an external voltmeter.

My mobo is the Crosshair Formula-V, I can't find a better board for delivering a clean and stable power line to the CPU.

And are you sure you are using the latest LinX patch? Latest version of LinX (11.0.1.005) can be downloaded with this link. (taken from here on XS)

When you run it you have to be sure and use a lot of memory, I use problem size 43122 which equates to ~14 GB. If you use less memory in the test then your system is taxed considerably less. LinX stable w/1GB mem setting is not the same as being LinX stable w/14GB in testing.

I am just really perplexed by your result. You are saying your system is stable at 4.6GHz using a voltage that I can't even use at stock 4Ghz clocks (which FWIW neither can guskline, his 8350 requires 1.45V for 4.5GHz operation, inline with mine)

Voltage conditioning by the mobo isn't going to make up for a whopping 0.2V difference like what you seeing. I've been doing this for a while and even on crappy DS3L mobos using Q6600's the Vdroop that LLC was countering was never 0.2V.

You either have a golden sample or you aren't challenging the CPU to the same extent that guskline or myself are doing when we are looking for minimum Vcc needed for stability at a given GHz IMO. I hope you have a golden sample, you could probably go above 5GHz with that thing and still keep the voltage under 1.5V.

 

[Idontcare]

Yeah I have an H100 but first I wanted to flesh out the OC capabilities of the processor at stock (without the aid of a $100 3rd party cooler ).

AMD's stock HSF is heads-and-shoulders above Intel's stock HSF and the max OC's with stock show that. I can't take my 3770K above 4.2GHz with the stock HSF (it starts thermal throttling), likewise I can't take my 2600K above 4GHz with its stock HSF.

But I have been running benchmarks at 4.5GHz with the piledriver on a stock HSF. Granted I didn't test for LinX stability, that is coming up, but still, pretty good for a HSF that comes with the CPU.

Well my efforts to run 4.5GHz with LinX were a total bust

Using the stock HSF and dialing in the voltage to 1.45V, temperatures rocketed to 83°C and power consumption climbed to 418W D:

And that was after just 3 minutes

So I think I might be done with fiddling around with the stock setup. I may try and see what kind of voltage/temps/power are needed for a mild OC of say 4.2GHz.

 

[CTho9305]

That is really cool, in many ways. Cool that AMD approved the release of that engineering data into the public domain, and the data itself is cool

From that article (I don't have access to the paper) can you tell me what the difference is between "worst case" and "best case"?

Are they showing the extremes of different chips, the bounds of the binning distribution that was coming from the wafers? Or is all the data for the same chip but running different application? Or is it a difference between a hot chip and a cold chip?

That particular graph was created by the author of the page I linked to, and is not from AMD (the way the charts are presented in the page make it hard to tell what AMD provided and what he created...everything is stock Excel stuff so it looks the same).

He just used the extreme ends of the lines from Figure 2 (note "Source" versus "Data Source" in the captions), labeled them "Best Case" and "Worst Case", and used the resulting two lines to fit a power model he created. AMD did not explain how the minimum and maximum voltage to show at each frequency were chosen (or what they mean), so I can't answer your question. Sorry.

 

[itsmydamnation]

My mobo is the Crosshair Formula-V, I can't find a better board for delivering a clean and stable power line to the CPU.

im using a GB-990FXA-UD3, not the best but shouldn't be bad, 2 phase power. I do tax my system quite hard, it does lots of 1080P transcoding for realtime streaming, it would have done this for maybe 100 hours since i have had it and no stability issues.

I am just really perplexed by your result. You are saying your system is stable at 4.6GHz using a voltage that I can't even use at stock 4Ghz clocks (which FWIW neither can guskline, his 8350 requires 1.45V for 4.5GHz operation, inline with mine)

i cant see my voltage in cpu-z because that info isn't passed to guests. but in BIOS it is hard set by me to 1.275. my chips stock was 1.300. I have no NB OC and ram isn't running that hard (1600mhz) with quite relaxed timings because i have 4 8gig dimms. I have a HTT of 230mhz and a multi of 20.

I haven't reached an OC limit on air, i just stopped once i started to get throttling.

 

[AnandThenMan]

Using the stock HSF and dialing in the voltage to 1.45V, temperatures rocketed to 83°C and power consumption climbed to 418W D:

Are you intentionally trying to kill the processor?

 

[Greenlepricon]

i cant see my voltage in cpu-z because that info isn't passed to guests. but in BIOS it is hard set by me to 1.275. my chips stock was 1.300. I have no NB OC and ram isn't running that hard (1600mhz) with quite relaxed timings because i have 4 8gig dimms. I have a HTT of 230mhz and a multi of 20.

I haven't reached an OC limit on air, i just stopped once i started to get throttling.

That's really low in my opinion too. My 8320 takes about 1.35v or something close to run at 4GHz and needs cranked up quite a bit for anything higher (which I hate to do). To be fair I have only cranked up the multiplier though. Still you got something amazing there if it was that simple for you.

 

[Gikaseixas]

Mine is running at 4.7ghz with 1.48v, could go higher but i think it's plenty fast as it is right now. I experimented a old Zalman 9700 cooler and maximum i could go was 4.6ghz at this same voltage, anythng higher it would throtle.
With the H100i i was able to get 4.9ghz with 1.5v stable but dialed down a bit for 24/7 use

 

[Vesku]

I'm not completely on board with their conclusion, but this does show how Piledriver isn't bad if you don't freak over powerdraw. ;p It brings up live streaming of games which seems to have gotten somewhat popular. I'm personally not into it but to each their own.

http://www.youtube.com/watch?v=eu8Sekdb-IE

The test systems - stock results only:

http://teksyndicate.com/videos/amd-...s-3820-gaming-and-xsplit-streaming-benchmarks

 

[inf64]

I'm not completely on board with their conclusion, but this does show how Piledriver isn't bad if you don't freak over powerdraw. ;p It brings up live streaming of games which seems to have gotten somewhat popular. I'm personally not into it but to each their own.

http://www.youtube.com/watch?v=eu8Sekdb-IE

The test systems - stock results only:

http://teksyndicate.com/videos/amd-...s-3820-gaming-and-xsplit-streaming-benchmarks

Well that was surprising result to say the least. In streaming it's quite shocking how it demolishes the 3570K. It's probably all the extra units 8350 has that helps it make such a big performance leap over the non-HT IB in their test scenario(24 or 32 if you count all FP EX units in "hybrid" 8 core 8350 Vs 12 EX units in SB/IB).

 

[AtenRa]

With CPU and CPU/NB LLC disable and voltages on Auto, I get max power usage in Linx at 227W with Turbo enable and CPU temps of up to 52c (Ambient at 22-23c).

Vcore fluctuated from 1.128v up to 1.413v

BUT, I get an error after 9+ minutes.

With the same settings, x264 HD 5.01 benchmark finished without any problems using up to 217W.

Now, OCed to 4.4GHz and set the voltage to 1.4125v, Linx would get an error BUT, x264 HD 5.01 benchmark show a big boost. Power consumption was up to 240W and CPU temp up to 59c (Ambient at 22-23c).

 

[inf64]

Thanks for the results AtenRa. I can't see your 4.4Ghz screenshot though (or I'm blind )

 

[AtenRa]

Ahh yeap, have a look again, last pic

 

[inf64]

Yep I see it now . Nice boost there and power draw is in line with perf. jump : 10% higher performance with 10% higher power draw.

 

[grimpr]

I'm not completely on board with their conclusion, but this does show how Piledriver isn't bad if you don't freak over powerdraw. ;p It brings up live streaming of games which seems to have gotten somewhat popular. I'm personally not into it but to each their own.

http://www.youtube.com/watch?v=eu8Sekdb-IE

The test systems - stock results only:

http://teksyndicate.com/videos/amd-...s-3820-gaming-and-xsplit-streaming-benchmarks

Good find, it clearly shows that the FX is a multitasking monster and shines where heavily pushed

 

[guskline]

grimpr: I agree. Interesting video. I've seen this reviewer before and he is quite interesting. I guess the FX 8350 is not that "bad" after all! I have nearly the same video cards in all three rigs and the 8350 plays games as smoothly as the 2500k rigs.

 

[Face2Face]

It's Logan. The old TigerDirect Reviewer. He was doing this well before the Newegg (Paul) and (Linus) NCIX guy. He is funny as crap...

 

[grimpr]

grimpr: I agree. Interesting video. I've seen this reviewer before and he is quite interesting. I guess the FX 8350 is not that "bad" after all! I have nearly the same video cards in all three rigs and the 8350 plays games as smoothly as the 2500k rigs.

In the video review he benchmarks with xsplit live streaming, a perfect multitasking scenario for the nature of the chip, xsplit uses x264 for encoding the stream and it clearly shows that the 8 integer cores of the FX carry more workload and higher fps than Intels quads and ht.

 

[SPBHM]

In the video review he benchmarks with xsplit live streaming, a perfect multitasking scenario for the nature of the chip, xsplit uses x264 for encoding the stream and it clearly shows that the 8 integer cores of the FX carry more workload and higher fps than Intels quads and ht.

it sounds like the perfect use for quick sync

 

[grimpr]

it sounds like the perfect use for quick sync

Sure why not, just convince xsplit developers to use it, piece of cake.