FX-8350 Benchmarks and review

Is it possible it's not reading the C-states correctly and it's running at a low multiplier/low voltage but the clock speed is being detected wrong and should read 800 mhz or whatever the lowest C-state is for the FX-8350?

That's what I'm thinking because that's pretty low voltage at that frequency.
 
The price is what's important anyway. $175 for the 8320? sub 200 for the 8350. That's low enough to make me consider upgrading my work machine.
 
Not anywhere near what it used to be. 8320/8350 is going to be at least $200 at launch, plus $150 for a decently nice 990FX. Less then $50 more would get a 3570K and Z77.

AMD's price lead was really in the lower end, before the Pentium and I3 lines got so great. I change my rig at least every 2-4 weeks just for the fun of it. I always try to get an AMD rig, but lately Intel combos have been cheaper, same price, or slightly more for much better performance. I have a MicroCetner I go to, and the AMD combo deals just arent as good as they were about a year ago, or more.

You forget the most important thing. People who are anticipating for piledriver already have a am3 + motherboard. So its wasting 400$ for no difference in performance instead of $200 direct drop.
 
I heard rumors that this chip was suppose to be released today, doesn't look like that will happen.

As per NDA, I can only say that it wasn't going to be released today, the rumor was based from a Teleconference on Vishera which happened this morning. I've got a pretty good idea when it'll come out so all I'll say is It'll probably come out within a week, so don't worry :D
 
As per NDA, I can only say that it wasn't going to be released today, the rumor was based from a Teleconference on Vishera which happened this morning. I've got a pretty good idea when it'll come out so all I'll say is It'll probably come out within a week, so don't worry

sigh....

its oct 23rd/24th. I dunno what this guy is saying. If he was under a NDA, then his last statement would not be legal.....
 
You forget the most important thing. People who are anticipating for piledriver already have a am3 + motherboard. So its wasting 400$ for no difference in performance instead of $200 direct drop.

Exactly. I've got two AM3+ mobo-based systems. One has a 1090T which I'm very happy with and won't look to replace. The other has a 720BE which is fairly begging to be replaced with a 6 or 8 core CPU. For the price of a bare-bones CPU (no cooler needed), I can upgrade that system. No memory $, no mobo $, no cooler $. No OS hiccup. It should take all of 15 minutes, including 10 minutes of coffee sipping.
 
What i want to know is the regular voltage that this CPU uses with turbo core on... my FX8150 with Turbo off load is at 1.3v and Turbo on is at 1.4... what about the FX8350?
 
What i want to know is the regular voltage that this CPU uses with turbo core on... my FX8150 with Turbo off load is at 1.3v and Turbo on is at 1.4... what about the FX8350?

Based on what we saw with Trinity, the voltages are likely very similar.
 
So, looks like even when overclocked it still gets beaten in single threaded applications by my Phenom II [email protected] Ghz, I got rid of a year ago.

5856232331_baebb4b725_o.jpg


And the FX-8350 just barely beats my old rig with its 8 overclocked cores, vs. my 6 cores at 4.1Ghz in mulithreaded tests...

5856787474_5462101979_o.jpg


Way to go AMD! :rolleyes:
 
The price is what's important anyway. $175 for the 8320? sub 200 for the 8350. That's low enough to make me consider upgrading my work machine.


they are getting better, but it's a shame the single threaded aka gaming performance on these sucks so bad.

if games start to get more multi-threaded, i think they might start to look good, but not yet.

in heavy threaded non gaming stuff, these can be pretty competitive with intel.
 
Zarathustra[H];1039242041 said:
So, looks like even when overclocked it still gets beaten in single threaded applications by my Phenom II [email protected] Ghz, I got rid of a year ago.

Phenom_II_1090_T.jpg


And the FX-8350 just barely beats my old rig with its 8 overclocked cores, vs. my 6 cores at 4.1Ghz in mulithreaded tests...

Way to go AMD! :rolleyes:

Mine actually does better and can clock higher than that. Depends on the board, ram, ram settings etc...
This is just some old tests i have done. It could be that i have done even better but they are not saved.
 
Zarathustra[H];1039242041 said:
So, looks like even when overclocked it still gets beaten in single threaded applications by my Phenom II [email protected] Ghz, I got rid of a year ago.

5856232331_baebb4b725_o.jpg


And the FX-8350 just barely beats my old rig with its 8 overclocked cores, vs. my 6 cores at 4.1Ghz in mulithreaded tests...

5856787474_5462101979_o.jpg


Way to go AMD! :rolleyes:

Herp derp my max overclocked phenom slightly beats a stock fx8350. They are a bunch of failures. Take your overpriced 3960x over to the intel thread.:rolleyes:
 
As per NDA, I can only say that it wasn't going to be released today, the rumor was based from a Teleconference on Vishera which happened this morning. I've got a pretty good idea when it'll come out so all I'll say is It'll probably come out within a week, so don't worry :D

Those of us who really have an NDA have known when the EXACT date the embargo would lift, and for quite some time.
 
Herp derp my max overclocked phenom slightly beats a stock fx8350. They are a bunch of failures. Take your overpriced 3960x over to the intel thread.:rolleyes:

I was comparing to overclocked figures. Maybe you should read before throwing a hissy fit.

I agree, my 3930k was more CPU than I needed (I should have picked up a 2600k or 2500k, don't know what I was thinking) but that doesn't make the fact that AMD's brand new flag ship CPU is unable to beat their own product they launched 2.5 years ago in per thread performance, and only barely able to beat it in multithreaded loads with two extra cores any less of a fail.
 
So, looks like even when overclocked it still gets beaten in single threaded applications by my Phenom II [email protected] Ghz, I got rid of a year ago.

the 81XX bulldozers beat thuban more often then not in most benchs. Pile-driver will just extend the lead....

for what its worth, my fx at 4.9 beats your single thread score, and your muti-thread score....
 
Last edited:
Zarathustra[H];1039243136 said:
I was comparing to overclocked figures. Maybe you should read before throwing a hissy fit.

I agree, my 3930k was more CPU than I needed (I should have picked up a 2600k or 2500k, don't know what I was thinking) but that doesn't make the fact that AMD's brand new flag ship CPU is unable to beat their own product they launched 2.5 years ago in per thread performance, and only barely able to beat it in multithreaded loads with two extra cores any less of a fail.

I needed to make the word stock large as you cant read. The 8150 beats phenom 2 processors max oc to max oc. The cost associated with a little performance boost going to the 8150 couldnt be justified from a six core and some cases a 4 core. Troll somewhere else.
 
I needed to make the word stock large as you cant read. The 8150 beats phenom 2 processors max oc to max oc. The cost associated with a little performance boost going to the 8150 couldnt be justified from a six core and some cases a 4 core. Troll somewhere else.

Implying a dedicated member of the community comparing legitimate numbers is a fanboy troll... Please be constructive or go hide under a bridge.
 
Implying a dedicated member of the community comparing legitimate numbers is a fanboy troll... Please be constructive or go hide under a bridge.

Why doesn't he compare overclock for overclock? A 8350 at near 5ghz scores almost 30% faster than his 1090t in multithreaded cinebench. http://www.techhum.com/amd-showcase-fx-8350-at-5ghz-near-idf/That would be the most fair comparison. But when you have an agenda you skew the results to whatever your viewpoint is.
 
Why doesn't he compare overclock for overclock? A 8350 at near 5ghz scores almost 30% faster than his 1090t in multithreaded cinebench. http://www.techhum.com/amd-showcase-fx-8350-at-5ghz-near-idf/That would be the most fair comparison. But when you have an agenda you skew the results to whatever your viewpoint is.

OP said he hit 4.5-4.6Ghz but not stable. I presumed this was approximately max overclock on typical enthusiast cooling based on that. If I was wrong that was my mistake.
 
OP said he hit 4.5-4.6Ghz but not stable. I presumed this was approximately max overclock on typical enthusiast cooling based on that. If I was wrong that was my mistake.


LOL what you smoking? some of the good Bulldozer chips clock to 5.2ghz under wicked cooling. Piledriver should be able to achieve the same results if not better, if AMD is right by saying power consumption has been reduced by 15%. Have to figure the orginal poster is on a ASRock board also, not exactly a overclockers board.
 
LOL what you smoking? some of the good Bulldozer chips clock to 5.2ghz under wicked cooling. Piledriver should be able to achieve the same results if not better, if AMD is right by saying power consumption has been reduced by 15%. Have to figure the orginal poster is on a ASRock board also, not exactly a overclockers board.

Not to mention who knows how stable the bios is with this new chip. Maybe it has stability problems when overclocked until they iron out bios issues.
 
OP said he got it to 4.5 stable on stock voltages. It's on page 4 I believe. Considering how OP also seems to be a beginner at overclocking, I wouldn't be surprised if he didn't have the optimal settings dialed in.
 
OP said he got it to 4.5 stable on stock voltages. It's on page 4 I believe. Considering how OP also seems to be a beginner at overclocking, I wouldn't be surprised if he didn't have the optimal settings dialed in.

Nah I was just going to fast then derped when I should have herped
 
LOL what you smoking? some of the good Bulldozer chips clock to 5.2ghz under wicked cooling. Piledriver should be able to achieve the same results if not better, if AMD is right by saying power consumption has been reduced by 15%. Have to figure the orginal poster is on a ASRock board also, not exactly a overclockers board.

Based on:

1.) Kyles review of Bulldozer at launch where he was able to hit 4.5ghz stable, and 4.6 was never quite stable.
2.) OP's statements that he was having trouble with stability at 4.5Ghz w. 4.6Ghz turbo

Guess I haven't been following others developments since then, as this is the first I've heard of users with non-extreme cooling hitting speeds above 4.6Ghz.

Not really interested in results from custom water loops, as these are not reflective of typical use.

With high end air coolers and corsair-style sealed water coolers, what is a typical overclock on a Bulldozer these days? I'm not interested in top overclocks like "this dude I know got a great chip and hit 5Ghz" kind of stuff. I'm interested in typical, which pretty much every chip you pick off the shelf will be able to hit?

Then it's a better comparison to my 1090T, as pretty much all of those were able to hit 4.0-4.1Ghz with high end air, or good sealed water cooling.
 
from what i have seen most 8120/8150's will clock between 4.6 and 5ghz. With high end air or sealed liquid units such as the antec 920, corsair h80, corsair h100 The smaller ones don't yield the best results.

My chip will hit 5.2ghz on a cold day with the window open. With the Antec 920. I could probably run this speed if I had a full custom loop to keep things nice a cool. With out having to aid my antec 920. I daily run at 4.9ghz

Bulldozer is a tricky beast though. Its very picky on what settings it likes and doesn't like. Most novice overclockers have a hard time hitting 4.6ghz+. With bulldozer its not just the vcore and multiplier you raise for the higher clocks. I have to raise the northbridge voltage and pill volts, to make things stable. I also have to relax the memory timings and speed. Ie stock speed i can run my ram a 2133 with 9-9-9-24 1t timing. At 4.9ghz I have to lower to 2000mhz, 9,9,9,27 1t ram timing. Not a huge difference but is the difference between a stable system and unstable one.

Head over to the Bulldozers owners club thread on OCN. http://www.overclock.net/t/1139726/amd-fx-bulldozer-owners-club

The first page has the CpuZ validations you can browse though which will give you an idea of what people are getting overclock wise on the Bulldozer.

Piledriver should improve on this front, how much is not known.
 
Bulldozer is a tricky beast though. Its very picky on what settings it likes and doesn't like. Most novice overclockers have a hard time hitting 4.6ghz+. With bulldozer its not just the vcore and multiplier you raise for the higher clocks. I have to raise the northbridge voltage and pill volts, to make things stable.


Yeah, I noticed that the hard way. I have a FX-8120 in my server running at stock clocks (it's a server, I just want stability). I plopped 32GB (4x8) of 1866 RAM in there, and it turns out that at SPD settings, the system hard locks at post with all slots populated. The Automatic BIOS recovery didn't work, so I actually had to take two of the sticks out, boot up, lower the settings and put the remaining two back in again.

Never had that level of finickiness with any system I've used/worked on before.
 
There's one other aspect you miss:

If you pay extra for the Intel quad core now, you won't be as tempted to upgrade a year or two down the line. I can't tell you how many people on this board have upgraded from Phenom II x4 to Phenom II x6 over the last couple years, and some have gone from that over to Bulldozer (and bought a new 990FX board to facilitate this). That's TWO upgrades inside of two years for some of these people (and for the many that did not bite on Bulldozer, they are STILL looking for a viable AMD upgrade path!)

You make a valid point. I would point out, however, that the period we've just been through for AMD processors has been essentially the worst-case scenario: marginal to no improvement between Phenom II X6 and Bulldozer. This is not the norm. Historically, AMD's chips have represented significant performance improvement over the previous one. I am making a speculation, but I would venture to guess that the performance improvement between Vishera (FX-8350) and the next generation Steamroller will be very significant. If a user can buy an AM3+ motherboard and an FX-8350 over the next few weeks, and then drop in a Steamroller chip into the same motherboard with the same ram, you may still be saving money while getting comparable or possibly better performance than Intel. If there's even a reasonable chance of this, then I'm going to grab an AM3+ mobo and either hold on to my Phenom II or spring for a Piledriver chip and overclock the hell out of it with my new water cooler. Holding on to the Phenom II and upgrading direct to Steamroller would probably provide a very dramatic performance improvement, but I'll probably break down and get a Piledriver long before Steamroller hits the shelves.

Frankly, the thought of giving Intel any of my money churns my stomach. I understand the performance argument - I really do. I love speed as much as anyone. However, AMD will probably either have to go completely bankrupt or get out of the performance x86 processor business before Intel will be able to pry a red cent out of my hands. I simply can't bring myself to give that company any money after what they did to AMD and by extension, to performance enthusiasts, by trying to threaten/blackmail major companies into avoiding AMD. Mafia tactics like those don't earn my respect or my dollars, and no, I haven't forgotten just because it happened a few years ago.
 
Last edited:
Historically, AMD's chips have represented significant performance improvement over the previous one.

Do you mind giving an example. I think ever since sledge hammer there hasn't been a significant performance improvement. Unless you count multi-core, but still...
 
They both have equal amounts of "bullshit" when purchasing, but AMD systems have a much longer upgrade lifespan. Your AM2+ CPU is, if you wanted, able to be placed in the same system as a Piledriver. That provides you the option to begin building a new system but not need to fork out $500 right off the bat. Alternative, and I won't deny it as an AMD fan myself, in doing so you are continuing to utilize hardware or buses that in the end may actually be hampering performance. This is a bad example and most likely quite unrelated, but to me it seems plausible to some degree: Look at the FM1 APUs, based on the Phenom II but did away with the Northbridge, and now you've got some pretty incredible memory performance on them after a bit of tweaking (I realize that AMD hasn't had the memory controller on the NB since A64). I'm curious what Piledriver on FM2 may be able to achieve with twiddling.

Intel had dropped 1156 for 1155 (which I'm not disregarding that AMD drop FM1 right away) and then changed how everything was connected with overclocking, limiting people quite a bit. Then they decided "oh alright, we goofed and should rework that a bit", so IIRC anyone with an 1155 who wanted the improved overclock options had to buy a new motherboard again. Not to mention if you wanted to go with Intel's crème de la crème you had to buy an insanely expensive motherboard along with an insanely expensive CPU (both 1366 and 2011). Yea you get some awesome performance, I'm not an AMD fanboy that is blind to that fact, but you sure have to pay for it that's for sure :(

At the end of the day, if gaming is your primary focus and you have budgeted your system out well enough to be able to purchase a really nice GPU, your choice in CPU isn't going to matter much. Again there are exceptions and tradeoffs, but when isn't there? :) Yet, f you have no budget... then you really aren't needing to worry about all these sorts of things are you? lol :p

Llano FM1 socket lasted 1 year :( BAD AMD!
 
the 81XX bulldozers beat thuban more often then not in most benchs. Pile-driver will just extend the lead....

for what its worth, my fx at 4.9 beats your single thread score, and your muti-thread score....

You have proof of that? Cinebench Screenshots or it didn't happen :)
 
Do you mind giving an example. I think ever since sledge hammer there hasn't been a significant performance improvement. Unless you count multi-core, but still...

Examples? Well, I'm not going to go back to the Intel reverse engineered period (286, 386, 486), but since about the K5 era, onwards, there's been noticeable improvement in performance from one generation to the next.

Athlon XP to Athlon 64. Athlon 64 to A64 X2, Athlon 64 X2 to Phenom I (B3 stepping - not the first B1 stepping), Phenom I (B3 stepping) to Phenom II. In all of these cases, the successor processor provided at least a 15-20% improvement in computational power (or more). Especially now with Jim Keller at the helm of x86 CPU design at AMD, I cannot believe the Steamroller chip will not exceed the performance of the Piledriver CPU by at least a similar, historical 15-20% in many to most metrics. He'll see to it the main components are hand optimized (unless he's prevented from doing so by management, in which case, all bets are off).
 
You have proof of that? Cinebench Screenshots or it didn't happen

do you have proof of your birth? Pictures or it didn't happen?

For the record I am not going to go out of my way to please some troll.
 
Examples? Well, I'm not going to go back to the Intel reverse engineered period (286, 386, 486), but since about the K5 era, onwards, there's been noticeable improvement in performance from one generation to the next.

Athlon XP to Athlon 64. Athlon 64 to A64 X2, Athlon 64 X2 to Phenom I (B3 stepping - not the first B1 stepping), Phenom I (B3 stepping) to Phenom II. In all of these cases, the successor processor provided at least a 15-20% improvement in computational power (or more). Especially now with Jim Keller at the helm of x86 CPU design at AMD, I cannot believe the Steamroller chip will not exceed the performance of the Piledriver CPU by at least a similar, historical 15-20% in many to most metrics. He'll see to it the main components are hand optimized (unless he's prevented from doing so by management, in which case, all bets are off).

I'm more familiar with the zero to performance "decrease". Like when the A64 went DDR2 or more recently the BD fiasco or trinity vs LLano.

And 15-20% performance increase is not enough to make AMD competitive. Nothing AMD has can even meet SB.
 
I'm more familiar with the zero to performance "decrease". Like when the A64 went DDR2 or more recently the BD fiasco or trinity vs LLano.

And 15-20% performance increase is not enough to make AMD competitive. Nothing AMD has can even meet SB.

This is totally weird I thought I had seen some benchmarks on linux and windows side which do beat Intel.
 
Frankly, the thought of giving Intel any of my money churns my stomach. I understand the performance argument - I really do. I love speed as much as anyone. However, AMD will probably either have to go completely bankrupt or get out of the performance x86 processor business before Intel will be able to pry a red cent out of my hands. I simply can't bring myself to give that company any money after what they did to AMD and by extension, to performance enthusiasts, by trying to threaten/blackmail major companies into avoiding AMD. Mafia tactics like those don't earn my respect or my dollars, and no, I haven't forgotten just because it happened a few years ago.

+1

Once you start down the dark path, forever will it dominate your destiny. Hard to see, the dark side is. :D
 
Back
Top