AMD FX 8350 Winning against i5/i7?

[Idontcare]

Yep, not sure if its in this vid, but in another one he mentions Intel compiled benchmarks and how they have flags to detect AMD CPUs and not run optimally. I knew right then that he had something more than these gaming demos that he wanted to get across.

I've never really understood the goal of making this excuse when people put in the effort to apologize for AMD's performance lackings compared to Intel.

Its not like the performance gap evaporates just because you can concoct a good excuse for the gap's existence. The gap is still there, regardless the reasoning why.

Having a plausible excuse to explain why it is ok for something one has in hand to be worse performing than something they do not have in hand is purely a psychological factor at that point.

The theory of cognitive dissonance in social psychology proposes that people have a motivational drive to reduce dissonance by altering existing cognitions, adding new ones to create a consistent belief system, or alternatively by reducing the importance of any one of the dissonant elements.

"My cpu sucks ballz but I know precisely why it sucks ballz so its all good"...uh, no, your cpu still sucks ballz, it is not all good. What is going on there is cognitive dissonance.

Cognitive dissonance theory explains human behavior by positing that people have a bias to seek consonance between their expectations and reality. According to Festinger, people engage in a process he termed "dissonance reduction", which can be achieved in one of three ways: lowering the importance of one of the discordant factors, adding consonant elements, or changing one of the dissonant factors.

"AMD is better, it MUST be better, but benchmarks don't support what I believe to be true (there is a disconnect between reality and my expectations), I will therefore lower the importance of benchmarks and substitute reality with an excuse for why AMD appears to be inferior"

:\

 

[Abwx]

I know AMD made some respectable improvements when going to Piledriver from Bulldozer, but some of these results look like too much for those improvements to even begin to cover that gap, even if we add 11% to the scores to make up the 3.6GHz -> 4GHz clock rate and then another 5-10% on top of that for architecture improvements I just don't see how some of these deltas are impacted at all (of which we'd also have to adjust for the fact that the latest intel CPUs in these tests are Sandy, not Ivy):


Here we begin to see where I think the work Teksyndicate has done might be seriously flawed, as these following results tend to agree with their findings in that AMD actually does surprisingly well in some games such as BF3 and FC3

Someone might buy an 8350 and be happy when playing Crysis 1 and 2, FarCry 3, and Battlefield 3, but as soon as they try out Planetside 2 or Crysis 3, their decision to go with AMD suddenly isn't looking so good.

You are still displaying the 8150 as exemple to sustain your point.

Architectural improvements yield 13.5% for games.

A short comparison at same frequency , green charts are
for games , the term "moyenne jeux" means "average for games".

http://www.hardware.fr/articles/880-6/bulldozer-vs-piledriver-4-ghz.html

 

[inf64]

18.1% faster than 8150(bot at stock) according hardware.fr average game fps chart . That's substantial increase in performance given the fact that 8150 usually runs at 3.9Ghz in many workloads(especially games).

I wonder what will happen when Kaveri launches. IF it brings big gains in single threaded/poorly threaded workloads, like AMD is claiming, it will practically become an ultimate desktop CPU. Why? Well for one thing a 2M Kaveri will probably outperform FX8350 by significant margin in games and will come close to 3570K. So scratch that one from the list of "bad" things in current FX. Second, power draw will probably be improved ,especially in pure CPU segment(whole APU will still consume similar amount of power tho,courtesy of very fast GPU). Third, Kaveri will support HSA and all the workloads in future that will have HSA implemented will be running much faster on Kaveri than on other non-HSA devices. Programming model is very easy and is similar to C++. Also AMD is giving away for free all the tools devs need to make their apps accelerated by HSA hardware(it runs seamlessly on CPU or GPU ,selection of code path is done in real runtime).

Will we be seeing the same comments when Kaveri is tested this way? I bet we will,maybe even worse since the better AMD does the attacking horde is more mad .

 

[frozentundra123456]

I've never really understood the goal of making this excuse when people put in the effort to apologize for AMD's performance lackings compared to Intel.

Its not like the performance gap evaporates just because you can concoct a good excuse for the gap's existence. The gap is still there, regardless the reasoning why.

Having a plausible excuse to explain why it is ok for something one has in hand to be worse performing than something they do not have in hand is purely a psychological factor at that point.



"My cpu sucks ballz but I know precisely why it sucks ballz so its all good"...uh, no, your cpu still sucks ballz, it is not all good. What is going on there is cognitive dissonance.



"AMD is better, it MUST be better, but benchmarks don't support what I believe to be true (there is a disconnect between reality and my expectations), I will therefore lower the importance of benchmarks and substitute reality with an excuse for why AMD appears to be inferior"

:\

Well said, especially your last paragraph. I might add that they also just ignore the benchmarks and say it "feels" faster, or project some future product/instruction set will be the magic bullet that makes AMD superior in every benchmark.

What I dont understand is why so many posters on both sides seem so vested in a particular company. Its not like either one really gives a rats *** about us except as someone to buy their product and make them money.

 

[Ferzerp]

What I dont understand is why so many posters on both sides seem so vested in a particular company. Its not like either one really gives a rats *** about us except as someone to buy their product and make them money.

http://arstechnica.com/science/2011...m-of-favorite-brands-as-threat-to-self-image/

It literally evokes the same response as their own personal failures. Some people resond to failure differently. What you see on message boards are the people who deny failure (e.g. No, ignore the preponderance of evidence and watch my youtube video that agrees with me instead), and/or blame their failures on others (e.g. It's the compiler's fault on the benchmark! It purposefully hobbles AMD).

 

[2is]

At stock AMD would actually have a slight advantage, I believe.

He reviewed them at stock.

But who leaves these processors at stock?

Not according to every other reputable review site including here at AT.

 

[piesquared]

18.1% faster than 8150(bot at stock) according hardware.fr average game fps chart . That's substantial increase in performance given the fact that 8150 usually runs at 3.9Ghz in many workloads(especially games).

I wonder what will happen when Kaveri launches. IF it brings big gains in single threaded/poorly threaded workloads, like AMD is claiming, it will practically become an ultimate desktop CPU. Why? Well for one thing a 2M Kaveri will probably outperform FX8350 by significant margin in games and will come close to 3570K. So scratch that one from the list of "bad" things in current FX. Second, power draw will probably be improved ,especially in pure CPU segment(whole APU will still consume similar amount of power tho,courtesy of very fast GPU). Third, Kaveri will support HSA and all the workloads in future that will have HSA implemented will be running much faster on Kaveri than on other non-HSA devices. Programming model is very easy and is similar to C++. Also AMD is giving away for free all the tools devs need to make their apps accelerated by HSA hardware(it runs seamlessly on CPU or GPU ,selection of code path is done in real runtime).

Will we be seeing the same comments when Kaveri is tested this way? I bet we will,maybe even worse since the better AMD does the attacking horde is more mad .

Agreed. Some reviewers will find a way to highlight any disadvantages AMD's platform has, while ignoring the same on intel platforms. As this review shows, the 8350 does in fact outperform the more expensive intel chippery in many cases in settings that most gamers would use while playing the game. Any review that benchmarks a CPU at 800x600 just to make a chart is stuck in the 90's and early 00's as those settings have zero relevance today. Well done Teksyndicate, showing real world conditions.

 

[Abwx]

. It's the compiler's fault on the benchmark!

It is as if you think that ignoring something will negate
its existence.

Intel compiler flags are a fact , as debunked recently by hardware.fr ,
with the instructions routing that are not using the CPUID but
detection of CPU brand.

http://www.hardware.fr/articles/847-16/conclusion.html

 

[Phynaz]

And yet the Intel compiler produces faster code for AMD CPU's than any other compiler.

 

[amdisstaying]

In my opinion, most of the benchmarks don't provide a complete picture of the AMD platform against the Intel platform.

In 2014, majority of the processors sold are going to be quad-cores or higher.

(1) Majority of users don't overclock their ptrocessors or systems
(2) Most users do multi-tasking today: Playing an HD movie, playing a game like Dirt3, Browsing using multiple browsers, downloading a file, and Editing a text document


"A double-blind test is a method that can be used in most any testing situation where there is a chance that the results of that test could be affected by the tester or the testee." I think the AMD FX 8350 could be compared to the Intel Core i5-3570K using the double-blind testing methodology.

A user will test each system for 10 minutes and the power draw is measured at the wall for each system and for each user. The user doesn't know which system belongs to which vendor. He/she records his/her experience. A simple of 50 would be very representative in this case.

This kind of testing will eliminate all biases and will represent a useful scenario for most users. We will also see the actual power used by each system in doing multi-taskings.

Reviewers should use 1920x1080P resolution or higher.

AMD FX 8350 OC vs i5 3570k OC Using an EVGA GTX 670

http://www.youtube.com/watch?feature=player_embedded&v=4et7kDGSRfc

I like this guy. He is a funny man.

 

[frozentundra123456]

Agreed. Some reviewers will find a way to highlight any disadvantages AMD's platform has, while ignoring the same on intel platforms. As this review shows, the 8350 does in fact outperform the more expensive intel chippery in many cases in settings that most gamers would use while playing the game. Any review that benchmarks a CPU at 800x600 just to make a chart is stuck in the 90's and early 00's as those settings have zero relevance today. Well done Teksyndicate, showing real world conditions.

So you are saying Anand's reviews, Tom's Hardware's review and many other sites that show Intel faster in games are biased while this one reviewer somehow is the only one who "tells it like it is"?

Interesting.

 

[Abwx]

And yet the Intel compiler produces faster code for AMD CPU's than any other compiler.

Not always , and sometime it is a caricature of a compiler
as proved by Anand.

If there was no unoptimizations there wouldnt be things like that :

http://www.anandtech.com/show/2978/amd-s-12-core-magny-cours-opteron-6174-vs-intel-s-6-core-xeon/7

 

[amdisstaying]

Here are some benchmarks of the AMD FX-8350 vs Intel core i5/i7 processors.

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

http://www.overclock.net/content/type/61/id/1255644/width/500/height/1000/flags/LL

http://www.overclock.net/content/type/61/id/1255643/width/500/height/1000/flags/LL

What is going on? Why is the AMD FX 8350 winning against the Intel core i7-3770K processor?

Here is another mutti-treaded benchmark where the AMD FX-8350 destroys the Intel core i7-3770K.

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=293f200&p=2

Are these results fake?

 

[piesquared]

So you are saying Anand's reviews, Tom's Hardware's review and many other sites that show Intel faster in games are biased while this one reviewer somehow is the only one who "tells it like it is"?

Interesting.

I think some sites are certainly biased. I don't think there is any question about that. My point though was that there are advantages to AMD's platform that reviewers go out of their way to deemphasize/ignore. The review at Teksyndicate shows AMD is definitely worth a look for a gaming CPU, but most people can cut through the marketing and PR and likely knew that already, so for many this is nothing really new. But it sure seems to bother intel fans. :shrug:

Speaking of Tom's, I remember that ridiculous stress test they ran quite a few years ago, killing intel board after intel board and making excuse after excuse for them and when they finally found a way to make intel wares survive the duration of the stress test (with intel engineers in tow), claimed intel hardware was superior. Meanwhile AMD's hardware powered right through the stress test without a single issue. Pretty hard to get a fair shake on the market when hardware reviewers go out of their way to step on the little guy. That hasn't seemed to have changed much over the years, and because of that bias, AMD's user base is only growing stronger so reviewers can expect continued backlash.

 

[Idontcare]

I think some sites are certainly biased. I don't think there is any question about that. My point though was that there are advantages to AMD's platform that reviewers go out of their way to deemphasize/ignore. The review at Teksyndicate shows AMD is definitely worth a look for a gaming CPU, but most people can cut through the marketing and PR and likely knew that already, so for many this is nothing really new. But it sure seems to bother intel fans. :shrug:

Speaking of Tom's, I remember that ridiculous stress test they ran quite a few years ago, killing intel board after intel board and making excuse after excuse for them and when they finally found a way to make intel wares survive the duration of the stress test (with intel engineers in tow), claimed intel hardware was superior. Meanwhile AMD's hardware powered right through the stress test without a single issue. Pretty hard to get a fair shake on the market when hardware reviewers go out of their way to step on the little guy. That hasn't seemed to have changed much over the years, and because of that bias, AMD's user base is only growing stronger so reviewers can expect continued backlash.

Hardware reviewers are in business to make money. They are not in business to step on the little guy...unless stepping on the little guy turns out to be a great way to make money

The challenge for reviewers and AMD's user-base alike is finding applications which actually do benefit from the microarchitectural nuances of AMD's CPUs.

All the hidden untapped hypothetical compiler-obfuscated performance in the world is of no relevance if at the end of the day the user-base can't access it.

It would be nice if a review was done which strove to find applications for any given app class (encoding, office, etc) that gave superior performance for Intel and those that gave superior performance for AMD. That way each user-base would know what specific application they ought to procure as best matching their hardware.

Even in gaming this happens. Years ago I was directed towards buying an AMD GPU instead of Nvidia because my primary game of interest was Oblivion and it made sense to me to tailor my purchase on the basis of my intentions for its end use. Reviewers have an opportunity to play a role in that kind of decision making.

 

[bunnyfubbles]

You are still displaying the 8150 as exemple to sustain your point.

Architectural improvements yield 13.5% for games.

A short comparison at same frequency , green charts are
for games , the term "moyenne jeux" means "average for games".

http://www.hardware.fr/articles/880-6/bulldozer-vs-piledriver-4-ghz.html

yeah, and? even if we gave it 20% like I said (of which that same site gives us an average number of 18.1% when going from 8150 -> 8350 for games) that's not enough to overcome any of those 50-100% deficits to Sandybridge chips in the examples I provided...

And I have to use the 8150 because that is one of the few sites that exists specifically to benchmark games and thus they're continually benchmarking GPUs and CPUs for each game as they bench it. Right now that site is, as far as I know, the best source for this kind of thing as it pretty much cannot suffer from software sampling bias because its always testing new games as they come out.

Sure, its not perfect, and we have to go ahead and self adjust the 8150's numbers by ~18% to get an idea of 8350 performance, which is definitely far from ideal, but its the best I've found so far

Again, while the AMD chips can do surprisingly well in some games (as evidenced by that very same site) there are examples (like the ones I've shown) where its an embarrassingly bad situation to where I'm not even thinking twice about whether or not an AMD chip could work for me.

Here are some benchmarks of the AMD FX-8350 vs Intel core i5/i7 processors.

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

http://www.overclock.net/content/type/61/id/1255644/width/500/height/1000/flags/LL

http://www.overclock.net/content/type/61/id/1255643/width/500/height/1000/flags/LL

What is going on? Why is the AMD FX 8350 winning against the Intel core i7-3770K processor?

Here is another mutti-treaded benchmark where the AMD FX-8350 destroys the Intel core i7-3770K.

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=293f200&p=2

Are these results fake?

the first link you posted is the in the OP of this thread, there was no need for you to repost it

the other benchmark results are selection bias - sure, there are occasions where BD/PD architecture can beat Intel, however its the exception, not the norm, and its certainly nothing new

 

[AnandThenMan]

Hardware reviewers are in business to make money. They are not in business to step on the little guy...unless stepping on the little guy turns out to be a great way to make money

Translated, money first the truth a causality? :|

 

[Abwx]

yeah, and? even if we gave it 20% like I said (of which that same site gives us an average number of 18.1% when going from 8150 -> 8350 for games) that's not enough to overcome any of those 50-100% deficits to Sandybridge chips in the examples I provided...

100% deficit is possible only with a turned off PC.....

the other benchmark results are selection bias - sure, there are occasions where BD/PD architecture can beat Intel, however its the exception, not the norm, and its certainly nothing new

Why a selection bias?.because it doesnt suit your point , surely.

Indeed , each time a bench show the FX under a good light
it is deemed as an exception , so now we re ending with tons
of exceptions but only the benchs showing an intel Cpu
as better must be assumed as being the rule....

 

[Idontcare]

Translated, money first the truth a causality? :|

Not like that IMO. No reviewer survives once it is known they can be bought and paid to give slanted reviews.

But it is the reality to say that we, the unpaying readership, get exactly what we pay for. A free review is worth nothing more.

If you want the truth then you need to pay in order to learn the truth. I did.

And if you think about it, what does that really mean? It means even AMD is willing to set aside "the truth" if it means making money.

AMD didn't offer me my FX8350 for free with the common goal that I could then divine the truth for myself. Their attitude was "money first, then truth shall come (if you insist on it)"...no different from Intel's attitude.

 

[bgt]

Being prejudiced is never a good thing to be. Thats why I like having both brands so I can see/feel/notice the differences..................if there are any.
Benchmarks are like religion. It can make people fanatic and make them judge blindly. Pity really:hmm:

 

[Idontcare]

yeah, and? even if we gave it 20% like I said (of which that same site gives us an average number of 18.1% when going from 8150 -> 8350 for games) that's not enough to overcome any of those 50-100% deficits to Sandybridge chips in the examples I provided...

100% deficit is possible only with a turned off PC.....

the other benchmark results are selection bias - sure, there are occasions where BD/PD architecture can beat Intel, however its the exception, not the norm, and its certainly nothing new

Why a selection bias?.because it doesnt suit your point , surely.

Indeed , each time a bench show the FX under a good light
it is deemed as an exception , so now we re ending with tons
of exceptions but only the benchs showing an intel Cpu
as better must be assumed as being the rule....

Not highlighting this particular convo to pick on either contributing member, but I swear we could teleport ourselves back to circa 2005 and the exact same back-and-forth posting argumentation would be found in these forums, only it would be about Prescott versus K8 :|

We have all aged 8 yrs but not a damn thing has changed otherwise when it comes to THE debate. Everyone champions their favored CPU for the niche applications it dominates, and bedevils the other camp with claims of review bias or benchmark cherry picking.

The more things change the more they stay the same.

To what end? The argument wasn't decided in 2005, it sure as hell isn't going to be decisively concluded in 2013.

AMD will suffer the consequences of building a Netburst analog until such time that they build themselves a Conroe analog.

That is all that history has to teach us, the rest is just us finding painful and frustrating ways to grow older year by year wiling away the time by tearing each other a new one in this turn-based RPG

 

[bunnyfubbles]

100% deficit is possible only with a turned off PC.....

no, when one game is double the frame rates that's 100% difference

by your logic a PC providing 50fps is 100% to a PC that is off, while a 3rd PC @ 100fps is also just 100% to the PC that is off...

whereas by my logic the PC that is providing 100fps is providing 100% more FPS than the PC providing 50fps

http://en.wikipedia.org/wiki/Percentage#Percentage_increase_and_decrease

Why a selection bias?.because it doesnt suit your point , surely.

because he's only posting results where AMD does well when there are far more results the suggest the contrary, ie he's the one selecting the results that suit his endevours, whereas I've actually posted results that show AMD isn't always bad (one of which is a BF3 result extremely similar to his) to go along with the results where they are doing poorly and then go on to talk about how it really doesn't even take that many negative results to make a very significant impact

again, I'll reiterate; someone who goes AMD might be just as happy as an Intel user when playing through games like Crysis 2 and Battlefield 3, but as soon as they jump into Planetside 2 the AMD user is suddenly at a massive disadvantage.

Indeed , each time a bench show the FX under a good light
it is deemed as an exception , so now we re ending with tons
of exceptions but only the benchs showing an intel Cpu
as better must be assumed as being the rule....

I posted 6 benchmarks where AMD is clearly inferior and only 2 where its keeping up. Just go to that gamegpu.ru site and look through all the benchmarks. Many of them show AMD doing just "ok" at best but often there are examples where they do poorly and only on occasion are they doing "well" and very rarely could we consider AMD a "winner" (and pretty much only ever to an i5)

And this is just gaming, and many of the situations are where a GPU bottleneck can help to even things out. When I talk about AMD doing well as an exception, I was primarily talking about all the other computing situations not including gaming.

 

[amdisstaying]

the first link you posted is the in the OP of this thread, there was no need for you to repost it

the other benchmark results are selection bias - sure, there are occasions where BD/PD architecture can beat Intel, however its the exception, not the norm, and its certainly nothing new

This guy has really opened a can of worms. So the first link is very very important because he was surprised with the test results. To be sure, he repeated the tests again.

I'm not surprised that the Words like "Cherry picking", "selection bias" and "the exception, not the norm" are used to downplay AMD winnings over and over...

Here is another interesting benchmark of the 95W AMD FX-8300:

http://dl.dropbox.com/u/14652801/Xt...0, 1,45V, Cinebench 11, 8 core, BIOS 1201.jpg

Nicolaus Copernicus showed the world that the Earth goes around the Sun. First, the people didn't believe him. The same is true here. The sentiment against AMD is very strong.

Soon, the AMD A10-6800K will have a base clock of 4.1 GHz, breaking the 4 GHz barrier.

Who is going to break the 5 GHz barrier on the base clock??? AMD or Intel

 

[NostaSeronx]

Most reviewers got paybacks for reviewing the FX series in negative light.

Then, I'll be surprised when the next FX series has 20 cores and everyone praises them.

 

[frozentundra123456]

Here are some benchmarks of the AMD FX-8350 vs Intel core i5/i7 processors.

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

http://www.overclock.net/content/type/61/id/1255644/width/500/height/1000/flags/LL

http://www.overclock.net/content/type/61/id/1255643/width/500/height/1000/flags/LL

What is going on? Why is the AMD FX 8350 winning against the Intel core i7-3770K processor?

Here is another mutti-treaded benchmark where the AMD FX-8350 destroys the Intel core i7-3770K.

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=293f200&p=2

Are these results fake?

No, they are not fake, the the two slides from review sites are obviously GPU bound and not a true measure of cpu performance. Ironically, BF3 is one of a very few games in which Vishera is quite competitive, so you didnt even need to link GPU limited scenarios. But according to the slides that you showed, a phenom II is also as fast as the 8350 or 3770K. Do you really think it is a valid test that shows this result?