CPU Charts 2012: 86 Processors From AMD And Intel, Tested

[kyuuketsuki]

I'd like to see Clover Trail and Brazos added to the charts.

 

[kyuuketsuki]

[citation][nom]Kyuuketsuki[/nom]I'd like to see Clover Trail and Brazos added to the charts.[/citation]Granted, I know that they'd be unable to be included in certain benchs (such as the gaming paired with a discrete GPU), and you can't swap them into a standardized test rig. I'd still like to see how they stack up with other CPUs, though.

 

[punnar]

AMD already said that they will no longer compete with Intel in the high performance arena but will concentrate more on mainstream and integrated systems since that's where most of the money is.

 

[Outlander_04]

AMD already said that they will no longer compete with Intel in the high performance arena but will concentrate more on mainstream and integrated systems since that's where most of the money is.

No , AMD have never said that .

An editor on a technology website speculated that that might be the case , but it has NEVER been AMD policy

Piledriver is sometimes a better performer than i7 , and dollar for dollar often much better

 

[luke904]

[citation][nom]amdfangirl[/nom]Sometimes I wish you updated legacy CPUs like the Core 2 Duo or even perhaps the Athlon 64 X2 series, just one or two models so that people upgrading can have an idea how much faster the CPU is in relation to their new purchase.[/citation]

Definitely agree. Instead they tested not 1, not 2, but 7 phenom II x6's that only vary by a couple hundred MHz. They should of done 2 cpus from each family to show clock differences in it and that's it.

 

[army_ant7]

What you have failed to account for is that the relationship is not linear . Depending on the game engine a difference in cpu power of less than 5% could double fps . Worse still is that running higher resolution and image settings loads the processor in a different way .
And the real world is more complicated . Computers are never just running a game benchmark . If you are playing online you will have firewall ,antivirus and network services wanting cpu cycles . But the way these benchmarks are run makes no account of that .
Anyone who believes a dual core i3 can match an i5 at the same clock speed is going to wind up very disappointed ... despite the results of many of these benches

I am aware of that. It is a complex matter with multiple factors. My example was just purely hypothetical.
I bet some games would show pretty much similar performance with i3's and i5's. For example, if the game can't effectively utilize more than 2 threads, or if the game is already GPU bottlenecked as is. Anyway, my point (that CPU that performs way better, even if it's in excess, still shows promise with more CPU intensive games) still stands.

No , AMD have never said that .

An editor on a technology website speculated that that might be the case , but it has NEVER been AMD policy

Piledriver is sometimes a better performer than i7 , and dollar for dollar often much better

“That era is done,” Rory Read said in an interview and added, “There’s enough processing power on every laptop on the planet today.” Uh oh. Those quotes leave room for speculation that is significant enough to upset an entire loyal customer base of enthusiast users and may not have been the smartest choice of words.

<b>AMD</b> Backing Out of <b>CPU</b> Speed Wars Against Intel
I just double-checked and it seems you're right. It is just "speculation."

I'm curious though. What test have you seen that showed a Piledriver beating an i7?

 

[Outlander_04]

[citation][nom]army_ant7[/nom]I'm curious though. What test have you seen that showed a Piledriver beating an i7?[/citation]

http://www.guru3d.com/articles_pages/amd_fx_8350_8320_6300_processor_4300_performance_review,5.html

Worth checking the limited gaming benches in that article too . For the reasons I described earlier user experience will be identical whether you paid under $200 for ann FX or over $300 for the latest i7 .
Arguably you could have better gaming performance with the FX because you'd have $100 more in your budget for a graphics card

And here is a BD versus an i7 2600K in gaming , high resolutions and image qualities
http://www.hardwareheaven.com/reviews/1285/pg1/amd-fx-8150-black-edition-8-core-processor-vs-core-i7-2600k-review-introduction.html

 

[army_ant7]

@Outlander_04
Oh, I see... I guess I shouldn't be surprised since the FX-8000 series CPU's do have 8 integer cores, and even though Intel's latest cores may be superior, Hyper-threading (with the i7's) may not really be enough to trump 8 FX CPU cores, that is if an application can efficiently utilize all 8 of those integer cores.

For the gaming benchmarks from Hardware Heaven, it seems like the test system was GPU bottle-necked, though the Bulldozer CPU did still beat the i7 by a bit. I wonder though if this could be because of Hyper-threading, like cases of bad scheduling so as to assign two threads to one physical core. It also could just be within the margin of error. (I don't know how many iterations they did.)

 

[Outlander_04]

@Outlander_04


For the gaming benchmarks from Hardware Heaven, it seems like the test system was GPU bottle-necked, though the Bulldozer CPU did still beat the i7 by a bit. I wonder though if this could be because of Hyper-threading, like cases of bad scheduling so as to assign two threads to one physical core. It also could just be within the margin of error. (I don't know how many iterations they did.)

The forum comments on the Hardware Heaven site are really interesting .
More than one person suggested "gpu limited" .........but this simply CAN NOT be the case or both processors would give identical results . Those tests quite clearly show that at those resolution and image settings , playing those games the BD was better than an i7

 

[army_ant7]

@Outlander_04
Though they are the same tests, haven't you ever heard of a "margin of error"? Different factors can influence test results mildly or wildly. I do wonder what their testing methodology is at Hardware Heaven, but like here at Tom's I have read they run the same test multiple times to combat variations in results. Things like maybe not walking in the same exact path in the game while benchmarking can cause variations, or different conditions in the game (like night and day). Also, Windows itself, like if a system process fired up or something.

I remember seeing GPU-bottlenecked tests where the CPU's that were compared still varied in performance slightly.

 

[drkamiya]

Q6600 plox!

 

[Outlander_04]

[citation][nom]army_ant7[/nom]@Outlander_04Though they are the same tests, haven't you ever heard of a "margin of error"? Different factors can influence test results mildly or wildly. I do wonder what their testing methodology is at Hardware Heaven, but like here at Tom's I have read they run the same test multiple times to combat variations in results. Things like maybe not walking in the same exact path in the game while benchmarking can cause variations, or different conditions in the game (like night and day). Also, Windows itself, like if a system process fired up or something.I remember seeing GPU-bottlenecked tests where the CPU's that were compared still varied in performance slightly.[/citation]

so if an intel wins a benchmark its superior , but if an AMD is better in a different situation then the test is faulty?

fanboy much?

 

[atminside]

bahhhhh my socket 478 P4 3.4GHZ extreme edition paired with my AGP HD3850 blows them all out of the water......

 

[JonnyDough]

Can you guys add an E6600 for scale and email me? Please? It's a super popular CPU.

 

[Guest]

Really wish you had the Socket 2011 CPUs in there. Was keen to see how quad channel memory affects the Adobe results.

 

[SirTrollsALot]

My ancient i7 930 at 4.1ghz on air with a ASUS P6X58-E PRO w/ SLI 460's , still holds well for the past couple of years. Runs anything!!

 

[scout62]

[citation][nom]klewlis1[/nom]It would be nice to see the 960T from AMD tested once in awhile.[/citation]


Agreed, I'd like to see that too.

 

[JonnyDough]

[citation][nom]SirTrollsALot[/nom]My ancient i7 930 at 4.1ghz on air with a ASUS P6X58-E PRO w/ SLI 460's , still holds well for the past couple of years. Runs anything!![/citation]

That's not even old, and any top of the line CPU from a couple of generations back can match today's low end CPUs.

 

[army_ant7]

so if an intel wins a benchmark its superior , but if an AMD is better in a different situation then the test is faulty?

fanboy much?

Please read (and understand) what I've said. Where I'm coming from is not whether the CPU is AMD or Intel. I proposed possible things. I didn't claim anything. I didn't say their methodology was "faulty" (just that I didn't know how reliable it is). I was skeptical though, based on benchmarks I remember previously seeing, but you did present data, which I acknowledged.

BTW, I honestly kinda got the feeling that you were a bit too much for AMD (i.e. fanboy) with previous posts of yours, but you didn't saying anything like you did, now did you? Because as long as you were giving reasons and data (both of which I acknowledged), why would I be so stupid as to claim you're a fanboy at once?

 

[blazorthon]

[citation][nom]dgingeri[/nom]It's funny. People accuse Intel of changing the sockets too often, yet, if you look at the same span of time, AMD has gone through just as many sockets as Intel:Intel: 775, 1156, 1366, 1155, 2011AMD: AM2+, AM3, AM3+, FM1, FM2At least Intel kept the value range server socket and the mainstream desktop socket the same. AMD didn't do that much.[/citation]

AMD's AM sockets are all inter-compatible (granted motherboard support depends on memory support and BIOS support). Intel's sockets are not inter-compatible. Besides, it's only since LGA 775 that Intel was bad about socket changing with modern CPUs. For example, I can put many AM3 CPUs in most AM2+ motherboards, most AM3+CPUs in most AM3 motherboards, AM2+CPUs in AM2 motherboards, and many AM2 CPUs in some socket 939 motherboards.

With some motherboards, CPU support simply depends on the BIOS and memory. For example, an AMD motherboard with both DDR2 support and DDR3 support can run almost every AMD CPU made in the last decade so long as BIOS support keeps up and it does with some such boards. The same can't be said about Intel ever since LGA 1156 launched.

Furthermore, you're bunching together some sockets with different purposes. LGA 1366 and LGA 2011 should be separate from the rest as should the FM sockets from AMD. Intel's 775, 1156, and 1155 sockets are in the same range as AMD's AM sockets (in their given time frames) and the rest in your lists are all niche sockets.

Intel keeping their consumer and server sockets the same doesn't seem very important to me. Also, AMD does have some low end server CPUs on their desktop sockets anyway, so it's not an *advantage* that AMD totally lacks.

EDIT: Also, most of AMD's consumer CPUs support a few features that Intel's consumer models don't such as ECC memory support.

 

[HVDynamo]

[citation][nom]amdfangirl[/nom]Sometimes I wish you updated legacy CPUs like the Core 2 Duo or even perhaps the Athlon 64 X2 series, just one or two models so that people upgrading can have an idea how much faster the CPU is in relation to their new purchase.[/citation]

I would love to see some sort of hierarchy chart like they have for GPU's. It would be much easier to see whether it is worth it or not to go for an upgrade on a particular CPU just the same as it is for a GPU.

 

[JonnyDough]

[citation][nom]HVDynamo[/nom]I would love to see some sort of hierarchy chart like they have for GPU's. It would be much easier to see whether it is worth it or not to go for an upgrade on a particular CPU just the same as it is for a GPU.[/citation]

Because I'm such a nice guy:

http://www.tomshardware.com/reviews/gaming-graphics-card-review,3107-7.html

 

[mapesdhs]

[citation][nom]amdfangirl[/nom]Sometimes I wish you updated legacy CPUs like the Core 2 Duo or even
perhaps the Athlon 64 X2 series, just one or two models so that people upgrading can have an idea
how much faster the CPU is in relation to their new purchase.[/citation]

[citation][nom]mayankleoboy1[/nom]Great benchmarks. But i want some processors which were
legendary overclockers, and representatives of their generation of CPU's, included with a nominal OC...[/citation]

I'm slowly accumulating results for such older CPUs compared to newer models, using the previous
generation of tomshardware tests and also a few of my own. I'm also testing how different CPUs
affect GPU performance using the same gfx card, though I'll merge the data into the main pages later.
I'm testing both stock and oc'd configs.

I have oodles more tests to run, but if you'd like to be kept informed of updates, especially of any
particular CPU I have in which you might be interested, then let me know via PM or email (Google
'Ian SGI' to find my site) and I'll email you info as and when I add it. For reference, I have Q6600,
E8400, QX9650, 6000+, Athlon II X4 635, Phenom II X4 965, etc. The full list is much larger. Still
not yet managed to obtain a Ph2 X6 though, but I will when I can, or any newer BD CPUs.


[citation][nom]A Bad Day[/nom]Throw in a Pentium 4 as a reminder to those folks who still think they can run BF3 on a 1.8 GHz Willamette. ...[/citation]

Indeed, I have a number of older CPUs to test aswell, including a P4/HT 3.4GHz (SL7PY), PentiumD
3.4 945 (D0), and some equivalent AMDs such as a Athlon64 3400+ (stock 2.4, oc 2.76), though I
draw the line at an AthlonXP.

[citation][nom]tomfreak[/nom]... There is no need to bench 2600K/2700K & Core i5-2300/Core i5-2310,
just one of the 2 set will do, because we all know their performance is close to identical.[/citation]

Yeah, some of the included CPUs do seem a bit superfluous, eg. the i7 870 is fine, no need for the 860
or 875K (especially the latter which is only useful for custom oc'ing). Ditto the 750/760; the 920/930;
and the 980 is plenty, no need for the 980X if it's not being oc'd, or the 990X given the marginal
difference. Same applies to the range of AMD CPUs.

However, I suppose those who happen to have one of the other CPUs will be happy. Personally though,
I would have left quite a few out and included some oc numbers instead, hence the choice of CPUs I
obtained for testing, eg. an i3 550 is an old chip now, but it oc's incredibly well. Likewise, how well does
an oc'd E8400 hold up today? (back in the day, the oc potential of Wolfdale was legendary) Or a Q6600?
Sometimes RAM matters too of course, though I've been surprised to find a number of S775 boards that
support DDR3 (not just the famous ones like the Striker II Extreme, but less well known boards such as
the Gigabyte GA-EP45T-UD3LR (quite good for a simple server).

Atm though I'm sorting out a 3930K system for someone. Will get back to testing in late Jan.

Ian.

 

[peter lau]

Socket FM1 Motherboard
MSI A75MA-G55, Chipset: => "Intel Z77" ??

please correct this minor mistake

 

[superkart]

I don't know if there was an overall better cpu than my my Phenom II 95w out there but I do know that it was drop in compatible with my 3 year old motherboard and happily replaced my Athlon x2.
I don't closely follow what Intel does, I've always been an AMD guy since the Palomino days, but I've read complaints about forcing their customers to get a new mobo every time they get a new chip out sort of.
Lower power consumption has lead traditional AMD fans switch to Intel even in the budget segment but overall AMD is still king in that segment.
Much as I love building new PCs,I haven't managed to found a good reason to do so three years in a row, just drop in more memory, get an SSD drive, replace the cpu and done.