FX Vs. Core i7: Exploring CPU Bottlenecks And AMD CrossFire

[Reynod]

Nice article crash - well written and informative, and importantly balanced.

Keep up the good work.

 

[ingtar33]

amd's 4/8 core design IS 8 cores... they just use a different structure to get there then intel would. And unlike Hyperthreading shows real world gains representing what you'd expect from an 8 core design. Hyperthreading is a software gimmick to mimic extra cores... and as a result only results in modest gains in performance in line with something akin to a turbo button on an old 486.

AMD's problem is their memory controller doesn't get data to all the cores efficiently. some people claim it's the core design... i think because bulldozer/piledriver is such a radical new design, it invalidates that claim. The truth is it's a northbridge/memcontroler problem. AMD's CPUs may or may not be slower then intel's... but we have no way of knowing. Because they're not being fed enough data to actually let them run at their full ability. they're being starved for info to process which is slowing them down.

 

[dkcomputer]

Looks like he tried his best to favor AMD with his review, picking an i7 and picking those games, and still amd failed miserably.

 

[quark004]

at higher resolution s games depend more on gpu power than cpu. modern games still do not fully utilise more than 2 cores effectively. what is the point of this benchmark anyway.

 

[Outlander_04]

[citation][nom]ingtar33[/nom]amd's 4/8 core design IS 8 cores... they just use a different structure to get there then intel would. And unlike Hyperthreading shows real world gains representing what you'd expect from an 8 core design. Hyperthreading is a software gimmick to mimic extra cores... and as a result only results in modest gains in performance in line with something akin to a turbo button on an old 486.[/citation]

An AMD FX core is not the same thing as an intel core , or a Phenom core . One small section of the core [ the integer unit ] is replicated and it shares all the rest of the core with its partner .
But even many reviewers failed to grasp the consequences of this and often said per core performance was lower than phenom. Its not because they are not comparing apples to apples .

As for Big Mack 70 ....hey man did you read the article? Unless you happen to be building a pc just to play skyrim you are wasting money on an intel build . It wont perform noticeably better . What I love about you intel fanboys is that what you lack in rationality you more than make up for with denial

 

[sugetsu]

What "fair and Balanced" review, the article ended up ignoring that games such as skyrim were not developed with PCs in mind and completely left off the part where such games are not well threaded. We all know that the FX processors do not as well as intel processors in single threaded applications, but as we all know, single threaded apps are on their way out. FX processors compete with the much pricier i7s when it comes to multi-threaded apps.

 

[Crashman]

[citation][nom]sugetsu[/nom]What "fair and Balanced" review, the article ended up ignoring that games such as skyrim were not developed with PCs in mind and completely left off the part where such games are not well threaded. We all know that the FX processors do not as well as intel processors in single threaded applications, but as we all know, single threaded apps are on their way out. FX processors compete with the much pricier i7s when it comes to multi-threaded apps.[/citation]Testing ignores nearly all political arguments and includes only benchmarks from 1.) Recent test suites 2.) are easy to repeat and 3.) that support or can be made to support EyeFinity. It probably could have been stated more explicitly in the article, that most of the value charts included the price of graphics because it's a test of graphics+CPU. A CPU-only value chart would be more appropriate for CPU-only benchmarks.

 

[sanilmahambre]

Where is this AMD's Escalator architecture CPU

 

[bystander]

[citation][nom]sugetsu[/nom]What "fair and Balanced" review, the article ended up ignoring that games such as skyrim were not developed with PCs in mind and completely left off the part where such games are not well threaded. We all know that the FX processors do not as well as intel processors in single threaded applications, but as we all know, single threaded apps are on their way out. FX processors compete with the much pricier i7s when it comes to multi-threaded apps.[/citation]
There are only a handful of games which utilize more than 4 threads. Whether they use 1, 2, 3 or 4 threads, the i7 has a large advantage, as you can see. Interestingly enough, he managed to find 1 of 2 games I'm aware of that even use more than 4 threads. Do you expect him to only include games which will use more than 4 cores in this benchmark suite? Wouldn't that be extraordinarily biased, considering how few games do that?

It may be a while before games use more than 4 threads on a regular basis. Game design doesn't naturally utilize a lot of threads. It is something that requires a ton more effort for very little gains.

 

[Gin Fushicho]

[citation][nom]bystander[/nom]3dmark11 shows improvements with hyperthreading enabled, it clearly will use more than 4 cores. It does so in the physics tests.Metro 2033, when advanced DoF is on, also uses more than 4 cores. You'd be pleased to note, that AMD did fair about the same as the i7 on the Metro 2033 benchmark when using advanced DoF, but was behind when it wasn't.[/citation]

Good to know. :3 I'll get the most Recent 3Dmark program that's out there. I always worry about the core usage.

 

[stairmand]

Poor choice of game engines IMO, I've got a 3770K and Crossfire Eyeginity setup and can tell you that Metro 2033 seems to run differently every time I bloody play it and is horribly setup for crossfire, AVP is way too old to be a decent benchmark.

From my testing the i7 seems to keep the crossfire setup more utilized, I actually had a decent improvement simply moving from an i7 920 (3.8) to an i7 3770K (4.8) framerates were up slightly but games certainly played smoother.

 

[stairmand]

Not in crossfire but gives a great CPU comparison in an engine that as worth testing with.

http://www.techspot.com/review/608-hitman-absolution-performance-benchmarks/page6.html

 

[stairmand]

http://www.techspot.com/review/608-hitman-absolution-performance-benchmarks/page6.html

 

[ojas]

[citation][nom]BigMack70[/nom]Has there been any meaningful performance increase shown from the move from a 3570k to a 3770k in games? I am under the impression, and all the info I've seen suggests, that they are basically identical for gaming purposes.So I don't necessarily think it's moving backwards to look at the 3570k rather than the 3770k. If there is a difference made by the jump from the i5 to the i7, that would be awesome to know about![/citation]
Yeah, the same reason i wanted to see an i5 in there as well.

Not a step backwards, it's just an additional step. Plus that will also help isolate games that use more than 4 threads and/or more cache. Memory bandwidth would be almost identical as long as you keep the same motherboard and RAM config, i guess. Of course, the 4.4 GHz OC setting would have ensured that clock speeds were equal too.

AND, we'd also know that if there's a point of the i7-3770k AT ALL, over the 3570K.

I mean, after all, in the "Best gaming cpus for the money" articles, you guys never include the 3770K. It's either the 3570K or the 3930K.

And these charts aren't helpful:
http://media.bestofmicro.com/Z/P/364165/original/game_crysis_ii.png
http://media.bestofmicro.com/Z/Q/364166/original/game_mafia_ii.png
http://media.bestofmicro.com/0/B/364187/original/res_syn_fritz.png

Because they suggest that games won't really need much above an i5 unless they're like Fritz, but then those are low res settings.

If you're exploring bottlenecks, shouldn't you be looking at stuff bit by bit, part by part? Like when you were testing MSAA bottlenecks on the 660 Ti.

 

[ojas]

[citation][nom]quark004[/nom]at higher resolution s games depend more on gpu power than cpu. modern games still do not fully utilise more than 2 cores effectively. what is the point of this benchmark anyway.[/citation]
The point of this was to see if the GPUs would be bottlenecked by the CPU.

 

[ojas]

[citation][nom]bystander[/nom]There are only a handful of games which utilize more than 4 threads. Whether they use 1, 2, 3 or 4 threads, the i7 has a large advantage, as you can see. Interestingly enough, he managed to find 1 of 2 games I'm aware of that even use more than 4 threads. Do you expect him to only include games which will use more than 4 cores in this benchmark suite? Wouldn't that be extraordinarily biased, considering how few games do that?It may be a while before games use more than 4 threads on a regular basis. Game design doesn't naturally utilize a lot of threads. It is something that requires a ton more effort for very little gains.[/citation]
I'm not sure using well threaded games would be biased, because since the next-gen consoles are coming out, and phones, tablets, Pentiums (heck even Atoms) are multi-core now, threaded applications and games should become increasingly common.

So if you were looking to buy a new PC right about now, or upgrade, you'd want to know how much you'd want to spend to stay relitively future proof.

So while IPC is very important in terms of gaming, it wouldn't be bad to test newer titles that demand more of both the GPU and CPU, and make use of multiple threads. Why? Because you're looking for bottlenecks.

I somehow feel in retrospect that this article became more of a performance comparison, and then a value comparison, when it started with an agenda of finding bottlenecks.

So i think that in addition to the 3770K and the 8350, maybe adding the 3570K and the 3930K wouldn't have been a bad idea.

 

[mrrez]

From the picture to the whole article, this smacks of Tom fanboyism at its best. How much did Intel pay you to write this drivel. And before it starts I run a z77 Intel based gaming rig. But after recently reading and watching some independent results I feel like a mug to have listened to all the Intel is best for gaming hype.

Let me be one of the first to say Sorry AMD I was wrong.

 

[oxiide]

I'm sure its been said, I didn't read all 5 pages of comments—but we already know the i5-3570K performs about the same as the -3770K in most of these games. i5-3570K vs. FX-8350 would have been a more realistic, price-controlled test.

 

[Crashman]

[citation][nom]oxiide[/nom]I'm sure its been said, I didn't read all 5 pages of comments—but we already know the i5-3570K performs about the same as the -3770K in most of these games. i5-3570K vs. FX-8350 would have been a more realistic, price-controlled test.[/citation]Scaling-down the CPU when running $800 in graphics is completely unrealistic. The 3570K wasn't missing at all, the only thing that was missing was a faster consumer-oriented AMD CPU.

 

[markem]

I clicked on a post that had 9+ positive remarks, that was critical of toms, and it never changed up.. but i bet if i clicked it down, it would have... something shady with the score system going on..

Wouldn't be surprised if this comment doesn't make it

 

[bystander]

[citation][nom]ojas[/nom]I'm not sure using well threaded games would be biased, because since the next-gen consoles are coming out, and phones, tablets, Pentiums (heck even Atoms) are multi-core now, threaded applications and games should become increasingly common.So if you were looking to buy a new PC right about now, or upgrade, you'd want to know how much you'd want to spend to stay relitively future proof.So while IPC is very important in terms of gaming, it wouldn't be bad to test newer titles that demand more of both the GPU and CPU, and make use of multiple threads. Why? Because you're looking for bottlenecks.I somehow feel in retrospect that this article became more of a performance comparison, and then a value comparison, when it started with an agenda of finding bottlenecks.So i think that in addition to the 3770K and the 8350, maybe adding the 3570K and the 3930K wouldn't have been a bad idea.[/citation]
Every game in the list is threaded. What are you going on about? Skyrim shows improvements up to 4 threads, and large improvements up to 3. The point is, the vast majority of the games in existence do not use more than 4 cores effectively and it would be extremely biased to only include those games. The idea is to get a typical cross section of games with some variety, which he has done, though it would have been nice to see more games in general.

 

[Crashman]

I clicked on a post that had 9+ positive remarks, that was critical of toms, and it never changed up.. but i bet if i clicked it down, it would have... something shady with the score system going on..

Wouldn't be surprised if this comment doesn't make it

Scoring system gets jammed all the time, but I can make you a tin foil hat if you'd like.

 

[tuffjuff]

[citation][nom]Haserath[/nom]Why not just use two computers?[/citation]

Lol.

 

[bystander]

I think I see a theme here. There are a bunch of people who believe that all comparisons are to be done in order to show AMD's strong points, but the truth is, AMD's CPUs don't perform well under all conditions.

This article is comparing CPU's for top end gaming systems. This isn't about making Intel or AMD look good. The purposes is simple, how does AMD's top end gaming CPU compare to Intel's top in gaming CPU with typical games.

If you want to see an article show AMD's strong points, look at production comparisons, or maybe budget gaming to some degree.

 

[ricardok]

Catalyst 12.10?? WTF??

The tearing is being addressed with 13.20 beta and you could have used 13.10 too if you wished. Why stay with an old 12.10 driver?

That's not even close to fairness..

I wish I could see the same tearing test done on 13.20!