AMD CPU speculation... and expert conjecture

[de5_Roy]

you guys are being lazy and arguing the benchmark instead of looking into it.
here are translated excerpts from the pclab test:

Of course, the frame rate depends on many things, including maps. It turns out that this beta was made available in one of the most challenging.

Wrote: on most computers, because, fortunately, we observed improved performance issues with certain processors. FX-6300/6350 is already faster than the FX-4300, as it should be, and the dual-core finally work. More on this topic write on the page dedicated to the testing of processors.

...

CPU test in Battlefield 4

The tests are divided into sections on processors and a section on graphics cards. The former test standard multiplayer map Siege of Shanghai as an option for the 64 players. Test conducted using two graphics cards: GeForce GTX 780 and Radeon R9 290x . It turns out that once again the relations between the processors depend on which of them will be found on your computer.

The test graphics cards in Battlefield 4

Because of the many difficulties and the time required to test each of the cards at the moment measurements performed in single player during a mission in Baku, while trying to escape from a hostile helicopter. We plan to test the card but also the multiplayer maps.

i looked at their test bench, nothing looks out of place e.g. underspecced, undercapacity system memory, cheaper mobo etc.
now look into gamegpu's test and compare.

edit3: haswell core i3 beating stock fx6350 in the full version is a bit hard to believe considering it's an 'amd gaming evolve' title and amd paid ea dice for mantle support. in fx's defense, the multiplayer bench is hard to replicate for each test.
edit4: reading further into the review, core i5, especially o.c. haswell 4670k holds considerable lead. to me it looks like amd's gfx card drivers had bigger cpu overhead which fx could not mitigate.
http://pclab.pl/zdjecia/artykuly/chaostheory/2013/10/bf4/charts/bf4_cpu_gpu.png
there was also fps improvement for fx cpus in multiplayer mode from (64bit) windows 7 to windows 8.1. pclab tested using windows 8.1 64bit.
edit5: amd sure is finding more room for improvement for supporting a game they hyped up so much:
http://techreport.com/news/25587/new-radeon-drivers-address-battlefield-4-instability
from here, looks worse than they did with bf3, lol.

 

[blackkstar]

you guys are being lazy and arguing the benchmark instead of looking into it.
here are translated excerpts from the pclab test:

Of course, the frame rate depends on many things, including maps. It turns out that this beta was made available in one of the most challenging.

Wrote: on most computers, because, fortunately, we observed improved performance issues with certain processors. FX-6300/6350 is already faster than the FX-4300, as it should be, and the dual-core finally work. More on this topic write on the page dedicated to the testing of processors.

...

CPU test in Battlefield 4

The tests are divided into sections on processors and a section on graphics cards. The former test standard multiplayer map Siege of Shanghai as an option for the 64 players. Test conducted using two graphics cards: GeForce GTX 780 and Radeon R9 290x . It turns out that once again the relations between the processors depend on which of them will be found on your computer.

The test graphics cards in Battlefield 4

Because of the many difficulties and the time required to test each of the cards at the moment measurements performed in single player during a mission in Baku, while trying to escape from a hostile helicopter. We plan to test the card but also the multiplayer maps.

i looked at their test bench, nothing looks out of place e.g. underspecced, undercapacity system memory, cheaper mobo etc.
now look into gamegpu's test and compare.

edit3: haswell core i3 beating stock fx6350 in the full version is a bit hard to believe considering it's an 'amd gaming evolve' title and amd paid ea dice for mantle support. in fx's defense, the multiplayer bench is hard to replicate for each test.
edit4: reading further into the review, core i5, especially o.c. haswell 4670k holds considerable lead. to me it looks like amd's gfx card drivers had bigger cpu overhead which fx could not mitigate.
http://pclab.pl/zdjecia/artykuly/chaostheory/2013/10/bf4/charts/bf4_cpu_gpu.png
there was also fps improvement for fx cpus in multiplayer mode from (64bit) windows 7 to windows 8.1. pclab tested using windows 8.1 64bit.
edit5: amd sure is finding more room for improvement for supporting a game they hyped up so much:
http://techreport.com/news/25587/new-radeon-drivers-address-battlefield-4-instability
from here, looks worse than they did with bf3, lol.

Battlefield has always been a game where Intel guys can cherry pick parts of the game and say that AMD sucks.

http://www.youtube.com/results?q=pentium+g+battlefield+3

Take a look. There are TONS of people in completely undemanding parts of the game going "look at how good my Pentium G runs BF3! PENTIUM G IS A GREAT GAMING CHIP!!!"

Now look what actually happens: http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/7

Notice how Pentium G was right in line with everyone else for FPS yet it gets completely destroyed in frame time?

You can cherry pick very well with Frostbite games to make your point about the number of cores used.

However, worst case, the CPUs with more cores are tied with a dual or quad whilst the best case are the CPUs with more than 4 thread walking away from everything else.

My issue with GamerK's argument is that he is saying that i3 can be just like FX 8350 in some situations so he comes to the conclusion that they must be equal chips.

It is like saying a Corolla and a Ferrari both go the same speed on a certain road so they must be the same speed everywhere.

The proper term for this is hasty generalization and I've ridden GamerK over this before and it made him stop, but apparently he seems to have forgotten that his entire argument of

1. Yes you have benchmarks where FX 8350 is competing with i7s and even Intel Hex, but here is a benchmark where FX competes with i3

2. Therefore, threads don't matter and all you need is core i3. Note that this changed to "FX performs more like i3"

Yes, FX can perform like i3 but it is when i3 is in an area when i3 is strong. i3 does not always stick with I7 and FX 8350 in all BF3 and BF4 benchmarks and if the chips were actually even in performance and the game didn't scale, there would be absolutely no difference in performance between core i3, i7 quad, i7 hex, and FX series between every benchmark in existence.

That's because if you want to make an argument, you need to have evidence that supports it overwhelmingly.

FX at best outperforms Intel Hex in BF4 and at worst competes with i3. If you buy an i3, you're not going to get those high performance times.

The FX guys are saying that BF4 CAN use those cores. They are not saying they use them 100% of the time.

Having a CPU that performs at worst a little better than i3 and at best a little worse than i7 hex is much better than just having an i3. I don't know what you're trying to argue, the FX would provide a superior experience as long as you have 3m/6 or more than anything Intel offers at those price points.

Overall, GamerK you need to stop with your hasty generalizations that you have some benchmarks that show BF4 doesn't scale to multiple cores and concluding that the games don't scale to multiple cores when there is evidence that it cant scale to multiple cores.

No one is saying BF4 will load an FX 8350 to all 8 threads 100% of the time. They are saying that those threads are very helpful in some situations and that the game can scale to tons of threads. You really, really need to work on your reading comprehension as well as your ability to logically reason, it's absolutely appalling that you have come into this thread multiple times with the same fallacious arguments thinking that you're some sort of super-clever-uber-elite game developer who knows more than anyone else.

I don't care if you were the lead developer for every major game engine and you were a lead developer in every blockbuster title for the last 15 years, you're making hasty generalizations, failing at reading comprehension (on multiple levels), failing at logical reasoning, etc. You argue and debate like a child fresh out of college who thinks they know everything because they've managed to surround themselves with fallacious logic that makes perfect sense to them.

 

[de5_Roy]

^^er... use gamerk316's post or no post in the reply if you're gonna talk about gamerk316...?

 

[gamerk316]

By your logic, according to TH's benches, the 2500k > 3960x.

I think that entire argument seems a bit flawed...especially considering minimum FPS.

I explicitly noted the 8350 had a higher minimum lower average. Hence why I said the FX-8350 and i5-2500k were essentially equal.

And yes, the 3960x is clearly better then the 2500k, based on a +6 minimum and <1 maximum difference. Blame Toms for how they ordered their chart.

Also remember this is a MP bench, so there is going to be more noise in the data.

Or you could take the other way and say that you often quote benchmarks biased against AMD.

Right now, the only ones I could find were GameGPU, PcLab, TomsHardware. Techspot has one, but its an obvious GPU bottleneck and pointless to this discussion.

That being said, I noted back when GameGPU did the Crysis 3 results that other sites tended to disagree with GameGPU's results, which showed AMD doing better then other sites did. So this isn't the first time this type of discrepancy has occurred. And as of yet, no ones been able to figure out why the results from GameGPU tend to deviate from other sites on the net, much less which results are "correct". Hence why I showed both GameGPU and PCLab (and later Toms).

edit3: haswell core i3 beating stock fx6350 in the full version is a bit hard to believe considering it's an 'amd gaming evolve' title and amd paid ea dice for mantle support. in fx's defense, the multiplayer bench is hard to replicate for each test.

That being said, if you averaged out the runs, you should expect the results to normalize somewhat. Granted, we don't know the exact test methodology (how many runs, etc).

Or, we could take my view that things like "Gaming Evolved" and "Intel Inside" and "TWIMTBP" are little more then marketing fluff, and there's actually very little developers can do to bias their titles one way or the other.

edit4: reading further into the review, core i5, especially o.c. haswell 4670k holds considerable lead. to me it looks like amd's gfx card drivers had bigger cpu overhead which fx could not mitigate.
http://pclab.pl/zdjecia/artykuly/chaostheory/2013/10/bf...
there was also fps improvement for fx cpus in multiplayer mode from (64bit) windows 7 to windows 8.1. pclab tested using windows 8.1 64bit.

8.1 should be benefiting AMD more then Intel though. In that case, I'd expect Pclab, not GameGPU, to be showing AMD ahead.

And OC'd SB/IB would tend to do better then OC'd BD/PD for the simple reason that Intel's higher IPC will lead to a higher rate of performance increase.

edit5: amd sure is finding more room for improvement for supporting a game they hyped up so much:
http://techreport.com/news/25587/new-radeon-drivers-address-battlefield-4-instability
from here, looks worse than they did with bf3, lol.

Is this shocking? You can't test every config, so I'm sure both NVIDIA and AMD will have bugfix drivers for the next release or two to fix some issue that went undetected in lab testing.

 

[gamerk316]

Having a CPU that performs at worst a little better than i3 and at best a little worse than i7 hex is much better than just having an i3

Just pointing out, the "It's almost as good, but cheaper!" argument has been the AMD tagline for a while now. Just saying, the i3 vs 8350 (or more realistically, i3 vs 6300) debate is kind of the same exact thing. You accept the worse case downsides in favor of more or less competitive performance at a cheaper price. I also note I've tended to compare i3's against the FX-6xxx line, except in cases where noting the FX-8xxx line doesn't pull significantly ahead.

My primary point is a simple one: If you have a CPU with good enough Clock + IPC, it doesn't matter if each core is running 6 threads, as you'll get roughly the same performance as a chip with significantly more cores, but lower IPC. Hence my argument, going back to the BD rumor thread, that BD was likely never going to be faster then Intel as a CPU architecture, even IF you have a situation where games start using more threads. [Nevermind the theoretical case where one overloaded core could actually result in worse performance, which seems to happen often in the case of the FX-4xxx lineup].

And if Intel released a BD like arch, I'd be making the same exact arguments, just like I bashed Intel's Larabee (which I said from day 1 would be a total failure).

And I again note, back in the 2009 timeframe, when everyone here was recommending the E8600 over the Q9550/Q9650, I was the one who quite loudly proclaimed "Duo's are dead". The fact the i3's hold up so well is a statement of how strong Intels IPC has gotten even since the Core 2 days.

Personally, I'd never recommend an i3 for a gaming rig, but if you are going to make the "Its almost as good but cheaper" argument, then you have to accept it as a viable gaming CPU compared to the FX-8350, which sometimes is slightly to significantly faster. [My personal recommendation is anything equal to or better then a 2500k for Intel, and only the FX-8350 for AMD. I do not view any other AMD CPU as a viable long term option.]

 

[de5_Roy]

8.1 should be benefiting AMD more then Intel though. In that case, I'd expect Pclab, not GameGPU, to be showing AMD ahead.

And OC'd SB/IB would tend to do better then OC'd BD/PD for the simple reason that Intel's higher IPC will lead to a higher rate of performance increase.

huh.
...
i googled "gamegpu.ru battlefield 4" and in the first link:
http://gamegpu.ru/action-/-fps-/-tps/battlefield-4-test-gpu.html
the test setup says:

Software Configuration
Operating system
Microsoft Windows 7 Ultimate Edition x64 SP1

Graphics Driver
Nvidia GeForce / ION Driver Release 331.65
AMD Catalyst 13.11 beta 7

The monitoring program
MSI Afterburner v beta 3 15
GPU Mist
FRAPS

pclab's:

System operacyjny:

Windows 8.1 64-bitowy.

Sterowniki:

AMD Radeon Catalyst 13.11 beta v7;
Nvidia GeForce 331.65 WHQL.

i am posting these only because people seem reluctant to look into the text. i don't care much about multiplayer benches since those are difficult to replicate.

 

[blackkstar]

My primary point is a simple one: If you have a CPU with good enough Clock + IPC, it doesn't matter if each core is running 6 threads, as you'll get roughly the same performance as a chip with significantly more cores, but lower IPC.

Yeah, and if that is true, then FX 8350 would only fall between i3 and i5 in multi-thread according to your logic.

But you seem to not be comprehending my point. My point is that there are situations where that happens, and then there are situations where FX 8350 overwhelms the competition because of the extra cores. You just seem to conveniently ignore when the more cores approach actually works and you just want to talk about when they don't.

You know, I can cherry pick examples too: http://hwbot.org/benchmark/hwbot_prime/rankings#start=0#interval=20

Scroll down to find 3930k (it's a few pages past all the FX 8350s)

 

[8350rocks]

My primary point is a simple one: If you have a CPU with good enough Clock + IPC, it doesn't matter if each core is running 6 threads, as you'll get roughly the same performance as a chip with significantly more cores, but lower IPC.

Yeah, and if that is true, then FX 8350 would only fall between i3 and i5 in multi-thread according to your logic.

But you seem to not be comprehending my point. My point is that there are situations where that happens, and then there are situations where FX 8350 overwhelms the competition because of the extra cores. You just seem to conveniently ignore when the more cores approach actually works and you just want to talk about when they don't.

You know, I can cherry pick examples too: http://hwbot.org/benchmark/hwbot_prime/rankings#start=0#interval=20

Scroll down to find 3930k (it's a few pages past all the FX 8350s)

Isn't it interesting there that the 8350 @ 4.48 outperforms the 3960x @ 5.10

 

[noob2222]

If that's your point then you need a different benchmark. The quad-core FX is markedly slower than the octo-core FX, despite running at the same clockspeed. More cores does matter for AMD.

Emphasis mine. More cores matter more for AMD because, with only four cores, they are getting totally overloaded due to their poor IPC. As a result, if even ONE core gets loaded, performance tanks. Hence why AMD's performance tends to scale linearly as you OC. Intel quads, by contrast, don't have that problem (for the most part) at base clocks, hence why OCing tends to offer less performance increase (no bottleneck to remove).

Hence why games like BF4, where AMD gains significant performance as more cores are added, yet Intel performance is still sorted largely by IPC + Clock performance, and also why Intel tends to maintain its gaming lead.

See:
http://www.bf4blog.com/battlefield-4-retail-gpu-cpu-benchmarks/
http://pclab.pl/art55318-3.html

FX-8350 ~ i5 2500k in BF4, despite how well threaded it is. That's how superior IPC is better for overall performance in a nutshell. For Intel, more cores essentially just lowers overall core loading, since its high tier i5's aren't bottlenecked.

This was my exact argument against BD from the very beginning, and three years in, AMD still can't beat SB i7's. Even in games using over a dozen threads, like BF4, AMD still looses, or at best, matches Intel, head to head.

Lets examine this shall we?

Using PClabs since it favors intel anyway.

I7 4770k @3.5 ghz = 78.7 fps
I7 4770k@ 4.5 ghz = 91.4 fps ... bottleneck ... no

lets check clock speed scaling

4.5/3.5 = 1.285 or 28.5% increase in clock speed
91./78.7 = 1.166 or 16.6% faster.
Ratio of clock speed to performance = 58%

8350, @4.0 ghz = 61.8 fps
8350 @ 5 ghz = 73.6
same math
skipping to ratio = 76%

sure, AMD scales faster because its slightly more bottlenecked, but that doesn't mean that Intel isn't bottlenecked

It could also mean that if none of them are close to being bottlenecked, that AMD will still scale better (8 semi-real cores vs 4+4 ht cores)

 

[gman97005]

Both the AM3+ and the 2nd Gen Richland APU will be backwards compatible with the new FM2+ Socket so if you already have either one it may behoove you to start shopping for a FM2+ Mobo..

 

[juanrga]

If that's your point then you need a different benchmark. The quad-core FX is markedly slower than the octo-core FX, despite running at the same clockspeed. More cores does matter for AMD.

Emphasis mine. More cores matter more for AMD because, with only four cores, they are getting totally overloaded due to their poor IPC. As a result, if even ONE core gets loaded, performance tanks. Hence why AMD's performance tends to scale linearly as you OC. Intel quads, by contrast, don't have that problem (for the most part) at base clocks, hence why OCing tends to offer less performance increase (no bottleneck to remove).

Hence why games like BF4, where AMD gains significant performance as more cores are added, yet Intel performance is still sorted largely by IPC + Clock performance, and also why Intel tends to maintain its gaming lead.

See:
http://www.bf4blog.com/battlefield-4-retail-gpu-cpu-benchmarks/
http://pclab.pl/art55318-3.html

FX-8350 ~ i5 2500k in BF4, despite how well threaded it is. That's how superior IPC is better for overall performance in a nutshell. For Intel, more cores essentially just lowers overall core loading, since its high tier i5's aren't bottlenecked.

This was my exact argument against BD from the very beginning, and three years in, AMD still can't beat SB i7's. Even in games using over a dozen threads, like BF4, AMD still looses, or at best, matches Intel, head to head.

Lets examine this shall we?

Using PClabs since it favors intel anyway.

I7 4770k @3.5 ghz = 78.7 fps
I7 4770k@ 4.5 ghz = 91.4 fps ... bottleneck ... no

lets check clock speed scaling

4.5/3.5 = 1.285 or 28.5% increase in clock speed
91./78.7 = 1.166 or 16.6% faster.
Ratio of clock speed to performance = 58%

8350, @4.0 ghz = 61.8 fps
8350 @ 5 ghz = 73.6
same math
skipping to ratio = 76%

sure, AMD scales faster because its slightly more bottlenecked, but that doesn't mean that Intel isn't bottlenecked

It could also mean that if none of them are close to being bottlenecked, that AMD will still scale better (8 semi-real cores vs 4+4 ht cores)

One can do other kind of math computations and see that the i7 has an overclocking efficiency of 91% whereas the FX has an efficiency of 95%. The difference is minimal. It can mean i7 bottleneck, or that the PD design is optimized for higher frequencies, or be a mere statistical effect.

 

[kansasboy001]

I talked to a tech and friend from ASUS and he told me that in Q4 2013 the successor to 990fx will be released. It only is logical that there will be a new cpu released at the same time. He also said that he thinks that there will be a new socket, so bye bye AM3+.

I think that AMD will release an 8 or possibly 10 core apu to replace the FX line.
Whatever happens next, you can be sure that it will be big, with AMD having Microsoft and Sony pouring cash in to them. AMD now has parts in all 3 major gaming consoles!

 

[de5_Roy]

^^ heh. there's a different way to 'read' into this. may be the 990fx successor is a88x. may be the new socket is fm2+. may be the new cpu(!) to be released is kaveri. the asus rep. will still be right.

IF there is indeed a new AM-class socket, current am3+ owners will be screwed. fx cpu owners(with low end am3+ mobos) might not be, though, if amd allows am3+ cpus run on the new socket. for example, upgrade from an fx6300+asus 760g am3+ combo to fx6300+new socket. otoh, if you have a high end 990fx am3+ mobo.... 😗

"AMD to Unveil Next-Generation APUs on November 11"
http://www.techpowerup.com/193616/amd-to-unveil-next-generation-apus-on-november-11.html

edit: i am having trouble reading the last post in the page. when i try to open it for the first time, i keep getting redirected to the first page. the stats page says that this thread has nearly 9000 answers and 179 pages, may be it's near it's limit?

 

[noob2222]

looks like xbone is going to lose

http://www.theguardian.com/technology/2013/oct/31/ps4-playstation-4-xbox-one-consoles-call-of-duty-ghostss

60 fps target, cut off graphics quality to achieve it, ps4 should look better, but still no PC quality.

 

[juanrga]

I talked to a tech and friend from ASUS and he told me that in Q4 2013 the successor to 990fx will be released. It only is logical that there will be a new cpu released at the same time. He also said that he thinks that there will be a new socket, so bye bye AM3+.

I think that AMD will release an 8 or possibly 10 core apu to replace the FX line.
Whatever happens next, you can be sure that it will be big, with AMD having Microsoft and Sony pouring cash in to them. AMD now has parts in all 3 major gaming consoles!

My bet is in a 8-core APU {*} as said before. I see no reason for 10 cores, because extra cores would remain unused for games, which has been AMD main problem with last gen games.

They could combine two of those APU for a top high-end 16-core APU for servers/HPC.

However, I see no sign this is going to happen soon. And leaked details that Carrizo will be ~65W eliminates any possibility for a 8-core version, in my opinion.

{*} With a CPU variant a la Athlon, i.e. APU with iGPU disabled.

 

[juanrga]

Well, we know Kaveri APUs will be named A10-7x00 and that will be accompanied by Athlons (as predicted)

http://www.planet3dnow.de/cms/5096-fm2-offiziell-nur-mit-neuen-chipsaetzen-moeglich/

Pay attention to the slideshow at the end of the page. There is a very very interesting slide #13 that, in my opinion, reveals which are AMD plans regarding the old FX-line. WARNING: Sensible HEDT enthusiasts better not to look at it.

Before I speculated two options for desktop: 1) Abandon FX-4 series and sell FX-6/8 up to end existences. And 2) Abandon FX-4 series and refresh FX-6/8 a la Warsaw. After seeing the slide #13, I think that it seems confirmed that AMD will abandon the FX-4 series.

Please don't kill the messenger if you don't like the message.

 

[juanrga]

I managed to isolate the slide #13

EDIT: They have deleted the link to the slide (the links given below by de5_Roy also ceased to work). Fortunately I saved a copy and have uploaded it now to my server

 

[de5_Roy]

relevant fm2+ related promo slides (thanks juanrga for the original link):
new platform will have 3 chipsets, sorted according to feature-richness: a88x, a78 and a55. seems kaveri will have 7x00 model numberings.

Spoiler

a88x is the full featured, has most number of sata ports and best raid support. it also supports crossfire but doesn't say which kind. i assume it's discreet card cfx otherwise they woulda said 'amd dual gfx'.
i managed to spot 2 mini itx motherboards from giglibytes and [strike]asus[/strike] asrock(! apologies. force of habit). looks good.

Spoiler

oh, here's the disclaimer slide that Everyone Forgets To Read.
http://www.planet3dnow.de/cms/wp-content/gallery/amd-apu-update-4q13/20-apu-4q13-update.png

the rest is usual promo b.s. look into them and you'll easily spot the lies. 😆

 

[gamerk316]

If that's your point then you need a different benchmark. The quad-core FX is markedly slower than the octo-core FX, despite running at the same clockspeed. More cores does matter for AMD.

Emphasis mine. More cores matter more for AMD because, with only four cores, they are getting totally overloaded due to their poor IPC. As a result, if even ONE core gets loaded, performance tanks. Hence why AMD's performance tends to scale linearly as you OC. Intel quads, by contrast, don't have that problem (for the most part) at base clocks, hence why OCing tends to offer less performance increase (no bottleneck to remove).

Hence why games like BF4, where AMD gains significant performance as more cores are added, yet Intel performance is still sorted largely by IPC + Clock performance, and also why Intel tends to maintain its gaming lead.

See:
http://www.bf4blog.com/battlefield-4-retail-gpu-cpu-benchmarks/
http://pclab.pl/art55318-3.html

FX-8350 ~ i5 2500k in BF4, despite how well threaded it is. That's how superior IPC is better for overall performance in a nutshell. For Intel, more cores essentially just lowers overall core loading, since its high tier i5's aren't bottlenecked.

This was my exact argument against BD from the very beginning, and three years in, AMD still can't beat SB i7's. Even in games using over a dozen threads, like BF4, AMD still looses, or at best, matches Intel, head to head.

Lets examine this shall we?

Using PClabs since it favors intel anyway.

I7 4770k @3.5 ghz = 78.7 fps
I7 4770k@ 4.5 ghz = 91.4 fps ... bottleneck ... no

lets check clock speed scaling

4.5/3.5 = 1.285 or 28.5% increase in clock speed
91./78.7 = 1.166 or 16.6% faster.
Ratio of clock speed to performance = 58%

8350, @4.0 ghz = 61.8 fps
8350 @ 5 ghz = 73.6
same math
skipping to ratio = 76%

sure, AMD scales faster because its slightly more bottlenecked, but that doesn't mean that Intel isn't bottlenecked

It could also mean that if none of them are close to being bottlenecked, that AMD will still scale better (8 semi-real cores vs 4+4 ht cores)

Ahhh, but you forget to realize that once you escape the effects of a CPU bottleneck, performance increases tend to be very minimal. You could make the argument that Intel's scaling is worse because the 4770k eventually escapes the effects of being a bottleneck, where the 8350 is still not at that point.

Isn't it interesting there that the 8350 @ 4.48 outperforms the 3960x @ 5.10

Some tasks naturally scale, and for those tasks, the FX-8350 would likely outperform most Intel CPU's. I've never argued against that particular point. [Go back to the BD thread; I predicted it would stink at gaming, but be a benchmark monster.]

For games, where most parallel processing is done on the GPU (Rendering + Physics), you are left doing mainly serial workloads, which favors a CPU with higher clock and higher IPC. THATS the point I'm arguing: For gaming, AMD does not have a particularly strong architecture. But for a subset of tasks, AMD's arch is better.

But for a gaming rig? I'd recommend Intel every time. The only attractive CPU AMD has for gaming is the FX-8350 [I do not think the 6300 is going to age well at all.]

 

[kansasboy001]

^^ heh. there's a different way to 'read' into this. may be the 990fx successor is a88x. may be the new socket is fm2+. may be the new cpu(!) to be released is kaveri. the asus rep. will still be right.

IF there is indeed a new AM-class socket, current am3+ owners will be screwed. fx cpu owners(with low end am3+ mobos) might not be, though, if amd allows am3+ cpus run on the new socket. for example, upgrade from an fx6300+asus 760g am3+ combo to fx6300+new socket. otoh, if you have a high end 990fx am3+ mobo.... 😗

"AMD to Unveil Next-Generation APUs on November 11"
http://www.techpowerup.com/193616/amd-to-unveil-next-generation-apus-on-november-11.html

edit: i am having trouble reading the last post in the page. when i try to open it for the first time, i keep getting redirected to the first page. the stats page says that this thread has nearly 9000 answers and 179 pages, may be it's near it's limit?

I think that there will be a completely new socket, A88x does not have enough pcie x16 channels to be on high end boards. ASUS tech thought AM4, but the APU option sounds more realistic.

 

[kansasboy001]

Possibly a phenom III

 

[de5_Roy]

I think that there will be a completely new socket, A88x does not have enough pcie x16 channels to be on high end boards. ASUS tech thought AM4, but the APU option sounds more realistic.

good call. i totally overlooked that little bit. amd does need a desktop platform for high end multi-gpu configs, with 30+ pcie lanes.
afaik, the current apus cannot address this issue without the help of a pcie lane mux chip or whatever it's called. actually, the pcie gen 2.0/3.0 controller is on-die, not in the chipset like the cpus (e.g. FX). unfortunately, it's nearly worthless (imo, from market segment p.o.v.) to implement such a switch on an apu mobo. intel's higher end mainstream mobos have them but those address high end, not entry level.
amd still needs a 'proper' hedt platform. unfortunately (again), i think they threw in the towel months ago. they still threw up a few centurion cpus and gouged a few clueless people. i dunno... i think amd itself is in a bit of limbo on whether to address the hedt segment or not. imo, the customers will respond very positively to competition should amd properly participate instead of crapping out worthless garbage like centurion.

edit3: apus are bad proposition for high lane count, multi-gpu platform due to the fact that the pcie controller is on the apu die. if amd goes that route, they'll have to turn the apu into a complex chip like intel's lga 2011 cpus. it'd be easier by updating the northbridge on AM-platform.

 

[kansasboy001]

I think that there will be a completely new socket, A88x does not have enough pcie x16 channels to be on high end boards. ASUS tech thought AM4, but the APU option sounds more realistic.

good call. i totally overlooked that little bit. amd does need a desktop platform for high end multi-gpu configs, with 30+ pcie lanes.
afaik, the current apus cannot address this issue without the help of a pcie lane mux chip or whatever it's called. actually, the pcie gen 2.0/3.0 controller is on-die, not in the chipset like the cpus (e.g. FX). unfortunately, it's nearly worthless (imo, from market segment p.o.v.) to implement such a switch on an apu mobo. intel's higher end mainstream mobos have them but those address high end, not entry level.
amd still needs a 'proper' hedt platform. unfortunately (again), i think they threw in the towel months ago. they still threw up a few centurion cpus and gouged a few clueless people. i dunno... i think amd itself is in a bit of limbo on whether to address the hedt segment or not. imo, the customers will respond very positively to competition should amd properly participate instead of crapping out worthless garbage like centurion.

edit3: apus are bad proposition for high lane count, multi-gpu platform due to the fact that the pcie controller is on the apu die. if amd goes that route, they'll have to turn the apu into a complex chip like intel's lga 2011 cpus. it'd be easier by updating the northbridge on AM-platform.

I agree, unless they put the pcie controller back on the north bridge for their enthusiast APUs. I think that AMD will be releasing some high end APU to replace the fx line. Or still possible, phenom III, they are reusing Athlon, so...

 

[de5_Roy]

I agree, unless they put the pcie controller back on the north bridge for their enthusiast APUs. I think that AMD will be releasing some high end APU to replace the fx line. Or still possible, phenom III, they are reusing Athlon, so...

detaching pcie controller for 'enthusiast apus' may have adverse affect. due to market positioning, apus benefit from more integration.

mobile version of athlon core was used in the first mainstream apu - llano. those didn't clock higher than 3 ghz. bd uarch is designed for easier integration with other components.

amd is reusing the athlon brand only, like they reused the fx brand for bd cpus. the underlying 'cpu's are apus with igpus lasered off.

 

[tracker45]

Should I sell my graphics card and get an APU ?