AMD CPU speculation... and expert conjecture

[Cazalan]

I really find it hard to defend an architecture when an AMD quad (FX-4300) at 3.8GHz is clearly CPU bottnecked, but an Intel quad at 3.3GHz (i5-2500k) clearly isn't. Or when an 8-core chip clocked at 4.0GHz (FX-8350) looses to a quad with HTT clocked at 3.4GHz (i7-2600k). Hell, in this case, the quad with HTT has lower average CPU usage! (Which again: I've been saying for years now)

If you don't include price into the equation then yes, the Intel chips almost always outperform. Pick yourself up a 12 core Xeon for $2500 and go to town.

 

[The Q6660 Inside]

Also, kinda surprised how the i7-920 and upwards still is a top-tier gaming CPU choice.

its actually quite amazing seeing a processor from 08 still game well and the I7 920 came out for just 280-300$.

By 2010 when I got it, it was like $220, man was it worth it!

 

[palladin9479]

Gamer remember there is a performance hit when both cores on a module are loaded and that's what your seeing with the AMD CPU's (~80% of actual performance). Yes there is obviously a main thread along with 5~6 additional threads that are being kept fairly busy. This is preferred over piling everything into two threads and calling it a day.

Normally I'd agree with you about codeing to low level API's except that API also exists on the upcoming consoles. Games developed for those consoles will have Mantle render paths anyway, the PC ports will be added in the DX11 paths. So even if some future update breaks your mantle path you still have your DX11 path. There really isn't a situation where the developer is losing anything here. Now if the game was developed for PC only then I don't expect there to be a Mantle path.

 

[Ags1]

Also, kinda surprised how the i7-920 and upwards still is a top-tier gaming CPU choice.

its actually quite amazing seeing a processor from 08 still game well and the I7 920 came out for just 280-300$.

It's not 1994 any more. The pace of change is a tenth of what it used to be. Hopefully Kaveri will buck the trend, and show a solid 20% jump over the previous generation!

 

[8350rocks]

And as predicted FX-8 > FX-6 > FX-4

FX-8350: 4GHz
FX-8150: 3.6 GHz
FX-6300: 3.5 GHz
FX-6100: 3.3 GHz
FX-4300: 3.8 GHz [AMD quads clearly bottlenecked at this point]

So yeah: The faster processor does better. Shocking!

Clock them at the same speed and call back; I'd wager the 6300 and 8350 would suddenly be performing about the same.

I really find it hard to defend an architecture when an AMD quad (FX-4300) at 3.8GHz is clearly CPU bottnecked, but an Intel quad at 3.3GHz (i5-2500k) clearly isn't. Or when an 8-core chip clocked at 4.0GHz (FX-8350) looses to a quad with HTT clocked at 3.4GHz (i7-2600k). Hell, in this case, the quad with HTT has lower average CPU usage! (Which again: I've been saying for years now)

Now OC the 2600k to 4GHz, and do the comparison again. Woops, Intel is likely a good 10-15 FPS faster then AMD at the same clock. So what do you think will happen if Intel ever releases a 4GHz @ stock chip? Oh, that's right: AMD loosing straight up.


I'll say what I said three years ago: AMD's CPU architecture is not well suited for games, even ones that scale reasonably well.

Hmm...I tend to disagree here...

The i7 is loaded to equal resources on it's physical cores...however, you'll notice the HTT usage is so high that you physically cannot possibly load the main cores anymore without overrunning the CPU. See the i3 for a moment:

The i3 is completely over run, and overwhelmed...(no surprise here...) and performs subpar as such. Speeding it up might make the FPS bump a bit, but you're not solving the inherent problem. Clockspeed can't fix that magnitude of bottleneck.

Intel designed HTT to operate using a maximum of 20-30% of core resources. In the instance of the i3, your HTT register stack is running all out, and you really don't have the resources on your core to do that and run the main core resources at 100%. Which means you are literally putting your "main 2 threads" on hold to run the extra threads.

In the 2600k, you're seeing just enough physical resources to keep from being completely bottlenecked. However, you're seeing the HTT maxed out, and the core resources being utilized as much as they can be without having a direct bottleneck.

Looking at the AMD architectures, we can conclude the following:

The FX 4300/4350 is not sufficient, it's over run entirely, much like the i3. This bodes poorly for 2M APUs.

The FX 6300 is just shy of a bottleneck, but only barely. This means an overclock would benefit it greatly, though it also means that BF4 will run 6-7 heavy threads when given the chance. So the FX 6350 would probably fare even better here (due to higher clockspeed).

The FX 8350 shows that BF4 scales well, but does not fully utilize 8 cores, which is mostly what we see from the i7x models in Intel...7 threads running, 6 on main cores and 1 HTT register stack.

The single core loaded excessively on the 8350 does speak toward potentially heavy FPU use in the main thread, which would explain the upshot in core usage for the first 2 cores, while the other cores are loaded relatively similarly to the Intel CPUs.

So, if you take all factors into consideration, the 2600k is just shy of bottlenecked, much like the 6300 is...where as the 8350 is likely suffering from the poorer FMAC performance of the PD architecture in the spike of core usage there. The FMAC is supposedly being tweaked for SR, IIRC, however, that may have been speculation from one of the many "theorists" about what is going on with SR.

EDIT: What would be truly telling is to see the i5 core usage numbers. That would show you the difference between 2 cores with HTT and 4 cores. Of course, looking at the data, it would suggest if the 2600k is all but bottlenecked, the 2500k would certainly be bottlenecked.

 

[8350rocks]

Gamer remember there is a performance hit when both cores on a module are loaded and that's what your seeing with the AMD CPU's (~80% of actual performance). Yes there is obviously a main thread along with 5~6 additional threads that are being kept fairly busy. This is preferred over piling everything into two threads and calling it a day.

Normally I'd agree with you about codeing to low level API's except that API also exists on the upcoming consoles. Games developed for those consoles will have Mantle render paths anyway, the PC ports will be added in the DX11 paths. So even if some future update breaks your mantle path you still have your DX11 path. There really isn't a situation where the developer is losing anything here. Now if the game was developed for PC only then I don't expect there to be a Mantle path.

Agreed...MANTLE is only a benefit to developers.

 

[8350rocks]

Looks like I was right. I5 2500K is about 20-25% faster then an fx6300. i7 2600k 68.375% cpu usage in bf4 vs 67.625% cpu usage for the fx8350 so the i7 2600k is similar on limit as the fx8350. Now whose ipc is right. Basically a sandy bridge i5 lets say at 4ghz is equivalent to an amd fx8320 at stock. i5 3570k at stock is about 35% faster then fx6300. i5 4670k stock settings is similar to an fx8350 multithreaded wise. 4 core steamroller juangra will be about 80% slower then haswell 4 core.

No, with HTT, this is how to properly calculate CPU load on an i7:

Add 4 physical cores % usage then divide by 4. Then add HTT numbers together and divide by 4, then multiply by 30%. Add Physical % + HTT% and you arrive at reality.

Your figures are arrived at by adding everything and dividing by 8, which is not reality. As the 2600k does *not* have 8 cores...

Reality comes out to 86.875% core usage on the 2600k, which is approaching a bottleneck. You're near maxed out on that CPU. That's not to say that overclocking wouldn't bring that number down closer to maybe 80% usage; however, it's still quite a bit higher than the 8 core AMD.

EDIT: BTW, cut out the BS where I bolded your post...that's so beyond old by now it isn't funny.

 

[8350rocks]

No, a high clocked haswell i3 would have to run at better than 4 GHz to match the 6300, it's just too overwhelmed. You also forget, for the money you'd spend on that i3, you can overclock that 6300 to ~4.5 GHz pretty easily. Which would give that CPU a pretty hefty boost. The i3 can't do it. You won't see a 4+ GHz i3 anytime soon either...

 

[Cazalan]

By 2010 when I got it, it was like $220, man was it worth it!

I got mine for free. BBuy marked them off $300 by mistake at launch.

 

[8350rocks]

http://www.3dmark.com/3dm11/5879474 3dmark11 physics score stock fx6300:5248
I think an i3 at 3.6ghz haswell would now be on par with an fx6300.

No physics is all FMAC/FPU ops. This means the FX6300 with 3 FPUs will be better than an i3 with 2 FPUs unless the i3 is clocked excessively higher. You can't overclock an i3, so you'll never get there.

 

[blackkstar]

http://gamegpu.ru/action-/-fps-/-tps/battlefield-4-beta-test-gpu.html

and

and

First off: Notice how much better the FX-8350 does in comparison to the FX-8150. BF4 Beta clearly likes per-core performance (as expected). About the same as BF3 in that regard.

It's also very worrying that even with twice the amount of cores, AMD's per-core workload is about the same as the i7-2600k. And one core is clearly overworked (95% load), which could very easily cause a deadlock situation, costing some significant FPS. I'd wager an OC would help this chip out a LOT.

I'm reasonably sure that is in fact what is happening to the FX-8150: One core being overworked, costing application performance. That's the downside to weak individual cores: If one gets overloaded, performance for the entire application can suffer.

Second: Note that the FX-6300 is clearly CPU bottlenecked. Competes against the i5-760. I'd be VERY interested to see how an OC'd FX-6300 holds up though.

Third: Game likes HTT a LOT. Helps keep the i3-2100 around FX-4300 level performance.

Would like to see how the i3-3220 holds up, given how well the i3-2100 does.


So, AS PREDICTED, overall performance is about the same as BF3, as far as CPU ordering goes.

You kind of dropped the ball. Windows scheduler after Windows XP can detect the difference between logical and virtual cores. You showed 3930k using all 6 real cores and then having another thread spill over to a virtual one because it had 7 threads to run and there were no more available physical ones.

Also, you completely derped with comparing FX 8150 to FX 8350. 8350 has an 11% clock speed advantage over 8150 yet was 17% faster.

Also, yet another derp on your behalf, 2600k 3% higher clock speed than 2500k, 2600k was almost 13% faster.

Also, because I like to see you squirm, 3930k has lower per core performance than 2600k thanks to lower 3930k clockspeed. 2600k has 6% faster clockspeed yet 3930k was 11% faster than 2600k with a 6% clock speed disadvantage.

GamerK please stop you're embarrassing yourself. Your evidence that BF4 wants more stronger cores says the exact opposite of what you think it means. 3930k with more cores and slower clocks beat 4 physical, 8 logical thread CPU with the same architecture. What do you think that means? 8350 had a 6% IPC increase according to this bench and 11% better clocks.

You seem to be really confused in thinking that adding more cores means that the cores have to be weaker. However it's nice to see Intel's "MOAR COARS SUX!!" FUD alive and well and being used to steer people like you away from 3930k and other Intel hex core CPUs.

 

[The Q6660 Inside]

By 2010 when I got it, it was like $220, man was it worth it!

I got mine for free. BBuy marked them off $300 by mistake at launch.

They did not care to charge afterwards?
@hafijur, plain and simple. FX6300>i3 Haswell

 

[jdwii]

Looks like I was right. I5 2500K is about 20-25% faster then an fx6300. i7 2600k 68.375% cpu usage in bf4 vs 67.625% cpu usage for the fx8350 so the i7 2600k is similar on limit as the fx8350. Now whose ipc is right. Basically a sandy bridge i5 lets say at 4ghz is equivalent to an amd fx8320 at stock. i5 3570k at stock is about 35% faster then fx6300. i5 4670k stock settings is similar to an fx8350 multithreaded wise. 4 core steamroller juangra will be about 80% slower then haswell 4 core at same clock speed.

Ok number 1 that game does not use 8 cores efficiently
Number 2 Amd is 20% slower per core than Intel's latest offering while probably being 40% slower in performance per clock.
Number 3 GPU bottleneck

Their you have your reasons for this now go home your drunk

 

[jdwii]

http://www.3dmark.com/3dm11/3300303 i7 3520m ib laptop cpu scores 4616 at 3.4ghz for two cores. So at 3.5ghz+ 7% ipc + desktop haswell is like scoring 5200 for a haswell i3 at 3.6ghz similar to an fx6300. Anyway I am off have fun calculating now what the equivalent i5 haswell to steamroller is. I predict a 6 core SR at 4.5ghz is equivalent to an i5 haswell at 3.3ghz.

That is just nuts the major improvement seems like the FPU in the module design for steamroller which is the main reason Amd is so far behind in certain games. Amd is only 40% slower in IPC and what you keep forgetting is Amd's module design does take a 20% hit when every 2 cores is used(in the module)

So for now on remember this for the currunt module design(steamroller is supposed to help this)
8 cores= 6.4 cores
6 cores= 4.8 cores
4 Cores= 3.2 Core
2 Cores= 1.6 Core

Now this is why the scaling looks so bad to you when comparing a 8 core fx to a I5 it seems like its getting the same if not just a bit more performance with double the cores. Now take this and the 20% hit per core since Amd is 20% weaker per core even with the high clock rates and now add up the issues with games not wanting to scale to the full 8 cores efficiently and you get a conclusion.

So for example if a game uses the 4 core processor very well you will see it perform like a 3.2 core BUT with 20% less Single threaded performance than Intel. So then you see Intel running a HT dual core processor and it runs like a 2.6 core since it has HT which at max adds 30% more performance per core so that is why you see the same performance or only slightly better performance between the 2 and the same goes for Amd's 6 core and 8 core processors.

This is why you need to stop comparing performance base on cores alone and instead compare it on end all performance and heat and power consumption and then cost.

Now what i am stating is actual facts and even can be proven by looking at Amd them self's the goal to their design was to get more multitasking performance with less die space did it pay off no i honestly think this design was a bad choice even some of Amd engineers agree on that point. With steamroller all their doing is adding things into their module to make it act more and more like bigger cores vs smaller ones. Also with HT you add free performance with CMT you pay for that performance in heat and power consumption.

Either way hafijur not trying to be mean or anything but read this and try to make some since out of it.

 

[juanrga]

http://gamegpu.ru/action-/-fps-/-tps/battlefield-4-beta-test-gpu.html

and

and

First off: Notice how much better the FX-8350 does in comparison to the FX-8150. BF4 Beta clearly likes per-core performance (as expected). About the same as BF3 in that regard.

It's also very worrying that even with twice the amount of cores, AMD's per-core workload is about the same as the i7-2600k. And one core is clearly overworked (95% load), which could very easily cause a deadlock situation, costing some significant FPS. I'd wager an OC would help this chip out a LOT.

I'm reasonably sure that is in fact what is happening to the FX-8150: One core being overworked, costing application performance. That's the downside to weak individual cores: If one gets overloaded, performance for the entire application can suffer.

Second: Note that the FX-6300 is clearly CPU bottlenecked. Competes against the i5-760. I'd be VERY interested to see how an OC'd FX-6300 holds up though.

Third: Game likes HTT a LOT. Helps keep the i3-2100 around FX-4300 level performance.

Would like to see how the i3-3220 holds up, given how well the i3-2100 does.


So, AS PREDICTED, overall performance is about the same as BF3, as far as CPU ordering goes.

FALSE. Effectively you predicted that the BF4 beta would be as the BF4 alpha, which was as BF3. However, data says otherwise:

BF4 alpha:
i5-2500k = 1.04x FX-8350 = 1.13x FX-6300 = 1.34x FX-4300

BF4 beta:
FX-8350 = 1.11x i5-2500k = 1.46x FX-6300 = 1.76x FX-4300

Therefore, the FX-8 has changed from being slower than the i5 to being faster than the i5. The difference between the FX-8 and the FX-6 was minimal (9%) but now it is of about a 50% (that is a brutal increase). The same when compared to the FX-4; it was 29% and now is 76%.

Therefore stop pretending that it is was as you said.

It is all just as we predicted. BF3 was optimized for 4-cores and barely used 6-cores. That is why BF3 recommended 4-cores AMD CPU. BF4 is optimized for 6-cores (95% load) and use 8-cores. That is why BF-4 recommend 6-cores AMD CPU.

During presentation of MANTLE it was promised that the MANTLE version of BF4 will use better the 8-cores

We will see what happens (I am enthusiast).

And as predicted FX-8 > FX-6 > FX-4

FX-8350: 4GHz
FX-8150: 3.6 GHz
FX-6300: 3.5 GHz
FX-6100: 3.3 GHz
FX-4300: 3.8 GHz [AMD quads clearly bottlenecked at this point]

So yeah: The faster processor does better. Shocking!

You snipped the part that follows to that quote:

In fact the 8-core offers about the double performance of the 4-core which suggest that the game scales well up to 8 cores.

Effectively, 4GHz/3.5GHz = 1.14 which implies that the FX-8350 would be about a 10% faster, but it is a 46% because the game uses more than six-cores.

I already said you that next gen games will be optimized for six-cores because six-cores of the consoles are exclusively used for games (I already shown you a PS4 CPU profile) and cores being clocked so low as 1.6 GHz implies developers cannot code for two of them and ignore the rest.

Your pretension that i3 would be top CPU for future gaming vanished in the air. Your predictions (LOL) about how engines couldn't really use more than two/four cores has vanished in the air as well. I'm sorry.

Similar stuff about MANTLE your nobody-will-use-it prediction will vanish in the air.

 

[juanrga]

^^ But again: You lose portability, even across vendors on the same platform. Portability and coding efficiency was the reason we stopped using hardcoded assembly for crying out loud!

Sure, if I coded to the metal, in pure assembly, violated every good programming practice, I'm sure I could double the performance of every application I write. Until the next hardware revision changes one of my assumptions (and it can be something REALLY trivial too) ends up costing me more performance then I gain. Devs HATE going back and re-optimizing for a different arch, and that's what API's like MANTEL force you to do every time the hardware changes.

If you continue attacking MANTLE at least spell it right. Besides that

MANTLE != coding to the metal level
MANTLE != assembly

Also DEVS and the tech. press are loving MANTLE. The only critic voices that I am reading on forums are from people who is very pro Intel/Nvidia.

Also, I'm awaiting BF4's first benchmarks for the Beta, which should be hitting sometime today. So far, all we have are older Alpha benchmarks back from June, though they do show similar performance to Bf3:

https://linustechtips.com/main/topic/30410-battlefield-4-alpha-gpu-and-cpu-benchmarks/

Already getting ready for the "crappy PC port" arguments to start up again.

Ok

And as predicted FX-8 > FX-6 > FX-4. In fact the 8-core offers about the double performance of the 4-core which suggest that the game scales well up to 8 cores. People who believed that an i3 would be top for next gen was dreaming.

Also interesting the GPUs

TITAN performance is disappointing. Any bet which will be the performance of the new R9 with the MANTLE version of BF4? Some developer said it will humiliate the TITAN.

hmm 4300 +20% for kaveri = 40 fps average.
8350 = 60 fps,
i5 760 = 45fps
how is kaveri a smart move to abandon high end cpus again and going to reach any i5 performance?

This seems to be another of your attempts to ignore what I say/said.

I said that Kaveri 4C ~ FX-6 ~ i5 and that game shows that Kaveri 4C could compete with a FX-6300.

I already shown my predictions on how Kaveri will be faster than an i5-2500k in some benchmarks and slower than the i5-2500k in others. I think you missed that as well.

I also said that kaveri 4C cannot compete with a FX-8350. I said people could continue purchasing FX-8 and FX-9 during 2014 before Carrizo is ready (2015). Also you missed that. The only way that a Kaveri 4C can compete with a FX-8350 is using HSA software. I also said this and showed a specific HSA workload, but again it was missed and/or ignored.

It seems that you also missed the news of last days. AMD announced something named MANTLE that will improve performance of CPUs by eliminating the DX draw call bottleneck. The MANTLE version of BF4 will run better in Kaveri than the DX version tested today.

Now say me, what part of this you will be ignoring in your next post? I would like to copy that part and save it for posting to you again the next time.

 

[juanrga]

The single core loaded excessively on the 8350 does speak toward potentially heavy FPU use in the main thread, which would explain the upshot in core usage for the first 2 cores, while the other cores are loaded relatively similarly to the Intel CPUs.

So, if you take all factors into consideration, the 2600k is just shy of bottlenecked, much like the 6300 is...where as the 8350 is likely suffering from the poorer FMAC performance of the PD architecture in the spike of core usage there. The FMAC is supposedly being tweaked for SR, IIRC, however, that may have been speculation from one of the many "theorists" about what is going on with SR.

3 pipe stages, improved cache, improved fetch, double decoder, and improved memory controller mean better FP performance than Piledriver.

 

[noob2222]

^^ But again: You lose portability, even across vendors on the same platform. Portability and coding efficiency was the reason we stopped using hardcoded assembly for crying out loud!

Sure, if I coded to the metal, in pure assembly, violated every good programming practice, I'm sure I could double the performance of every application I write. Until the next hardware revision changes one of my assumptions (and it can be something REALLY trivial too) ends up costing me more performance then I gain. Devs HATE going back and re-optimizing for a different arch, and that's what API's like MANTEL force you to do every time the hardware changes.

If you continue attacking MANTLE at least spell it right. Besides that

MANTLE != coding to the metal level
MANTLE != assembly

Also DEVS and the tech. press are loving MANTLE. The only critic voices that I am reading on forums are from people who is very pro Intel/Nvidia.

Also, I'm awaiting BF4's first benchmarks for the Beta, which should be hitting sometime today. So far, all we have are older Alpha benchmarks back from June, though they do show similar performance to Bf3:

https://linustechtips.com/main/topic/30410-battlefield-4-alpha-gpu-and-cpu-benchmarks/

Already getting ready for the "crappy PC port" arguments to start up again.

Ok

And as predicted FX-8 > FX-6 > FX-4. In fact the 8-core offers about the double performance of the 4-core which suggest that the game scales well up to 8 cores. People who believed that an i3 would be top for next gen was dreaming.

Also interesting the GPUs

TITAN performance is disappointing. Any bet which will be the performance of the new R9 with the MANTLE version of BF4? Some developer said it will humiliate the TITAN.

hmm 4300 +20% for kaveri = 40 fps average.
8350 = 60 fps,
i5 760 = 45fps
how is kaveri a smart move to abandon high end cpus again and going to reach any i5 performance?

This seems to be another of your attempts to ignore what I say/said.

I said that Kaveri 4C ~ FX-6 ~ i5 and that game shows that Kaveri 4C could compete with a FX-6300.

I already shown my predictions on how Kaveri will be faster than a i5-2500k in some benchmarks and slower than the i5-2500k in others. I think you missed that as well.

I also said that kaveri 4C cannot compete with a FX-8350. I said people could continue purchasing FX-8 and FX-9 during 2014 before Carrizo is ready (2015). Also you missed that. The only way that a Kaveri 4C can compete with a FX-8350 is using HSA software. I also said this and shown a specific HSA workload, but again it was missed and/or ignored.

It seems that you also missed the news of last days. AMD announced something named MANTLE that will improve performance of CPUs by eliminating the DX draw call bottleneck. The MANTLE version of BF4 will run better in Kaveri than the DX version tested today.

Now say me, what part of this you will be ignoring in your next post? I would like to copy that part and save it for posting to you again the next time.

you completely missed the point of AMD abandoning high end according to your self delusional expert opinion.

you also completely missed the point that your delusional and change your story constantly.

you missed the point that Mantle was just announced and you know more about it than anyone as if you ever even wrote a program. Thats not even mentioning that Mantle isn't APU specific as you just implied.

you also completely missed the point that most people here think your an idiot.

Now that your done with your stupid tirade, explain how abandoning high end for a middle quality part is a good idea?

Next time you want to attempt to insult me instead of answering a simple question, suck my $#@%. ANSWER THE @%$# QUESTION and quit actling like a child when you don't have the answer.

Aside from that I was giving you the benefit of a doubt by using the 4300. Truthfully, kaveri is based off richland performance, wich is actually 10-20% slower than the 4300 in games. http://www.ultimatehardware.net/amd/amd_a10_6800_vs_amd_fx_4300_page4.htm

so kaveri +20% performance -20% APU performance puts kaveri = 4300 FX or 34 fps. But since it wasn't on the BF4 test, I was being nice.

As a matter of fact :

trinity fps : 26.5 +20% for kaveri = 31.8 fps

so again ... answer the question.

 

[juanrga]

^^ But again: You lose portability, even across vendors on the same platform. Portability and coding efficiency was the reason we stopped using hardcoded assembly for crying out loud!

Sure, if I coded to the metal, in pure assembly, violated every good programming practice, I'm sure I could double the performance of every application I write. Until the next hardware revision changes one of my assumptions (and it can be something REALLY trivial too) ends up costing me more performance then I gain. Devs HATE going back and re-optimizing for a different arch, and that's what API's like MANTEL force you to do every time the hardware changes.

If you continue attacking MANTLE at least spell it right. Besides that

MANTLE != coding to the metal level
MANTLE != assembly

Also DEVS and the tech. press are loving MANTLE. The only critic voices that I am reading on forums are from people who is very pro Intel/Nvidia.

Also, I'm awaiting BF4's first benchmarks for the Beta, which should be hitting sometime today. So far, all we have are older Alpha benchmarks back from June, though they do show similar performance to Bf3:

https://linustechtips.com/main/topic/30410-battlefield-4-alpha-gpu-and-cpu-benchmarks/

Already getting ready for the "crappy PC port" arguments to start up again.

Ok

And as predicted FX-8 > FX-6 > FX-4. In fact the 8-core offers about the double performance of the 4-core which suggest that the game scales well up to 8 cores. People who believed that an i3 would be top for next gen was dreaming.

Also interesting the GPUs

TITAN performance is disappointing. Any bet which will be the performance of the new R9 with the MANTLE version of BF4? Some developer said it will humiliate the TITAN.

hmm 4300 +20% for kaveri = 40 fps average.
8350 = 60 fps,
i5 760 = 45fps
how is kaveri a smart move to abandon high end cpus again and going to reach any i5 performance?

This seems to be another of your attempts to ignore what I say/said.

I said that Kaveri 4C ~ FX-6 ~ i5 and that game shows that Kaveri 4C could compete with a FX-6300.

I already shown my predictions on how Kaveri will be faster than a i5-2500k in some benchmarks and slower than the i5-2500k in others. I think you missed that as well.

I also said that kaveri 4C cannot compete with a FX-8350. I said people could continue purchasing FX-8 and FX-9 during 2014 before Carrizo is ready (2015). Also you missed that. The only way that a Kaveri 4C can compete with a FX-8350 is using HSA software. I also said this and shown a specific HSA workload, but again it was missed and/or ignored.

It seems that you also missed the news of last days. AMD announced something named MANTLE that will improve performance of CPUs by eliminating the DX draw call bottleneck. The MANTLE version of BF4 will run better in Kaveri than the DX version tested today.

Now say me, what part of this you will be ignoring in your next post? I would like to copy that part and save it for posting to you again the next time.

you completely missed the point of AMD abandoning high end according to your self delusional expert opinion.

you also completely missed the point that your delusional and change your story constantly.

you missed the point that Mantle was just announced and you know more about it than anyone as if you ever even wrote a program.

you also completely missed the point that most people here think your an idiot.

Now that your done with your stupid tirade, explain how abandoning high end for a middle quality part is a good idea?

Next time you want to attempt to insult me instead of answering a simple question, suck my $#@%. ANSWER THE @%$# QUESTION and quit actling like a child when you don't have the answer.

 

[gamerk316]

Something else i found REALLY interesting:

Using a 770. FX-8350 slightly ahead of the 2600k, matching what we saw above. However:

Using a 7970. A FX-8350, at stock, falls behind the 2500k, also at stock.

Really, GPU choice shouldn't be affecting where each CPU is placing, just how close they are to eachother. Nothing like this...Something looks REALLY screwy with the 7970 numbers, so I'm hoping its a driver problem or issue with the Beta. Something to watch going forward...That being said, given how BF4 will use 2.5GB RAM or so, its possible memory throughput could be causing what we are seeing (though I doubt it...).

Still, goes to show, you need multiple benchmark sets to see what's going on.

 

[The Q6660 Inside]

Something else i found REALLY interesting:

Using a 770. FX-8350 slightly ahead of the 2600k, matching what we saw above. However:

Using a 7970. A FX-8350, at stock, falls behind the 2500k, also at stock.

Really, GPU choice shouldn't be affecting where each CPU is placing, just how close they are to eachother. Nothing like this...Something looks REALLY screwy with the 7970 numbers, so I'm hoping its a driver problem or issue with the Beta. Something to watch going forward...That being said, given how BF4 will use 2.5GB RAM or so, its possible memory throughput could be causing what we are seeing (though I doubt it...).

Still, goes to show, you need multiple benchmark sets to see what's going on.

Yeah, it looks a biit odd, chances are it is due to drivers, but we can never be so sure. It seems as if the 8350 actually does better with nVidia GPUs for some reason. It is safe to say, @1080p, 965BE gaming is DEAD.

 

[-Fran-]

Yeah, it looks a biit odd, chances are it is due to drivers, but we can never be so sure. It seems as if the 8350 actually does better with nVidia GPUs for some reason. It is safe to say, @1080p, 965BE gaming is DEAD.

Even more... Notice how the 965 scales crappy to 4Ghz, while the FX4350 (a faster 4300) jumps pass it easily. That's a nice surprise, even though I agree that the graph looks suspicious against the 770 graph where the 965 is still faster, which seems about right. Still, for the price, the 4300 looks like a real bargain and the 6300 even more tempting for a new build.

Cheers!

 

[Cazalan]

Well December is just 2 months away now. Hopefully there will be more leaks. AMD's next earnings announcement is on October 17th. There should be at least some official news on Kaveri but probably not before then.

My main concern with Kaveri is clock speeds on the Bulk process. Their core design fixes themselves should be the biggest bump in performance AMD has seen in a long time. I'm not expecting anything stellar with only 4C but budget buyers could be excited by a lower cost part due to the bulk process. Maybe even $100 for ~6300ish performance.

 

[blackkstar]

Something else i found REALLY interesting:

Using a 770. FX-8350 slightly ahead of the 2600k, matching what we saw above. However:

Using a 7970. A FX-8350, at stock, falls behind the 2500k, also at stock.

Really, GPU choice shouldn't be affecting where each CPU is placing, just how close they are to eachother. Nothing like this...Something looks REALLY screwy with the 7970 numbers, so I'm hoping its a driver problem or issue with the Beta. Something to watch going forward...That being said, given how BF4 will use 2.5GB RAM or so, its possible memory throughput could be causing what we are seeing (though I doubt it...).

Still, goes to show, you need multiple benchmark sets to see what's going on.

How many times must I tell you that "our game engine uses 8 threads" doesn't mean it uses 8 threads all the time?

Congratulations, you've found a benchmark where 6 to 8 threads are being used and then you found another one where less threads are used.

However, I'm docking points because you are yet again trying to use this evidence to imply that AMD can not make up for its lack of single threaded performance by making a rig with more weaker cores.

The problem with that is that AMD is forcing game developers to go for many weak cores by getting XBone and PS4 CPUs with 8 weak cores. You're also forgetting that this is a massive shift in gaming and that big shifts like this take time to fully adjust.

I'm going to repeat myself, but this happens every console generation and with with console ports. Developers are still learning how to get the most out of the system, so it has problems. The problem with this generation of consoles will definitely be game engines which don't scale to as many cores as possible as much as possible.

I'm going to repeat myself again, since you seem to fail in reading comprehension, but there's basically this big bounty out there where whoever writes the best performing game engine that can scale the best to the number of weak cores available in XBone/PS4 is going to rake in massive cash. The company that does this first is going to have the engine for people to develop games on and they're going to make a ton of money off of licensing their engine out.

So I don't really understand why you think the game industry is going to sit passive writing games for old consoles and old PC CPUs when there's a ton of money in licensing fees.

Why do you think DICE is embracing Mantle with Frostbite 3? Because the game developers who use Frostbite 3 and get Mantle support in their console games are going to slaughter the competition using Xbone DX. Suddenly, DICE/EA has the game engine to go to for beautiful XBone and PS4 games and everyone else is behind.

This is how it works from the business side, but I have a feeling that you're one of those "self declared game developers" that the internet is full of, which conjures images of the claimant working on AAA titles, but "game developer" can be as low as writing a game in TI-BASIC for TI-83 calculators.

 

[8350rocks]

Just heard Tom clancy has died at the age of 66. Farewell and RIP.

Wow, I can say I never saw that coming, not right now anyway. I read many of his books, LOVED some of the movies. The Hunt for Red October will always be one of my favorites...Rainbow Six was a great series to read as well.

That's pretty crazy...