amd fx cpus were future proof...

[TheELF]

When scientific computing (which is what we've been discussing for the last 2 days over several threads) is nearly always parallel workloads.

I think you are mixing up the threads you are responding to.
In this thread OP talked about DX12/mante, so gaming not scientific computing,and almost the whole of the first page is the typical intel vs amd and lot vs strong cores slander that you get on all such threads.

And I didn't tell you what you can and can't use,I told you that you can't call someone heavily biased and then go and refer to a benchmark that focuses purely on parallel workloads because that's heavily biased asswell.

 

[BSim500]

However, we should be cognizant of the fact that Star Swarm is not a real game either, so the results may be different in real games.

^ This +1,000,000. I remember seeing exactly the same "+200% miracle gain" Star Swarm charts last January when in reality Mantle turned out to be more like only a 5-30% in most actual games:-
http://www.maximumpc.com/pitting_amd’s_mantle_against_direct3d_11_2014
http://www.hardwarepal.com/dragon-age-inquisition-benchmark-mantle-vs-directx/

DX12 probably will help the slower CPU's (AMD's + i3's) to some extent, but anyone using Star Swarm as a predictor of real game performance boosts for DX12 is being more than a bit naive after its similar "triple your fps with Mantle" charts barely 13 months ago...

 

[NTMBK]

Star Swarm is a visual benchmark where the only real CPU load is the thousands of draw calls; in a real game, there is CPU load from the actual game engine logic. I won't jump to any conclusions until I see results from actual DirectX 12 games (which I expect will look a lot like the results from Mantle games).

 

[TheELF]

DX12 probably will help the slower CPU's (AMD's + i3's) to some extent, but anyone using Star Swarm as a predictor of real game performance boosts for DX12 is being more than a bit naive after its similar "triple your fps with Mantle" charts barely 13 months ago...

Basically it will help the ps4/xbox where it will free up a whole core more or less,so that developers will be able to make games with more stuff going on.

This is based on the fact that mantle's main accomplishment is getting rid of the driver thread which for most desktop CPUs is not all that big,so no big improvements there.

Star swarm is a demo where they are only doing draw calls, so more cores and/or stronger cores gives you even more draw calls,in a real game you will have to be able to do something more then just watch, you want to be able to control things and at that moment the faster core will give you more speed.

 

[Gikaseixas]

I use both Intel i7 which i love and AMD FX which i love too
Ofcourse Intel is faster and better more often than not but the FX is no slouch, it's an all round good chip only let down by the ancient mobo platform (I've said it many times)
People using power consumption and heat are really just using these KPIs to justify their cause and nothing more.

As for the poll in question, the answer is no simply because it just matches Intel in DX12, it doesn't beat (in the bench used in OP). Being futureproof is being able to outlast the competition and what we're seeing here is A10 and i3 trading blows whereas before the i3 simply killed it. So if AMD is futureproof, so is Intel for being able to stay ahead

Zen has a huge mountain to climb due to the fab advances Intel has in their favour. The true enthusiast will be rooting for it to succeed since having more choices is better than just having a single option.

 

[ShintaiDK]

Basically it will help the ps4/xbox where it will free up a whole core more or less,so that developers will be able to make games with more stuff going on.

This is based on the fact that mantle's main accomplishment is getting rid of the driver thread which for most desktop CPUs is not all that big,so no big improvements there.

Star swarm is a demo where they are only doing draw calls, so more cores and/or stronger cores gives you even more draw calls,in a real game you will have to be able to do something more then just watch, you want to be able to control things and at that moment the faster core will give you more speed.

I wouldnt expect any miracles on the consoles. MS also said this for the Xbox One. And PS4 already uses libgcm.

Another example even on DX11 could be Tomb Raider with the infamous 800 vs 4700Mhz Haswell. In the benchmark they score almost the same. But ingame its a factor ~4x difference.

 

[AtenRa]

^ This +1,000,000. I remember seeing exactly the same "+200% miracle gain" Star Swarm charts last January when in reality Mantle turned out to be more like only a 5-30% in most actual games:-
http://www.maximumpc.com/pitting_amd’s_mantle_against_direct3d_11_2014
http://www.hardwarepal.com/dragon-age-inquisition-benchmark-mantle-vs-directx/

DX12 probably will help the slower CPU's (AMD's + i3's) to some extent, but anyone using Star Swarm as a predictor of real game performance boosts for DX12 is being more than a bit naive after its similar "triple your fps with Mantle" charts barely 13 months ago...

Slower CPUs gained tremendously with Mantle, same will happen with DX-12.
For example, FX6300 is faster in Thief with Mantle than Core i7 3770K with DX-11 both in minimum and average fps. Even Core i7 920 at 2.66GHz is almost as fast as 3.5GHz Core i7 3770K.

http://forums.anandtech.com/showpost.php?p=36301071&postcount=136

 

[AtenRa]

As for the poll, i have voted yes. With Mantle and DX-11 the FX83xx is becoming better all the time even in games with low thread performance like Thief.

 

[svarog19]

People that voted "hell no" should be ashamed of them self for being out of touch with reality when we see that Mantle and DirectX 12 API gives considerably more boost in performance than compared to Intel.

Mantle and DirectX 12 API off-loaded some of the work and better utilized CPU's and we can see from gains that AMD's hardware was being better utilized.

 

[PC Master Race]

As for the poll, i have voted yes. With Mantle and DX-11 the FX83xx is becoming better all the time even in games with low thread performance like Thief.

Still even with mantel optimizations that;s just putting AMD FX 8 Core on par with an Intel Core i3 as far as minimum framerates go.

 

[ShintaiDK]

Thief uses an ingame benchmark tool that is prescripted. So the benchmarks are pretty much useless CPU wise.

 

[svarog19]

Still even with mantel optimizations that;s just putting AMD FX 8 Core on par with an Intel Core i3 as far as minimum framerates go.

Developers have mastered usage of dual cores and quad core CPU's...

Hex core CPU's aren't being utilized to the fullest and Octa core's a far from being utilized fully while Intel has majority of x86 market share then developers put most of the effort for optimizing games for Intel CPU's and using Intel's compilers.

 

[frozentundra123456]

Slower CPUs gained tremendously with Mantle, same will happen with DX-12.
For example, FX6300 is faster in Thief with Mantle than Core i7 3770K with DX-11 both in minimum and average fps. Even Core i7 920 at 2.66GHz is almost as fast as 3.5GHz Core i7 3770K.

http://forums.anandtech.com/showpost.php?p=36301071&postcount=136

In this particular case, mantle allowed FX 6300 to reach similar performance to 3770k because the game became gpu limited. With more powerful gpus or in a different game, the results would be different. Not to mention that by the time DX 12 becomes mainstream we will be on Skylake and nearing 10nm for intel, while AMD will still be soldiering on with basically the same chips. So we will have to wait and see, but one cherry picked example certainly does not prove that AMD will catch up to intel in gaming performance.

 

[PC Master Race]

Developers have mastered usage of dual cores and quad core CPU's...

Hex core CPU's aren't being utilized to the fullest and Octa core's a far from being utilized fully while Intel has majority of x86 market share then developers put most of the effort for optimizing games for Intel CPU's and using Intel's compilers.

So then why are FX 4300 quad core CPUs falling flat on their face and getting creamed by Dual Core Intels ?

 

[cbn]

Slower CPUs gained tremendously with Mantle, same will happen with DX-12.
For example, FX6300 is faster in Thief with Mantle than Core i7 3770K with DX-11 both in minimum and average fps. Even Core i7 920 at 2.66GHz is almost as fast as 3.5GHz Core i7 3770K.

http://forums.anandtech.com/showpost.php?p=36301071&postcount=136

I see the Pentium G3240 didn't gain anything in terms of avg FPS in Mantle for Thief. That is interesting.

But some of the weaker processors like Q9450 with better a higher multi-threading to single thread ratio like the Q9450 made ~50% gains in avg FPS.

Looking ahead it will be interesting to see how DX12 compares to these results and what type of processor comparisons show the greatest gains. (Logic follows processors that have some degree of untapped multi-threading will show the largest gains, but I wonder where the threshold will be)

 

[PC Master Race]

In this particular case, mantle allowed FX 6300 to reach similar performance to 3770k because the game became gpu limited.

LOL those graphs you posted showed the AMD FX CPU leading only in power consumption. The FX CPUs as shown in the graphs you posted are still behind the Intel Core i3 in performance.

 

[PC Master Race]

But some of the weaker processors like Q9450 with better a higher multi-threading to single thread ratio like the Q9450 made ~50% gains in avg FPS.

Hes try's to claim the 6300 was better performance than the i7 3770K but yet the FX 6300 was still behind the performance of a Core i3 LOL. Troll detected.

 

[cbn]

Looking at AtenRA's Thief results again it looks like the two AMD processors stand out the most are the FX-6300 and FX-8350:

FX-6300: 46.70 avg FPS to 70.50 avg FPS (51% gain)
FX-8350: 48.50 avg FPS to 72.20 avg FPS (49% gain)

The quad cores still did very well though:

A10-5800K: 43.20 avg FPS to 61.10 avg FPS (41% gain)
A10-7700K: 43.80 avg FPS to 60.70 avg FPS (38% gain)
Athlon x4 750K: 39.90 avg FPS to avg 55.20 FPS (38% gain)

 

[TheELF]

I see the Pentium G3240 didn't gain anything in terms of avg FPS in Mantle for Thief. That is interesting.

It was already at 50FPS average with DX11 ,faster than/on par with the fx-8xxx
it's obvious that there is a threshold, and that to get over 50fps you need fast enough cores but also enough of them,and only the I3 and the I7 fulfill this quota.
The pentium also gained 10FPS on the minimum which is nice enough of a gain the moment the CPU is already that fast.

 

[cbn]

Hes try's to claim the 6300 was better performance than the i7 3770K but yet the FX 6300 was still behind the performance of a Core i3 LOL. Troll detected.

He mentioned the FX-6300 in Mantle was faster than i7-3770K in DX11.

However, it is just amazing to me how Intel's Core i3 (even with these Mantle results) continues to be the standard by which AMD must beat.

 

[BSim500]

I see the Pentium G3240 didn't gain anything in terms of avg FPS in Mantle for Thief. That is interesting.

Thief is rumored to have some fairly broken code that results in i5's scoring higher than i7's (which the above i5-less charts don't show). It's rumored to be one of those games that "hard codes" certain threads to arbitrary core numbers resulting in threads being assigned to HT cores before the real cores are used. Lower frame-time (in ms) charts with i5 vs i7 also reveal a similar effect (ie, the i5-4670K has 15% lower frame-times than the i7-4770K and less than half that of the FX's).

 

[poofyhairguy]

He mentioned the FX-6300 in Mantle was faster than i7-3770K in DX11.

However, it is just amazing to me how Intel's Core i3 (even with these Mantle results) continues to be the standard by which AMD must beat.

Yeah makes you wonder if they will ever pass Intel's 2500k from 2011 (especially when you count a modest over clock). I mean without going to 32 cores or something nuts.

 

[PC Master Race]

He mentioned the FX-6300 in Mantle was faster than i7-3770K in DX11.

However, it is just amazing to me how Intel's Core i3 (even with these Mantle results) continues to be the standard by which AMD must beat.

3fps LOL that's within margin of error of performance parity ... That said did the Intel plarfoem not recive a boost as well in Mantle ? Lets use apples to apples here. AMD FX is still a slug and will only get even worse with time because of dead platform. On a postive note it's nice to see that when under Mantle AMD FX is doing pretty well as good as a Core i3.

 

[MiddleOfTheRoad]

I think you are mixing up the threads you are responding to.
In this thread OP talked about DX12/mante, so gaming not scientific computing,and almost the whole of the first page is the typical intel vs amd and lot vs strong cores slander that you get on all such threads.

And I didn't tell you what you can and can't use,I told you that you can't call someone heavily biased and then go and refer to a benchmark that focuses purely on parallel workloads because that's heavily biased asswell.

Not really. Shintai made a blanket statement that the FX was obsolete at debut.... I simply pointed out there are still several tasks that FX chips still do extremely well nearly 3 years after their release.

Anyone who has run Virtual Machines, Rendering or Scientific Computing on an AMD FX knows that Shintai's idea of obsolete is downright hilarious. It really is a warped sense of reality.

 

[MiddleOfTheRoad]

Hes try's to claim the 6300 was better performance than the i7 3770K but yet the FX 6300 was still behind the performance of a Core i3 LOL. Troll detected.

Talk about the pot calling the kettle black.