Who here has Metro 2033 and an FX-83## or FX-81##

[humbug]

And and AMD GPU, preferably similar to mine, or any will do.

I want to compare a Metro 2033 Game bench with Advanced Physx.

 

[humbug]

But it will still be inferior to an Nvidia setup seeing as Physx in Metro is an Nvidia implemented feature.

It also runs off the CPU if you don't have an Nvidia GPU.

 

[humbug]

ok

 

[humbug]

That's with a 7970 and an 8150, which is running at 4.1Ghz with 4.4/4.6 turbo.

Looks like my results are pretty poor compared.

Yeah your MIN and Avg FPS are some 30% lower than mine.

That's really not good

Thanks for your contribution.

I'm still looking for some FX-83## results.

 

[humbug]

Thanks ^^^^ but its the CPU performance i'm interested in, your running Physx off your GTX 680 SLI, not the CPU

You might be interested in this http://forums.overclockers.co.uk/showthread.php?t=18451929

 

[humbug]

I dont see why you are so interested in this, I doubt it runs very well. Nvidia have made sure of that.

By the time there are cpu's fast enough to run physx, we will have even faster gpu's or no cpu's at all.

Actualy so far it runs perfectly fine, i run it in game and its smooth enough.
I'm interested to see is the FX-83## can deal with it better than mine, as a measure of its performance.

 

[humbug]

I can turn off Phsyx on the GPU and do another run.
Just went into the NV control panel and set Physx to CPU.

Performance as expected is terrible running Physx on the CPU.
Minimum frame rates well under 10 and averages in the 30's why would you want that.

I don't know if that's any better than me or not.

 

[humbug]

A comparison with your GPU's to that would be good but its impossible to compare Avg FPS when your running SLI GTX 680 against a 7870.

But your Min and Avg FPS are lower than mine, yet higher than X3T who is running an FX-8150 @ 4.1 / 4.4Ghz with a 7970 @ 1070 / 1400 which is not a to big gain from mine at 1200 / 1500

 

[humbug]

Completely unscientific, a complete mess of GPU variables but the results still speak for themselves.

 

[humbug]

Physx runs in x87 on the CPU, PD has to do x87 in software emulation.............and you were expecting what exactly?

Ah... i did not know Physx runs on x87

 

[humbug]

I think humbug does

Nothing wrong with Physx, but its no big deal to me, its a few extra dust particles, a few extra sparks, if i can run it i will, if not i wont miss it.

I used it to compare FX-83## CPU performance to mine, but since AMD don't support x87 the comparison is out of the window.

Besides, it does not run well on Intel either, so there is no point in one or the other.

 

[humbug]

Actually it runs quite well on the Intel chips.
I had no trouble averaging in the 70's on the test with a 3820

CPU Physics

I can see from that chart that your FPS dropped to similar levels and in the same place as i have, the reason you have much higher Avg FPS is because your running two GTX 680's vs my one 7870. (hence 400 FPS MAX)

Its not performing any better than my CPU in this, put them side to side and look.

Let me show you something http://www.tomshardware.co.uk/nvidia-physx-hack-amd-radeon,review-32051-5.html

Performing these benchmarks with a powerful main graphics card and a dedicated PhysX card was a deliberate choice, given that a single Nvidia card normally suffers from some performance penalties with GPU-based PhysX enabled. Things would get quite bad in this already-GPU-constrained game. In this case, the difference between CPU-based PhysX on a fast six-core processor with well-implemented multi-threading and a single GPU is almost zero.

Phenom II x6 1090T @ 4Ghz; CPU Physx = 32 FPS

GPU Physx 39 FPS

That X6 1090T is running @ 4Ghz and probably stock memory / CPU-NB, by comparison i'm running 1700Mhz memory, 3000Mhz CPU-NB and 4.1Ghz, my 1090T runs circles around that one at my clocks.

In this game the difference between the Nvidia GPU and the Phenom II 6 core for Physx is not huge, there is literally just a few FPS in it.

Metro 2033 it very GPU, because of that switching between GPU and CPU Physx makes little difference when using a decent CPU.

Either way its playable, acceptably smooth, on my CPU and i have no doubt on yours to.

 

[humbug]

Stop trying to draw a conclusion that suits you.

Simple fact.
I used 2 GPU's, ran on AMD average frame rate low 30's
Same 2 GPU's ran on Intel average frame rate low 70'

You have a bottleneck in your GPU that is masking the weakness in your CPU.
I simply removed the bottleneck and allowed the CPU to be tested.

Which after all was your intention.

No, your running two GTX 680's vs my one 7870 and tried to claim your avg FPS were so much higher because of the Intel CPU. Its utter garbage.

I just proved it isn't with your own data, the proof is in the graphs, anyone with half a mature brain and no agenda can see that as plain as day. your performance increase is purely down to your GTX 680 SLI, its taken the Avg FPS from 3 time more GPU grunt from when when your CPU is not bottle-necking those GPU's at the end of the bench.

Other the the high end FPS your PFS are the same as mine (in the Graph) you run your bench again only this time with the equivalent of a 7870 and the results will look just like mine.

Get off your high horse kid

 

[humbug]

Erm.... Humbug..... His post makes it sound like he's used his FX8320 with the 680's too for this.

You know, since he has an FX8320 and says ;

Why are these guys only getting 30 and 40 Avg FPS on overclock 2500K and 3770K?

http://forums.overclockers.co.uk/showpost.php?p=22993838&postcount=64

http://forums.overclockers.co.uk/showpost.php?p=22998343&postcount=107

Is his Intel CPU 2x more powerful?

 

[humbug]

I don't really know or care, has nothing to do with what I posted.
But you went off on the guy for absolutely no reason whatsoever, it's not his fault you're unable to read he used his 680 SLI on both his AMD set up and Intel set up.

His Intel setup is getting 2x more Avg FPS than other more powerful Intel setup's.

 

[humbug]

How's a 680 SLI and higher clocked 3820 slower than a 2500k and 680? (Your first link).

3770K @ 4.8Ghz 7970 = 40 FPS Avg.

performance difference between 3820 and 3770K http://www.anandtech.com/bench/Product/523?vs=551

The 3770K @ 4.8Ghz is 15 FPS faster than me.

He is running a slower CPU + GTX 680 SLI and getting 70 FPS.

If that 3770K @ 4.8Ghz was also running GTX 680 SLI what do you think his Avg FPS would be?

 

[humbug]

Meh.

Exactly. His 70 FPS is because of the GTX 680 SLI, not his CPU.

 

[humbug]

I think you're forgetting where he used the 680's with the FX8320...

I really can't be fussed with this, all you'll do is move the goal posts, ignore something, make another mistake, misread another thing and so on and so forth.
I mean, you've done most of that this thread already.

I'm not forgetting it, if you look back my Thuban is about 40% faster than that FX-83## on a much slower GPU, that says more about the FX CPU than his CPU and mine.

 

[humbug]

I didn't realise i would have to spoon feed humbug but here goes.

The test you are running is hard on GPU's
A single GPU such as a 680 or 7970 will only average high 40's. maybe low 50's on a highly overclocked card.

Now you wanted to test CPU Physx.

I did the test using 2 cards which are capable of 90+ frames per second.
The result when using the CPU to do the Physx, a miserable 30 FPS with the FX 8320.

For comparison i used an Intel 3820 running at the same clock speed as the FX 8320 4.6GHz.
Actually 4625 because you have to use the 125 strap as they have only partially unlocked multis.

That when doing the Physx was able to sustain a much higher FPS as i showed over 70.

Now had i used one card the result would have been in the 40's.
The GPU would have been the throttle not the CPU.

Understand now ?

One GPU and you will not see the max Physx capability of an Intel CPU using your test settings.

Your turning in to a joke now.

Put the chart up for that test and i will show you something, something you already know.

You; http://forums.overclockers.co.uk/showpost.php?p=23096540&postcount=11

Did a metro run on my FX 8320. 4.6Ghz
680 SLI

So the FX-83## works fine without Physx.

I can turn off Phsyx on the GPU and do another run.
Just went into the NV control panel and set Physx to CPU.

Performance as expected is terrible running Physx on the CPU.
Minimum frame rates well under 10 and averages in the 30's why would you want that.

FX8320 4.6Ghz Physx running on CPU

You see that? see how flat that is? the FX-83## can't deal with Physx.

It also runs off the CPU if you don't have an Nvidia GPU.

Thuban can. its looks just like your bench with GPU Physx.

If you put up the single GTX 680 on the FX-83## CPU Physx i bet you it still looks exactly like the bench you did in FX-83## + SLI GTX 680, and nothing like mine.

Whats more your Intel + CPU Physx 'does' look just like mine, including the 10 / 15 FPS bit in the middle.

Actually it runs quite well on the Intel chips.
I had no trouble averaging in the 70's on the test with a 3820

CPU Physics

Can you see where the CPU bottleneck is for both our CPU's? look at it again with GPU Physx vs CPU Physx.

Your 70 FPS is absolutely nothing at all to do with your Intel CPU, your 'averaging' 70 FPS because of higher FPS with SLI GTX 680's vs my one 7870 out side of the CPU bottleneck.
Your Intel CPU bottleneck part is every bit a prevalent as it is with mine, its not better in this situation, this is confirmed by 2 other completely independent Intel CPU benches which have similar results as me. (you yourself have similar results to me with just one GTX 680 on your Intel i7)

One GPU and you will not see the max Physx capability of an Intel CPU using your test settings.

No, your FX CPU can't deal with Physx