Overclock.net banner
41 - 60 of 64 Posts

· Registered
Joined
·
1,238 Posts
If you are running at 3.0ghz...then yes, you are bottlenecking the gpu. If you could overclock to around 3.8 the bottleneck is alot less. Games love ghz...it doesnt matter how many cores you have (99% of the games anyways). Try to overclock if you can
 

· Premium Member
Joined
·
6,680 Posts
Quote:
Originally Posted by GrandMax;14227350
I have two 6870s with a 1055t at 3.5, which is 50-70% faster than your 480.

The only games I ever been bottlenecked were Crysis 1 and GTA IV cause they use only one thread.

DX11 allows rendering on multiple threads so the 1055t is futureproof.
Crysis uses 2 threads AFAIK and GTA4 uses at least 4 threads.

DX11 has nothing to with allowing multiple threads. We've had multithreading for ages.
 

· Premium Member
Joined
·
15,625 Posts
There's so much bad information in this thread that it's not even funny. His cpu is properly overclocked and will easily outperform any PHII or Thuban at 4ghz with a stock cpu-nb. Your setup is fine and isn't behind by any significant amount of a SB/GTX480 combo.
 

· Padawan Overclocker
Joined
·
2,868 Posts
Quote:
Originally Posted by B!0HaZard;14239002
Crysis uses 2 threads AFAIK and GTA4 uses at least 4 threads.

DX11 has nothing to with allowing multiple threads. We've had multithreading for ages.
Lol non-scaling multithreading means nothing. There's no multi-threading. There's 2-thread optimized or maybe very rare 3/4-thread optimized games. They never intented to develop multithreading as they see current cpu speeds are enough to utilize most gpus.

When you see below %50 cpu or %50 gpu utilization, i think we can say GFX engine is the bottleneck
biggrin.gif


The truth is programmers focus so much on compatibility that games can not take advantage of anything because the architectures are so different.
 

· Premium Member
Joined
·
6,680 Posts
Quote:


Originally Posted by Catscratch
View Post

Lol non-scaling multithreading means nothing. There's no multi-threading. There's 2-thread optimized or maybe very rare 3/4-thread optimized games. They never intented to develop multithreading as they see current cpu speeds are enough to utilize most gpus.

When you see below %50 cpu or %50 gpu utilization, i think we can say GFX engine is the bottleneck


The truth is programmers focus so much on compatibility that games can not take advantage of anything because the architectures are so different.

No, we can't, that makes no sense. If my CPU utilization is below 50% because my GPU is crap then the engine is not at fault. If my GPU utilization is below 50% because my CPU is crap then the engine is not at fault.

Also, WTH do you mean by non-scaling multithreading? If it's multithreaded, it's using more cores, so logically, it would be performing worse on fewer, equally fast cores.
 

· Registered
Joined
·
5,847 Posts
Discussion Starter · #47 ·
Quote:


Originally Posted by yoyo711
View Post

You get the Nd-14 cooler and push your cpu overclocks up

I'm going to get the Corsair H80 as the reviews say that it does better than a NH-D14
 

· Registered
Joined
·
431 Posts
Quote:


Originally Posted by B!0HaZard
View Post

Crysis uses 2 threads AFAIK and GTA4 uses at least 4 threads.

DX11 has nothing to with allowing multiple threads. We've had multithreading for ages.

Not for rendering. We have had multiple threads for physics, AI and many other stuff.

Quote:


Direct3D 11 was released as part of Windows 7. It was presented at Gamefest 2008 on July 22, 2008 and demonstrated at the Nvision 08 technical conference on August 26, 2008.[15][16] AMD previewed working DirectX11 hardware at Computex on June 3, 2009, running some DirectX 11 SDK samples.[17] Its features include:

Tessellation - to increase at runtime the number of visible polygons from a low detail polygonal model
Multithreaded rendering - to render to the same Direct3D device object from different threads for multi core CPUs
Compute shaders - which exposes the shader pipeline for non-graphical tasks such as stream processing and physics acceleration, similar in spirit to what OpenCL, Nvidia CUDA, ATI Stream achieves, and HLSL Shader Model 5 among others.[15][16]

Source: http://en.wikipedia.org/wiki/Direct3D#Direct3D_11

My point is that DX11 allows to use all six cores of the 1055t. It won't bottleneck a 480 nor 2 x 6870. The per core performance is not very good, but the 6 threads compensate partially for that. Don't get me wrong, it is not close to a sandybridge.
 

· Premium Member
Joined
·
6,680 Posts
Quote:


Originally Posted by GrandMax
View Post

Not for rendering. We have had multiple threads for physics, AI and many other stuff.

Source: http://en.wikipedia.org/wiki/Direct3D#Direct3D_11

Oh, I see, I must've missed the rendering part of your post.
 

· Banned
Joined
·
9,061 Posts
Quote:
Originally Posted by GrandMax;14275348
My point is that DX11 allows to use all six cores of the 1055t. It won't bottleneck a 480 nor 2 x 6870. The per core performance is not very good, but the 6 threads compensate partially for that. Don't get me wrong, it is not close to a sandybridge.
Except almost no DX11 game uses 6 cores (in fact outside of the Dirt Series I'm not sure if I know of any that do), as far as I can tell Crysis 2 for instance uses 3 cores, with the third thread being very weak.

The other problem is that Phenom II integer performance is quite slow, and the physics performance is as well.

An example of part of that can be seen in 3Dmark11's physics test which is multithreaded and runs on all 6 cores maxed out.

A 1090T @ 4,252MHz scores 7273 points in the physics test

While a i5-2500k @ 5,264MHz scores 10414 in the same test.

That is a rather large difference in performance despite the 1090T having 50% more cores which are used to their fullest in that test.
 

· Registered
Joined
·
431 Posts
Quote:
Originally Posted by sleepergsr;14227564
i highly doubt it. can we see some proof?
Here is a good example in DX11.

CPU usage = 60-70%
GPU = 100% on both

View attachment 220544

As far as my experience goes: The GPUs are still the limiting factors in most of my recent games. It is almost never my 1055t at 3.5. In other words, at maximum fps, my cpu will not be at full capacity/usage but my cards will.

And that has been the case in Dirt 3, Heaven 2.5, BFBC2, Portal 2 with SSAA 8x, Far Cry 2, Mass Effect 2 with SSAA 4x, Metro 2033, Shogun 2, L4D2 with 4x SSAA, etc... I know those are either DX11 or at a crazy high res due to SSAA but this is still what I play. Crysis 2 is also behaving the same.

Edit: And Crysis can use up to 8 cores.

http://translate.google.com/translate?js=n&prev=_t&hl=en&ie=UTF-8&layout=2&eotf=1&sl=auto&tl=en&u=http%3A%2F%2Fwww.pcgameshardware.de%2Faid%2C768480%2FCrysis-2-Alles-ueber-DirectX-11-3D-ohne-Leistungsverlust-und-8-Kern-Optimierung%2FAction-Spiel%2FTest%2F
 

· Premium Member
Joined
·
5,100 Posts
Quote:
Originally Posted by BallaTheFeared;14276605
Except almost no DX11 game uses 6 cores (in fact outside of the Dirt Series I'm not sure if I know of any that do), as far as I can tell Crysis 2 for instance uses 3 cores, with the third thread being very weak.

The other problem is that Phenom II integer performance is quite slow, and the physics performance is as well.

An example of part of that can be seen in 3Dmark11's physics test which is multithreaded and runs on all 6 cores maxed out.

A 1090T @ 4,252MHz scores 7273 points in the physics test

While a i5-2500k @ 5,264MHz scores 10414 in the same test.

That is a rather large difference in performance despite the 1090T having 50% more cores which are used to their fullest in that test.
1 word for this. "Architecture"
 

· Banned
Joined
·
9,061 Posts
Quote:
Originally Posted by GrandMax;14276630
Here is a good example in DX11.

CPU usage = 60-70%
GPU = 100% on both

Edit: And Crysis can use up to 8 cores.
Heaven is a gpu benchmark. Do you see any physics? Any AI? Any thing other than simple image rendering? No? Probably because its a gpu test, right?

My cpu usage is almost non existent in Heaven, in fact it down clocks to power saving mode during most of the test.

CryEngine may use up to 8 cores, that doesn't mean Crysis 2 does. I've done several tests in SLI and came to the conclusion it is dual threaded with third thread that is independent of rendering and much weaker than the other two in terms of demand.

Better read your link bro.
 

· Registered
Joined
·
2,002 Posts
Quote:
Originally Posted by GrandMax;14276630
Here is a good example in DX11.

CPU usage = 60-70%
GPU = 100% on both

View attachment 220544

As far as my experience goes: The GPUs are still the limiting factors in most of my recent games. It is almost never my 1055t at 3.5. In other words, at maximum fps, my cpu will not be at full capacity/usage but my cards will.

And that has been the case in Dirt 3, Heaven 2.5, BFBC2, Portal 2 with SSAA 8x, Far Cry 2, Mass Effect 2 with SSAA 4x, Metro 2033, Shogun 2, L4D2 with 4x SSAA, etc... I know those are either DX11 or at a crazy high res due to SSAA but this is still what I play. Crysis 2 is also behaving the same.

Edit: And Crysis can use up to 8 cores.

http://translate.google.com/translate?js=n&prev=_t&hl=en&ie=UTF-8&layout=2&eotf=1&sl=auto&tl=en&u=http%3A%2F%2Fwww.pcgameshardware.de%2Faid%2C768480%2FCrysis-2-Alles-ueber-DirectX-11-3D-ohne-Leistungsverlust-und-8-Kern-Optimierung%2FAction-Spiel%2FTest%2F
try running 3dmark or vantage and report back to us. lets see if you can still get the same results. =)
 

· Registered
Joined
·
773 Posts
Overclock your NB @ 2600/2800 MHz.
 

· Registered
Joined
·
5,847 Posts
Discussion Starter · #57 ·
Quote:
Originally Posted by sleepergsr;14277040
try running 3dmark or vantage and report back to us. lets see if you can still get the same results. =)
Quote:
Originally Posted by Benz;14279841
Overclock your NB @ 2600/2800 MHz.
I'll see what I can do myself, I will give 3DMark 11, Vantage and Heaven and will post some links up with my scores for your curiosity.

Note, that I'm using the AMD Stock cooler, so it hinder's my ability to get 3.7GHz and over.
 
41 - 60 of 64 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top