AMD Zambezi news, info, fans !

Printable View

Show 100 post(s) from this thread on one page

10-02-2011, 01:50 AM
Olivon

Quote:

Originally Posted by muziqaz

Olivon, when you have your own system with 3.8ghz thuban with stock 6970 doing such low numbers, then maybe I would believe. I don't really care what some website got on their system. I had fraps recording gameplay while I was playing it was 55-60fps. If it was 31fps min I would have noticed it with jerky gameplay. Too bad I deleted that POS of a beta from my PC :/

I got a GTX 470 running 850MHz and I do around 35-40fps on max settings. The 6970@stock give 70% better FPS :confused:

I don't think so ... :p:
10-02-2011, 02:18 AM
muziqaz

Olivon, sorry that I cannot verify my numbers, as I did not record fps with fraps. But I am 100% sure I did not see number 3 in my corner of the screen not even once :)
10-02-2011, 02:28 AM
Mechanical Man

Quote:

Originally Posted by Mechanical Man

No, you saw all CORES used, there is lot more threads in there.

Windows switches thread from core to other core, that is why it does not show 4*100% & 2*5% load on taskmanager when game uses only four cores worth of heavy threads.

ps. THERE IS NO ULTRA IN BF3 BETA. HIGH IS MAX. yes u can select ultra, but it does not change anything compared to high.

Yeah, ~60% cpu usage for 1024*768 all MED + no AA, 8*Ani. Filter, no ambient occlusion.
10-02-2011, 02:42 AM
imamage

Quote:

Originally Posted by liberato87

http://www.overclock.net/15136591-post6839.html

I guess I can just save the money and wait for Trinity then
wasted the budget to grab Crosshair V :( (Well the board is fun to OC with)
10-02-2011, 03:38 AM
gosh

Quote:

Originally Posted by Mechanical Man

No it does not scale over 4.

I have 1100T and i can easily look at task manager showing usage of ~66% That would be easily calculated to 4 cores at max usage. I have DX11 gpu. I'll do some tests on lower resolutions & settings than 1680*1050.

If you had a 2560x1600 monitor you would probably see usage of 30% or lower.
10-02-2011, 03:44 AM
gosh

Quote:

Originally Posted by JumpingJack

Yep... Also the data set above shown is not wise to use to demonstrate core scaling. BF3 appears to stop at 4 cores (as you observed).

Scaling perfectly for a game to more cores than three is very hard (impossible today) becauase there are so many different type of workloads. Using 100% of three cores in playable resolutions is an achievement

Slide 33 in this presentation: http://www.slideshare.net/DICEStudio...-battlefield-3
10-02-2011, 04:02 AM
Mechanical Man

Quote:

Originally Posted by gosh

If you had a 2560x1600 monitor you would probably see usage of 30% or lower.

Somehow you did miss this:

Quote:

Originally Posted by Mechanical Man

Yeah, ~60% cpu usage for 1024*768 all MED + no AA, 8*Ani. Filter, no ambient occlusion.
10-02-2011, 04:11 AM
gosh

Quote:

Originally Posted by Mechanical Man

Somehow you did miss this:

And you wouldn't see 100% even if you went down to 640x480, that doesn't mean that there is no improvement with more cores on those low resolutions. Applications like games are complex. Hardware needs to keep up with the cpu. Just feeding the cpu with data will be very difficult when more cores are working. If the cache has a higher hitrate more cores can be fed if memory use is big

Why do you think servers have such high memorybandwidth
10-02-2011, 04:17 AM
Mechanical Man

Quote:

Originally Posted by gosh

And you wouldn't see 100% even if you went down to 640x480, that doesn't mean that there is no improvement with more cores on those low resolutions. Applications like games are complex. Hardware needs needs to keep up with the cpu. Just feeding the cpu with data will be very difficult when more cores are working. If the cache has a higher hitrate more cores can be fed if memory use is big

Why do you think servers have such high memorybandwidth

Originally we were talkin about cores. Not cpu arch. And that is what I did discuss.

I demonstrated that it does not scale over 4 cores (ppl with 4 cores do actually report 100% usage). That does not mean it could not be faster on different arch (this arch having more cores or less).
10-02-2011, 04:31 AM
gosh

Quote:

Originally Posted by Mechanical Man

Originally we were talkin about cores. Not cpu arch. And that is what I did discuss.

I demonstrated that it does not scale over 4 cores (ppl with 4 cores do actually report 100% usage). That does not mean it could not be faster on different arch (this arch having more cores or less).

I am not talking about final game performance (the worst quad will run battlefield 3 almost as fast as the best). Frostbite 2 is the first game engine I have seen that seems to chunk up workloads in smaller pieces. This is something new and that means a shift in how games are beeing developed. More games will follow and they move away from one single render thread and starts to use available cores on the cpu.
10-02-2011, 04:52 AM
Mechanical Man

Quote:

Originally Posted by gosh

I am not talking about final game performance (the worst quad will run battlefield 3 almost as fast as the best). Frostbite 2 is the first game engine I have seen that seems to chunk up workloads in smaller pieces. This is something new and that means a shift in how games are beeing developed. More games will follow and they move away from one single render thread and starts to use available cores on the cpu.

And it is max of four cores now. There is so much dependency between threads it cant extract more than 4 cores worth of parallerism out of threads. So more clock is better than more cores once you have four cores.

That basically means 8-core BD should be in max turbo all the time.
10-02-2011, 05:01 AM
muziqaz

Ok, just redownloaded the game, just for you folks to see if I am delusional or something else is going on.

Code:

Frames Time (ms) Min Max Avg 3154 60000 46 70 52.567

This is what fraps showed me.
If anyone of you want frame by frame numbers I can send you a file.
This is 6970@stock on thuban@3.8ghz with 8gb 1600Mhz cas 7 ram with velociraptor and win 7 64bit
I don't know what those guys at that website were testing but clearly there is some bottlenecks. And my system has 44 days of uptime and had chrome up with dozen tabs open, Origin doing god knows what, winamp, skype running.

Also I have little something something for Mechanical Man:
http://www.youtube.com/watch?v=A-enG0JCIH8
If that is not using all the cores, I have no idea what is :/ In my books 'using all the cores' means they are all loaded and being used. I am sure BD 8 cores will be used the same way.
besides, i was playing one map without any heavy machinery or explosions, though heavy fight. I believe that in that Alpha version they also have some of the frostbite features disabled.
10-02-2011, 05:06 AM
Olivon

You got magical HD6970 ... :ROTF: :rofl:
10-02-2011, 05:13 AM
muziqaz

No, but seriously :/ maybe they have something else set in drivers?
It is easier to explain nvidia, as it is different card from mine and there might be driver things going around. But what about 6970 scores :/ which are clearly on faster CPU and with newer drivers(I have 11.8 I think). And my GPU is stock HIS Radeon.
10-02-2011, 05:25 AM
muziqaz

Quote:

Originally Posted by Olivon

You got magical HD6970 ... :ROTF: :rofl:

On a second thought, maybe I am running 6 core BD?
That would be more believable than magical HD6970
:ROTF: :rofl:
10-02-2011, 06:00 AM
Mechanical Man

Quote:

Originally Posted by muziqaz

Ok, just redownloaded the game, just for you folks to see if I am delusional or something else is going on.

Code:

Frames Time (ms) Min Max Avg 3154 60000 46 70 52.567

This is what fraps showed me.
If anyone of you want frame by frame numbers I can send you a file.
This is 6970@stock on thuban@3.8ghz with 8gb 1600Mhz cas 7 ram with velociraptor and win 7 64bit
I don't know what those guys at that website were testing but clearly there is some bottlenecks. And my system has 44 days of uptime and had chrome up with dozen tabs open, Origin doing god knows what, winamp, skype running.

Also I have little something something for Mechanical Man:
http://www.youtube.com/watch?v=A-enG0JCIH8
If that is not using all the cores, I have no idea what is :/ In my books 'using all the cores' means they are all loaded and being used. I am sure BD 8 cores will be used the same way.
besides, i was playing one map without any heavy machinery or explosions, though heavy fight. I believe that in that Alpha version they also have some of the frostbite features disabled.

It was never about "using all cores", it was about getting more performance from more than four cores. Heavy singlethread program that spawns 5 more very light threads can also use all cores, its just not getting any real performance boost from it.

I said before, BF3 can take only advantage of four cores for performance. Extra two wont add any significant performance boost at same clock.
10-02-2011, 06:17 AM
muziqaz

Does anyone here have two identical systems with just different cpus? like deneb and thuban at same freq?
It would be interesting to see if there is no perf benefit out of more cores. Even if there is a small benefit going to more cores, prices at least in AMD side are negligible.
I just played the game with affinity set to 4 and 3 cores in both instances those selected cores were never loaded at 100%.
10-02-2011, 06:22 AM
FlanK3r

I have some CPUs (some Denebs, some Thubans, SandyBridge, Gulftown, Llano), but I have not BF3 :)
10-02-2011, 06:26 AM
muziqaz

Quote:

Originally Posted by FlanK3r

I have some CPUs (some Denebs, some Thubans, SandyBridge, Gulftown, Llano), but I have not BF3 :)

It is free on Origin(EAs Steam) just download 4gb of the beta, launch it through Origin and have some 'fun' :D

I have deneb laying around somewhere here as well, it just takes ages for me to change the cpu.
10-02-2011, 06:43 AM
FlanK3r

hm, Il busy next week :(....
10-02-2011, 06:48 AM
muziqaz

Quote:

Originally Posted by FlanK3r

hm, Il busy next week :(....

That's ok, as we are waaaay off topic :D
Though I wouldn't mind an explanation why my system is kicking around much more expensive systems in that chart :D
10-02-2011, 08:09 AM
MaddMutt

2 Attachment(s)

Multicore Speed UP

Here I have a Chart..... Sorry it's not BF3. It's Chinebench 10 with my Deneb x4 955 and my Thuban x6 1100T both are OCed to the same speed. You can see that AMD did some under-the-hood tweaking with the x6.

Attachment 120731

I read in here somewhere in a post that one of our guys busted his Thuban by adding to much voltage to the uncore.:shakes::rofl: CHEW loaned/let him have one of his extras to hold him over until BD came out. Here is some testing that I did on this problem using AIDA64. I hope this helps :cool:

Attachment 120725
10-02-2011, 08:32 AM
Mechanical Man

Quote:

Originally Posted by MaddMutt

Here I have a Chart..... Sorry it's not BF3....

Our discussion was about BF3 multicore scaling. Ofc, it was OT in this thread.
10-02-2011, 08:39 AM
MaddMutt

Sorry :(

I just got off from pulling an all nighter.... I work 3rd shift
10-02-2011, 09:40 AM
zoomee

If its of any use - My system (stock crossfire 5850 BE's) with 4Ghz on thuban is scoring the following in BF3 - which I am super happy with btw@ 1920x1200

ULTRA/HIGH (As high as it will go) - But no AA - AF set to x16:
Metro - outside 80 > 120FPS, Inside 90>150fps

ULTRA/HIGH 2xAA, AF x16
Metro - outside 35>50FPS, Inside 45>70FPS.

Didn't stress any further,, as I've been trying to play the game - someone kill those darn flashlight flashers!! aaaarggghhhh

Show 100 post(s) from this thread on one page