Re: CPU bottleneck article
That does suprise me, good job I only buy middle of the range hardware anyway isn't it? What a waste to have a monster GFX card with a mid-range CPU!
Interesting is the (mostly) tiny amount of difference 5870s make to the 440 over a 5850 in what has to be an 'average res' of 1680*1050.
~1fps in FC2
~40 in Stalker (fair enough!)
~ -1.5 in Crysis
2fps in WiC
Madness I tell you!
Re: CPU bottleneck article
I found this article very interesting especially about visual perception thanks
Re: CPU bottleneck article
Toms Hardware and Bit tech are among the two most Intel biased sites out there so take their results with a pinch of salt.
Far Cry2 runs better on Intel processors ffs.
Of course titles like Crysis obviously scale very well with Crossfire! :rolleyes: What the fact that there maybe Crossfire scaling issues then??
Idiotic and pointless comparison.
Re: CPU bottleneck article
Quote:
Originally Posted by
CAT-THE-FIFTH
...Of course titles like Crysis obviously scale very well with Crossfire! :rolleyes:....
Not on the 440 it doesn't :D
Re: CPU bottleneck article
Quote:
Originally Posted by
Rob_B
Not on the 440 it doesn't :D
Crysis does not scale well with two cards. On top of this it also runs better on Intel processors too. The Core i7 920 does have very strong single threaded performance.
The fact is that the Core i7 is a faster processor than a Athlon X3 but remember that for the price of a Core i7 920 and an X58 motherboard you could get an Athlon X3 440,4GB of DDR3,an AM3 motherboard and an HD4850 1GB.
So basically there are 4 games tested:
1.)Far Cry2 which runs better on Intel processors and also prefers more cores
2.)Crysis which runs better on Intel processors
3.)WIC which scales well with more cores and Intel hardware.
4.)Stalker which runs better on AMD and ATI hardware. Funnily enough this game seems to run fine on the AMD processor.
The minimum framerates are what make this comparison pointless.
1.)With a pair of HD5870 cards
http://media.bestofmicro.com/5/K/246...e_min_1920.png
http://media.bestofmicro.com/5/L/246...e_min_2560.png
2.)With a HD5850
http://media.bestofmicro.com/5/4/246...e_min_1920.png
http://media.bestofmicro.com/5/5/246...e_min_2560.png
So basically they are saying games which run better on Intel hardware run better on Intel hardware, games which run better on AMD hardware perform very well even on cheaper AMD hardware and games which like more cores will run better on a CPU which has more cores! :rolleyes:
Why isn't a processor which costs over 2.5 times more(£66 versus £170) producing much higher minimum framerates then?
What about cheaper Intel processors under £100 then?? Why didn't they test a Core i3 530 or G6950 at stock speeds then?
Re: CPU bottleneck article
I'm quite suprised by your assessment of the article CAT, I'd always thought of Toms as being amd biased if anything as their best cpu for the money charts always barely feature Intel.
Also I'm confused by your comments on xfire scaling. Eg crysis shows a 10fps gain from the 5870s over the 5850, whereas the x3 shows none. Are you saying that xfire scaling can be affected by cpu manufacturer?
I also wasn't aware that some games inherently favour one manufacturer over another to such an extent. I knew that you should take into account what games you play when choosing a gfx card, but didn't realise the same applied to cpus, so I've learnt that if nothing else!
Re: CPU bottleneck article
Quote:
Originally Posted by
Random_guy
I'm quite suprised by your assessment of the article CAT, I'd always thought of Toms as being amd biased if anything as their best cpu for the money charts always barely feature Intel.
I never read the CPU charts for any website as pricing changes. For example the Core i7 920 has dropped by £50 to £60 making it relatively good value for people who need a platform which can do Crossfire and SLI at PCI-E 2.0 16X/16X and processor intensive tasks.
Also the AMD test rig uses a 790X motherboard which does Crossfire at PCI-E 2.0 8X/8X.
Quote:
Originally Posted by
Random_guy
Also I'm confused by your comments on xfire scaling. Eg crysis shows a 10fps gain from the 5870s over the 5850, whereas the x3 shows none. Are you saying that xfire scaling can be affected by cpu manufacturer?
I also wasn't aware that some games inherently favour one manufacturer over another to such an extent. I knew that you should take into account what games you play when choosing a gfx card, but didn't realise the same applied to cpus, so I've learnt that if nothing else!
What I am more actually more surprised about is that the a cheapo Athlon II X3 is producing significantly higher minimum framerates in Stalker than a Core i7 920! WTF?
I would expect the Core i7 920 to have higher minimum framerates. It has higher average framerates but you would expect the Core i7 920 not to have such huge framerate drops in parts of the game.
However Stalker is meant to run better on AMD hardware and even AMD plug the game on their website:
http://game.amd.com/us-en/play_info.aspx?p=0&id=49
Likewise Crysis has a lovely Intel splash screen at start up as you know! ;)
Intel plugging Crysis:
http://game-on.intel.com/eng/games/crysis/default.aspx
Intel plugs FarCry2 on their website and even offers additional mission maps:
http://game-on.intel.com/eng/games/f...d/default.aspx
Intel plugging WIC:
http://game-on.intel.com/eng/games/wic/default.aspx
This is why I would have preferred the following test instead:
1.)Athlon II X3 versus more expensive AMD processors.
2.)Core i3 530 and Pentium G6950 versus more expensive Intel processors.
Re: CPU bottleneck article
Here is a Toms Hardware article comparing the X2 555BE,Pentium G6950 and the Core i5 750:
http://www.tomshardware.com/reviews/...55,2540-8.html
An HD5850 1GB was used. The two games which ran considerably better on the Core i5 750 were WIC and FarCry2 and these game prefer more CPU cores! ;) At stock speeds the Core i5 750 is faster than a Core i7 920 in a single card setup due to more aggressive TurboBoost and the PCI-E controller being in the CPU.
Fallout 3 and H.A.W.X. run better on AMD hardware IIRC! ;)
The reduced L3 cache and the lower clockspeed is not helping the G6950 though.
I will now run away since the Intel Core i7 owners are now probably priming their HT canons at the poor Moose!
Re: CPU bottleneck article
So basically for each game, Tom's have 2 unrelated data points. I don't understand how you can learn anything from that. If the budget point was generally faster than the expensive one, then it would be worth an article.
Many years ago Anandtech did a similar CPU scaling article, except they chose one CPU and varied the multiplier. Must have taken them ages, but that is the scientific way to go.
Here it is, was a brilliant article in its day:
http://www.anandtech.com/show/941/6
As for the perception stuff...
http://www.anandtech.com/show/2794/3
There was a time when Tom's Hardware was really good, and then Anand left :D
Re: CPU bottleneck article
I guess the article just challenged my assumption that when it comes to games if you have a mid-range cpu then you'll pretty much always be gpu limited. Makes me wonder how much my c2d is holding back my gtx 275...
I seem to remember an article a while back where one of the big review sites looked at could core scaling by disabling cores on Phenom II x4, but I can't find it now.
Re: CPU bottleneck article
The idea itself to compare an i7 to an Athlon makes me wonder about the author's state of mind
:D
Re: CPU bottleneck article
You would imagine it might have been better to compare a low end AMD and a high end AMD, and a low end Intel then a high end Intel.. atleast it would have been a bit more like for like. Rather than Ferarri to skateboard
Re: CPU bottleneck article
I think they have a point that the average gamer doesn't necessarily need a high end CPU. But then with something like the single player campaign in ARMA2, it will bring most PC's to their knees. There are so many calculations going on in the background, and the CPU is struggling to control all the hundreds of units in the massive area. The graphics card doesn't make any difference at this point, it's all about the CPU. So if you happen to be wanting to play very CPU intensive games, then really you pretty much need the fastest thing you can afford. There is also Flight Sim 10 and whatever else. I also saw a video of Mafia 2 and they were showing off the fancy 'physics'. If you don't happen to have a physx card or an nVidia card which can help out, then I'm assuming most of that stuff is going to be offloaded to your CPU too. So personally going in to the future, I'll still be wanting the best CPU I can afford.
Although I think it's true what people say about the XBOX 360 holding gaming back. When so many games these days are designed to run on all platforms, the XBOX 360 is the weakest link so is holding things back. Stuff like Battlefield Bad Company 2, if it was exclusively for PC, I bet they would have pushed things far more.
Re: CPU bottleneck article
was an interesting read, something that I didn't know before was the fact that the refresh rate is limited to the screen to the point that anything past 60FPS is negligible.. it's a good point for anyone looking to build their first high end PC on a budget to keep in mind