when i had my amd phenom II x4 965, at like 3.7ghz, i would be getting say 30fps in a random game, and now i have upgraded to a 2600k at 4.5ghz, im getting like 40fps this is to be expected but what i dont get is why. i look on task manager mid game, and the cpu usage of the amd chip was at like half way to 3/4 way up on all cores. now obviously the 2600k is a little lower, but what i dont get is why does the 2600k offer a fps boost when the amd chip wasnt even apparently working at full load in the first place the obvious explanation as to the better fps when upgrading cpus is that it is bottlenecking the gfx card. but i dont see how the 965 could have been (for a start i only have a 4870 and thats no gtx 590) and besides it wasnt even at full load. like; where has this extra fps come from?
From looking at mobo reviews on here it looks like the motherboard has an impact also, so that might be one reason...
I would like to know the answer to that as well. I thought it was to do with bandwidth. Then I realised I didn't know what the hell I was talking about.
The processor has a tiny segment of it's time portioned off for system tasks, part of that is stuff like PCI-E lane management (Along with the north and South bridges). The faster the processor; The larger the portion that's retasked (Usually done in singular cycles. And we're talking millions of cycles a second) Motherboards also have an Impact. PCI-E 1.0 was no faster than AGP (8x PCI-E is 8X AGP in terms of speed. At last check). Since then; all versions have doubled it. PCI-E 3.0, by extension, could theoretically support a PCI-E 1.0 card on a 1x socket and still hit near-full speed. In addition to that; the shorter the distance and the less stuff in the way on PCI-E lanes between them, the Northbridge and Processor; the lower the latency, thus; Higher the FPS. This is how Motherboards can have an impact on FPS. I think. At least that's my understanding on it.
Its probably the Instructions per Cycle, as on Intel chips it is way higher. This means it's kicking through data at a much higher rate, so although task manager reports the same CPU usage the Intel CPU is processing data at a much higher rate...
I assume your talking about average FPS. There are lots of things that come to my mind... The best bet is that your GPU could be waiting for some data from the processor to start working on the current frame. That means your processor would be fully stressed for a short while and then it would do another non critical tasks while the GPU does its work. You could see it as processing bursts. That would explain why higher IPC brings higher FPS and why you dont see your CPU usage at 100%.... Weeelll, that's just my theory...
More often than not the CPU won't have to do half as much as the GPU. so an overclocked cutting edge CPU won't be stretched much beyond 50%, the graphics is generally more important, unless the CPU is too slow to support the card at a decent potential.
It depends on the efficiency of the architecture, and we already know that the Sandy Bridge quad core CPUs kick serious AMD ass (not a fanboy statement; hard fact) where efficiency is concerned. That's why you got your performance boost without an apparent increase in CPU usage. Also memory has a part to play - a few years ago when 2GB-4GB of RAM was the norm for gamers, it was revealed that using 8GB significantly raised the min frame rate in graphics intensive games, which of course affects the average frame rate. Did you have 16GB in your AMD system? If not, I think you can safely assume that there are a number of factors involved in your improved frame rates besides "just" the CPU.
It really depends on games. See battle field 3 review .... wether it was a dual core@2GHz or a Quad@3+GHz, the fps were still the same. Some games are more CPU dependant, like RTS for examples. Others are more GPU dependant. The motherboard is an important factor, as well as RAM speed.
Tom's Hardware did an article on something related to this - i.e. For gaming, is the maximum you need in terms of CPU is an Athlon x3? linky
