This topic is locked from further discussion.
The frequency and the severity of CPU bottle necking against GPU performance is highly over-exaggerated on this forum. The few tests done (THG has done some, if you want the URL, ask) to determine the effects have resulted in extremely small differences. The phenomenon does exist, but is at the far limits of what can even work together. If you had a 1.3 GHzP4 in a relatively new (AGP3) s778 motherboard, then it would put a crimp in what an X1950 XT AGP could do, certainly. But at 2.8 GHz for the CPU, it starts being so comparatively small that human senses can't tell it's happened. Any CPU and MB new enough for PCI-e is also new enough to eliminate the very wide range of that particular AGP example.
(Edit) Incidentally, I have had my share of credulity with regard to quite a few of the THG results over the years, but the charts they have are just about the only resource of the type we have (CPU, GPU). This time, I didn't immediately see anything seriously "wrong" with what the article has as a conclusion/ summary, and I don't think they are saying that there is no loss of performance, merely that the effect is rather small. An in that aspect, I am in agreement.
Right this minute, we have a momentary richness of CPU performance at some tremendous pricing (due to be better tomorrow) that seems to make last year's hardware seem really ancient and slow, but in fact that isn't the case (last year's A64s, at least, are still quite respectable). The THG example above involved an AMD 3800 (I believe an X2, I think that's the only A64 3800), matched to the latest high zoot video. Like I said, I have the URL. Remember, it's only Tom's Hardware, and there always seems to be some agenda behind their frquently slanted stories. What the heck here it is:
http://www.tomshardware.com/2007/05/11/system_builder_marathon/page11.html
The frequency and the severity of CPU bottle necking against GPU performance is highly over-exaggerated on this forum. The few tests done (THG has done some, if you want the URL, ask) to determine the effects have resulted in extremely small differences. The phenomenon does exist, but is at the far limits of what can even work together. If you had a 1.3 GHzP4 in a relatively new (AGP3) s778 motherboard, then it would put a crimp in what an X1950 XT AGP could do, certainly. But at 2.8 GHz for the CPU, it starts being so comparatively small that human senses can't tell it's happened. Any CPU and MB new enough for PCI-e is also new enough to eliminate the very wide range of that particular AGP example.
Right this minute, we have a momentary richness of CPU performance at some tremendous pricing (due to be better tomorrow) that seems to make last year's hardware seem really ancient and slow, but in fact that isn't the case. The THG example above involved an AMD 3800 (I believe an X2, I think that's the only A64 3800), matched to the latest high zoot video. Like I said, I have the URL. What the heck here it is:
http://www.tomshardware.com/2007/05/11/system_builder_marathon/page11.html
Kiwi_1
Comparing a gtx with a gts? :roll: even with that difference of processors should be obvious for games.
The Firing Squad article didn't cover the full range of currently popular CPUs that gamers are relying on to last another year or so; it was limited to only the stuff from the immediate last few months, and the particular CPU they were concentrating on is still out of the financial reach of most of us. Anyway, rather few of the PC game playing folks that I am familiar with are willing to accept Low Resolutions to game with. If their display will do it, and their video card can push it, they are at their upper limit.
One of my systems has an inexpensive 17" LCD panel that has a native resolution of 1024 by 768, and I have relegated that one to the Living Room for the grandkids to play on (X850 Pro video card, old XP 3200 CPU Overclocked a ton, etc.) its visuals are decent, but the others are better!
Please Log In to post.
Log in to comment