I play ut2004, favourate game of all time and that has been cpu bottlenecked since the nvidia 6800's came out.
I used to have a p4 @ 3.5, on a large 32 player server in a big battle that could drop into the mid 20's for fps. Nothing to do with my graphics (6800GT), I could set the settings to whatever I liked and I'd still get the same fps. Even a top end athlon FX would drop under 40 fps at the same point. UT is a game where fps matters - you really want at least 60.
I now have an E6600 @ 3.2 and that is the first processor that can play the game at close to 60 fps at all times.
So there are popular games that are cpu and not gpu bottle necked.
I used to have a p4 @ 3.5, on a large 32 player server in a big battle that could drop into the mid 20's for fps. Nothing to do with my graphics (6800GT), I could set the settings to whatever I liked and I'd still get the same fps. Even a top end athlon FX would drop under 40 fps at the same point. UT is a game where fps matters - you really want at least 60.
I now have an E6600 @ 3.2 and that is the first processor that can play the game at close to 60 fps at all times.
So there are popular games that are cpu and not gpu bottle necked.
Aikouka
Lifer
- Nov 27, 2001
- 30,383
- 912
- 126
Alraito! I agree with that one guy.. after looking back up.. anthrax.
There's a curve relating to performance / processor "power". Power being the relative ideaology of a processor being faster than another. At a certain point (processor), yes, while just playing the game the game's performance curve will begin to wane and flatline. This is completely true as the game cannot saturate the processor anymore.
Now, let's stop with this perfect world hogwash. I run a metric ass-ton of extra programs in the background... I think around 60 processes at once on my dual-core machine. Just this alone would show that switching from my older Athlon 64 2800+ or my older Athlon 64 3200+ to my Athlon 64 X2 4400+ or my Core 2 Duo E6600 would show a huge performance increase since the processor would be able to handle more than one task at once (and further reduce the need to perform the register switches and stuff (I forgot the proper term)). I buy a faster processor so I can do more than game at once. PC's are multi-tasking machines and we eventually become used to such.
If someone asks whether or not their single-core processor (a relatively newer.. say K8) would be a bottleneck compared to a newer dual-core... I'm gunna tell them yes. The ability to off-load excess mandatory yet mundane tasks to another core is by far more beneficial than not and could easily help show improvement in games.
There's a curve relating to performance / processor "power". Power being the relative ideaology of a processor being faster than another. At a certain point (processor), yes, while just playing the game the game's performance curve will begin to wane and flatline. This is completely true as the game cannot saturate the processor anymore.
Now, let's stop with this perfect world hogwash. I run a metric ass-ton of extra programs in the background... I think around 60 processes at once on my dual-core machine. Just this alone would show that switching from my older Athlon 64 2800+ or my older Athlon 64 3200+ to my Athlon 64 X2 4400+ or my Core 2 Duo E6600 would show a huge performance increase since the processor would be able to handle more than one task at once (and further reduce the need to perform the register switches and stuff (I forgot the proper term)). I buy a faster processor so I can do more than game at once. PC's are multi-tasking machines and we eventually become used to such.
If someone asks whether or not their single-core processor (a relatively newer.. say K8) would be a bottleneck compared to a newer dual-core... I'm gunna tell them yes. The ability to off-load excess mandatory yet mundane tasks to another core is by far more beneficial than not and could easily help show improvement in games.
People like to have their parts in the same 'end' or 'league.' I know it'd bother me if I have a 7900GTX and only a 3200 AMD64.
Higher ended CPUs do give better performance. Plain and simple. And some engines like the Source engine are CPU-bound.
(rant will continue after I get home from class)
Higher ended CPUs do give better performance. Plain and simple. And some engines like the Source engine are CPU-bound.
(rant will continue after I get home from class)
I think we need to revise the statement.
"Any MODERN (by modern I mean Celeron 326 and greater) cpu is not the bottle neck as long as the resolution is high enough, and IQ settings and AA/AF are turned up enough on a modern graphics card."
Hows that one?
And when I'm talking about bottleneck, I'm not talking about going from 100fps to 150fps with a faster CPU. That doesn't really matter.
"Any MODERN (by modern I mean Celeron 326 and greater) cpu is not the bottle neck as long as the resolution is high enough, and IQ settings and AA/AF are turned up enough on a modern graphics card."
Hows that one?
And when I'm talking about bottleneck, I'm not talking about going from 100fps to 150fps with a faster CPU. That doesn't really matter.
Aikouka
Lifer
- Nov 27, 2001
- 30,383
- 912
- 126
Then your definition of a bottleneck is off. It doesn't matter if you don't think it matters, a bottleneck is defined as what constricts the flow (hence the term bottleneck, as a neck of a bottle is smaller than the bottle and constricts flow) of data. If going from an Athlon 64 2800+ to an Athlon 64 3500+ shows a 15FPS improvement, then the game is somewhat CPU bound (increasing graphics may also improve the framerate at this point as well).
Yup, your keyboard is bottlenecking it all.. ultimately. Esp if your on PS/2!! OMGZORZ!
As far as this thread: Games are largely GPU bound at the high resolutions that people play at. /end
You know, I agree with the general wisdom of the OP's premise (GPU>CPU) in most applications. We might note AT's own review of CPU performance in Oblivion, however, specifically with regard to CPU/GPU scaling tests done in town:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2747&p=4
Outdoors, where the GPU demands are higher, only crossfire x1900xt's required the use of anything more than the lowest-end Athlon 64 (remember the time frame in which the article came out, pre-Conroe). But in towns, with the much higher concentration of NPCs, AT found that even an X1800xl paired with a 2.6Ghz Athlon64 would soundly beat crossfire x1900xt's paired with a 1.8Ghz model.
BUT, the difference between losing the ~9fps from the low 40s to the low 30s in the 2.6Ghz/x1800xl vs. 1.8Ghz/x1900xt "town" comparison, isn't nearly as important, of course, as the the difference in the "outdoor" benchmark: ~12fps from the upper 20's to the upper teens.
Thus the GPU>CPU wisdom generally holds true in even CPU-intensive titles.
Still, I would *not* want to play Oblivion on a Celeron processor. Sure, you can get semi-acceptable framerates if you pair it with ultra-high end graphics cards, but even then it will be seriously sluggish (19.9 fps in the Oblivion Gate bench, and 21.5 in the town bench with crossfire x1900xts). Considering just how much you can stretch the performance of even a single x1900 with better processors, a Celeron owner would have been better served ditching the second x1900 and upgrading their CPU to at least an Athlon64 3000+. That setup would have beaten the performance of x1900crossfire by 10fps in every benchmark and would probably have been cheaper, even taking new motherboard & RAM into consideration.
Situation, situation, situation. There is no single, magic bullet for upgrading your GPU/CPU. Generally, your GPU is likely to be more of a bottleneck in most games. If you can find the information, however, it just may be that your favorite game needs a slightly better balance of CPU & GPU.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2747&p=4
Outdoors, where the GPU demands are higher, only crossfire x1900xt's required the use of anything more than the lowest-end Athlon 64 (remember the time frame in which the article came out, pre-Conroe). But in towns, with the much higher concentration of NPCs, AT found that even an X1800xl paired with a 2.6Ghz Athlon64 would soundly beat crossfire x1900xt's paired with a 1.8Ghz model.
BUT, the difference between losing the ~9fps from the low 40s to the low 30s in the 2.6Ghz/x1800xl vs. 1.8Ghz/x1900xt "town" comparison, isn't nearly as important, of course, as the the difference in the "outdoor" benchmark: ~12fps from the upper 20's to the upper teens.
Thus the GPU>CPU wisdom generally holds true in even CPU-intensive titles.
Still, I would *not* want to play Oblivion on a Celeron processor. Sure, you can get semi-acceptable framerates if you pair it with ultra-high end graphics cards, but even then it will be seriously sluggish (19.9 fps in the Oblivion Gate bench, and 21.5 in the town bench with crossfire x1900xts). Considering just how much you can stretch the performance of even a single x1900 with better processors, a Celeron owner would have been better served ditching the second x1900 and upgrading their CPU to at least an Athlon64 3000+. That setup would have beaten the performance of x1900crossfire by 10fps in every benchmark and would probably have been cheaper, even taking new motherboard & RAM into consideration.
Situation, situation, situation. There is no single, magic bullet for upgrading your GPU/CPU. Generally, your GPU is likely to be more of a bottleneck in most games. If you can find the information, however, it just may be that your favorite game needs a slightly better balance of CPU & GPU.
Aikouka
Lifer
- Nov 27, 2001
- 30,383
- 912
- 126
Yes, I'm sure those USB headers on the bottom of your motherboard are causing that lag that makes you die in CS. Tell that to someone next time you die.
Don't pull out illogical crap. Jeez, what kind of argumentative skills are you youngens learnin' these days? Oh that's right, Bill Gates said public schools suck, you're excused ^_^.
Correct! the biggest / slowest bottleneck on a computer system is the user (especially dumb ones).
- Feb 5, 2006
- 2,778
- 1
- 81
I upgraded my 3500+ up to a 4400+ both with my BFG 7800 GTOC. and saw a pretty pronounced improvement on BF2. I get that it's not a bottleneck to the GPU in most cases, but I have to say that it most certainly improved my performance.
BF2 is a RAM hog. You could go back to your 3500 with 2 gigs of RAM and surpass your current performance... at least when loading levels... this is the reason I can't stand this game... gotta wait a couple minutes for every lame level to load. :roll:
And to the guy who thought he needed an E6600 to play UT2004, your videocard is still punking you because an X850xt with an E6400 will give you a couple hundred frames per second. I don't know where your bottleneck is but I am guessing it's that videocard. My recent 7900GS has UT2004 up to something like 300 frames per second or something ridiculous like that. Might as well be talking about Q3 fps's! :laugh:
You obviously just read the silly little tests review sites use for ut2004 (16 bots, deathmatch, average fps, whatever). Try playing the game for real on a 32 player ons server (e.g. titan 32 player one) and watch your frame rates dive. Min fps is as I said in my previous mail (and min fps happens in the middle of a fire fight when you need those fps the most). I am an experienced ut2004 player, I know lots of other experienced ut2004 players, I didn't make this up, it simply is a fact that the game gets cpu bound on big servers if you have a modern (6600GT or better) graphics card.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter