Discussion in 'General Hardware' started by EastCoasthandle, Apr 12, 2011.
Were the vantage runs with PhysX disabled I hope (looks like, just confirming).
The results are without using physx.
Thanks, EastCoasthandle... Nice thread...
Hope it produces good info, without any BS.
Correct, but those that can afford a proper SLi/Crossfire setup that would show a noticeable difference between the two can also probably afford a SB setup to support it as well.
But if you are still on C2Q, you're also likely looking for a single GPU setup, in which case the GPU is going to be the limitting factor, and any game that isn't limitted by the GPU will be getting well over 60FPS anyway, so the CPU limitting the game won't matter.
NP, I was looking forward to adding dank1983man420 results but I have no idea what happened.
Put like that I, see your logic Guess I'm just impressed at the benefits I have seen with a single card setup, great chips. But yes, a nice C2Q with a single card will still give more than adequate framerates.
My friend went from a Q9xx at 3.4 ghz to a 2500k at 4.5ghz and doubled his wow frame rate in busy areas. So more than just clock for clock, you have to also account for the significant overclock advantage as well. That's what really widens the gap.
SB is more able to handle both specific cpu rendering, AI, particles, physics, culling, etc (anything else a particular game may use the cpu for) and still feed the GPU. Doing a better job then the C2Q. Even though the C2Q CPUs are capable of handling the same.
The question that I have that remains is how much of that performance boost is due to the faster memory subsystem?
I've added some results from winrar.
So.. Is HT enabled on the SB in your tests? Sorry if you mentioned it and I just can't read.
If HT was on during the WinRar test I wonder what the difference would be with it off.
HT is on during those benchmarks/test. I suppose it would have an effect if it's disabled. In any case, I have one more test to add and that's the Super PI Mod results. Enjoy.
I did my SB rebuild last night and the difference is huge between my q9550 and 2500k. Crysis and BC2 are now running much smoother and at higher FPS. The Q9x50 chips are good for their time and unless you're running a highend GPU it might not be worth it to upgrade. However, if you're running decent graphics and especially Crossfire/SLi the difference is HUGE.
I didn't get much benching done before I did the swap but @ stock (3.3Ghz) the SB system score 2000pts more in 3DMark06 vs my q9550 @ 4ghz.
Some games are showing a noticeable difference (IE: BC2).
BC2 is extremely CPU limitted, so a large difference isn't a surprise, but I wonder how noticeable it really is.
HUGE! Using 11.4 with my 5970 I woud average around 45fps in smaller scale 32 player maps and using older drivers (10.5) I would average around 55fps with most settings on medium. Since upgrading I can use 11.4 at the highest possible settings and the game is unbelivably smooth. I now average 80fps (ish) and I havent even reinstalled Windows yet.
The difference is large and measurable.
Crysis on the other hand didnt have a massive increase in measurable performance but the perceived difference and fluid gameplay is amazing.
I had a few BC2 fraps logs saved that I can compare against the 2500k. I'll try to get them posted later today.
With BC2 the Q9650 OC'd can handle it with no problems until the scene gets hectic with smoke, explosions, teammates shooting about, artillery raining down, etc. As long as you keep the above to a minimum the Q9650 OC'd didn't have much of a problem. But when you start to see the examples above in a any combination of 2 or more is when frame rate dips were noticed (resulting in some stuttering, etc) while the 2600K didn't have a problem.
I also updated the OP showing an example from Starcraft 2.
If you happen to try out Serious Sam HD that would be nice. I saw a nice improvement with a moderate overclock on my Lynnfield. Oblivion (especially modded) or even Fallout New Vegas would be another good one. Darkplaces (Quake source port) is another good one where I saw some serious gains while overclocking especially with the right mods like pretty water.
Anandtech's cpu gaming comparisons are awful. Who is running a $300 cpu at 1680x1050 or lower resolutions and low quality settings?
@ 1680x1050 any CPU limitation will be more obvious. That's why you see some websites using even lower resolution like 1024x768.
I don't think that is necessarily true. In some games yes but not all. Not by a longshot. The games that I mentioned in that quote are an example of that. GTA4 and FSX are two well known examples of games where the more resolution and IQ settings that you use the more cpu limited that you can become.
I don't even look at Anand's cpu reviews because of that. It makes the comparison about as useful as any other synthetic benchmark.
Before and after FRAPS log from random BC2 before and after upgrading to the 2500k. Most recent scores are at even higher settings.
My max became my average. This speaks for itself.
2011-04-11 21:04:00 - BFBC2Game
Frames: 3788 - Time: 60000ms - Avg: 63.133 - Min: 49 - Max: 76
2011-04-13 22:14:36 - BFBC2Game
Frames: 4144 - Time: 60000ms - Avg: 69.067 - Min: 45 - Max: 106
2011-04-15 20:38:15 - BFBC2Game
Frames: 3341 - Time: 60000ms - Avg: 55.683 - Min: 18 - Max: 173
2011-04-21 18:22:17 - BFBC2Game
Frames: 6234 - Time: 60000ms - Avg: 103.900 - Min: 61 - Max: 144
2011-04-21 18:24:22 - BFBC2Game
Frames: 6472 - Time: 60000ms - Avg: 107.867 - Min: 65 - Max: 172
2011-04-21 18:28:50 - BFBC2Game
Frames: 7297 - Time: 60000ms - Avg: 121.617 - Min: 70 - Max: 169
I've added BC2 and SCII comparisons without HT to the OP.
So what im seeing here is that the 2600K isnt realy any faster in games unless it is very CPU intensive, even the older Q9650 is holding its own still in all those games and gives enough grunt.
I honestly dont see a point upgrading if you still have a Q9650, not worth it at all.
And, really, once you overclock the Q9650 to the 3.6GHz range, which they all pretty much do, it isn't going to hold you back any noticeable amount in any game.
Yes, Sandybridge is faster and overclocks better, but in real world usage the difference won't really be noticeable.
One reason why I still love my Q9650 ! This is one solid CPU . I do not see a future build from what I have now any time soon . One thing I see happening faster and faster is the tech is moving far too fast for me to keep up with . Every time I get some thing "New" I find that in a month it is OLD and slow . I am really sick of trying to keep up and have settled for just keeping what I have .
Separate names with a comma.