Sandy Bridge experience on Arma2 (from Q6600)

[GeezerButler]

do people really choose which games to play based on how well they are coded? If so, they must have a limited selection of titles. I play games because I like them and if they are slow because of poor coding I either have to wait til the dev fixes it (yeah right) or I make the best of it by throwing hardware at the problem. I have a 2500K coming this week to help improve my rig's performance with arma2 and fsx. My q6600 was never stable over 3.4Ghz.

 

[DBT85]

do people really choose which games to play based on how well they are coded? If so, they must have a limited selection of titles.

Sadly there are. There are also people who refuse to play games that don't run at 60fps, that don't make their rig cry, and that don't get 90%+ in reviews. Its one of the oddest kinds of eliteism ever.

 

[Columbo]

q6600 old technology.
sure its a good cpu, but it wont do 4.3ghz in 1.3vcore.
it also wouldnt do 4.5ghz.
or 4.8ghz.

I use a i7 950, at 4ghz, a sandy makes it likely 4.6ghz at same voltage.
that is 600mhz more with a more effective ipc clock for clock.
the new cpu smokes the old ones.

The point for me is does it offer value for money over and above a current set-up?

And looking at most benchmarks, user reviews the resounding answer is no.

I was hoping for a much larger jump in performance but the Q9550 will be staying a while longer by the look of things.

 

[ThomasTheTankEngine]

do people really choose which games to play based on how well they are coded? If so, they must have a limited selection of titles. I play games because I like them and if they are slow because of poor coding I either have to wait til the dev fixes it (yeah right) or I make the best of it by throwing hardware at the problem. I have a 2500K coming this week to help improve my rig's performance with arma2 and fsx. My q6600 was never stable over 3.4Ghz.

Arma 2 isn't badly coded. It has a 10km draw distance.

 

[cjph]

Arma 2 isn't badly coded. It has a 10km draw distance.

. . . and covers a shedload of other processing in terms of AI etc. I'm no expert on coding but their attention to getting out patches and supporting older games is up there with the best of them, which is why there is a strong community producing a lot of addon material.

If I could only get my Q9650 to 4GHz on a P5Q board I have lying around . . . .

cjph

 

[PCM2]

Just for reference to newcomers to this thread. Some people seem to be misinterpreting the data that's infront of them here and selectively reading things. Comparing to the Q6600 clocked as high as it would go it is important to look beyond the figures alone (come on now - don't be lazy). I've done the groundwork for you:

This was a particular game title where, during actual gameplay, the Q6600 @3.2Ghz presented a bottleneck...In heavily built up urban areas and the hills overlooking them, in particular, the frame rate would dip below 25fps

The performance in the places that proved problematic on the Q6600 was much, much better with the 2500k. Instead of being still in the doldrums at a miserable 20-25fps the frames were soaring away at 45fps+.

If you read the accompanying text you will see that the differences during normal gameplay are much more profound. There is something strange in the way that benchmark is made. Minimum framerate of 25fps was frequently encountered during actual online play and now in the same areas is up to around 45 or more making it actually comforable to play. It makes a very profound difference indeed.

I have just had a look at the Fraps counter on the Arrowhead benchmark since a lot of you seem to be ignoring the accompanying text and just looking straight at the figures. Something is seriously buggered with that benchmark because the minimum framerate of around 26fps is only there for literally a fraction of a second. This is when it pans into an area for the first time very rapidly and this is not something you would ever encounter during the actual gameplay. As testiment to this, within the same second it's running at over 40fps!!! So don't look too much into the minimum frame rates there. As I've said it makes a huge difference (from unplayable ~25 to playable 45+) in a lot of 'real' situations. Afraid I need to go now and will be able to update on Sunday evening at the earliest.

I've also since pushed the 2500k to 4.4Ghz comfortably and performance is even better again. In actual play it doesn't make a difference, though, as it's as smooth as butter now anyway. It never was before and that's a fact. Bear in mind I have no interest whatsoever in 'selling' this product to you but be sure to look a little deeper at the data and don't just eyeball some largely unrepresentitive figures and leave it at that.

 

[iamtheoneneo]

PCM2 your doing a good job mate, ignore the trolls..they would rather look at a stupid benchmark then actually understanding real world data

 

[Doppleganger]

It seems to me some are partly trying to convince themselves that there's little difference. I will build my SB system tomorrow and see for myself. Plan is to clock it initially at 3.5Ghz and compare against my Q6600 just for giggles.

 

[Scougar]

In my mind it has to be clock for clock to see the worth otherwise it's a naff comparison. At least it should give you a rough percentage of improvement.

Obviously... it's good to know that the SB chip can clock much higher and provide better minimums, but it would be nice to know that you didn't waste the money and that you could have got the performance from buying a Q9xxx chip @ 4Ghz or more.

 

[DBT85]

In my mind it has to be clock for clock to see the worth otherwise it's a naff comparison.

I still don't understand this. If you test the Q6600 at its best speed why would you want to see what your new chip can do at 70%? Even if clock for clock performance is identical, the simple fact is that you can get more clocks in in the first place.

As i said before, if my Q6600 could get to 3.6 or 3.8 i might not have bothered but at 3.0 the difference looks to be more substantial. Also bearing in mind that i gave up on clocking since i could only get 600mhz extra so mine has actually been running at 2.4 most of the time!

 

[DrewBrees]

Just thought I`d throw my 2p in here. I play ARMA2 a lot too, and I know its a very CPU intensive game. Would the newer chip not be able to process AI cycles faster/ more efficiently or both?

If you have too many units in play at once, the AI starts to get dumber and dumber as each new unit reduces the amount of CPU power available to generate their responses to player actions.

So, benchmarks are one thing, but being able to play warfare or co-op games with lots more (or even just smarter) ai units would also be a bonus.

Just thought I`d chip in coz I`ve just spent megabucks on a new Sandy system and want to convince meself it was worth it. Roll on Friday.

 

[GeezerButler]

I didn't say that ARMA2 was badly coded - the point about badly coded (or poorly optimised) games was a general one. I want the new CPU for ARMA2 and FSX because they are particularly CPU intensive.

 

[Rofflay]

Sadly there are. There are also people who refuse to play games that don't run at 60fps, that don't make their rig cry, and that don't get 90%+ in reviews. Its one of the oddest kinds of eliteism ever.

People have little digs at me all the time about this, you should see the backlog of games i have and i always do this and i admit its a hassle and very odd but i simply cannot stand a game that dips below 60.Not only does it stutter to my eagle eyes it just feels laggy and not "normal".

I will play crysis and the other 15 odd games at max settings when i get my new 2600k and gtx 580 though! and yes to find those games i simply filter gamespot by rating and anything below 8 gets dissed.TF2 and l4d2 at 8.5 are the lowest rated games i have played i think, i just find it hard to enjoy games when it doesnt blow me away.

Probably grown tired of it after all i have owned every console since the snes and i had a c64 as well with simon the sorcerer!

 

[Rofflay]

Originally Posted by PCM2 View Post
I have just had a look at the Fraps counter on the Arrowhead benchmark since a lot of you seem to be ignoring the accompanying text and just looking straight at the figures. Something is seriously buggered with that benchmark because the minimum framerate of around 26fps is only there for literally a fraction of a second. This is when it pans into an area for the first time very rapidly and this is not something you would ever encounter during the actual gameplay. As testiment to this, within the same second it's running at over 40fps!!! So don't look too much into the minimum frame rates there. As I've said it makes a huge difference (from unplayable ~25 to playable 45+) in a lot of 'real' situations. Afraid I need to go now and will be able to update on Sunday evening at the earliest.

I get this in arkham asylum on my rig and i want to play arma when i get my new one so im interested as to why this happens.Do you own an SDD? and how much ram have you got? It sounds like an SSD would solve that initial fps dip as its obvious it sounds like its rapidly freezing while it caluclates the whole new scene.

 

[MaxP779]

In my mind it has to be clock for clock to see the worth otherwise it's a naff comparison. At least it should give you a rough percentage of improvement.

Obviously... it's good to know that the SB chip can clock much higher and provide better minimums, but it would be nice to know that you didn't waste the money and that you could have got the performance from buying a Q9xxx chip @ 4Ghz or more.

Exactly this.

 

[PCM2]

I get this in arkham asylum on my rig and i want to play arma when i get my new one so im interested as to why this happens.Do you own an SDD? and how much ram have you got? It sounds like an SSD would solve that initial fps dip as its obvious it sounds like its rapidly freezing while it caluclates the whole new scene.

Yes - two SSDs as stated in the previous post (don't blame you for not seeing it buried in the rest of the information): "The system also uses X25M 80GB and Vertex 2E 60GB SSDs." I never notice stutter or any type or any sudden unexplained dips during actual gameplay, even if I visit the areas where the dips occur in the benchmark. It's the same place all the time and it seems to have something to do with the engine not liking the camera panning or something. Not sure but from the player's perspective it never happens.

P.S. Scougar and max. I had money to burn so I couldn't care less what a Q9xxx would do for me. This thing runs at 4.4Ghz at around 1.28vcore and never exceeds 63 celcius in full load. Much better than any 'Q' chip and frankly the rest of the system was limiting overclocking potential anyway.

 

[Stryda09]

I have upgraded to a 2500K from an AMD 965. These benchmarks aren't really solid proof but here is my 2p worth of how it has increased arma performance for me.

Specs:
Both using all same components apart from mobo, ram and CPU
Graphics Card: 6950 unlocked
SSD: Crucial M225 128GB

AMD:
Phenom II 965 @ 3.8GHz
OCZ 4GB 1600MHz AMD Black Edition
MB: MSI 790FX GD-70

Intel:
i5 2500K @ 3.8GHz
Geil 8GB Value stuff 1600MHz
MB: MSI P67A-GD65

Settings I can print screen for anyone who wants, but they were the same for both tests. I was only able to get avg FPS as that is what the benches gave me and I didn't want to watch Afterburner FPS counter for highs and lows. This is in no way solid proof on performance because the benchmark is realtime simulation so the same stuff doesn't happen everything. I therefore ran twice on each (Would have done 3 but I was a bit lazy).
Oh and I do not have Operation Arrowhead so could not do that one.

Benchmark 1, Day
AMD Setup avg
Run1 - 31FPS
Run2 - 32FPS

Intel Setup avg
Run1 - 44FPS
Run2 - 45FPS
(Note: VSync was enabled for both and intel test did hit 60FPS for atleast 4 seconds at one point during the benchmark.)

Benchmark 2, Night
AMD Setup avg
Run1 - 13FPS
Run2 - 12FPS

Intel Setup avg
Run1 - 21FPS
Run2 - 21FPS

These are both at same clock so should give you more of an idea.

As I said do not take this as pure fact as it isn't really the best thing to test it on but it appears ArmA II likes intel a bit more. I have run some other benchmarks but not many as I didn't really know what to run and is too late to do now. BUT I have seen a massive CPU performance increase on the benches I have done, such as the PassMark one and Cinebench. It has also increased graphics scores aswell but not by massive amounts in other benchmarks.

i.e in Heaven it has massively increased my min fps from 3.9 to 21.5 and high from 94 to 98 but only 3FPS better average.

Hope this is useful to someone.

 

[Doppleganger]

Built my S/B system and done a quick overclock to 4.5Ghz, still on stock volts.

Playing Civ 5 (a cpu intensive game fair enough) and the difference in smoothness and responsiveness is phenomenal compared to my Q6600 clocked at 3.5Ghz. My frame rates have gone from about 14-25 to around 58-60 (v-sync enabled) on a huge map.

So far, this upgrade has been well worth it.

 

[flopper]

I have upgraded to a 2500K from an AMD 965. These benchmarks aren't really solid proof but here is my 2p worth of how it has increased arma performance for me.

Specs:
Both using all same components apart from mobo, ram and CPU
Graphics Card: 6950 unlocked
SSD: Crucial M225 128GB

AMD:
Phenom II 965 @ 3.8GHz
OCZ 4GB 1600MHz AMD Black Edition
MB: MSI 790FX GD-70

Intel:
i5 2500K @ 3.8GHz
Geil 8GB Value stuff 1600MHz
MB: MSI P67A-GD65

Settings I can print screen for anyone who wants, but they were the same for both tests. I was only able to get avg FPS as that is what the benches gave me and I didn't want to watch Afterburner FPS counter for highs and lows. This is in no way solid proof on performance because the benchmark is realtime simulation so the same stuff doesn't happen everything. I therefore ran twice on each (Would have done 3 but I was a bit lazy).
Oh and I do not have Operation Arrowhead so could not do that one.

Benchmark 1, Day
AMD Setup avg
Run1 - 31FPS
Run2 - 32FPS

Intel Setup avg
Run1 - 44FPS
Run2 - 45FPS
(Note: VSync was enabled for both and intel test did hit 60FPS for atleast 4 seconds at one point during the benchmark.)

Benchmark 2, Night
AMD Setup avg
Run1 - 13FPS
Run2 - 12FPS

Intel Setup avg
Run1 - 21FPS
Run2 - 21FPS

These are both at same clock so should give you more of an idea.

As I said do not take this as pure fact as it isn't really the best thing to test it on but it appears ArmA II likes intel a bit more. I have run some other benchmarks but not many as I didn't really know what to run and is too late to do now. BUT I have seen a massive CPU performance increase on the benches I have done, such as the PassMark one and Cinebench. It has also increased graphics scores aswell but not by massive amounts in other benchmarks.

i.e in Heaven it has massively increased my min fps from 3.9 to 21.5 and high from 94 to 98 but only 3FPS better average.

Hope this is useful to someone.

twice the fps, its just awesome.

 

[DBT85]

Built my S/B system and done a quick overclock to 4.5Ghz, still on stock volts.

Playing Civ 5 (a cpu intensive game fair enough) and the difference in smoothness and responsiveness is phenomenal compared to my Q6600 clocked at 3.5Ghz. My frame rates have gone from about 14-25 to around 58-60 (v-sync enabled) on a huge map.

So far, this upgrade has been well worth it.

When you say stock volts, is that leaving the board to regulate the supplied voltage or does it actually stick to a set voltage?

I only ask since the Asus P67 boards have an "auto" setting which ramps up vcore as and when the chip needs it.