CPU bottlenecking ? upgrade to intel ?

[Zinth]

Running a phenom quad core @ 4.0ghz with x2 6950's and they're usage always seems to drop when massive explosions occur on bf3 and they aren't always on 99%, could my cpu be bottlenecking me or is this normal ?

 

[slade]

I would like to know this as well. I have now gone CrossFire with 2x 6970 @980 and they never go above 80% in MSI AB. PII 955 @ 4.2

 

[Marine-RX179]

Running a phenom quad core @ 4.0ghz with x2 6950's and they're usage always seems to drop when massive explosions occur on bf3 and they aren't always on 99%, could my cpu be bottlenecking me or is this normal ?

If your frame rate drops a lot when your GPU usage goes down (or usage is not close to 99%), then your CPU is bottlenecking. Let's say your CF6950 is capable of deliver minimum frame rate of 60fps but your CPU only capable of 30-40fps at specific scenes, you would only get 30-40fps it is simple as that.

 

[Zinth]

Cheers for reply.
Yeah in non graphically demanding parts of bf3 they will run @ 99% with like 100fps or something. But then soon as loads is going on fps can sink and usage sinks sometimes to as low as 50% :|. If i were to move to an i5 processor, what sort of frequencey would i have to clock to to avoid bottlenecking ???.

 

[MOOGLEYS]

I have an I5 2500k at 4.6ghz and I can keep 2 6950's at %100 GPU usage. My 960T running 4ghz also kept them at near enough %100 and tbh I don't see any massive benefits from the I5 to the Phenom II ingame..? I'm sure others will disagree.

 

[Marine-RX179]

If i were to move to an i5 processor, what sort of frequencey would i have to clock to to avoid bottlenecking ???.

Whatever clock speed and can keep your CF6950 at close to 99% usage at all time?

You gonna have to try it out yourself...but you give you an rough idea, a Phenom II X4 980BE at 3.7GHz is only as fast as i3 2100 at 3.10GHz (dual-core) on gaming performance, so with the i5 2500K at stock speed 3.30GHz with 4 cores, it is pretty much already twice the gaming performance of the Phenom II X4 980BE

 

[MOOGLEYS]

Just a shame you don't see anywhere near twice the speed other than synthetic benchmarks...

 

[Marine-RX179]

Just a shame you don't see anywhere near twice the speed other than synthetic benchmarks...

Because it is always the case of either GPU bound would hit first...and the issue of lots of games still uses less than 4 cores

 

[MOOGLEYS]

Because GPU bound would always hit first...

Ingame I notice very little difference going from my 4ghz 960 to the setup I have now which is an I5 2500K @ 4.6ghz with or without crossfire. I see big differences in benchmarks but to me noticeable differences in game are what I wan't and there just not there.

 

[tommybhoy]

Had a 3.6GHz q6600 crossfire 6950>70, wasn't performing to well tbh: compared to the [email protected]:

[email protected]:

AMD 555BE on a Sabertooth(stop gap while waiting for BD, then got sold on to get the intel setup) [email protected]:

Then I changed to the 2500K:

[email protected] using OC genie button:

Now for the best indication of cpu bottleneck with CrossFire Scaling:

[email protected]:

[email protected]:

That's what you need for proper CrossFire Scaling!

These are all my results from my setups by the way, not from the net!

You have got to weigh the pros and cons, whether you can live with it or bite the bullet and shell out the cash for an intel setup.

If you can't be bothered with oc'ing then the MSI Z68A-GD55-G3 is what I have and has the OC Genie button that will take your 2500K to 4.2GHz stable and it runs very, very cool.

I can run up to 4.7GHz, but i leave it at 4.5GHz.

Hope this helps guys.

 

[CAT-THE-FIFTH]

Crysis 2 uses upto 4 threads so the 4GHZ Phenom II X4 the OP has will do better than a 3.9GHZ Phenom II X3. Crysis only really uses around two threads IIRC.

 

[Mattuk69]

Would a phenom X4 @3.8ghz bottleneck crossfire 5850's? Because i sold one of my 5850's because i couldn't find out why some games would lagg more than a single 5850 in crossfire. It would only work with a couple games like Dirt 2,3.

 

[slade]

removed

 

[LtMatt]

I have an I5 2500k at 4.6ghz and I can keep 2 6950's at %100 GPU usage.

Ok. Now im worried. My gpu usage never gets to 99%, maybe it does one one gpu for a short while. My cpu is at 4.6. Hmmmm.

I was getting severe fps drops last night, playing a 32 man rush server. I even turned aa off and i was still getting these drops so its got to be something else. Suggestions? Does anyone care? lol

 

[MOOGLEYS]

Ok. Now im worried. My gpu usage never gets to 99%, maybe it does one one gpu for a short while. My cpu is at 4.6. Hmmmm.

I was getting severe fps drops last night, playing a 32 man rush server. I even turned aa off and i was still getting these drops so its got to be something else. Suggestions? Does anyone care? lol

Bf3 was having serious issues last night I kept getting disconnected and I gave up in the end.

 

['Wes]

I had this problem last night, I run an AMD 965 at stock and a 5870 and 5970 [factory overclocked] in tri-fire, i'm thinking i'll need to overclock my CPU back to 4GHz.

 

[eyetrip]

The other night I was suddenly getting horrible performance on my 2500K with my crossfire toxics. Frame rate was all over the place.

Finally remembered had reset my BIOS, so the CPU was running at 3.4 as opposed to the 4.5GHz I had it at before. Went back to 4.5 and was smooth.

So I would say it is certainly possible you are experiencing a bottleneck on CPU.

 

[LtMatt]

The other night I was suddenly getting horrible performance on my 2500K with my crossfire toxics. Frame rate was all over the place.

Finally remembered had reset my BIOS, so the CPU was running at 3.4 as opposed to the 4.5GHz I had it at before. Went back to 4.5 and was smooth.

So I would say it is certainly possible you are experiencing a bottleneck on CPU.

Yeah thats what i was thinking dude, im sure i had it at 4.6. Today its been fine so god knows what the problem was. I'll have to check out a rush server though as thats where the problem occured.

 

[spixelspixel]

edit - posted in wrong thread

 

[almighty151986]

I was running 5850 Crossfire overclocked to 850Mhz on a 4.1Ghz Phenom 2 x6 1075T

On Crysis my average frame rate was 55fps with a minimum of 38fps and looking at MSI afterburner it showed that neither of my GPU's were running at 99% load so I was CPU limited.

The cards were then moved over to a 2500k, Using the same clocks on the 5850's and the same drivers the average fps moved to 58fps but the minimum jumped up to 48fps and I was GPU limited as both cards were 99% loaded throughout the entire benchmark.