Video Card + CPU ... do you really need a 3ghz+ CPU???

So I finally came back to the computer hardware/overclocking hobby, after being gone for 2 ½ years, and I really needed to upgrade my computer to play todays games… I recently picked up Left 4 dead, gears of war, Crysis warhead, COD4 and COD world at war… and ive noticed that my system wasn’t cutting it anymore, even at really low res, some of the games Ive been playing have been really slow… so I decided its time to do a upgrade;

Option 1:
Video card: Radeon HD 4850 512MB (hopfully i can get up to 4870 speeds)
Processor: Intel Core 2 Duo E8400 (hopfully i can get up to 4-4.2ghz)
total cost: 440ish (candian dollors)

Option2:
Video card: eVGA e-GeForce GTX 260 SuperClocked 896MB
Power supply: PC Power and Cooling Silencer 750 Quad Power Supply
total cost: 420ish (candian dollors)

Now the majority of you guys leaned towards option two, because their logic was after 3ghz the cpu is no longer a bottleneck, and games now are relying heavily on GPU (altho I figured the overclocked E8400 @ 4ghz would help more then a better video card) So I wanted to put this to the test;

The test consists of me running CS: stress test (MAXED everything settings at 1605x1080) x3 and getting the average of the 3 run's

Hardware setup: P5B deluxe / intel E6300 / 7800gt and 8800gts / 2x 1gig PC8500 patriot ram

7800GT config
E6300 @ 1.86ghz = FPS 88
E6300 @ 3.0ghz = FPS 88.1
E6300 @ 3.29ghz = FPS 88.3
E6300 @ 3.5ghz = FPS 88.4

(looks like in this instance, the CPU has already maxed out GPU)

8800GTS config
E6300 @ 1.86ghz = FPS 182
E6300 @ 3.0ghz = FPS 213
E6300 @ 3.29ghz = FPS 214
E6300 @ 3.5ghz = FPS 215

8800GTS config
Q6600 @ 2.4ghz = FPS 182
Q6600 @ 3.0ghz = FPS 213
Q6600 @ 3.2ghz = FPS 214

(looks like in this instance, there was a 31FPS jump from 1.86ghz to 3.0ghz, BUT there less then 2FPS from 3ghz to 3.5ghz)

As you can see, at around 3ghz, you start to near the MAX of your cpu….. in my case, those gamers who are trying to decide on CPU/Video card combo’s for a system, this info might be able to help you out…. After this benchmark, ive decided I don’t need to upgrade my CPU to an E8400…. hopfully this helped some of you guys out there...

I’ll be getting a GeForce 260GTX in a week, and I’ll be doing the same test to see if it follows the same pattern.

when it comes to a gaming machine, the CPU is about 30% as important as a power supply or GPU, or display. these items directly relate to the depth of your experience. the CPU, not so much.

I think you'll find that both of the vidcards under test bottlenecked the cpu at about 3.0GHz. Once you get the 260 you'll discover that point moving much higher, and might even be bottlenecked by your cpu. I look forward to your testing.

I absolutely agree with you that super high CPU speeds get diminishing returns, and are just silly. We all know today that a getting a powerful GPU impacts 3d gaming more than CPU. And from a CPU perspective, I think that recently released games (like Left 4 Dead), and games coming out in the future, will actually benefit more from multithreading than from high CPU clock speeds.

This is a large reason why I chose to get a quad core CPU, keep it at stock speeds (2.4Ghz), undervolt it to keep heat generation low, and keep all of your fans down to low speeds for a quiet PC. To anybody who hasn't tried this, I highly recommend you give it a shot to see if the decrease in fan noise outweighs the very small drop in framerates. After all, if you can't get much performance benefit out of overclocking your CPU, why not try and get benefit from a silent PC?

One area in which cpu does matter is multi-gpu configurations like Crossfire and SLI. Scaling on such systems is an area where i7 does offer meaningful gain for gaming.

i saw almost zero improvement in fps when i overclocked my core2duo from 2.8ghz to 3.5ghz

when it comes to gaming i see no point in spending more than $100 on the cpu if you are playing at any respectable resolution

This thread is good food for thought.

i got my hand on a Q6600 and did some tests, and the results are very interesting

8800GTS config
E6300 @ 3.0ghz = FPS 213
E6300 @ 3.29ghz = FPS 214
E6300 @ 3.5ghz = FPS 215

8800GTS config
Q6600 @ 3.0ghz = FPS 213
Q6600 @ 3.2ghz = FPS 214

if you look here, Q6600 @3.0ghz and the E6300 @3.0ghz both get 213FPS

and Q6600 @3.2ghz and the E6300 @3.249hz both get 214FPS

im looking foward to my gtx 260 results!

Quote:

---

Originally Posted by sharkypinoy

i got my hand on a Q6600 and did some tests, and the results are very interesting

8800GTS config
E6300 @ 3.0ghz = FPS 213
E6300 @ 3.29ghz = FPS 214
E6300 @ 3.5ghz = FPS 215

8800GTS config
Q6600 @ 3.0ghz = FPS 213
Q6600 @ 3.2ghz = FPS 214

if you look here, Q6600 @3.0ghz and the E6300 @3.0ghz both get 213FPS

and Q6600 @3.2ghz and the E6300 @3.249hz both get 214FPS

im looking foward to my gtx 260 results!

---

Not really that surprising, since it just means that the game you're running cannot take advantage of 2 of the cores in the Q6600.

Quote:

---

Originally Posted by qmul

Not really that surprising, since it just means that the game you're running cannot take advantage of 2 of the cores in the Q6600.

---

but also keep in mind that the Q6600 has 4mb cache, where the E6300 has only 2mb... so that might show the trend for those who have 6mb cache with the E8x00 processors...

Quote:

---

Originally Posted by sharkypinoy

but also keep in mind that the Q6600 has 4mb cache, where the E6300 has only 2mb... so that might show the trend for those who have 6mb cache with the E8x00 processors...

---

That won't work either. The 4MB cache in the Q6600 is split right down the middle with each "pair of cores" only seeing 2MB each. Only the Phenoms and now the i7 have a true unified cache that could be accessed by all its cores.

Good info. But what if the proc is a couple of years old? I have a 4870 on the way (if ups ever gets around to delivering it) and it will be paired up to my same 'ol X2 4400+.

CPU bottleneck, in this case?

Quote:

---

Originally Posted by r'aggro

Good info. But what if the proc is a couple of years old? I have a 4870 on the way (if ups ever gets around to delivering it) and it will be paired up to my same 'ol X2 4400+.

CPU bottleneck, in this case?

---

I was in a similar situation, but I have an opteron running at 2GHz. I upgraded to a 4850 from a 7800GT and saw a huge difference. With the 4870, I would expect the cpu to be the bottleneck.

Quote:

---

Originally Posted by r'aggro

Good info. But what if the proc is a couple of years old? I have a 4870 on the way (if ups ever gets around to delivering it) and it will be paired up to my same 'ol X2 4400+.

CPU bottleneck, in this case?

---

a 4400 runs at what 2.2ghz? i think it would be sufficient if you could overclock it to around 2.6ghz

I would be awsome if you ran the same benchies but with more up to date games. throw crysis in there, farcry2, CoD4, ut3, lets get some games that actually hurt your system, it would be good to see the results then. epsecially with crysis with dual vs quad would be interesting to see.

It really depends on the game still. Games based on Valve's engine do like processor speed as well as video card speed.

Also, vertex calculations, physics, and AI are still processed by the CPU. Depending on how heavy the game is with these factors can still have you in a CPU limited situation where the video card is waiting on data.

Quote:

---

Originally Posted by kpxgq

a 4400 runs at what 2.2ghz? i think it would be sufficient if you could overclock it to around 2.6ghz

---

That would be good, I've run this X2 @ 2.6 since new. If not I'll look into a platform upgrade.