[TBG] The Best Gaming CPUs: Pentium vs. Core i3 vs. Core i5 vs. Core i7

[Deders]

They should have tested Metro: Last Light and Arkham Origins with max PhysX. I found that in Arkham City at least that even Hardware PhysX is quite demanding on the CPU.

I played it first on my old 560TI and if I set PhysX to high it was unplayable unless my CPU was overclocked.

 

[BrightCandle]

Then problem with all the CPU comparisons I have seen is they just don't show enough games to show the real variety of results you can get. Only about 5% of games show an advantage to a 6 core CPU over the latest, but then they may show up to a 50% gain. Then you have a lot of games that really don't care, 2 cores is enough and they aren't even dependent on clock speed. You have the games that in certain areas really need as much performance as possible and in others and much easier on the CPU.

One day I am hoping some site over than gamegpu.ru will take up the challenge of frametimes in multiple games with different CPUs and GPUs and finally start to build a picture of genuinely what this looks like. But I think most reviews are doing us a disservice right now and driving us to the wrong conclusions due to the limited data they are looking at.

 

[sm625]

I've built two G3258 systems so far, one is 4.4GHz the other is 4.5GHz. What's with the paltry 3.7?

Also, when certain games score badly on the pentium, eg Crysis 3 and BF4, what effect would reduced settings have on the overall experience. How bad would you have to make the game look in order to bring the FPS up?

 

[Deders]

I've built two G3258 systems so far, one is 4.4GHz the other is 4.5GHz. What's with the paltry 3.7?

Also, when certain games score badly on the pentium, eg Crysis 3 and BF4, what effect would reduced settings have on the overall experience. How bad would you have to make the game look in order to bring the FPS up?

This is a clock for clock comparison meant to emphasise differences in architecture like Cache and Hyperthreading etc.

 

[wand3r3r]

It appears they were trying to balance the playing field. I believe Tech Buyer's Guru clocked all the CPU's to 3.7 Ghz so the clock speed was identical -- thus making the performance boost of extra cores and hyperthreading more easily identified.

Personally I think their conclusion was dumb. They said skip the G3258 and get an i3 4158. I can't imagine why any gamer should buy a locked CPU.
People really need to step up for the 4690K or 4770K -- or settle for a G3258 to get started.

Exactly, the others were already at 3.7 so they just bumped up the clocks to even the comparison. Now the inherit product differences will stand out, cache/HT/cores etc.

As for the i3 vs G3258 it obviously depends on your finances, you are free to interpret the results yourself. If you value $50 over the gains that the i3 would bring feel free, just be aware of what you are potentially compromising. I guess I would agree with their conclusion because the gap becomes much smaller between all the more expensive CPUs, and the 50% price increase sure doesn't bring much in the selected games. Personally if money were the main concern I would still hate to know that I was compromising too much.

The explanation on the 3.7 GHz.

We considered using one of the fixed-clock models (the fastest is 3.5GHz), but instead went with the enthusiast-favorite G3258. Now, it’s well-documented that this chip can reach 4.5GHz, but given the existing symmetry in the Intel lineup, we decided to simply clock it to 3.7GHz and provide our readers a thorough clock-for-clock comparison of all four main Intel processor families.

 

[AtenRa]

I've built two G3258 systems so far, one is 4.4GHz the other is 4.5GHz. What's with the paltry 3.7?

Also, when certain games score badly on the pentium, eg Crysis 3 and BF4, what effect would reduced settings have on the overall experience. How bad would you have to make the game look in order to bring the FPS up?

I can speak for BF4 MP and tell you that the game with dual core CPUs is unplayable even if you play at low settings 720p at 60fps.

 

[crashtech]

I can speak for BF4 MP and tell you that the game with dual core CPUs is unplayable even if you play at low settings 720p at 60fps.

That's certainly not been my experience, though perhaps we do not share the same definition of playable. I suppose since this seems to be a contentious issue I might have to try and record some gameplay on his machine.

edit: I'm wondering if you have tried this lately, with a worthy GPU like a 280x or better.

 

[AtenRa]

That's certainly not been my experience, though perhaps we do not share the same definition of playable. I suppose since this seems to be a contentious issue I might have to try and record some gameplay on his machine.

edit: I'm wondering if you have tried this lately, with a worthy GPU like a 280x or better.

Im not talking about 16 player maps, try at 48 players and higher up to 64 player maps and see how the game behaves.

 

[MiddleOfTheRoad]

That's certainly not been my experience, though perhaps we do not share the same definition of playable. I suppose since this seems to be a contentious issue I might have to try and record some gameplay on his machine.

edit: I'm wondering if you have tried this lately, with a worthy GPU like a 280x or better.

Well, BF4 pushes the CPU on multiplayer maps -- A dual core G3258 and i3 will jitter badly when you've got 32+ players. No way is it good for them to recommend a dual core for "Best Gaming CPU" -- not even an Haswell. If you can't afford an i5, then at the minimum get an unlocked Athlon X4..... Even an overclocked G3258 will stutter on the newer multiplier games. Don't get me wrong, I absolutely love the G3258 -- it's a wonderful chip for playing single player games.

it just can't handle enough threads to support modern games with 32 - 64 players on the map. No dual core should be named "Best Gaming CPU" in 2014. If this were 2012, sure.

 

[cbn]

I can speak for BF4 MP and tell you that the game with dual core CPUs is unplayable even if you play at low settings 720p at 60fps.

What dual cores are you using?

I'm using Pentium G3258 @ 4.5 GHz and R7 250X at 1080p low on 64 player (BF4 and BF3) and the game plays great IMO.

 

[AtenRa]

What dual cores are you using?

I'm using Pentium G3258 @ 4.5 GHz and R7 250X at 1080p low on 64 player (BF4 and BF3) and the game plays great IMO.

Can you post a picture with both Render.DrawScreenInfo and PerfOverlay.DrawGraph enabled using the Pentium G3258 and 250X ??

 

[cbn]

Well, BF4 pushes the CPU on multiplayer maps -- A dual core G3258 and i3 will jitter badly when you've got 32+ players. No way is it good for them to recommend a dual core for "Best Gaming CPU" -- not even an Haswell. If you can't afford an i5, then at the minimum get an unlocked Athlon X4..... Even an overclocked G3258 will stutter on the newer multiplier games. Don't get me wrong, I absolutely love the G3258 -- it's a wonderful chip for playing single player games.

it just can't handle enough threads to support modern games with 32 - 64 players on the map. No dual core should be named "Best Gaming CPU" in 2014. If this were 2012, sure.

I think we need more testing.

I don't know where the ideas of jittering and stuttering come from?

Maybe this comes from people using GPUs and detail settings that impose too much overhead on the processor?

 

[crashtech]

Well, BF4 pushes the CPU on multiplayer maps -- A dual core G3258 and i3 will jitter badly when you've got 32+ players. No way is it good for them to recommend a dual core for "Best Gaming CPU" -- not even an Haswell. If you can't afford an i5, then at the minimum get an unlocked Athlon X4..... Even an overclocked G3258 will stutter on the newer multiplier games. Don't get me wrong, I absolutely love the G3258 -- it's a wonderful chip for playing single player games.

it just can't handle enough threads to support modern games with 32 - 64 players on the map. No dual core should be named "Best Gaming CPU" in 2014. If this were 2012, sure.

I'd like to examine hard info that shows the Athlon X4 750 or 760 will eclipse the performance of an overclocked G3258 in a variety of games. Most sites have stopped recommending them now after head to head testing. The review by Tom's in particular made it look like the G3258 was better overall, with the added benefit of getting into the newer LGA 1150.

http://www.tomshardware.com/reviews/pentium-g3258-overclocking-performance,3849.html

 

[crashtech]

Im not talking about 16 player maps, try at 48 players and higher up to 64 player maps and see how the game behaves.

I'm willing to admit I may be mistaken, since I was only observing the frame rate counter and asking for user feedback as to the fluidity of the game, but I would suggest you test again with the latest BF4 version, it seems to me that something may have been patched that improves performance.

 

[AtenRa]

I'd like to examine hard info that shows the Athlon X4 750 or 760 will eclipse the performance of an overclocked G3258 in a variety of games. Most sites have stopped recommending them now after head to head testing. The review by Tom's in particular made it look like the G3258 was better overall, with the added benefit of getting into the newer LGA 1150.

http://www.tomshardware.com/reviews/pentium-g3258-overclocking-performance,3849.html

Well this is what I'v been saying about dual cores. You may have high fps but you have terrible gameplay.

The G3258 has higher fps than 750K

but then have a look at Frame Times,

I have also experienced this behavior in my Thief Mantle review with the Pentium G3420.

Same happens in BF4 even in Single mode, just imagine what happens in 64 Player maps in MP mode.

 

[AtenRa]

I'm willing to admit I may be mistaken, since I was only observing the frame rate counter and asking for user feedback as to the fluidity of the game, but I would suggest you test again with the latest BF4 version, it seems to me that something may have been patched that improves performance.

I have said this in another topic, im experiencing lagging even with my Core i7 3770K @ 4.44GHz with HD7950 @ 1GHz at 64 Player maps. I dont even dare to disable HT and 2 Cores to see what will happen.

edit: I have the Pentium G3258 and ill do more testing when ill return from my vacations

 

[crashtech]

Well if we mean any drop below 60 to be lagging then I ought to immediately concede. Also, it's normal to post graphs that agree with one's premise, but readers should look at all the games tested and come to their own conclusions. It's telling that Tom's now recommends the G3258 on the low end for OCers, not the X4 750K.

 

[AtenRa]

Well if we mean any drop below 60 to be lagging then I ought to immediately concede. Also, it's normal to post graphs that agree with one's premise, but readers should look at all the games tested and come to their own conclusions. It's telling that Tom's now recommends the G3258 on the low end for OCers, not the X4 750K.

For older games the Pentium G3258 is faster than 750K, but newer games need 4 threads or more and I cannot recommend any dual core CPU today.

Edit: When i say im experiencing lagging it has nothing to do with fps drops since i always have more than 60fps constantly. No, it is like you have high pings but in reality my pings are low.

 

[MiddleOfTheRoad]

I'd like to examine hard info that shows the Athlon X4 750 or 760 will eclipse the performance of an overclocked G3258 in a variety of games. Most sites have stopped recommending them now after head to head testing. The review by Tom's in particular made it look like the G3258 was better overall, with the added benefit of getting into the newer LGA 1150.

http://www.tomshardware.com/reviews/pentium-g3258-overclocking-performance,3849.html

They may not match the frames per second of an overclocked g3258, but the quad core FM2 Athlons stutter a lot less than any of the dual core CPU's i've ever used in Battlefield 4. The stuttering is VERY bad on large player maps. BF4 will try to push 8 threads if your CPU can support it.... A dual core is poorly suited for that game outside of single player mode. People really need to spend the extra money on an i5/i7 if they are planning to play multiplayer BF4. A dual core is going to lead to a poor experience with that game.

A modern gamer really needs a quad core. The G3258 is a wonderful CPU for the price, but it's probably 2 years too late to be the legend that it could have been.

 

[MiddleOfTheRoad]

Well if we mean any drop below 60 to be lagging then I ought to immediately concede. Also, it's normal to post graphs that agree with one's premise, but readers should look at all the games tested and come to their own conclusions. It's telling that Tom's now recommends the G3258 on the low end for OCers, not the X4 750K.

The 750k is old now..... Nobody should go there for a FM2 build at this point. If you are building an FM2 system, the 760K is the wiser investment.

I really would take Tom's conclusion with a grain of salt. It makes very little sense to be testing a 3 generation old AMD chip (circa 2012 Trinity) against a brand new Intel chip. He should have tested the 760K for an accurate reflection of what budget gamers are using right now for AMD builds.

Holy cow, AtenRa -- that Frame Rate Variance chart is ugly.... I didn't realize the dual cores were quite that bad.

 

[cbn]

Can you post a picture with both Render.DrawScreenInfo and PerfOverlay.DrawGraph enabled using the Pentium G3258 and 250X ??

Here are two screenshots (G3258 @ 4.5 GHz/R7 250X/1080p low) on Goldmund Railway with 64 players:



In the second shot, I'm taking some fire:

 

[cbn]

They may not match the frames per second of an overclocked g3258, but the quad core FM2 Athlons stutter a lot less than any of the dual core CPU's i've ever used in Battlefield 4. The stuttering is VERY bad on large player maps. BF4 will try to push 8 threads if your CPU can support it.... A dual core is poorly suited for that game outside of single player mode. People really need to spend the extra money on an i5/i7 if they are planning to play multiplayer BF4. A dual core is going to lead to a poor experience with that game.

.......but the game seems to run just fine for me with G3258.

Worst case scenario I've run across (so far) is vehicles in the air on large maps and even that is not really that bad IMO.

With that mentioned, I am open to the idea of one day testing a quad core AMD alongside the G3258. But I wonder just how much improvement would I really see?

 

[MiddleOfTheRoad]

.......but the game seems to run just fine for me with G3258.

Worst case scenario I've run across (so far) is vehicles in the air on large maps and even that is not really that bad IMO.

With that mentioned, I am open to the idea of one day testing a quad core AMD alongside the G3258. But I wonder just how much improvement would I really see?

From what I've seen, the AMD 760K will probably drop a couple FPS versus the G3258 but even out the variance -- making the actual gameplay smoother/more consistent.

It's again becoming more of a personal preference. If you are playing older games and want maximum frames per second, the overclocked G3258 wins big time. If you want to play modern games in large multiplayer maps, then the X4 760k is probably the better CPU.

 

[Makaveli]

You mean SimCity 2? Because that game uses a many threads as Starcraft 2 does...
http://www.bit-tech.net/hardware/cpus/2010/08/18/how-many-cpu-cores-does-starcraft-2-use/2

ya sc2 is locked to two cores and only seems to care about IPC.

Not the best choice.

 

[cbn]

From what I've seen, the AMD 760K will probably drop a couple FPS versus the G3258 but even out the variance -- making the actual gameplay smoother/more consistent.

What resolution and graphical detail settings (along with GPU) do you use when your G3258 vs. AMD quad core comparisons in 64 player BF4?

P.S. If there is significant frame time variance with my set-up, I rarely notice it. Stuttering only happens on extremely rare occasions and it doesn't appear to be consistent at all.