Game-Off: Seven Sub-$150 Processors Compared

[triculious]

nice article... proving once again that you don't need to spend an arm and a leg when purchasing a cpu for your gaming rig

looking forward for a part II where OC comes into game

it also would be nice to have a similar article where the budget goes up to 200... in my very personal opinion over 200 for gaming is worthless, but that's just me and my wallet speaking, and I would like to see if paying some 30-40 extra bucks would be worth it

 

[zorky9]

The multitasking experiment's a great addition! Hope you keep this test in your future reviews.

 

[megamanx00]

Interesting article. The Dual Core Pentium certainly did worse than I thought it would. It's interesting to see how more cores have an effect and the impact of the L3 cache on AMD CPUs. Honestly though I really want to know about Bulldozer

 

[MU_Engineer]

[citation][nom]liquidsnake718[/nom]I guess this is a symbolic article where-in Toms is also telling us that we have to all eventually let go of those dusting core2 chips...[/citation]

Not necessarily. The faster Conroes and slower Wolfdales are roughly on par with the Athlon IIs and the Pentium G6950, the faster Wolfdales are more in line with the slower Clarkdales, so they are certainly usable. The midrange and faster Core 2 Quads match up with the quad-core CPUs and faster Clarkdales tested here today.

What you would want to let go of is an early, slower Core 2 Duo setup, particularly one that can't take 45 nm chips. Yes, the E6850, Q6700, and such are competitive chips here, but they are very expensive compared to what performance they provide. Even the better 45 nm chips are overly expensive today as well, and you will quite possibly have more performance for the dollar in going for one of these CPUs in a new board rather than buying a faster Core 2 unit.

One question, what if these chips were compared to a Phenom X6 or even a core i7 1366 socket chip? If not, then one can truly compare the 5870 vs the gtx480 head to head.

Applications that are heavily multithreaded would perform considerably better on those CPUs than the ones tested today, but lightly-threaded and GPU-bound applications would show little benefit since the clock speeds of these CPUs is pretty similar to the LGA1366 i7s and Phenom II X6s.

 

[saint19]

Like always I say if u want a quad core or tri core, buy a REAL quad core or tri core

 

[failboat]

I wish this whole article was done overclocked. Most people who read this and buy sub 150 cpu's OC.

On another note, the amd x3 435, on newegg, is 65 oem. =) No heatsink, but if you were going to OC; you might as well get a better one.

 

[boogalooelectric]

I got my Phenom II X4 945 Black edition a couple of weeks ago for $117, it was an OEM version but still much cheaper than this chart shows, thus improving its impressive price/performance ratio. At least in my opinion.

 

[logainofhades]

Now to figure out if I want to sell the Athlon II X4 630 I have lying around and just go i3, get an AM2+ board and reuse the ddr2 ram in my file server, or new board and ram for the Athlon II. I was hoping this article would clear that up for me but it seems to be as clear as mud. I guess I will need to wait for the OC results.

 

[porksmuggler]

Top notch review Don. Now lets see it, the i3-530 vs Phenom II X4 940 overclocked. Is the AMD worth the extra $10, we're not leaving these builds stock you know.

 

[coldmast]

[citation][nom]luke904[/nom]sorry but i must disagree...the core i3 530 was 8% faster than the athlon X4 and costs $5 lessits a great processor it seems, a nice change from intel. but i admit, my heart sunk after seeing amd's athlon X4 get beat. its like sports, i root for AMDplease dont quote the multitasking benchmark as no sane person compresses stuff while gaming...yes the athlon would probably be better overall for most people, but not for gaming[/citation]

Certain Motherboards don't allow for SLI or Crossfire with the i5 600 or 500 series w/ integrated graphics. So which processor is 'not' for gaming?

 

[Stardude82]

[citation][nom]Cleeve[/nom]As we mention in the article, we stayed away from the LGA 775 platform. It doesn't really have a future.[/citation]

Neither does the 1156 or 1366 really. Intel just rolled out new 775 CPU's (E5500 and E3500) this month, too.

 

[Guest]

I bet Corei3 540 would best the Phenom in power specs.

 

[cleeve]

[citation][nom]Stardude82[/nom]Neither does the 1156 or 1366 really. Intel just rolled out new 775 CPU's (E5500 and E3500) this month, too.[/citation]

Hmmm. I guess we'll have to agree to disagree when it comes to 1156 and 1366, I anticipate more high-end stuff to come from those platforms in the future for sure.

I personally wouldn't consider the E5500 and E3500 CPUs that will turn LGA 775 into a viable platform for the future, but whatever floats your boat.

 

[Alvin Smith]

Yeah ... Would be good to see all these chips pushed to a "conservatively agressive" clock-rate of ~3.7GHz, ... They should all be able to exceed 3.6GHz "conservatively", with decent air cooling (no?).

 

[killerclick]

[citation][nom]Cleeve[/nom]If you're a corporation deploying thousands of PCs it might make a difference, but for the home user $3 a month isn't going to change your world.[/citation]

Well yeah $3 a month would be only $36 a year but why is it different from a CPU that performs the same and costs $36 more? Would you recommend one CPU over another in that case? I agree that the green thing has gotten a bit out of hand but if you combine a more efficient GPU, a more efficient CPU, 80plus PSU and a LED monitor, that could add up to a respectable amount in savings over the lifetime of the computer. Paying attention to such things would also stimulate hardware manufacturers to make more efficient products.

 

[pinkfloydminnesota]

I'd be interested in a survey of readers and the percentage who plan to overclock. Are we in the majority? Without the overclocking information, I feel I missing important information on the value differences between these parts.

Even in a single review when they overclock one part, it's rare to overclock the parts being compared.

I guess you feel most readers aren't going to overclock their components.

 

[randomizer]

[citation][nom]luke904[/nom]no sane person compresses stuff while gaming[/citation]
I've run ray tracing renders in the background while gaming, I guess I must be insane wanting to get work done while having fun.

 

[cleeve]

Well yeah $3 a month would be only $36 a year but why is it different from a CPU that performs the same and costs $36 more? Would you recommend one CPU over another in that case?

I pulled $3 a month out of my butt as an example of how futile the difference is, but if it seems significant to you let's look at the actual numbers:

The highest TDP of any of the CPUs is the Phenom II X4 940 with 125 W, the lowest TDP of any of the CPUs are the Clarkdales with a 73W max. That's a theoretical difference of 52W under full load.

Electricity is about 15 cents/killowatt hour in the most expensive parts of the US. that means each hour it costs 15 cents for 1000 watts of usage. So in the worst case scenario, under full load, the per hour cost difference between a Core i3 540 and a Phenom II X4 940 is 0.78 cents ((52w/1000w=0.052)x15 cents=0.78 cents/hour). Not even a penny.

Assuming someone games four hours a day, putting the CPU to full load, every day, for a 31-day month. The cost difference under full load is 0.78 cents x 4 hours, or 3.12 cents per day times 31 days, which equals 96.72 cents.

Almost a dollar. That's almost a dollar at the maximum theoretical limits, worst case scenario, gaming for four hours a day, every day. no days off, always full CPU load. Not even 12 dollars a year. And that's using the most expensive electrical rates in the country.

In real life, no CPU is pushed to it's max limits for four hours a day, even if the user games for four hours a day. With normal desktop usage past experience has shown us that we would expect to see a 10-20 watt difference because all of these CPUs will throttle down and sip the juice unless they're pushed. And it all becomes irrelevant when the PC is put to sleep.

And there are other factors. Which chipset do I use? What peripherals? What power supply? You can't easily apply a calculation to a CPU vs. CPU and say "this one costs less!".

All of this is why I have a hard time being concerned about power usage differences with CPUs, especially sub-$150 models.

 

[cleeve]

[citation][nom]pinkfloydminnesota[/nom]I guess you feel most readers aren't going to overclock their components.[/citation]

Frankly, I know I'd personally agree with that. The percentage of people who overclock are a tiny fraction of the people who run at stock speeds.

Having said that, I always add overclocking if it's reasonable to do so. I value it and I believe a lot of our readers do, too, even if they don't plan to overclock personally.

But you have to understand that we have a job to do. I have to produce a reasonable amount of content. You can't overclock a CPU in an hour and declare it stable. Overclocking takes time, stress runs to test it out, more time, more stress runs. We don't want to report a useless overclock. We want to report the limit of what can be reasonably accomplished, and the time investment is astronomical when you introduce overclocking.

For this reason I usually try to limit overclocking to a 1-on-1 comparison where I have time to do it properly. My schedule simply did not allow enough time to accurately and reliably overclock 5 CPUs for this article, so I chose to leave it out instead of doing a half-assed job. That's the unfortunate reality.

 

[cleeve]

Yeah ... Would be good to see all these chips pushed to a "conservatively agressive" clock-rate of ~3.7GHz, ... They should all be able to exceed 3.6GHz "conservatively", with decent air cooling (no?).

You might have to push some of the Athlons/Phenoms pretty hard to get there. A couple might not make it. The Clarkdales would probably all make it there pretty easy tho.

 

[seerwan]

As the article says in the end, you need to overclock them to arrive at the most conclusive results on which is best.

The i3-540 would probably defeat the AMD 940/945.

This is especially true as these are cpus which are likely to be overclocked as they are for the budget gamer, who need to get as much bang for the buck as they can.

 

[alcalde]

"please dont quote the multitasking benchmark as no sane person compresses stuff while gaming..."

This isn't 1999 when my 450MHz K6-III+ took longer than real-time to create MP3 files. The whole promise of multicore IS to be able to do stuff at the same time. Nowadays, I don't even think twice when I start a batch video compression job then fire up a game. The multitasking chart should have made it clear to you that it is indeed something you can do.

These gaming results really make me feel lucky... in December I got a 2.9GHz Athlon II X3 for $79.99 shipped after coupon, and I was able to unlock the fourth core and overclock it to 3.4GHz without adding any voltage. That should put its results about 13% higher than the Athlon II X4 on this chart....

 

[JonnyDough]

Wouldn't it make more sense to use the Athlon II X3 440 (3.0ghz, which appears to be the average/median but don't quote me because I didn't bother doing the math) rather than the 445? 100mhz makes virtually no difference and it is $15 cheaper.

 

[aihyah]

needs rar creation time, for all we know how the cpu is divided between the apps might vary.

 

[arkadi]

The Multitasking test is a must have...finally we can see it here. Still i think you can think of a better way to implement it. Anal benchmarks not necessarily show the full picture. Good thinking on this one.