Who's Got Game? Twelve Sub-$200 CPUs Compared
-
Page 1:The Sandy Bridge Factor
-
Page 2:Test System And Benchmarks
-
Page 3:Benchmark Results: Metro 2033
-
Page 4:Benchmark Results: Lost Planet 2
-
Page 5:Benchmark Results: Aliens Vs. Predator
-
Page 6:Benchmark Results: F1 2010
-
Page 7:Benchmark Results: Just Cause 2
-
Page 8:Benchmark Results: StarCraft 2
-
Page 9:Benchmark Results: Multitasking
-
Page 10:Conclusion: Sandy Bridge Has Game
Test System And Benchmarks
Testing LGA 1156-, LGA 1155-, and AM3-based CPUs requires three separate platforms. All use the same memory and hard drive to eliminate those factors as performance variables.
We're using Gigabyte's GA-MA790XT-UD4P in our AMD-based platform. This board features the 790X chipset, which should have no trouble delivering similar performance as any of the 800-series chipsets.
Intel's LGA 1156-based system employs Gigabyte's H55M-UD2H, a board that we've seen perform well in a number of articles, including the previous sub-$150 gaming CPU comparison.
A single GeForce GTX 480 drives all three of these builds. Just to be clear, we selected this card not because it's something we recommend pairing with a low-end CPU, but because it's a powerful board that should serve to alleviate any potential graphics bottleneck in our quest to explore CPU performance.
| Simulated CPU Chart | |||||
|---|---|---|---|---|---|
| AMD Athlon II X3 455 | Intel Core i3-550 | Intel Core i3-560 | Intel Core i3-2120 | Intel Core i5-2300 | |
| Simulated With: | AMD Athlon II X3 450 | Intel Core i3-540 | Intel Core i3-540 | Intel Core i3-2100 | Intel Core i5-2500K |
| Target Clock Speed: | 3300 MHz | 3200 MHz | 3330 MHz | 3200 MHz | 2800 MHz (3100 Turbo) |
| Actual Clock Speed: | 3304 MHz | 3197 MHz | 3335 MHz | 3186 MHz | 2807 MHz (3097 Turbo) |
| Target Memory Speed: | 669 MHz @ 9-9-9-20-1T | 669 MHz @ 9-9-9-20-1T | 669 MHz @ 9-9-9-20-1T | 669 MHz @ 9-9-9-20-33-1T | 669 MHz @ 9-9-9-20-33-1T |
| Actual memory Speed: | 688 MHz @ 9-9-9-20-1T (19 MHz o/c) | 695 MHz @ 9-9-9-20-1T (36 MHz o/c) | 725 MHz @ 10-10-10-22-1T (56 MHz o/c) | 707 MHz @ 9-9-9-20-33-1T (48 MHz o/c) | 646 MHz @ 9-9-9-20-33-1T (23 MHz u/c) |
Because some of our CPUs have to be simulated, memory clocks are a little tricky. We're staying as close to the base specification (669 MHz memory with 9-9-9-20-1T timings) as possible.
Despite the small memory speed inconsistency for these simulated processors, the results fall precisely where we would expect them to be at standard clocks. Any performance differences appear to be well within the margin of error.
| Intel LGA 1156 Test System | AMD Test System | Intel LGA 1155 Test System | |||
|---|---|---|---|---|---|
| CPU | Intel Core i3-550 | AMD Athlon II X3 455 3.3 GHz (Rana)** **simulated by overclocking an Athlon II X3 450 | Intel Core i3-2100 3.1 GHz (Sandy Bridge) Intel Core i3-2120 3.3 GHz (Sandy bridge)*** Intel Core i5-2300 2.8 GHz (3.1 GHz Turbo) (Sandy Bridge)**** Intel Core i5-2400 3.1 GHz (3.4 GHz Turbo) (Sandy bridge) ***simulated by overclocking a Core i3-2100 ****simulated by underclocking a Core i3-2500K | ||
| Motherboard | Gigabyte H55M-UD2H LGA 1156 | Gigabyte GA-MA790XT-UD4P Socket AM3 | Asus P8P67 Pro Chipset: Intel P67 | ||
| Networking | Onboard Gigabit LAN controller | ||||
| Memory | OCZ PC3-16000 | ||||
| Graphics | GeForce GTX 480 Reference | ||||
| Hard Drive | Western Digital Caviar Black 750 GB | ||||
| Power | ePOWER EP-1200E10-T2 1200W | ||||
| Software and Drivers | |||||
| Operating System | Microsoft Windows 7 x64 | ||||
| DirectX version | DirectX 11 | ||||
| Graphics Drivers | GeForce 266.58 WHQL | ||||
| Benchmark Configuration | |
|---|---|
| 3D Games | |
| Metro 2033 | Version 1.0.0.1, DirectX 11, benchmark tool, High Quality, No AA, 16x AF, advanced physX disabled, tesselation enabled, DOF disabled |
| Aliens vs Predator | Version 1.0.0.0, Aliens vs Predator DirectX 11 benchmark Default Settings, No AA, 16x AF |
| Lost Planet 2 | Version 1.0.1.129, DirectX 11, Highest Settings, 4x AA, In-game benchmark Test A |
| F1 2010 | Version 1.1.1.129, DirectX 11, in-game benchmark Ultra Detail Settings, 4x AA |
| Just Cause 2 | Version 1.0.0.2, DirectX 10, Concrete Jungle Benchmark Highest Details, 4x AA, 16x AF, Bokeh Filter & GPU Water Simulation off |
| StarCraft 2 | Version: 2.1.2.2105, Ultra-High Settings Tom's Hardware Guide Benchmark |
- The Sandy Bridge Factor
- Test System And Benchmarks
- Benchmark Results: Metro 2033
- Benchmark Results: Lost Planet 2
- Benchmark Results: Aliens Vs. Predator
- Benchmark Results: F1 2010
- Benchmark Results: Just Cause 2
- Benchmark Results: StarCraft 2
- Benchmark Results: Multitasking
- Conclusion: Sandy Bridge Has Game
o_O I'd like me one of those
a pc will never be gaming only, unless you have more than one, in that case, for for the cheaper dual core hyperthreaded, but if you do anything else, get a real quadcore and don't even take into consideration the logical cores.
AMD build w/AMD Athlon II x3 455 w/Asus 870 based mobo:
$89 for Athlon II x3 455
$90 for AMD mobo (Asus) w/6xSATA 3, 6 USB 2.0, 2 USB 3.0, 4 x DDR3 slots.
ASUS M4A87TD/USB3 AM3 AMD 870 SATA 6Gb/s USB 3.0 AMD Motherboard
http://www.newegg.com/Product/Product.aspx?Item=N82E16813131651
Intel build w/i3 550 w/P55 based mobo (Asus also):
$130 i3 550
$150-$10 MIRc Comparable mobo (Asus also) 6xSATA 3, 6 USB 2.0, 2 USB 3.0, 4 DDR3 slots.
ASUS P7P55D-E LX LGA 1156 Intel P55 SATA 6Gb/s USB 3.0 ATX Intel Motherboard
http://www.newegg.com/Product/Product.aspx?Item=N82E16813131634
**These are all Newegg prices**
AMD build (using the same Case/PSU/RAM/DVD parts in both systems)
$179 + shared parts.
Intel build (same parts shared w/AMD build)
$280 + shared parts.
This equals out to ~$100 price difference between the 2 builds, which to me is quite a bit!
So in general when trying to factor in "Value" for the gaming buck I still see the AMD based system being the better buy. Assuming your using mobo's with about the same features. If you notice the Intel based mobo's will cost you more for similar AMD based mobo's. This is where a lot of the value comes from AMD. Don't get me wrong here, the Intel based system is very good system too, you just have to pay more for them.
Yes the amd build is pretty cheap, but swap you're i3 550 with the i3 2100 and the p55 mobo with the p67 mobo, and you have a build that is now worth the $100 over the amd build
also I think this conclusion summarizes well AMD's predicament in a months time:
"the Core i3-2100 performs as well as (or slightly better than) AMD's Phenom II X4 970 flagship."
I almost missed this typo an AMD pentium hmmm something seems wrong.
If the point is that Intel has the best budget system at the current prices - then yes, the point is made. But it looks more like you're trying to prove Intel is better just before AMD launches a new generation of CPU's. While I can't speak for anyone else, I'm at least going to give their next generation a chance.
Certainly no one needs quad cores for web browsing and word lol.
quote from the conclusion page:
"Because our charts are arranged in order of processor price, with the most expensive at the top and the least expensive at the bottom,"
but the first page shows the X6 being the most expensive CPU !!!
Also I miss the price/performance graph:
X3 455 = 88.99USD
i5 2400 = 189.99USD
from the average game performance graph:
i5 2400 offers 148% of X3 455 performance but it is priced at 213% of the X3 455 !!!
how about graphing the performance progression and cost progression in one graph to actualy see the best value for money ? the last graph only focuses on relative performance but price increase is not considered (at least only in the relative more/less expensive way).
and ... do NOT test HT on I/O intensive workloads (winrar). run something like super pi or a physics simulation on the CPU. I/O limited workloads are the most favorable HT scenario.
But they still have nothing to combat AMD's APU + GPU combo.
Wait yes they do, i3..
I wonder if AMD's planning to go out in a bombshell..
Or will they make a comeback to remember them by?
Its stupid compare CPUs in games where the graphic card scores under 50pfs...
would have been interesting to see some overclocked benchmark though, to see how the unlocked muliplyer of the Phenoms would do against Sandy bridge with no unlocked muliplyer.
And I know it's an older game, but GTA has the reputation of being a real cpu killer, so would have been interesting with that ass well.
And what did i learn? that my next cpu, will be a Phenom II 955, an upgrade from my current Athlon II 445.
Why not do a max OC comparison of these CPU's?? Especially given that at this price range most people who are buying these chips for their own rigs are trying to maximize their dollar and are probably at least considering doing so.
Too many "simulated" cpu's??
Amd supporters are finally willing to talk about OCed performance considering non-K SandyBridge is not OCable.
How about paying $15 dollars more for the 2500K and OCed it to 4.4GHz? 2500K can achieve 4.4GHz easily with stock cooler which gives you a lot more performance without requiring paying extra cost ($80+) for a high end cooler. You will then end up with a much better performance for less cost than the PII-X6@4GHz+high end cooler.
My old ass E4300 (which I got at the time for CHEAPER than the current SB procs) OCed to 3.5 ghz trashes the dual-core SB procs that can't overclock
Outside those who are willing to spend money on the SB mobos and are in the price range ($250 ish) for the 2500k/2600k, SB is completely pointless for overclockers of any stripe.
Since the AMD chips atm are still complete failures, people like me who are stuck with OCed core 2 duos and core 2 quads have nothing to look forward to in affordable high performance procs that are worth upgrading to until Bulldozer from the looks of it.