With the wealth of reviews coming online over the last few months on AMD’s Trinity APU I thought it time to give the platform a onceover myself to see what the flagship FM2 chip has to offer. With the A10-5800K currently being AMD’s biggest offering it’s time to push the chip through the ringer and compare it to it’s main competitor. Having done the research at the cash register it seems unfair to pit the 5800K against a 3770k as AMD’s offering is under half the price so I chose to go the price-point route and will compare against Intel’s i3 3220. On the surface it may seem cheeky to compare a Quad core with a Duo with HT baring in mind that FM2 is currently not AMD’s 'big boy' socket the pricing route should be a good match. The A10-5800K is AMD's second-generation mainstream APU platform. Codename Trinity the APU sits in AMD's roadmap as the linchpin to it's main target markets - notebook & general purpose desktop, (to keep things tidy I'm not going into AMD's conformation of being at the heart of the new PS4 or for that matter that rumours have the A10-5800K is in the PS4 development kit. Speculation I know but I highly doubt it'll be a Trinity APU in the PS4 but having gone this far to mention it, with those sort of sales on the horizon you can certainly see the incentive for companies to explore alternative avenues). I could go on about the technical’s of AMD bringing on support for SSE4, AVX, AES-NI and all the rest but unless you’re an electrical engineer, regular enthusiasts like me want to know the simple things… how do the dam things perform? For the purpose of fairness between the two rivals I chose to stick with as similar a tech bench as I could muster. Same Rams, same PSU, same HDD and of course the same prepped OS, (Win. 7 Ultimate 64-Bit). With the obvious motherboard issue in hand I’m lucky enough to pit Gigabyte’s best of both sockets against each other. Test Benches: AMD A10-5800K w/stock HS & Fan assembly Gigabyte F2A85X-UP4 Team Xtreem 2400MHz Patriot Pyro 60GB SSD Corsair HX1050w PSU Intel I3 3220 w/stock HS & Fan assembly Gigabyte Z77X-UP7 Team Xtreem 2400MHz Patriot Pyro 60GB SSD Corsair HX1050w PSU Using a mix of 2D & 3D benchmarks I hope to be able to demonstrate the difference in performance, if there is one between the two chips as well as discuss the benefits and potential applications each platform may have. All benchmarks will be run on respective platforms at full stock settings, on their stock heatsink assemblies, as well as using XMP for the rams. Interestingly with both systems built neither was able to operate the rams at their stock rated XMP speeds of 2400MHz @ 9-11-11-27 2T with the required 1.65v without a certain amount of tinkering. So for the purpose of running the benchmark process fairly I reduced the ram speeds to best fit for each platform: AMD - 2130MHz @ 8-10-10-25 1.65v Intel - 2200MHZ @ 9-11-11-26 1.65v It is also worth noting that the A10-5800K, when planted in the UP4 defaulted to a 40 multi / 4GHz core clock speed. With it boxed as 3.8GHz and having a turbo boost to 4.2GHz I chose to leave the 4GHz alone and press on.. For a good proportion of buyers looking at this platform I surmise that this is what would likely happen in the real world. At a later date I will analyse the A10s overclocking abilities and will address the increased clock speeds and performance. The benchmarks used for this stage of the review in order are: 3DMark 06 3DMark 11 3DMark Vantage Aliens VS Predator Cinebench 11.5 Lost Planet 2 DX9 & 11 Super Pi 1M & 32M Stalker Call of Pripyat Unigine Heaven MS Windows Experience Wprime I've also chosen to monitor power consumption, being pulled from the wall, throughout the testing window and will show the final result. Behind each graph I've also linked the lead screen shot for reference. Benchmarks: 3DMark06 (Higher is better) With the first benchmark completed it is obvious to see that the A10s discreet GPU, the 7660D, is likely to be a far superior offering. 3DMark 11 (Higher is better) This result is pretty interesting to me for all the wrong reasons. Yet again the onboard graphics from AMD's unit significantly pulls ahead but if you look closely at the scores you'll notice that physics is very similar. With 3DMark 11 being based on the Bullet physics engine I had assumed before running the benchmark that AMD would have been able to pull ahead, if only on a driver level, due to their ability and historical proficiency in being able to manipulate Tessellation. The fact that Intel are level-heading here strikes me that either they have pulled the rabbit from the hat or AMD's left hand is not talking to the right? Alien Vs Predator (Higher is better) The AMD unit smashes it's way through this benchmark with four times the average FPS. A strong result indeed. Cinebench Open GL (Higher is better) Once again the 5800K delivers a very respectable score for onboard graphics. This is really starting to show where AMD is shining. Cinebench CPU (Multi-Thread) (Higher is better) Interestingly both CPUs come in with the same score. I will cover this a little more in my conclusion. Lost Planet 2 DX9 Test A (Higher is better) Lost Planet 2 DX9 Test B (Higher is better) Lost Planet 2 DX11 Test A (Higher is better) Lost Planet 2 DX11 Test B (Higher is better) As we see above the A10 performs admirably even when compared to entry level graphics cards like Nvidia's 630. I have to admit that for an onboard unit I am impressed. Super Pi 1M (Lower is better) Super Pi 32M (Lower is better) Ouch! This is where things start to change. With Super Pi being a single threaded it's a good benchmark to highlight CPU's core speed. As we can see Intel has the edge, the blade, the hilt and is encroaching on the handle. Stalker Call of Pripyat (Higher is better) Another solid return on FPS from AMD. Not only is the Max and Average frame rate consistently solid but unlike some Ati graphic cards the minimum frame rate is well under control. Unigine Heaven (Higher is better) Knowing that Unigine Heaven uses Tessellation to the point that it has brought low-end Ati graphics cards to it's knees I had assumed before running the benchmark that the 5800K would have been given a hard time here. Interestingly this was not the case. The 5800K scoring over double the overall FPS of the 3220 and having a very acceptable minimum frame rate. This result strikes me as at odds with the 3DMark 11 result but it's great to see AMD's ability to build on previous architecture. 3DMark Vantage (Higher is better) Now this result really surprised me. Expecting to see AMD pull ahead on the CPU result, being a Quad core unit over Intel's Duo it was disappointing to see the Duo move ahead in a multithreaded benchmark. Saying that we can see the 5800Ks discreet GPU absolutely crushed the graphics side of the benchmark and ultimately the overall score. Windows Experience Unsurprisingly AMD comes out on top but it was interesting to note that the Windows Experience Memory test didn't prefer the tighter timings over the extra MHz on the rams but it did like the two extra true cores the 5800K has over Intel's Duo with HT. Wprime 32 (Lower is better - i3 3220 linked) Wprime 1024 (Lower is better - A10-5800K linked) These results really reflect where my gut feeling when looking at this review. Clock for clock the i3 3220 is the stronger unit but its dominance is not linier in multi-threaded applications. Power Consumption (Lower is better) As you can see the 5800K has good low power draw at idle increasing steeply under load unlike the 3220 that remains fairly consistent throughout. Personally speaking, power draw for me is a low priority unless I am working with machines that are permanently left running or have a mobile application effecting battery time so it's is good to see AMD has built on Llano. Conclusion: Sitting at the bench building this review, my final thoughts are a little troubled. The FM2 / A10 platform responded well, felt snappy to operate and has kickass onboard graphics. Price wise it's a very competitive unit. To get the Intel rival to meet the same gaming potential a buyer is looking at spending another $100 on a GFX card to bring it up to muster. Thing is.. for AMD's leading current Gen Quad the 5800K just doesn't have the raw clock speed I personally want to see. AMD just seems to have abandoned the competition for brute strength with Intel leaving consumers with a one horse race if that's what they are looking for. Furthermore it is worth noting that to date the FM2 platform has not adopted Pcie3. There is a very valid argument that the Gen 2 bus is still not being exhausted in current use but for most it is a feature expected and it's inclusion goes somewhat towards the argument of future proofing. That said this APU has be aligned with the budget conscious. Not that it won't be done but I doubt many FM2 rigs will be put together with Crossfire 7970s or $1000 Gen 3 raid cards but for consumers it would be nice to have the choice. Having built too many systems to consider remembering there is most certainly a place for Trinity in the desktop market. If you are looking for a home server or 24/7 encoder this is not the platform for you, (although from personal experience it's faster on Handbrake than the i3). When I'm next asked to put together a general purpose desktop for small business, a HTPC or arguably a budget LAN gamer I wouldn't be looking further than the A10-5800K. As for the i3 3220, it is disappointing that Intel have not released an unlocked multi Duo since 1156. No doubt Ivy bridge packs a punch with is raw processing power and the iGPU onboard the 3220 comes with is more than enough to watch Blu-ray, online or terrestrial TV. If your a bit of a retro gamer is owns Command & Conquer but.. if you wanted to play something a little more modern at 1080p you'll be wanting to add 3rd party GFX. The upside to both platforms is the price and performance delivered for what would ultimately cost you less than half a single top-end GFX card on today's market is excellent. If I had $300 in my pocket I know which machine I would be putting together. Thumbs up AMD.