AMD FX: Round 2
It ha been almost a month since the launch of the FX processors from AMD. During that time AMD has been taking a beating on the review sites and even here we found the chip less than inspiring in it’s performance. But looking back at our review I realized we had strayed from our normal review formula. Rather than taking a more subjective look at the performance of the FX we had instead gone with a traditional benchmarking approach. Now the reason for this was time limitations on getting the review done as well as this being a totally new chip designing and my desire to a comparison of the new design to the old. With this in mind shortly after the first review was completed I began our normal subjective batter of tests.
First a quick refresher, the FX processor is AMDs new “Bulldozer” based chip that uses a modular core design putting two cores on a module. The modules share various resources including the FPU. This design is aimed squarely at heavy multithreaded applications an multitasking usage.
Okay now to the testing, the first thing we look at at was what to make the comparison against in our subjective testing. The chip we have in the lab is the FX 8150, while this was supposed to go for a lower cost the current cost of this chip is near $280. This puts it from a cost view point very close to the 2600K. The FX 8120 which is the same chip with a lower clock speed is going for $219 and is from a cost view point the direct competition to the 2500K. With that in mind I spoke to some people and AMD and was assured that I could adjust the speeds down to equal those of the 8120 and get results that would be identical to buying an 8120 chip.
For testing I wanted to put the systems on equal footing and so I chose for both systems to use a 6850 video card, 8 gigs of Kingston Hyper X 1600, and both systems ran at stock speeds. For desktop settings I choice 1080 as the working resolution and both systems had a clean install of Windows 7 Professional x64 with the latest drivers and updates.
For the first round of subjective comparison I chose the most common usage of a computer, I played with internet browsing. I did everything from Hulu and Netflix to watch video to Pandora for music and even pushed things by using Office Live and Photoshop Online. The idea was to spend some time on each system, essentially doing the same tasks and seeing what the results would be. Using a few other people to do this same test as well with them having no idea which system was which I compared the findings and drew my conclusions. Round one was a solid tie with neither system leaping out and showing us anything outstanding. Even when pushed with heavy multitasking, i.e. several instances of IE running as well as Office applications in the back both systems delivered smooth and strong performance.
For round two we did some more advanced work with photo editing and resizing as well as file compression, opening and creating various compressed files. For purposes of this test we used the built in Windows compression tool for zip files and opened other compression formats using 7-Zip. Now I left the other testers to their own personal methods to replicate a true real world usage for this testing. I also added some batch picture editing, 10 at a time as well as some larger compression jobs, 750 meg of files. The results where identical to the first test with no one being able to really see a difference in everyday usage of the systems. Even my harsher testing which is still closer to everyday use than benchmarks typically are could not reveal any difference in actual use.
I threw in another round here that was purely me but then again I do not know many people that do a lot of audio editing. Using Adobe Audition I did a series of file conversions and editing on some of the show archives. This ranged from changing the audio format to applying various filters and adjustments to the entire hour long archived files. The 8120 clocked FX was again in almost a dead heat with the 2500K.
The final round of subjective testing was gaming, for this we used Champions-Online, Supreme Commander, Demi-God, Dirt 3, Dragon Age and a game that is in beta and cannot be named. All of the games where set to 1080 resolution and then to high on their detail settings in game. With the games all setup we fire them up and went to playing. Again with all the games the game play was smooth, no stutter, and the experience very enjoyable.
Now reading this you might think, god this is the most boring review I have ever read, I do not blame you. However I think there is a solid point to be made from these findings. The FX processor may not win any races but it does not completely crap our. For a bonus round I bumped the FX speed back to those of the 8150 and redid some of the testing and found the same results, so these results are not all good for the FX lineup. In actual use, the way most people will use a PC the FX lineup is just middle of the road, it does not deliver any BANG.
I wanted to try and get more insights into AMDs direction with the FX processor so I request on interview be taped where we could discuss the chip and what the target was for todays show. I was greeted with an enthusiastic response early in my request and waited for the for the scheduling of the interview to get done. After over a week of waiting I was surprised to be told that AMD was declining my offer for an interview. I was however allowed to email in some questions to get answers. I put together four base questions to get the conversation started. I am tempted to post the responses here but I am not going to, the reason is they where none responsive responses.
You see the questions I posed where about the target the chip was aimed at for sale, what needed to happen for the FX chips potential to be seen, why pushing overclocking and not actually support it and finally the fate of the Phenom II. The responses I got where boiler plate PR BULLSHIT! Yes I said the word! I had asked for a chance to do a candid interview and they declined that and then offered a written one and instead of candid I got the typical PR crap. I did some follow up questions but a week later those have gone unanswered. So I will not insult my listeners and readers with the answers given, you deserve better.
That I think sums up the entire FX lineup at this point. We have had a lot of PR hype and promises of the moon but what we got was a decent chip. This is a shame because if AMD had not put the hype machine into full gear I think the reviewers would have been a little less harsh but the hype builds expectation and when you purposefully build expectation and cannot deliver then people are going to call you on it.
Based on the results of my subjective testing as well as our initial testing I stand by my original positions. If you already have a Phenom II X4 based system, say a 955 or better, or you have an Intel i5 first generation or better then this chip is NOT an upgrade, it is a sideways step into the future AMD hopes will come to pass. It is NOT worth the cost to move from what you already have. However the good news is that there really is no reason even from Intel to move from what you already have. While the newer chips show great in benchmarks in real life usage they are not getting the big gains people crave, save your money and hold tight, see what the future brings for now your computing experience should still be outstanding.
If you are using an older system or want to build a new system the waters are a bit muddier. AMD historically has done a solid job of not throwing out the baby with the bath water every time they make a new chip. This means the AM3+ platform may have another generation of potential in it. I have tried to get AMD to discuss if this platform was going to move as others have through a generation or two at least but cannot get a response. While the Intel offerings look awful attractive against the FX they carry with them the burden of the Intel toss the whole thing, buy again marketing strategy. These waters are further muddied by the fact that AMD is still making the Phenom II processors, which have some very attractive price offerings and still deliver a great computing experience.
At the end of the day we are left with an FX chip that looks to the future but is lackluster in the present having to fight against a solid champion of the present, the i5 and a solid value of the present the Phenom II. The gamble is will the future that the FX is designed for be coming soon? If you are worried about benchmarks and saying you have the biggest and baddest then the i5s from Intel are the way to go. If you want the best value then the Phenom II still rocks the house. If you want to buy an FX then you can get a solid computing experience and a decent machine but for right now I would look in other directions if today is your focus.
I’m confused with the direction AMD is taking on the FX. It very much sounds like a “if you built it, they will come” type of proposition for software developers. Currently, consumers don’t need the additional cores AMD is providing, and it seems that by the time the extra cores will be supported by the OS / applications Intel will be expanding their architecture as needed. Hopefully AMD can bounce back with future versions and build on their framework, we need good competition in the processor market.
I have to agree this design makes no sense when you look at the performance it delivers. While it is adeqaute it does not excel at anything in the market place. My personal thought is AMD is better on the future and wants to have time to mature their platform in the wild before it’s full potential is realized so they hit the ground running as the software moves to heavy threading.