System Builder Marathon, Dec. 2011: System Value Compared

[Guest]

The major problem with the builds this month is that the $1200 build uses a $160 CPU whereas the $600 build uses a $190 CPU.

The 6100 was an extremely poor choice for a Dual GPU build. It's quite obvious that it would be extremely CPU limited. If you put an equivalent intel CPU(i3) in that system it would've likely still had a CPU bottleneck, albeit less.

In fact even comparing the $1,200 build this month to last month's is a joke. $220 i5 vs $160 BD. Who in their right mind would compare the two?

 

[Crashman]

[citation][nom]elbert[/nom]I think its a bit more as the HD scores comparing just the WD black to the $600 green drive should produce the scores we see all but the starting of apps.[/citation]The HD scores you see are actually for the SSD's, except in the case of the $600 system which has no SSD.

 

[cadder]

Maybe we can eliminate AMD from consideration in future builds for awhile and concentrate on the good stuff from Intel.

 

[elbert]

[citation][nom]cadder[/nom]Maybe we can eliminate AMD from consideration in future builds for awhile and concentrate on the good stuff from Intel.[/citation]
Yes and for an SSD should it be much higher than a green drive? This is a very bad performing SSD in the $1.200 build. Doesn't this drive require a tweak to never going into idle with windows 7 or else it starts corrupting the drive?

 

[Crashman]

[citation][nom]elbert[/nom]Yes and for an SSD should it be much higher than a green drive? This is a very bad performing SSD in the $1.200 build. Doesn't this drive require a tweak to never going into idle with windows 7 or else it starts corrupting the drive?[/citation]Page 12 shows the combined results of the four drive tests at 262% greater performance for the $1200 PC compared to the $600 PC. I know 262% might be spare change to some users, but I still think that qualifies as "much higher".

 

[ilikemacandpc]

If a person upgraded that $600 build with an SSD, an i5 2500k, a better motherboard, and a faster GPU, and an OS, they would be at about $1000 and would annihilate the $1267 FX 6100 build.

 

[elbert]

[citation][nom]Crashman[/nom]Page 12 shows the combined results of the four drive tests at 262% greater performance for the $1200 PC compared to the $600 PC. I know 262% might be spare change to some users, but I still think that qualifies as "much higher".[/citation]
Wait I see the green drive was on the $2,400 system. The $600 build has a much faster 500GB HD. Kind of makes sense against a 7200 rpm drive. Also this $600 build is a monster because at 1920X1080 its only 15.6 fps slower than Sept's i5 2500k with 2 460's in F1 2010. The September build scores a 81.9 and this little $600 build 66.3 fps.
http://media.bestofmicro.com/1/9/318861/original/F1%20Ultra.png
http://media.bestofmicro.com/8/U/319134/original/image012.png

 

[noob2222]

[citation][nom]Krnt[/nom]For some reason the $1200 build is the worst performing bulldozer build I've ever seen.There is definitively something wrong with the memory configuration.I made some test with a FX 4100 in my pc that normally uses a Phenom II 965 using the same memories of the $1200 build, and the FX 4100 reached 16.8 GB/s at SIS sandra memory bandwith, while my 965 achieves only 15.8 GB/s.I think the problem is that Biostar motherboard, cutting off the memory bandwidth, and maybe even a complete channel.I know this will not change the fact that the BD is a bad performer, but things like these can change the results, giving you at least more competitive results.[/citation]


I didn't even notice the memory bandwidth. looks like you may be right, its only using one memory channel. http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.technic3d.com%2Freview%2Fcpu-s%2F1305-amd-fx-bulldozer-architektur-im-test%2F5.htm%23sisoft_sandra

memory = 18.3, Toms memory = 8.5.. wonder how much memory is reported in windows.

one thing is for certain, the entire build is screwed, not just the cpu

 

[Onus]

There are SO many things to test...
The $1200 build asks the most questions over how badly it sucks, and I'd like to know why. I don't expect any of these tests to "rescue" the FX, but some of the numbers raise questions that I don't think can be explained just by a poor CPU. Below, when I say "retest," I don't necessarily mean the full suite; even a couple of the synthetics should show if "something" in addition to a slow CPU should share the blame.
How about a retest with all the same parts EXCEPT the mobo? Rule out the Biostar as a possible cause of [RAM / channel / firmware] issues.
Then rule out the chipset by retesting with a single GPU, possibly a X2 so the Crossfire/SLI is being handled on the board, not by the chipset.
Retest with a X4 980 or a X6 1100T, or even a FX-8120.
Use the results from the above abbreviated testing to decide if more thorough tests are warranted.

 

[demonhorde665]

ouch i'll take the 600 build any day over that mess of 1200 build , hel teh 1200 build would have done much better if they swaped the bull dozer for a phenom 2 4 or 6 core (both of which can be gotten at a similar price range as the dozer). WHAT THE HELL was AMD thinking with this chip design.

 

[demonhorde665]

[citation][nom]grody[/nom]Yes it does have major problems. It has bulldozer.[/citation]


LOL funny shit , and i'm even an AMD fan usually, but that is usually when thier processors deliver n3ear intel performance hat great value , and bulldozer doesn't deliver , apprently some haters gave you a thumbs down . my next build wil certainly include eitehr intel i5 or an older phenom 2 .. event eh high end bull dozer's suffer baddly compared to similar priced intel chips Amd needs to get on top of the situation fast , becasue i can't recomand an AMD bulldozer chip to any one cept a non gamer, non workstation build with the way things are

 

[doron]

[citation][nom]hmp_goose[/nom]So the "oopsie" build this quarter with be replacing the $1200 with a i5-2500k?[/citation]

I hope they'll revise this build with a 2500k as well as an overclocked phenom II x6 for a solid comparison, especially gaming wise.

 

[iam2thecrowe]

[citation][nom]doron[/nom]I hope they'll revise this build with a 2500k as well as an overclocked phenom II x6 for a solid comparison, especially gaming wise.[/citation]
there i no need to revise it, im sure they just put a FX chip in there to make people realise what a poor choice it is. Point made, fx chips are indeed crappy and poor value for money, something most people knew, now everyone knows. There are always a few ignorant people who are blind to results though and will still go buy an fx cpu, i feel a little sorry for them.

 

[Onus]

I'd still like to know if something else in addition to (not "instead of") the CPU is partly responsible for that utterly miserable showing. Companies release suck products from time to time, but this was so bad as to make no sense; nobody likes delays, but I have no idea why AMD would release a CPU this bad.

 

[kinggraves]

AMD already said that Bulldozer isn't going to be good at gaming and low thread applications. You might as well complain about how your toaster doesn't make good spaghetti. It should be good for the server type applications it's designed for, especially after OS that are better able to handle multi threading comes out. AMD left the gaming enthusiasts to Intel, they didn't screw it up, they released exactly what they wanted to. Stick with Intel for decent gaming performance.

Still, the $1200 build was bad all around, not just the CPU. The SSD was lousy, memory bandwidth is clearly gimped, GPU power far outpaced CPU power. Remove the second GPU and spend the money on an 8150 and a better SSD/motherboard. You still wouldn't have great graphics processing but it'd be better than this mess.

What I'd like to see is a $600 build featuring a Llano proc and a hybrid Xfire compatible GPU. I think this combo could prove better than anything a low end BD would do.

 

[g4114rd0]

-There's always hope...
Phenom II X8 3020: 3.00GHz, 125W

 

[Guest]

Okay, in regards to the 1200 build. The RAM is wrong and Crossfire is over kill with that big of a GPU. Personally, on air, I have the same set up, with 8 gigs of 1866 G-skill RAM and am showing ZERO bottle necks. The MoBo adjusts as needed through the Toverclocker utility. (I know this is ghetto, but it works surprisingly well.) 40 degrees OCed to 4.4, again, on air. The biggest mistake AMD seems to have made is that the released the chips too low for the stock speed. The natural stock speed should have been .7-1 higher, as it actually runs more naturally that way.

Second, my build only cost 700 bucks with it, so yea, the secondary stuff outside of the CPU, MoBo, and RAM are what killed that test. The box just isn't set up right at all. BUT, with all the viral poo-pooing that's gone on, no ones actually sat down and taken the time to do what I've experimented with.

Third, with the SLI bios update on that MoBo, SLI 550-TIs are stupid efficient. The key is to not only OC the CPU, but the NB and RAM as well. It all works in concert. OCing the CPU alone will cause bottlenecks left and right, and that's my assumption with several of these reviews I've seen.

BTW, it's stable darn near 5.5 with just an H60. The thing is stupid easy and efficient to OC.

 

[Guest]

you say haters gonna hate and that`s right

1200 pc in weaker than 600 pc ... c`mon man ... instead of cr or sli buy one videocard and the result will be better

 

[pacioli]

After all the Bulldozer hype and my own high hopes I am truly disappointed by the $1200 machine.

 

[silverblue]

Nicely done DuncanEllis.

 

[blazorthon]

Some of the commentors here noticed that bulldozer is not the only problem with that $1200 machine.

But on bulldozer... AMD should have taken the improvements over the Phenom II cores made in Llano (only like 6% performance improvement, but hey it helps) and done a die shrink with some more improvements. Would have been much less work than making such a different architecture like bulldozer and you could still fit 8 cores on a CPU... 8 cores that are improvements over the older Phenom IIs, not a step backwards in performance. Could have been much faster than bulldozer and AMD would have only needed to slightly modify what they already have, much easier.

 

[g4114rd0]

 
- ...most of us would be more than happy with a m@rch 12 $600 system .

 We Wish You aMD Christmas!

 

[brucek2]

In real life, a bulldozer is a slow moving vehicle with the capability to demolish even a perfectly good structure into a pile of rubble.

It seems that this latest generation AMD CPU was unfortunately if accurately well named.

Oh well. Hopefully this will provide the motivation for a massive comeback in the next generation.

 

[blazorthon]

[citation][nom]brucek2[/nom]In real life, a bulldozer is a slow moving vehicle with the capability to demolish even a perfectly good structure into a pile of rubble.It seems that this latest generation AMD CPU was unfortunately if accurately well named.Oh well. Hopefully this will provide the motivation for a massive comeback in the next generation.[/citation]

That's a good point lol.

 

[youssef 2010]

This review is missing something: MINIMUM FPS. Personally, as a man who is sensitive to micro-stuttering, prefer a smooth gameplay experience to raw FPS. To that end, I suggest that a minimum FPS chart accompanies every game benchmark run on this venerable website. I suggest making the reviewer who is most sensitive to this phenomenon comment on the smoothness of each game benchmark. It's taken for granted now that AMD's latest CPUs are frustrating, but ,when it comes to gaming performance, we need to assess the smoothness as well as the FPS scores.