System Builder Marathon, Dec. 2011: $1200 Enthusiast PC

[dalethepcman]

[citation][nom]BLACKSCI[/nom]They also stated that they went with that ssd cuz it was cheep! and why would you use a 5400 rpm drive? i mean yeah ill put games on my mechanical, but at least a 7200, i want to see my load times improve with a build, not stay the same.[/citation]

The Corsair Nova is twice as fast and costs less, and its one generation behind the times.

Additional as some of the other posters have mentioned the Bulldozers love fast ram. If you don't use at least 1866, preferably 2133 ram then you are crippling your system.

 

[51l3n5t]

[citation][nom]king_maliken[/nom]*Thought this was for people who didn't really know what to build, or how to build a computer? I wouldn't recommend this build.[/citation]

+1

 

[fb39ca4]

$180 for the hard drive? holy $4!7

 

[dragonsqrrl]

So lower performance and higher power consumption, sounds about right.

The AMD fanboy go-to response: You don't need more than 40 fps anyway. People who think "more performance = better" are stupid. And the productivity/encoding results? You never need more performance then what AMD currently offers. You really think you'll notice the difference between 67 sec and 93 sec encode time?

... lol

It was an interesting experiment, and I applaud the authors willingness to try something different, but I don't think we'll see another Bulldozer based build for a while.

 

[bradkman]

I don't agree with this build at all. I would have picked the Asus EVO M5A99X motherboard with at least 8GB of ram. In all the years I've built computers, I know that Asus gives you the best performance. Given the components that were used, the bench mark results are kinda useless to me. What about the recent windows hotfix that microsoft just released for the AMD Bulldozer chips? Was that installed? I'm not saying that it should have been, just curious that's all.

 

[Crashman]

[citation][nom]bradkman[/nom]I don't agree with this build at all. I would have picked the Asus EVO M5A99X motherboard with at least 8GB of ram. In all the years I've built computers, I know that Asus gives you the best performance. Given the components that were used, the bench mark results are kinda useless to me. What about the recent windows hotfix that microsoft just released for the AMD Bulldozer chips? Was that installed? I'm not saying that it should have been, just curious that's all.[/citation]I just tested the patch with most of these apps, and it had no benefit. Perhaps other apps benefit, but these apps are the ones used in this series.

 

[cmcghee358]

I have a question about Crossfire.

Do 2 1GB cards equate to 2GB of available VRAM?

I've heard only the primary cards VRAM is used.

Furthermore if you paired a 2GB 6950 with a 1GB 6950 should the 2GB version be in the top slot(assuming only the primary cards VRAM is used)

 

[cleeve]

I don't agree with this build at all. I would have picked the Asus EVO M5A99X motherboard with at least 8GB of ram. In all the years I've built computers, I know that Asus gives you the best performance. Given the components that were used, the bench mark results are kinda useless to me.

That's strange because the experience I've gotten from years of benchmarking different motherboards has shown me that motherboard manufacturer is irrelevant when it comes to raw performance.

But instead of merely disagreeing, I'll prove my point with actual data (courtesy of Crashman, actually):

http://www.tomshardware.com/reviews/sabertooth-990fx-990fxa-ud7-990fxa-gd80,3068-13.html

The most expensive ASUS performance = cheapo ECS performance. The CPU and chipset are the performance differentiators, not the name on the box. If you want to talk features, reliability, and overclocking ability that's a different thing, but these bench results are as valid as an ASUS premium board.

I can also show you a ton of benches that demonstrate 8 GB will offer no frame rate performance increase over 4 GB, if you like.

 

[cleeve]

I have a question about Crossfire.

Do 2 1GB cards equate to 2GB of available VRAM?

I've heard only the primary cards VRAM is used.

Furthermore if you paired a 2GB 6950 with a 1GB 6950 should the 2GB version be in the top slot(assuming only the primary cards VRAM is used)

both cards use their own RAM, it's not shared. Since it's not shared things like textures have to be copied to each GPU's dedicated RAM, so all the RAM is used, but there's no benefit to having more.

In a case where 1GB and 2GB cards are paired in Xfire, both only use 1 GB no matter where they sit. So it's like having two 1 GB cards in Xfire.

 

[cmcghee358]

Well I guess that's what I thought my situation might be a bit more complex.

I have a 2GB 5870 and a 2GB 5970. GPU-Z reports 1GB per GPU.

So do I have 1 GB of VRAM available or 2GB

Basically is it the total per card, or total available per GPU?

 

[mikeangs2004]

VRAM is common knowledge. I would prefer you choose dual fan CPU cooler w/ slightly higher price. The RAM slot thing is a common problem.

 

[murambi]

I dont get why you would use dual gpus when you say that there is a problem with micro stuttering why not use the gtx 580. Ever since you guys did an article on that I find it difficult to recommend to anyone to buy an sli or crossfire gpus

 

[cleeve]

I dont get why you would use dual gpus when you say that there is a problem with micro stuttering why not use the gtx 580. Ever since you guys did an article on that I find it difficult to recommend to anyone to buy an sli or crossfire gpus

Our article said it was a problem with cards slower than the 6950, and these are 6950s.

 

[cinergy]

Once again, Tom's proved to be stupid building custom PC's. What good is it to have two high end GPUs if the CPU is low end? But it's really surprising how big bottleneck FX processor is with single or dual threaded applications.

 

[gam0reily]

[citation][nom]Cleeve[/nom]Our article said it was a problem with cards slower than the 6950, and these are 6950s.[/citation]

So Xfire/SLI with a slower card is not recommended?
Like a single gtx 570 up against two gtx 460s? the prices are almost the same.

 

[gam0reily]

and what if the FX 6100 was replaced by a phenom II 1100 T??

 

[Krnt]

and what if the FX 6100 was replaced by a phenom II 1100 T??

That might be good but the Phenom II X6 1100 T is not at the same price so you will need to sacrifice something.
A Phenom II X4 975, would do better job at the same price.

 

[Uni-duni-te]

Despite the choice of bulldozer that I would not make (although I understand why you tried it), I would not get the Xigmatek cooler for $30 when only $10 more could get you the Thermalright True Spirit that can play with the $60-70 big boys for a very low price. To be that is the best budget cooler around. Maybe try it in the next build.
Would be nice to see if you could unlock the shaders on the 6950 1Gb and get even more performance (there are lots of reports of successfull unlocks on the 1Gb version boards like on the 2Gb ones).

 

[grody]

[citation][nom]Crashman[/nom]I spent a day testing it with Tom's Hardware's motherboard test suite and saw 0% increase... A day wasted during the holiday season, a day behind on my other articles.[/citation]
Wait, why can't your experience with the patch have become an article? I think this is worth informing people about.

 

[Crashman]

[citation][nom]grody[/nom]Wait, why can't your experience with the patch have become an article? I think this is worth informing people about.[/citation]Because it would have been a boring article. To spice it up I would have had to ask AMD for an application to "prove" the patch, benchmarked more and added a little AMD spin.

I don't spin, sorry, and "unspinning" would have made poor AMD look even worse.

 

[Keiki646]

There is no way that I would have gone with Mushin Ram with only 4GB kit when I can get a G.Skill Sniper (2 x 4GB) 8GB kit for 49.99 at newegg switch are compatible with AMD platforms

 

[Keiki646]

Mushkin Enhanced Redline 4 GB (2 x 2 GB) is a DEACTIVATED product, why add a DEACTIVATED. product to a build when and as I said before G.Skill Sniper 2 x 4GB (8GB) kit is only 46.99 at newegg. Poor judgement call there if I must say so

 

[silverblue]

So let me get this straight... Bulldozer performs better with a single card than Crossfire?

Doesn't anybody do internal testing at AMD anymore? Why do the review sites have to do their work for them?

Now I can see why they said that the experience on the 980X and FX8150 looks the same; if you ramp everything up to its highest, the processor isn't the bottleneck anymore, thus hiding inefficiencies.

AMD's main weakness seems to have transferred from integer calculations to floating point, probably why gaming is so damned weak now. I'm almost scared to see the comparison article.

 

[Guest]

I wish TH would make the GIVEAWAY INTERNATIONAL. I wouldn't mind paying for the shipping and customs.

 

[clonazepam]

[citation][nom]cmcghee358[/nom]Well I guess that's what I thought my situation might be a bit more complex.I have a 2GB 5870 and a 2GB 5970. GPU-Z reports 1GB per GPU.So do I have 1 GB of VRAM available or 2GBBasically is it the total per card, or total available per GPU?[/citation]

That's the norm. For dual gpu cards, they report total memory on the board. So its 1GB per gpu on the board, so that is why you see 1GB. You can look up a good Tom's article comparing the 590 and 6990 for more detail.