Cyberpower’s Gamer Dragon: Can AMD Bring The Game?

[cleeve]

[citation][nom]osse[/nom]The setup u just put up for the AMD rig cleeve, well no builder with knowlage would say that is the most cost effective solution, the most costeffective is phii940 ddr2 ram, 790fx am2+ motherbored, and vertex SSD drive, that do atleast here match the excat cost as the SBN build.[/citation]

I considered that. But even though we're going to get a mere 1-3% bump from DDR3 and a 3-5% bump from the 790FX, this will still represent the best case scenario for AMD.

Let's say you save another $100 and go for the AM2+/DDR2 setup - what're you going to do with that cash? It's not enough for a pair of GTX 285's or 4870 X2s.

Might get you the SSD, and from what I've seen SSD's don't work miracles, and they sure don't hold much unless you've got more than $100 to invest in one.

 

[blevsta]

Excellent. I look forward to the new article. I think you lose some of the price benefit of chosing AMD when you seek out an SLI or Crossfire board. Corei7 forces you to go that route, where as with AMD you can go with a single card motherboard for even cheaper than you've listed.

On another note...where are the temperatures tests? You guys do this for most other machines you build, but not with the machines from the system makers (you've failed to include them for Alienware, Ibuypower, and now Cyberpower). Is there some sort of agreement to leave this info out? Its simple testing, and important info to have. How hard is Cyberpower really pushing that AMD chip to get 3.6ghz?

 

[osse]

Total 604,96 On newegg

Phii 940 199,99
ASUS M4A79 Deluxe 127,99
OCZ 4GB (PC2 8500) 47,99
OCZ Vertex Series OCZSSD2-1VTX60G 2.5" 60GB SATA II MLC 199
Sunbeam CR-CCTF 120 mm 29,99
so yes u can afford an vertex, even thou not as fast as an itel x25, it reduces one of the main thing for an gamer, it reduces min framrate . And compared to a 7200 rpm spindisk, there is a world of diffrense in real word use, that not even the i7-920 can compaste. But true avragae framerate will not vary that much.

But there is also the optioun



2x4890 = 199,99x2 = 399,98

1x gtx 295 = 529,00 - the vertex 199 = 330,99, add 4 gb more ram + 49.99 = 380,99

This 2 eksamples should tell u not to use the phii 955 and am3 in a rigg shoutout between an AMD and an Intel rigg.

What do u use the ekstra 4 gb ram for, yes u use it to disabel disk swap file to reduce framedrops.

 

[cnox]

Cleeve,
Don't second guess yourself, you hit it right on the head: take that 80 bucks and shift it towards the gpu subsystem on the AMD system. That's a great place to start. Additionally, take advantage of the tricks on the dragon platform such as the gaming profiles (I'm talkin out of my butt a bit on this one but I thought I read that the machine can be put into various modes that cater towards the type of use the system is doing). Leverage the different aspects of both platforms. If you're going with the triple channel ddr on i7, take advantage of the bells on dragon.

And 4gb DDR2 vs 6gb ddr3? I donno...I thought I read you don't get much at 6gb, but depends on the game. And while I think that not a single FPS should be sacrificed on this build-off, if you can go with DDR2 on AMD and get an even better gpu setup, then that's worth serious consideration.

 

[cleeve]

[citation][nom]Cnox[/nom]Cleeve, Don't second guess yourself, you hit it right on the head: take that 80 bucks and shift it towards the gpu subsystem on the AMD system.[/citation]

Well, that's my plan. And I can see where osse is coming from, but since I'll be recording average framerates for these benchmarks this time around, the SSD wouldn't show us a tangible benefit.

If folks call for it I might consider an AM2+ build in a future article, but this time around it's a simple i7 920/DDR3/4870 Crossfire vs. Ph2 955/DDR3/4890 Crossfire piece.

 

[osse]

but i see littel to gain going for a gtx 295, but as a rigg a vertex would do a lot, and hey u have 12 seconds headstart when benching with a vertex ws an wd caviar from u start up the rigg

 

[osse]

I understand that u want this speical test cleeve, it will prove u right conserning ur findings in this article, and i would love to see a real review again where u max ur dollars and do riggs vs riggs test.

 

[hotfuzz]

I see one thing everybody else seems to have missed in this test. The Intel Motherboard has two X16 PCI express slots. The AMD board only has one X16 and one X8. I suspect this is why you see the big difference in gaming performance and not because of the Processors. Would be intesting to see the same test but instead of the GA-MA790XT-UD4P you use the GA-MA790FXT-UD5P (has 2 X16 PCI-E)

 

[cleeve]

[citation][nom]hotfuzz[/nom]I see one thing everybody else seems to have missed in this test. The Intel Motherboard has two X16 PCI express slots. The AMD board only has one X16 and one X8. [/citation]

nah, it's been mentioned a couple of times. Frankly, 8x vs 16x PCIe 2.0 doesn't make a big difference, pretty much no difference at all actually as we've demonstrated in past tests.

Even so, I'm using the 790FX chipset in the follow-up article to eliminate that as a factor.

 

[osse]

4870 vs 4890 is round 10-15 % on stock speed, but 4890 has a lot more headroom wich is the main reason to recomend 4890 instead of 4870, so u should also use round 940 hz for the core on 4890 and round 780 for the 4870.

 

[cleeve]

[citation][nom]osse[/nom]4870 vs 4890 is round 10-15 % on stock speed, but 4890 has a lot more headroom wich is the main reason to recomend 4890 instead of 4870, so u should also use round 940 hz for the core on 4890 and round 780 for the 4870.[/citation]

Well, I'm trying to keep the variables simple. To fabricate an artificial clock-speed handicap based on how people might overclock the cards doesn't appeal to my sense of fairness.

 

[osse]

its unfair against the 4890 not to use the ekstra headroom a 4890 have, compared to the 4870. I havnt seen any review that hasnt come to that conclusion when it comes to overclocking. It is the main reason to buy a 4890 instead of a 4870.

 

[Schip]

[citation][nom]Cleeve[/nom]Here's what I've come up with...
Intel:Core i7 920 279.99
G.Skill 6GB PC3 10666 94.99
Mobo: DFI X58T3H6 199.99 (Cheapest: MSI X58M @ $169.99 - $30 delta)
Cooler: Xigmatek Drkngt 39.98
TOTAL i7 CORE COMPONENTS: $614.95
AMDhenom II 955 245.00
G.Skill 4GB PC3 10666 64.99
Mobo: M4A79T Deluxe 179.99 (cheapest: DFI Lanparty DK 790FX-M2RS @ $134.99 - $45 delta)
Cooler: Zerotherm NV120 49.99
TOTAL PhenomII CORE COMPONENTS: $539.97[/citation]

I personally like what you've put together, especially the memory. The only thing I would change would be the cooler on the AMD rig. You could use the same cooler from the i7 rig, save ten bucks and have quantifiable better cooling according to this article at FrostyTech. Of course, it may not be a big difference, I just don't think its a great value to spend more for less performance. Cheers!

 

[Schip]

Sorry for the funky citation Cleeve. Also, I agree with cnox that you should also use AMD's Fusion software at least with some of the testing to see if it can provide an appreciable difference. Thanks

 

[noob2222]

[citation][nom]osse[/nom]I understand that u want this speical test cleeve, it will prove u right conserning ur findings in this article, and i would love to see a real review again where u max ur dollars and do riggs vs riggs test.[/citation]

I will sum it up for you, i7 975EE wins with any max vs max destktop setup you can possibly do.

DDR3 vs DDR2 is a whole different story all together. With any tweaking other than memory size, these systems need to be as close as possible because what you can do with system A to save a few dollars, can most likely be done with the other system as well.

This article should lead to the conclusion with raw facts, price for price, wich offers a better solution?

Would help also to split the conclusion to two categories, gaming and productivity. We all know and accept that Intel owns the productivity category, but we have all seen mixed results when it comes to gaming with the I7 and AMD machines. If not this arguement would have never made it past page 5.

Yes, there have been comparisons with x16 vs x8 when using crossfire, but often left behind as the cards get faster and faster, we forget we weren't pushing any bandwith issues way back when.

http://www.tomshardware.com/reviews/crossfire-meets-pci-express,1761-13.html

hd3870 was top of the line then, 6-20% increase between x8cf and x16 cf

Found another one with the 4850 being used, you will think again before saying x8 is the same as x16 in crossfire mode.

http://www.tweaktown.com/articles/1472/intel_p45_vs_x48_crossfire_performance/index10.html

 

[cleeve]

[citation][nom]osse[/nom]its unfair against the 4890 not to use the ekstra headroom a 4890 have, compared to the 4870. [/citation]

We'll have to agree to disagree on that one, bro. I'm not putting an artificial handicap on the 4870.

 

[cleeve]

[citation][nom]noob2222[/nom]Found another one with the 4850 being used, you will think again before saying x8 is the same as x16 in crossfire mode.http://www.tweaktown.com/articles/ [...] dex10.html[/citation]

Well, good thing we're using the 790FX in the follow-up, innitt?

 

[noob2222]

[citation][nom]Cleeve[/nom]Well, good thing we're using the 790FX in the follow-up, innitt?[/citation]
There will definately have some interesting results to compare for sure between all of the systems.

save a buck here, lose 10-15% performance, possibly more.

anyway, peace out, looking forward to the results.

 

[cleeve]

[citation][nom]noob2222[/nom]3870 was top of the line then, 6-20% increase between x8cf and x16 cfFound another one with the 4850 being used, you will think again before saying x8 is the same as x16 in crossfire mode.http://www.tweaktown.com/articles/ [...] dex10.html[/citation]

Actually, I thought there was a newer one... we used some 4870s:

http://www.tomshardware.com/reviews/crossfire-pci-express,2095-3.html

And the conclusion was that PCIe 1.0 would handicap crossfire, but PCIe 2.0 slots - even 8x slots - wouldn't noticeably slow things down at all.

 

[jennyh]

If you use the same games, the results will be as follows.

Crysis - Fairly clear win for the i7
WiC - Should be close, probably a small win for the i7
FC2 - Close but win for the Phenom II
Stalker - Big win for the Phenom II

Throw in CoD4, Grid, Age of Conan (note all 3 of these are some of the most popular games of the past year) and the Phenom will win those pretty easily. It's all about choosing a decent selection of games to bench instead of constantly choosing the very few games that massively favour cpu's over gpu's.

 

[cleeve]

[citation][nom]jennyh[/nom]If you use the same games, the results will be as follows.[/citation]

Testing's mostly done, but your predictions aren't holding up well.

[citation][nom]jennyh[/nom]It's all about choosing a decent selection of games to bench instead of constantly choosing the very few games that massively favour cpu's over gpu's.[/citation]

If I've 'constantly chosen CPU favored games', then why are you predicting that exactly half the games will be won by the Phenom II? According to your predictions, I've chosen as fairly as possible.

 

[jennyh]

Because the Phenom II has been given two superior gpu's?

If you chose games that actually cared about what gpu you put in it, you'd get results where better gpu's actually mattered.

So far, the only game on your list that really works best with better gpu's is stalker. And you might want to look at your results on that one again btw - 20% more performance is a pretty big lead and no offence but you aren't gonna spend that amount of cash on a system and not have 4xAA.

If you did a real bench of farcry 2 instead of running the cpu bench, you'd see that faster gpu's help a lot more in that game too.

Crysis will always run better on an i7 and WiC is a very cpu intensive game (also strongly nvidia favouring) and everyone knows this.

So no, I dont consider 1/4 benchmarks which actually care about your gpu to be fair as possible. Crysis, WiC and Farcry 2 cpu benchmark is about as far in favour of cpu-intensive games that you could have went actually.

 

[cleeve]

So on one hand we've chosen games that "massively favour cpu's over gpu's", however you predict the Phenom II will win exactly half of them "Phenom II has been given two superior gpu's".

Well, which is it? Do half of them favor the Phenom II's superior GPUs, or do the games massively favor CPUs?

You can't complain about both, they sort of cancel each other out...

 

[jennyh]

Sorry let me be clear.

WiC and Farcry 2 favour Nvidia, however with the swap to 4870's (or your underclocked 4890's), the lead for the i7 in WiC should be lessened, and I also believe that the 4890's would be enough to push the Phenom II over the i7 in Farcry 2.

Also, if I may ask another question - why no AA in Crysis? I believe that if you had put on 2xAA or 4xAA in that then the i7 system would have plummetted below the phenom II system, especially on very high. The rest got AA so why not Crysis?

 

[thebatman]

Wow 13 pages. Cleeve i'm so glad your doing this new article and can't wait for you to finish it.