- Oct 30, 2004
- 32,252
- 14
- 99,960
Most of the components used are actually ones I use in many of my builds, however, many times not in the same builds. I have never used twin GFX cards bigger than the 560 Ti in an Antec 300 for example.
I wouldn't spend $30 to move from CAS 9 to CAS 8 when for another $19 I could have CAS 7.
I do wish that some more time was spend on the GFX card selection, or at least in qualifying the justification. Later on in the comments section it's mentioned that :
"Yes, we should probably include a 2560x1600 display with each system, even the cheap one so that comparative tests can be standardized".
And with that criteria, I'd agree wholeheartedy with the GFX selection. But if your res is 1920 x 1200/1080, I'd have to say "Ya haven't made ya case"....at least not based upon the reference to the recent THG SLI / CF article.
I hope this is taken in the constructive spirit in which it was given as I have been following THG for more years than I care to admit. But I have to voice my disappointment in THG's recent trend in doing reviews which includes just 3 or 4 actual game benchmarks. It also omits the current "Crysis of our day" which is Metro 2033. Metro shows amazing scalability ... 197 % with 2 cards and 288% with 3way SLI
"Lies and Statistics" is a common cry in politics and limiting the tests to so small a sampling gives the impression that there's a predetermined conclusion in mind, especially when similar articles have come to strikingly different conclusions.
Looking at Guru3D's test on the two cards compared for this review for example, total fps performance in COD-MW, Bad Company 2, Dirt 2, Far Cry 2, Metro 2033, Dawn of Discovery and Crysis Warhead. The 570's in SLI scored 873 total fps; the CF'd 6950's scored 751....that 16% difference is bigger than anything we see in the THG article referenced in the build review. Incidentally, it's 873 in SLI to 825 for CF'd 6870's. Individually,
COD-MW SLI'd 570's get 230 to CF'd 6950's 150
Bad Company 2 SLI'd 570's get 103 to CF'd 6950's 96
Dirt 2 SLI'd 570's gets 138 to CF'd 6950's 131
Far Cry 2 SLI'd 570's gets 153 to CF'd 6950's 147
Metro 2033 SLI'd 570's gets 54 to CF'd 6950's 50
Dawn of Discovery SLI'd 570's gets 109 to CF'd 6950's 90
Crysis Warhead SLI'd 570's gets 86 to CF'd 6950's 87
It must be pointed out that there is no fps difference between the 1GB and 2 GB models of the 6950....at least not at 1920 x 1200.
http://www.guru3d.com/article/radeon-hd-6950-1gb-vs-geforce-gtx-560-ti-review/8
http://www.guru3d.com/article/radeon-hd-6950-crossfirex-review/5
In conclusion, w/o that 2560 x 1600 qualifier in the actual article, when looking at the breadth of tests available on various sites, I see little to support the conclusion that "Magic happens when two Radeon HD 6950s are paired, as superior multi-GPU scaling allows these to overtake the “more powerful GeForce GTX 570 when both products are configured in pairs."
By comparison, I'm seeing 166.60% scaling the 570's in SLI on Guru3D's tests versus 156.78% for the CF's 6950's at 1920 x 1200 . I also should point out that even the THG article referenced does not support the above quoted "overtake" statement except at 2560 x 1200. The THG link shows:
1650 x 1050 (No AA / 4 AA)- No "overtaking" here.
CF'd 6950's (107/91)
SLI' 570's (110/92)
1920 x 1200 (No AA / 4 AA)- No "overtaking" here.
CF'd 6950's (99/84)
SLI' 570's (101/84)
1920 x 1200 (No AA / 4 AA)- Here we finally have a win for the 6950 but less than half the 16% we saw when Guru3D did it.
CF'd 6950's (69/57)
SLI' 570's (64/53)
Again, the 2560 x 1600 qualifier in the article would have left me silent on this topic as at 2560 x 1600, I'm choosing the 69xx over the 5xx unless the user has specific application req'ts that favor nVidia (i.e. 3D rendering, CS5, 2D drafting, etc).
Now this is the part that I hope is read before the flames start, and it's more a comment on the referenced SLI / CF article than the build article. I am not suggesting that THG's tests were invalid and Guru3D's are solid ... I am not saying that the SLI'd 570's are a superior choice than the 6950's/6970's. I am saying that it's common knowledge that different games perform differently on different GFX hardware. Simply put, one's GFX card choice may well depend upon what games you're actually playing and it certainly depends on what resolution one is playing at. The article would have been more useful I think had the intended resolution been included. At 1920 x 1200 looking at this build I think I'd give serious consideration to a a pair of the factory OC'd GTX 560 Ti's cards from this article, as the best of them beat both the 6950 and 570 in single card competition presented here:
http://www.tomshardware.com/reviews/geforce-gtx-560-ti-roundup-asus-engtx560-graphics-card-overclocking,2858-15.html
would have been nice to see how they did in twin card configurations....as to use the words in the build article the factory OC'd 560 Ti's are "selling for a far lower price".
Similarly, using tests on just 3 or 4 "select" games to form the basis of such a wide sweeping conclusion, could leave the reader who made choices based upon these recommendations a bit disappointed as no where is it hinted at "your mileage may vary" if using anything other than the 3 or 4 games in the test.
All in all, the suggested build is very much one I'd do for someone with a 2560 x 1600 monitor, except for the RAM as noted above.
At 1920 x 1200, I'd be saving $80 on twin OC'd 560 Ti's and investing that in a full tower case / PSU combo (10.0 rated on jonnyguru) with swappable storage bays, USB 3 front panel, better air circulation and other modern features.
I wouldn't spend $30 to move from CAS 9 to CAS 8 when for another $19 I could have CAS 7.
I do wish that some more time was spend on the GFX card selection, or at least in qualifying the justification. Later on in the comments section it's mentioned that :
"Yes, we should probably include a 2560x1600 display with each system, even the cheap one so that comparative tests can be standardized".
And with that criteria, I'd agree wholeheartedy with the GFX selection. But if your res is 1920 x 1200/1080, I'd have to say "Ya haven't made ya case"....at least not based upon the reference to the recent THG SLI / CF article.
I hope this is taken in the constructive spirit in which it was given as I have been following THG for more years than I care to admit. But I have to voice my disappointment in THG's recent trend in doing reviews which includes just 3 or 4 actual game benchmarks. It also omits the current "Crysis of our day" which is Metro 2033. Metro shows amazing scalability ... 197 % with 2 cards and 288% with 3way SLI
"Lies and Statistics" is a common cry in politics and limiting the tests to so small a sampling gives the impression that there's a predetermined conclusion in mind, especially when similar articles have come to strikingly different conclusions.
Looking at Guru3D's test on the two cards compared for this review for example, total fps performance in COD-MW, Bad Company 2, Dirt 2, Far Cry 2, Metro 2033, Dawn of Discovery and Crysis Warhead. The 570's in SLI scored 873 total fps; the CF'd 6950's scored 751....that 16% difference is bigger than anything we see in the THG article referenced in the build review. Incidentally, it's 873 in SLI to 825 for CF'd 6870's. Individually,
COD-MW SLI'd 570's get 230 to CF'd 6950's 150
Bad Company 2 SLI'd 570's get 103 to CF'd 6950's 96
Dirt 2 SLI'd 570's gets 138 to CF'd 6950's 131
Far Cry 2 SLI'd 570's gets 153 to CF'd 6950's 147
Metro 2033 SLI'd 570's gets 54 to CF'd 6950's 50
Dawn of Discovery SLI'd 570's gets 109 to CF'd 6950's 90
Crysis Warhead SLI'd 570's gets 86 to CF'd 6950's 87
It must be pointed out that there is no fps difference between the 1GB and 2 GB models of the 6950....at least not at 1920 x 1200.
http://www.guru3d.com/article/radeon-hd-6950-1gb-vs-geforce-gtx-560-ti-review/8
http://www.guru3d.com/article/radeon-hd-6950-crossfirex-review/5
In conclusion, w/o that 2560 x 1600 qualifier in the actual article, when looking at the breadth of tests available on various sites, I see little to support the conclusion that "Magic happens when two Radeon HD 6950s are paired, as superior multi-GPU scaling allows these to overtake the “more powerful GeForce GTX 570 when both products are configured in pairs."
By comparison, I'm seeing 166.60% scaling the 570's in SLI on Guru3D's tests versus 156.78% for the CF's 6950's at 1920 x 1200 . I also should point out that even the THG article referenced does not support the above quoted "overtake" statement except at 2560 x 1200. The THG link shows:
1650 x 1050 (No AA / 4 AA)- No "overtaking" here.
CF'd 6950's (107/91)
SLI' 570's (110/92)
1920 x 1200 (No AA / 4 AA)- No "overtaking" here.
CF'd 6950's (99/84)
SLI' 570's (101/84)
1920 x 1200 (No AA / 4 AA)- Here we finally have a win for the 6950 but less than half the 16% we saw when Guru3D did it.
CF'd 6950's (69/57)
SLI' 570's (64/53)
Again, the 2560 x 1600 qualifier in the article would have left me silent on this topic as at 2560 x 1600, I'm choosing the 69xx over the 5xx unless the user has specific application req'ts that favor nVidia (i.e. 3D rendering, CS5, 2D drafting, etc).
Now this is the part that I hope is read before the flames start, and it's more a comment on the referenced SLI / CF article than the build article. I am not suggesting that THG's tests were invalid and Guru3D's are solid ... I am not saying that the SLI'd 570's are a superior choice than the 6950's/6970's. I am saying that it's common knowledge that different games perform differently on different GFX hardware. Simply put, one's GFX card choice may well depend upon what games you're actually playing and it certainly depends on what resolution one is playing at. The article would have been more useful I think had the intended resolution been included. At 1920 x 1200 looking at this build I think I'd give serious consideration to a a pair of the factory OC'd GTX 560 Ti's cards from this article, as the best of them beat both the 6950 and 570 in single card competition presented here:
http://www.tomshardware.com/reviews/geforce-gtx-560-ti-roundup-asus-engtx560-graphics-card-overclocking,2858-15.html
would have been nice to see how they did in twin card configurations....as to use the words in the build article the factory OC'd 560 Ti's are "selling for a far lower price".
Similarly, using tests on just 3 or 4 "select" games to form the basis of such a wide sweeping conclusion, could leave the reader who made choices based upon these recommendations a bit disappointed as no where is it hinted at "your mileage may vary" if using anything other than the 3 or 4 games in the test.
All in all, the suggested build is very much one I'd do for someone with a 2560 x 1600 monitor, except for the RAM as noted above.
At 1920 x 1200, I'd be saving $80 on twin OC'd 560 Ti's and investing that in a full tower case / PSU combo (10.0 rated on jonnyguru) with swappable storage bays, USB 3 front panel, better air circulation and other modern features.
, 6Gb Ripjaws 1600 cl 8, i7 920, Asus x58 Deluxe, 3x 1Tb Hard Drives, Card Reader.
Facebook
Twitter
Instagram