Best CPU for a Single ATI Radeon HD 5870 and ATI Radeon HD 5870 CrossFireX Configurat

[GunterFalstaff]

Im runnin a single 5870 with a 955 at 3.8ghz on a 26inch 1920*1200... according to this... if i ever want to go crossfire im ONLY going to get about a 5 fps increase? wtf

[Mungri]

Actually, the difference in the OP charts is around 10-15 FPS, seeing gains from around 30 FPS to 40+.

The total FPS gain is not how to measure crossfire scaling, but rather the percentage difference, which will also vary from game to game.

Oh, wait, youre using a phenom :p.

[Manicdan]

i would have liked that review so much better if they had 5870s since its 2x 5770s. in many cases the 5850 by itself was close enough to be equal to the 5770 xfire. and the price of a 5850 is better than 5770 xfire too. its availability that sucks

[orangekiwii]

except far cry 2 sucks - so its basically a benchmark

[zanzabar]

way to use an old 8x8 ultra low end board for amd but use full 16x16 boards for intel thats not stacked at all

it looks like it dosnt matter solong as u have a good MB

[kite7]

way to use an old 8x8 ultra low end board for amd but use full 16x16 boards for intel thats not stacked at all

it looks like it dosnt matter solong as u have a good MB

I don't think PCIe 2.0 8x vs 16x has much of an impact. techPowerUp did a review of the PCIe scaling on a 5870, but no word on minimum fps

Example of a 5870 on PCIe 2.0 8x vs 16x on crysis

http://www.techpowerup.com/reviews/A...Scaling/7.html

But it's probably the crossfire overhead eh?

[zanzabar]

I don't think PCIe 2.0 8x vs 16x has much of an impact. techPowerUp did a review of the PCIe scaling on a 5870, but no word on minimum fps

Example of a 5870 on PCIe 2.0 8x vs 16x on crysis

http://www.techpowerup.com/reviews/A...Scaling/7.html

But it's probably the crossfire overhead eh?

it dosnt matter all that much but when u are compairing platforms that are all native 16x16 but then u take a gimped lowend board for the amd part there is something wrong. especially when there is 1-2 fps from the best to worst in most of the tests

[tajoh111]

Xbitlabs is the last place I would expect AMD biased against results.

This is quite the surprise.

[grimREEFER]

well, there are no core 2 quad results, but im guessing thats because no one buys core 2 quads anymore. why there are core 2 duo results, i dont know.
anyway, there are phenom results, thats basically gonna be pretty close to a core 2 quad at the same frequency.

[LedHed]

well, there are no core 2 quad results, but im guessing thats because no one buys core 2 quads anymore. why there are core 2 duo results, i dont know.
anyway, there are phenom results, thats basically gonna be pretty close to a core 2 quad at the same frequency.

if so then why does the C2D out pace the Phenom in almost every test?

[grimREEFER]

if so then why does the C2D out pace the Phenom in almost every test?

the phenom does beat the core 2 duo in most tests

[kite7]

ATI HD 5870 Scaling Performance: X58 vs P55 Showdown
http://www.anandtech.com/mb/showdoc.aspx?i=3649&p=3

Barely any performance difference between 16X CF and 8X CF (X58 vs P55 respectively) , though I do agree xbitlabs should have went with a 790FX board anyways and more ram for the AMD setup.

[gosh]

if so then why does the C2D out pace the Phenom in almost every test?

It doesn't, but games have one renderthread where the "action" is. DX11 will scale better

[initialised]

Anyone wh tries to ru(i)n an 5970 with a dual core is a retard (Hangover pronunciation)!

[LedHed]

Anyone wh tries to ru(i)n an 5970 with a dual core is a retard (Hangover pronunciation)!

that entire review would beg to differ:

Now, I’ll try to sum everything up. First of all, the single Radeon HD 5870 does not depend as much on the CPU as it is supposed to. According to my tests (the left part of the diagrams), the Radeon HD 5870 is quite satisfied with an overclocked dual-core CPU manufactured two years ago. And you can even leave the CPU at its default frequency at the high-quality settings and 1920x1200. The only exceptions are Left 4 Dead and Warhammer 40000: Dawn of War II. The former game is not a problem for modern top-end Radeons while the latter, on the contrary, calls for a quad-core CPU, preferably from Intel.

he only later mentions that the dual core supposedly is more "jerky", but to me that is a message added by sponsors pushing new hardware. Because jerkiness effects min fps, we would be able to see it if it was happening. Like the Phenom shows, sometimes only pulling 6fps, while the rest of the CPUs are around 20fps.

the phenom does beat the core 2 duo in most tests

not where it matters, look at the 1920x1200 Full IQ tests (where owners of 5870's will be playing).

[blindbox]

not where it matters, look at the 1920x1200 Full IQ tests (where owners of 5870's will be playing).

5870, 1920x1200 Full IQ settings(the very bottom graphs), 4.1 GHz, Top c2d, Bottom Ph II x4

10dddd0d1dddddd
01dddd1d0dddddd
C2D=2 wins 2 losses, rest draws
Ph II x4=2 wins 2 losses, rest draws.

5870 CFX, 1920x1200 Full IQ settings, 4.1GHz,

1100d00d1dd000d
0011d11d0dd111d
C2D=3 wins,7 losses, rest draws.
Ph II x4=7 wins, 3 losses, rest draws.

I don't know how you managed to come up with that. Even a skim through is enough to see Ph II wins. Well, it's quadcore vs dualcore after all.

Draws are when the fps difference is only 1 fps. 2 fps isn't a draw. The GPU seems to be the bottleneck more frequently on PhII than C2D though.

Why didn't they use C2Q anyway? Why didn't they test Phenom II about the jerkiness test either? Anandtech reported "smoothness" on Core i7 and Phenom II compared to C2Q and C2D ages ago.

*Takes cover behind flameshield*

Techpowerup reviews tested PCI-e bandwidth with a single card setup, not a crossfire setup. That's why we have CFX dongles, because the bandwidth required for inter-card communication on PCI-e is simply not enough. Legionhardware tested 4870x2 CFX before on 8x 8x setup and it shows performance drop.

I conclude the review is fail (overall. It's good on some matters).

[LedHed]

5870, 1920x1200 Full IQ settings(the very bottom graphs), 4.1 GHz, Top c2d, Bottom Ph II x4

10dddd0d1dddddd
01dddd1d0dddddd
C2D=2 wins 2 losses, rest draws
Ph II x4=2 wins 2 losses, rest draws.

5870 CFX, 1920x1200 Full IQ settings, 4.1GHz,

1100d00d1dd000d
0011d11d0dd111d
C2D=3 wins,7 losses, rest draws.
Ph II x4=7 wins, 3 losses, rest draws.

I don't know how you managed to come up with that. Even a skim through is enough to see Ph II wins. Well, it's quadcore vs dualcore after all.

Draws are when the fps difference is only 1 fps. 2 fps isn't a draw. The GPU seems to be the bottleneck more frequently on PhII than C2D though.

Why didn't they use C2Q anyway? Why didn't they test Phenom II about the jerkiness test either? Anandtech reported "smoothness" on Core i7 and Phenom II compared to C2Q and C2D ages ago.

*Takes cover behind flameshield*

Techpowerup reviews tested PCI-e bandwidth with a single card setup, not a crossfire setup. That's why we have CFX dongles, because the bandwidth required for inter-card communication on PCI-e is simply not enough. Legionhardware tested 4870x2 CFX before on 8x 8x setup and it shows performance drop.

I conclude the review is fail (overall. It's good on some matters).

I guess you missed the whole point of terrible min fps (as in game stutter) for the Phenoms compared to the C2D or the i7 at 1920x1200 full IQ.

Haha I just saw a C2D commercial; I don't think I've ever seen one except when retailer mentions it in specs (but that is normally sale paper).

[skugpezz]

I don't think PCIe 2.0 8x vs 16x has much of an impact. techPowerUp did a review of the PCIe scaling on a 5870, but no word on minimum fps

Example of a 5870 on PCIe 2.0 8x vs 16x on crysis

http://www.techpowerup.com/reviews/A...Scaling/7.html

But it's probably the crossfire overhead eh?

doesnt matter , apples vs apples ftw

[blindbox]

I guess you missed the whole point of terrible min fps (as in game stutter) for the Phenoms compared to the C2D or the i7 at 1920x1200 full IQ.

Haha I just saw a C2D commercial; I don't think I've ever seen one except when retailer mentions it in specs (but that is normally sale paper).

Well there's only 7 games there that shows minimum fps data. 3 among them Phenom II wins, and 2 draws. That makes two C2D wins. There's no conclusion here, just not enough data.

You can't simply say min fps is the factor of smoothness. You'll need graphs for this. I can have an average of 45 fps but my game is a 60fps~30 fps saw graph or I can have an average of 45 fps constant with a single moment dip to 30 fps.

Something that some reviewers still need, user experience description.

[Manicdan]

min FPS can sometimes be a simple HDD limitation. graphs show smoothness, or a standard deviation chart might if anyone cared to make those anymore

[damtachoa]

The difference between Core2Duo 4.1 vs i7 4.1 is almost 50% every case, ...

The core2duo has over 50% power consumption than i7.

[Luka_Aveiro]

The core2duo has over 50% power consumption than i7.

@ load, idle, average...?

you think i care about power consumption when I happen to use a C0 e8400 1,46v to 4ghz under H2O?

guess what... I don't.

lol

[railer]

The 5770s have legendary crossfire scaling. You will find in a lot of tests that using two 5770s in crossfire can give a 100% and over performance increase

How many times it has to be repeated that you can not get more then 100% in CF or SLI. There is no over 100% increase period , not possible .
It just boggles me that people could believe that 5+5 in some instances could equal 11.

[zanzabar]

How many times it has to be repeated that you can not get more then 100% in CF or SLI. There is no over 100% increase period , not possible .
It just boggles me that people could believe that 5+5 in some instances could equal 11.

if u have rounding or a nonlinear score u can get over 100%

or if u have a game with frame smoothing a target fps that is set to low for the single card then it set alot higher or properly for the multi card u can get over 100%

[railer]

How? lets say hypothetically i have 2x5870 with each scoring 20K in 3dmark. Explain to me how its mathematically feasible to get 41K in CF. And lets drop artificial "rounding" when 3dmark for some reason is "set" to give me the most 19K with one card. I might be missing something but i just cant compute this in my head.