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

[skugpezz]

How strong a processor does the fastest single-chip graphics card need to show its best? And what processor would be necessary for two cards like that in CrossFireX mode? You’ll find the answers in this review.





http://www.xbitlabs.com/articles/vid...u-scaling.html

[Luka_Aveiro]

The difference between Core2Duo 4.1 vs i7 4.1 is almost 50% every case, so Crysis does take advantage of SMT.

Such a "bad coded" game, huh?

Interesting to see also that adding a second card offers performance increases below 30%, something wrong with GPU scaling, but CPU scaling seems ok. Drivers, perhaps?

[Jamesrt2004]

The difference between Core2Duo 4.1 vs i7 4.1 is almost 50% every case, so Crysis does take advantage of SMT.

Such a "bad coded" game, huh?

Interesting to see also that adding a second card offers performance increases below 30%, something wrong with GPU scaling, but CPU scaling seems ok. Drivers, perhaps?

yeah x-fire is still fail on the 5 series :/

[kiwi]

So, the numbers don’t always provide the full picture

So true

[DjKlax]

yeah x-fire is still fail on the 5 series :/

I wouldn't call it a failure, i would just say that in this particular games XFire doesn't scale to good. Just look at the Stalker Clear Sky and Far Cry 2 numbers and you will see a different picture.

At least in Stalker the numbers are up by around 90% for CrossFire and core number and speed make almost no difference.

It all depends on drivers, on the game developers ability to code for a platform and other stuff like that.

[jam2k]

Comparing i7 (Quad Core) against Core2Duo (Dual core) is pretty much retarded.

They should compare i7 against Core2Quad.

[skugpezz]

The difference between Core2Duo 4.1 vs i7 4.1 is almost 50% every case, so Crysis does take advantage of SMT.

Such a "bad coded" game, huh?

Interesting to see also that adding a second card offers performance increases below 30%, something wrong with GPU scaling, but CPU scaling seems ok. Drivers, perhaps?

I take this with a huge bag of salt. They always get results that no one ever gets

[informal]

I take this with a huge bag of salt. They always get results that no one ever gets

Indeed,it's Xbitfail after all.

[demonkevy666]

no L3 cache speeds listed......
cas 4 with 4 slots filled doesn't happen on Amd, this isn't new either >_>... all 4 stick memory are relaxed slightly.

[LedHed]

Comparing i7 (Quad Core) against Core2Duo (Dual core) is pretty much retarded.

They should compare i7 against Core2Quad.

I agree it is somewhat skewed since they are comparing 2 cores to 4 cores and like you said C2Q would be a better comparison. My question is how does the C2D@5ghz+ (since many extreme OCers push the 45nm Wolfdale through the 5ghz wall) do against the i7 at stock? I'm really surprised the C2D@4.1ghz pulled the numbers it did.

The difference between Core2Duo 4.1 vs i7 4.1 is almost 50% every case, so Crysis does take advantage of SMT.

I guess you are ignoring the fact that at 1920x1200 where the GPU is finally getting stressed the C2D@4.1ghz is only 2-3fps different than the i7 when the GPU is under maximum stress (Fully IQ@1920x1200). This is the main reason I went with a highly clocked dual core system, many games just don't use all 4 cores effectively and the higher clocked C2D pulls ahead for that reason. Now down the road yes the quads will pull away but right now the C2D@4ghz+ are pretty good CPUs even at the high end market. If you play at 1920x1200 it really depends on your GPU a lot more than your CPU.

[Behemot]

Agree, pointing on difference at 1280×1024 is really stupid. Everyone knows i7 is better in clock-to-clok comparison with Phenom II. But normal user has some cheap Phenom II and HD 4870/HD 5770 (at maximum) for 1280×1024 resolution. But almost no difference (single GPU) between Phenom II and Core i7 (clock-to-clock!!) at 1920×1200. For that is almost every 2,8 GHz+ K10/Lynnfield or whatever enough. So what want authors of articles like this point to?? I still don't get it.

[Manicdan]

id like to see article try to answer questions like, whats the best resolution for me, or how much should i spend on my cpu, or is my duel core getting too old. interesting niche articles that show how one aspect is affected by all the others. too many reviews just do the fastest stuff, at cookie cutter settings, which in the end gives us 100 tests all identical, with different answers.

[Mungri]

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 (I get 23 FPS in heaven benchmark with one 5770, and 48 with both at the same clock speeds).

Core I7 with a pair of 5770s is IMO currently the most efficient setup.

[LedHed]

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 (I get 23 FPS in heaven benchmark with one 5770, and 48 with both at the same clock speeds).

Core I7 with a pair of 5770s is IMO currently the most efficient setup.

benchmarks are obviously going to use SLI/CF better than games, but no one really cares about benchmark performance except benchers. The question is how much FPS do you get in an actual game between one 5770 and two.

[zalbard]

Seems suspicious.
Coming from Core 2 Duo to i7 I gotta say there is a noticeable overall performance boost, though.
So they are right, at least to some extent.

[LedHed]

Seems suspicious.
Coming from Core 2 Duo to i7 I gotta say there is a noticeable overall performance boost, though.
So they are right, at least to some extent.

not at 1920x1200, it's barely noticeable in most games. Now newer games that use Quads may pull the gap further, but it's only 2-3fps at 1920x1200 Full IQ.

[Oliverda]

"One more pitfall for the AMD platform in this test session was PCI Express. The Gigabyte GA-MA790GP-DS4H mainboard is based on the AMD 790GX chipset which supports two ATI Radeon cards in CrossFireX only as PCIe x16 + PCIe x8"

It's x8+x8 in CF and not x16+x8.

They should have used a high end 790FX AM3 board for this test and not a AM2+ mainstream mobo.

i7 - X58 + DDR3 + x16+x16

C2D - X48 + DDR3 + x16+x16

PHII - 790GX + DDR2 + x8+x8

[SamHughe]

I agree. Why would anybody who has i7 @ 4.1 Ghz and 5879 crossfire wants to play Crysis at 1280x1024 res. with no AA/AAF?

They should have found out the highest "playable" settings (i.e. 1680x1050 2xaa, No AAF) on the lowest clock speeds (i.e. 2.7 ghz) and take that setting as a starting point in their comparison.

[Manicdan]

cant wait for monitors that support 240FPS (even though that is a very long ways off)

[Mungri]

benchmarks are obviously going to use SLI/CF better than games, but no one really cares about benchmark performance except benchers. The question is how much FPS do you get in an actual game between one 5770 and two.

You can google a review for this information, I do not need to provide it .

Crossfire gain in games on the 5770 is so good that every review recommends buying them.

[LedHed]

You can google a review for this information, I do not need to provide it .

Crossfire gain in games on the 5770 is so good that every review recommends buying them.

already seen reviews and the performance gain in Games is not 100%, 75% at the most.

[Xoulz]

"One more pitfall for the AMD platform in this test session was PCI Express. The Gigabyte GA-MA790GP-DS4H mainboard is based on the AMD 790GX chipset which supports two ATI Radeon cards in CrossFireX only as PCIe x16 + PCIe x8"

It's x8+x8 in CF and not x16+x8.

They should have used a high end 790FX AM3 board for this test and not a AM2+ mainstream mobo.

i7 - X58 + DDR3 + x16+x16

C2D - X48 + DDR3 + x16+x16

PHII - 790GX + DDR2 + x8+x8

Yep, like the article was purposely skewed.. Phenom II, no DDR3, lol.

[Spectrobozo]

yes... quite strange, no core 2 quad, phenom II without using 790FX (16x+16x) and using ddr2... also, wouldn't 2x2gb be more appropriate than 4x1gb?...

at the end I don't think it would make a huge difference, but...

[Manicdan]

yes... quite strange, no core 2 quad, phenom II without using 790FX (16x+16x) and using ddr2... also, wouldn't 2x2gb be more appropriate than 4x1gb?...

at the end I don't think it would make a huge difference, but...

the only people who probably have 4x1gb of ddr2 are probably still using a PII 940, like myself, for the last year now. but even i have a 2x 16x PCIe slots, and my mobo is older than any PII out there.

[Mungri]

already seen reviews and the performance gain in Games is not 100%, 75% at the most.

Here:

http://www.tweaktown.com/reviews/298...re/index9.html

Probably 95-110% (estimate) performance gain in Far Cry 2.