SLI GeForce GTX 460 @ 775/1000 versus Radeon HD 5970 @ 800/1150 !!!

[bakalu]

My system
CPU Core i7 Extreme 965 @ 3.6GHz
Mainboard ASUS Rampage II Gene
RAM 3x2GB DDR3
ASUS GTX 460 768MB core/shader/memory 675/1350/900
ASUS GTX 460 768MB Direct CU core/shader/memory 700/1400/920
ASUS HD 5970 core/memory 725/1000
Windows 7 Ultimate 64bit RTM
Driver NVIDIA 258.96 beta
Driver ATI 10.7 WHQL + 10-7_ATICatalyst_AppProfiles

3DMark Vantage Performance

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

Crysis Warhead. 1920*1080, AA=4X, Enthusiast DX10, ambush map

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

I played a short and used Fraps to benchmark, map Train. 1920*1080, AA=4X, Enthusiast DX10

I begin to benchmark here

And stop here

ASUS HD 5970 @ 800/1150

2010-07-28 10:45:11 - Crysis
Frames: 14320 - Time: 274358ms - Avg: 52.195 - Min: 20 - Max: 89

SLI ASUS GTX 460 768MB @ 775/1000

2010-07-28 14:03:04 - Crysis
Frames: 11787 - Time: 276624ms - Avg: 42.610 - Min: 14 - Max: 80

Call of Juarez DX10. 1920*1080, AA=4X Super_sampling, Max Settings

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

S.T.A.L.K.E.R Call of Pripyat DX11. 1920*1080, AA=4X, Max Settings, Tessellation=ON

Settings in S.T.A.L.K.E.R Call of Pripyat DX11

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

Demo benchmark Dirt 2 DX11. 1920*1080, AA=8x MSAA, Ultra Settings. POST PROCESS=HIGH

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

Benchmark Bad Company 2 DX11. I play the first map. Use Fraps. 1920*1080, AA=8X, AF=16X, Max Settings

I begin to benchmark here

And stop here

ASUS HD 5970 @ 800/1150

2010-07-28 12:04:04 - BFBC2Game
Frames: 55874 - Time: 575774ms - Avg: 97.042 - Min: 33 - Max: 147

SLI ASUS GTX 460 768MB @ 775/1000

2010-07-28 15:24:42 - BFBC2Game
Frames: 53637 - Time: 556597ms - Avg: 96.366 - Min: 44 - Max: 160

Benchmark Metro 2033. I play the first scene. Use Fraps. 1920*1080, NoAA, AF=16X, Very High, DX11

I begin to benchmark here

And stop here

ASUS HD 5970 @ 800/1150

2010-07-28 11:51:21 - metro2033
Frames: 8712 - Time: 241089ms - Avg: 36.136 - Min: 15 - Max: 60

SLI ASUS GTX 460 768MB @ 775/1000

2010-07-28 15:14:42 - metro2033
Frames: 8052 - Time: 239472ms - Avg: 33.624 - Min: 18 - Max: 54

Devil May Cry 4 DX10. 1920*1080, AA=8X, Super High

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

Resident Evil 5 DX10. 1920*1080, AA=8X, Max Settings

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

Heaven Demo benchmark DX11. 1920*1080, AA=4X, AF=16X, Tessellation=Extreme, Max Settings

ASUS HD 5970 @ 800/1150

SLI ASUS GTX 460 768MB @ 775/1000

Warhaead Crysis gameplay, HD5970 is much smoother than SLI GTX460 768MB. In conclusion, victory belongs HD5970.

Performance of SLI GTX460 768MB @ 775/1000 is more stronger GTX480. You can see the benchmark's result of GTX480 in the same system of SLI GTX460 768MB @ 775/1000 here : http://forum.amtech.com.vn/overcloc...asus-hd-5870-matrix-2gb-gddr5-950-1250-a.html

Source : http://forum.amtech.com.vn/overcloc...sus-radeon-hd-5970-800-1150-a.html#post700994

 

[overclocking101]

not bad at all man the dual gpu nvidia card is going to come close or beat the 5970 and it shows. plus dual 460's are cheaper. oh and OC that damn cpu man its an unlocked proc push it! those scores will gain tons with more cpu, (I know you did it for review purposes im just sayin)

 

[eidairaman1]

I could only assume from the benchmarks that the 5970 which is sandwiched 5850s is still beating out cards that are In SLI despite being bandwidth limited due to being on a single slot, also being 6-8 months older than the competition. Good bench run I say. I could only Imagine how 2x would run. Now only if we Could get a 5975 (Non ASUS 5870X2 2GB/4GB Model). Also reason why AMD parts havent really gone down in price is they feel pretty comfortable at the price range they are at currently, reminds me of the G80 8800GTX/Ultra.

 

[newtekie1]

So...for ~$300 more the HD5970 gives less than 10% better performance...nice. I think if nVidia could get two of these GF104 cores with all the shaders enabled(remember the GTX460 has 48 Shaders disabled and the ones in this test had a memory controller disabled on the core as well) they might have a card that can really push the HD5970.

I could only assume from the benchmarks that the 5970 which is sandwiched 5850s is still beating out cards that are In SLI despite being bandwidth limited due to being on a single slot, also being 6-8 months older than the competition. Good bench run I say. I could only Imagine how 2x would run. Now only if we Could get a 5975 (Non ASUS 5870X2 2GB/4GB Model). Also reason why AMD parts havent really gone down in price is they feel pretty comfortable at the price range they are at currently, reminds me of the G80 8800GTX/Ultra.

Bandwidth limitted? When a single HD5870 isn't bandwidth limitted from dropping to an x8 slot, why would two underclocked HD5870s be bandwidth limitted when sharing an x16?

 

[eidairaman1]

Forcing 2 cards to operate on a single bus.

So...for ~$300 more the HD5970 gives less than 10% better performance...nice. I think if nVidia could get two of these GF104 cores with all the shaders enabled(remember the GTX460 has 48 Shaders disabled and the ones in this test had a memory controller disabled on the core as well) they might have a card that can really push the HD5970.



Bandwidth limitted? When a single HD5870 isn't bandwidth limitted from dropping to an x8 slot, why would two underclocked HD5870s be bandwidth limitted when sharing an x16?

 

[adj408]

Wow nice read on your awesome review!

 

[bakalu]

Additional results of the benchmark SLI GeForce GTX 460 1GB @ 775/1000

My system
CPU Core i7 Extreme 965 @ 3.6GHz
Mainboard ASUS Rampage II Gene
RAM 3x2GB DDR3
ASUS GTX 460 1GB Direct CU core/shader/memory 775/1550/1000
GIGABYTE GTX 460 1GB OC core/shader/memory 715/1430/900
Windows 7 Ultimate 64bit RTM
Driver NVIDIA 258.96 beta

3DMark Vantage Performance

SLI GeForce GTX 460 1GB @ 775/1000

Crysis Warhead. 1920*1080, AA=4X, Enthusiast DX10, ambush map

SLI GeForce GTX 460 1GB @ 775/1000

I played a short and used Fraps to benchmark, map Train. 1920*1080, AA=4X, Enthusiast DX10

I begin to benchmark here

And stop here

SLI GeForce GTX 460 1GB @ 775/1000

2010-07-29 11:17:59 - Crysis
Frames: 14089 - Time: 273362ms - Avg: 51.540 - Min: 15 - Max: 87

Call of Juarez DX10. 1920*1080, AA=4X Super_sampling, Max Settings

SLI GeForce GTX 460 1GB @ 775/1000

S.T.A.L.K.E.R Call of Pripyat DX11. 1920*1080, AA=4X, Max Settings, Tessellation=ON

Settings in S.T.A.L.K.E.R Call of Pripyat DX11

SLI GeForce GTX 460 1GB @ 775/1000

Demo benchmark Dirt 2 DX11. 1920*1080, AA=8x MSAA, Ultra Settings. POST PROCESS=HIGH

SLI GeForce GTX 460 1GB @ 775/1000

Benchmark Bad Company 2 DX11. I play the first map. Use Fraps. 1920*1080, AA=8X, AF=16X, Max Settings

I begin to benchmark here

And stop here

SLI GeForce GTX 460 1GB @ 775/1000

2010-07-29 11:56:18 - BFBC2Game
Frames: 56880 - Time: 555258ms - Avg: 102.439 - Min: 48 - Max: 166

Benchmark Metro 2033. I play the first scene. Use Fraps. 1920*1080, NoAA, AF=16X, Very High, DX11

I begin to benchmark here

And stop here

SLI GeForce GTX 460 1GB @ 775/1000

2010-07-29 11:24:11 - metro2033
Frames: 8871 - Time: 239294ms - Avg: 37.072 - Min: 20 - Max: 60

Devil May Cry 4 DX10. 1920*1080, AA=8X, Super High

SLI GeForce GTX 460 1GB @ 775/1000

Resident Evil 5 DX10. 1920*1080, AA=8X, Max Settings

SLI GeForce GTX 460 1GB @ 775/1000

Heaven Demo benchmark DX11. 1920*1080, AA=4X, AF=16X, Tessellation=Extreme, Max Settings

SLI GeForce GTX 460 1GB @ 775/1000

Warhaead Crysis gameplay, SLI GTX460 1GB is much smoother than SLI GTX460 768MB. Frame rate of SLI GTX460 768MB will be dropped down a lot when turning or having the explosion since 768MB is not enough to play Crysis Warhead at resolution 1920*1080, Enthusiast DX10, AA = 4X.

The benchmark result of Stalker DX11 Call of Pripyat, the min frame rate of SLI GTX460 1GB is much higher than HD5970.

In terms of performance/price, SLI GTX460 1GB is much better than HD5970.

 

[happita]

I haven't really checked too many 460 reviews, but I especially haven't seen them compared to a 5970 in SLI. Very interesting to see. Me thinks 5970 should drop in price

Great job man!

 

[newtekie1]

Forcing 2 cards to operate on a single bus.

Yeah, I get that, but it still doesn't make any sense.

In an x8 slot, which gives 4GB/s of bandwidth a full HD5870 shows no real performance drop. So why would you think that two, underclocked, sharing 8GB/s of bandwidth in an x16 slot would be bandwidth limitted by the slot?

 

[bakalu]

To view the performance of SLI GTX 460 1GB and CrossFireX HD 5850 improve how much to compare with one VGA, I benchmark 5850 and GTX 460 1GB at the same resolution and clock of 5970 and SLI GTX 460 1GB.

My system
CPU Core i7 Extreme 965 @ 3.6GHz
Mainboard ASUS Rampage II Gene
RAM 3x2GB DDR3
ASUS GTX 460 1GB Direct CU core/shader/memory 775/1550/1000
ASUS HD 5850 core/memory 725/1000
Windows 7 Ultimate 64bit RTM
Driver NVIDIA 258.96 beta
Driver ATI 10.7 WHQL + 10-7_ATICatalyst_AppProfiles

3DMark Vantage Performance

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

Crysis Warhead. 1920*1080, AA=4X, Enthusiast DX10, ambush map

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

I played a short and used Fraps to benchmark, map Train. 1920*1080, AA=4X, Enthusiast DX10

I begin to benchmark here

And stop here

ASUS HD 5850 @ 800/1150

2010-07-31 08:23:10 - Crysis
Frames: 8767 - Time: 274972ms - Avg: 31.883 - Min: 15 - Max: 53

ASUS GTX 460 1GB @ 775/1000

2010-07-30 09:05:29 - Crysis
Frames: 7653 - Time: 275306ms - Avg: 27.798 - Min: 11 - Max: 46

Scale SLI and CrossFireX 2-way over single VGA:

Call of Juarez DX10. 1920*1080, AA=4X Super_sampling, Max Settings

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

S.T.A.L.K.E.R Call of Pripyat DX11. 1920*1080, AA=4X, Max Settings, Tessellation=ON

Settings in S.T.A.L.K.E.R Call of Pripyat DX11

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

Demo benchmark Dirt 2 DX11. 1920*1080, AA=8x MSAA, Ultra Settings. POST PROCESS=HIGH

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

Benchmark Bad Company 2 DX11. I play the first map. Use Fraps. 1920*1080, AA=8X, AF=16X, Max Settings

I begin to benchmark here

And stop here

ASUS HD 5850 @ 800/1150

2010-07-31 09:49:11 - BFBC2Game
Frames: 32271 - Time: 553377ms - Avg: 58.316 - Min: 13 - Max: 92

ASUS GTX 460 1GB @ 775/1000

2010-07-30 09:23:39 - BFBC2Game
Frames: 30677 - Time: 560596ms - Avg: 54.722 - Min: 26 - Max: 91

Scale SLI and CrossFireX 2-way over single VGA:

Benchmark Metro 2033. I play the first scene. Use Fraps. 1920*1080, NoAA, AF=16X, Very High, DX11

I begin to benchmark here

And stop here

ASUS HD 5850 @ 800/1150

2010-07-31 10:00:43 - metro2033
Frames: 5025 - Time: 237458ms - Avg: 21.162 - Min: 11 - Max: 35

ASUS GTX 460 1GB @ 775/1000

2010-07-30 10:09:52 - metro2033
Frames: 4631 - Time: 238197ms - Avg: 19.442 - Min: 11 - Max: 34

Scale SLI and CrossFireX 2-way over single VGA:

Devil May Cry 4 DX10. 1920*1080, AA=8X, Super High

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

Resident Evil 5 DX10. 1920*1080, AA=8X, Max Settings

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

Heaven Demo benchmark DX11. 1920*1080, AA=4X, AF=16X, Tessellation=Extreme, Max Settings

ASUS HD 5850 @ 800/1150

ASUS GTX 460 1GB @ 775/1000

Scale SLI and CrossFireX 2-way over single VGA:

Clearly, improved SLI performance is a lot better than CrossFireX, so not surprisingly performance of GTX 460 1GB lost HD 5850 a lot but the performance of SLI GTX 460 1GB is the same as Radeon HD 5970.