Bought an Asus 5970, Benchmarking PCI-E 1.0 vs PCI-E 2.0 5970vs5870, 5970/5870/5850CF

[happy medium]

I don't get it. You do realize two of these
http://www.newegg.ca/Product/Product...-351-_-Product

Would be faster and $150 cheaper right?

I think you are wrong. A 5970 is clocked the same as a 5850 ,but has more sp's and more TMU's.
With a small overclock they should easily beat highly overclocked 5850's in crossfire.

Two ASUS 5850's would cost you over 600$

It's a fine purchase and they should hit 900 core easy and smoke any overclocked combo of 5850's.

 

[Zebo]

I think you are wrong. A 5970 is clocked the same as a 5850 ,but has more sp's and more TMU's.
With a small overclock they should easily beat highly overclocked 5850's in crossfire.

Two ASUS 5850's would cost you over 600$

It's a fine purchase and they should hit 900 core easy and smoke any overclocked combo of 5850's.

Every review shows stock xfire 5850s faster than stock 5970 and the cards I pointed him to are OCed stock further lengthening lead.
http://www.anandtech.com/show/2877/13

If you want to OC both 5850 also has advantage by having a separate HSF for each core.

 

[RussianSensation]

Nice videocard I like the understated look. Fire her up and show us some benches against a single 5870 in BF:BC2

However, what camera were you using? Massive corner-edge blurriness everywhere (Canon SD1100-1200?)

 

[happy medium]

Every review shows stock xfire 5850s faster than stock 5970 and the cards I pointed him to are OCed stock further lengthening lead.
http://www.anandtech.com/show/2877/13

If you want to OC both 5850 also has advantage by having a separate HSF for each core.

I guess the 5970 just sucks with low quality games that get 200 fps because even this review has the 5850's beating the 5970.

http://www.guru3d.com/article/radeon-hd-5970-review-test/20

But if you use a game with some high quality grapgics like Crysis warhead the 5850's get smoked. In fact it's beating 5870's in crossfire here.

http://www.guru3d.com/article/radeon-hd-5970-review-test/16

Overall they seem about even, when both at stock, depending on the review, resolution, and drivers used. The latest drivers give the 5970 a nice jump.

The one advantage the card has is you can crossfire it with another 58xx series card in the future or another 5970.

 

[1h4x4s3x]

Nice.
If you've enough time, can you do some benches and log the fps and maybe compare it to your 5870? I wonder if there's any microstuttering.

 

[Apocalypse23]

Thank you everyone for the warm comments, I'm conducting benchmarks atm, will let you guys know soon. Stay tuned.

The benches will be based on:

3dMark Vantage
Dirt2
BFBC2

 

[Will Robinson]

Awesome setup dude...you rock.
Thanks for the pics.

 

[Apocalypse23]

Thanks Will Robinson.

Okay folks, I ran the initial benchmarks on my current system:

E8400@3.9Ghz
Asus P5KE-Wifi AP mobo PCI-E 1.0 16x
8gb DDR2 PC2 6400
Windows 7 Pro 64
CCC driver 10.3a
CCC settings at default and AA/AF on application control

The results indicate a clear CPU and PCI-E 1.0 bottleneck, I'll be testing this on my new i7 setup next week hopefully to compare.
My Benchmarks:

 

[Tempered81]

word, nice results. 5970 level power is worthless without a good >3.5g i7 + X58 to support it.

 

[Apocalypse23]

Nice videocard I like the understated look. Fire her up and show us some benches against a single 5870 in BF:BC2

However, what camera were you using? Massive corner-edge blurriness everywhere (Canon SD1100-1200?)

Hehe, the camera I'm using kinda sucks when taking still images without flash...it's a pain with flash as well as it takes a lot of time to focus on an object..It's a Sony cybershot 7.2 MP DSC-W55.

 

[Apocalypse23]

word, nice results. 5970 level power is worthless without a good >3.5g i7 + X58 to support it.

Yup, I was a bit disappointed really , anyhow we will see how it performs on the new setup..I'll try to get some benchmark comparison's from my brother's PCI-E 2.0 E8400 setup soon, once he lets me on it.

 

[Apocalypse23]

Are there any Mods that I can pm? I need this thread's title changed...please let me know MODS

 

[BTRY B 529th FA BN]

Once you click edit on the opening post, click Go Advanced and you should be able to change the title.

Nice desk btw, I think I have the same.

 

[Apocalypse23]

Once you click edit on the opening post, click Go Advanced and you should be able to change the title.

Nice desk btw, I think I have the same.

Roger that, thanks bro!

 

[RussianSensation]

word, nice results. 5970 level power is worthless without a good >3.5g i7 + X58 to support it.

Exactly. Look at those Dirt2 results. How can you have a 52% higher maximum framerates and conclude PCIe 1.0x bottleneck? Many reviews have clearly shown that PCIe 16x Gen1 results are about 2-3% worse than PCIe 16x Gen2. It's either CF profiles are not enabled properly, or there is a CPU bottleneck if anything. Those BF:BC2 results just don't make any sense!

 

[Apocalypse23]

Exactly. Look at those Dirt2 results. How can you have a 52% higher maximum framerates and conclude PCIe 1.0x bottleneck? Many reviews have clearly shown that PCIe 16x Gen1 results are about 2-3% worse than PCIe 16x Gen2. It's either CF profiles are not enabled properly, or there is a CPU bottleneck if anything. Those BF:BC2 results just don't make any sense!

Yeah, the BFBC2 scores are the most puzzling. I could not believe my CPU cores were throttling at 98-99% usage....It's like my dual core gets raped while running dual-gpus...My CPU usage was around 85-90% with the 5870 alone running BC2...

 

[edplayer]

Are there any Mods that I can pm? I need this thread's title changed...please let me know MODS

just start a new thread.

This one is all cluttered up anyways. Link back to this one if you think there is any good info

 

[RussianSensation]

Yeah, the BFBC2 scores are the most puzzling. I could not believe my CPU cores were throttling at 98-99% usage....It's like my dual core gets raped while running dual-gpus...My CPU usage was around 85-90% with the 5870 alone running BC2...

I know it's hard to believe that an E8400 @ 3.9ghz is not that fast anymore!

* 1920x1200 is about 11% greater resolution than your 1920x1080 setup. You are testing with 4AA/16AF. Let's assume for simplicity sake that 2AA->4AA and from 4AF-->16AF increases load on the graphics card by 11%. Check out CPU scaling on a Core i7:
http://www.techspot.com/article/255-battlefield-bad-company2-performance/page7.html

5870s in CF on a Core i7 920 @ 1920x1200 2AA/4AF
Core i7 3.7ghz = 100.3 frames average (+30% faster)
Core i7 2.96ghz = 77.1 frames

They dont' reach GPU bottleneck until 2500x1600 using 5870s in CF

Since C2D/Q architecture was such a breakthrough from Pentium 4/D era, a lot of people don't realize just how much faster Core i7 is over C2D/Q. I had a Q6600 @ 3.4ghz before and check this out:

Resident Evil 5 Benchmark with 4890:
Q6600 @ 3.4ghz 1680x1050 0AA = 57.8 fps
Q6600 @ 3.4ghz 1920x1080 4AA = 55.1 fps
Core i7 860 @ 3.9ghz 1920x1080 8AA = 75.8 fps

That's comparing quad to a quad not dual to a quad and with a 4890 not 5970!

 

[Apocalypse23]

I know it's hard to believe that an E8400 @ 3.9ghz is not that fast anymore!

* 1920x1200 is about 11% greater resolution than your 1920x1080 setup. You are testing with 4AA/16AF. Let's assume for simplicity sake that 2AA->4AA and from 4AF-->16AF increases load on the graphics card by 11%. Check out CPU scaling on a Core i7:
http://www.techspot.com/article/255-battlefield-bad-company2-performance/page7.html

5870s in CF on a Core i7 920 @ 1920x1200 2AA/4AF
Core i7 3.7ghz = 100.3 frames average (+30% faster)
Core i7 2.96ghz = 77.1 frames

They dont' reach GPU bottleneck until 2500x1600 using 5870s in CF

Since C2D/Q architecture was such a breakthrough from Pentium 4/D era, a lot of people don't realize just how much faster Core i7 is over C2D/Q. I had a Q6600 @ 3.4ghz before and check this out:

Resident Evil 5 Benchmark with 4890:
Q6600 @ 3.4ghz 1680x1050 0AA = 57.8 fps
Q6600 @ 3.4ghz 1920x1080 4AA = 55.1 fps
Core i7 860 @ 3.9ghz 1920x1080 8AA = 75.8 fps

That's comparing quad to a quad not dual to a quad and with a 4890 not 5970!

Yeah dude, it's nice to see you got that extra breathing room with your i7 860 , I'm really hoping I can expect some good stuff.

 

[BFG10K]

Your PCI Express 16x 1.0 lane is going to hurt performance for sure.

Not to any meaningful degree it won’t.