Overclock.net banner
Cancel

5850 xfire vs. 5870 xfire

4199 Views 22 Replies 13 Participants Last post by  gerardfraser, 
Robilar
I own sets of 5870's and 5850's.

I was curious as to the differences in performance. I have benched the 5870's thoroughly and now working with paired 5850's.

Cost on the 5870 is currently $435 CAD each, 5850 is $295 CAD each.

I have a 5970 ordered as well (cost of $700 CAD). From what I have seen, it matches a pair of 5850's in terms of performance at lower resolutions (without overclocking).

I'll post results from both as I progress through this.

For frame of reference, I run 1920x1200 default res (except in 3dmark benches where I leave it at default). CPU is clocked at 4ghz, ram at 1600mhz (8-8-8-21 1T timings). Cards are strictly at stock clocks.

First off is results of 3dmark06 (all using identical system and driver 9.11).

(Note: HT is on for this test all 3 slides)

First is single 5870, second is dual 5850's and finally dual 5870's.

Notice anything interesting about the score?



See less See more
4
Reply
Save
Like
1 - 20 of 23 Posts
myresolution_72
Quote:


Originally Posted by Robilar
View Post

I own sets of 5870's and 5850's.

I was curious as to the differences in performance. I have benched the 5870's thoroughly and now working with paired 5850's.

Cost on the 5870 is currently $435 CAD each, 5850 is $295 CAD each.

I have a 5970 ordered as well (cost of $700 CAD). From what I have seen, it matches a pair of 5850's in terms of performance at lower resolutions (without overclocking).

I'll post results from both as I progress through this.

For frame of reference, I run 1920x1200 default res (except in 3dmark benches where I leave it at default). CPU is clocked at 4ghz, ram at 1600mhz (8-8-8-21 1T timings). Cards are strictly at stock clocks.

First off is results of 3dmark06 (all using identical system and driver 9.11).

First is single 5870, second is dual 5850's and finally dual 5870's.

Notice anything interesting about the score?


The 5850 Crossfire are close to a 5870 Crossfire?
See less See more
2
Reply
Save
Like
Robilar
Exactly.

My gaming FPS tests (so far testing out COD WAW) shows them to be virtually the same at 1920x1200 with 4x AA enabled.
Reply
Save
Like
S
Quote:


Originally Posted by myresolution_72
View Post

The 5850 Crossfire are close to a 5870 Crossfire?

well according to 3dmark yes.

If possible could we get some actual game benchies?
See less See more
Reply
Save
Like
myresolution_72
Quote:


Originally Posted by Robilar
View Post

Exactly.

My gaming FPS tests (so far testing out COD WAW) shows them to be virtually the same at 1920x1200 with 4x AA enabled.

They do have the same core?

And is the 5870 just a 5850 overclocked?

If they have the same core and are the same card then it's kinda a market scam seeing how closely they perform. And charging more for a similar performing card. Like charging $250 for a GTS 250 and $150 (Being dramatic) for a 9800GTX+.

EDIT: Newegg - $309 for a 5850 and $419 for a 5870 is not even worth the price difference. Wow... Lol AMD scammed !
See less See more
Reply
Save
Like
N
Quote:


Originally Posted by Robilar
View Post

I own sets of 5870's and 5850's.

I was curious as to the differences in performance. I have benched the 5870's thoroughly and now working with paired 5850's.

Cost on the 5870 is currently $435 CAD each, 5850 is $295 CAD each.

I have a 5970 ordered as well (cost of $700 CAD). From what I have seen, it matches a pair of 5850's in terms of performance at lower resolutions (without overclocking).

I'll post results from both as I progress through this.

For frame of reference, I run 1920x1200 default res (except in 3dmark benches where I leave it at default). CPU is clocked at 4ghz, ram at 1600mhz (8-8-8-21 1T timings). Cards are strictly at stock clocks.

First off is results of 3dmark06 (all using identical system and driver 9.11).

First is single 5870, second is dual 5850's and finally dual 5870's.

Notice anything interesting about the score?


One of those runs, aka the 5850 run seems to have HT disabled. HT disabling in 06 will result in a much higher sm2 score at the same speeds. Seems to be throwing off the final scores there.
See less See more
2
Reply
Save
Like
Robilar
Perhaps the 5870's will have more overclocking headroom? (although the reviews I read on the 5850's show them to overclock really well).
Reply
Save
Like
Robilar
Quote:


Originally Posted by NCspecV81
View Post

One of those runs, aka the 5850 run seems to have HT disabled. HT disabling in 06 will result in a much higher sm2 score at the same speeds. Seems to be throwing off the final scores there.

Good point, forgot about that.

I'll reenable it and run it.
See less See more
Reply
Save
Like
Gill..
Quote:


Originally Posted by myresolution_72
View Post

They do have the same core?

And is the 5870 just a 5850 overclocked?

If they have the same core and are the same card then it's kinda a market scam seeing how closely they perform. And charging more for a similar performing card. Like charging $250 for a GTS 250 and $150 (Being dramatic) for a 9800GTX+.

EDIT: Newegg - $309 for a 5850 and $419 for a 5870 is not even worth the price difference. Wow... Lol AMD scammed !

No - they're inherently different cores - 5850's have 1440 threads, 5870 1600. They're essentially the same - but entire areas of the chip was disabled for the 5850. And yes, the 5850's are clocked much lower.

This isn't Nvidia here...
See less See more
Reply
Save
Like
myresolution_72
Quote:


Originally Posted by Gill..
View Post

This isn't Nvidia here...

Lol!!


Well that shows my knowledge of AMD. Guess I'll stay in my little nVidia circle.
See less See more
2
Reply
Save
Like
raisethe3
So i guess that means the 5850 would be the better deal then? Because according to your results, crossfiring the 5850 vs the 5870 made little difference?
See less See more
Reply
Save
Like
asuindasun
Quote:


Originally Posted by myresolution_72
View Post

They do have the same core?

And is the 5870 just a 5850 overclocked?

If they have the same core and are the same card then it's kinda a market scam seeing how closely they perform. And charging more for a similar performing card. Like charging $250 for a GTS 250 and $150 (Being dramatic) for a 9800GTX+.

EDIT: Newegg - $309 for a 5850 and $419 for a 5870 is not even worth the price difference. Wow... Lol AMD scammed !

The 5870 has more shader cores than the 5850, (1440 shaders on the 5850 vs 1600 on the 5870) but the clock speeds are pretty easy to match on a 5850. Have mine at 900/1200 (5870 is 850/1250 stock I believe?) so with a little tinkering you can make up for the lost shader cores

BUT, if you overclock the 5870s, which i have no personal experience overclocking but have seen numerous people hitting 1ghz on them, it will obviously still beat the 5850. These xfire benchies are interesting though... woulda though there would be a larger gap between the 5850 and 5870 xfire's. Glad i went 5850
See less See more
3
Reply
Save
Like
slytown
Looks like the only difference is the stream processing units in terms of specs, but if they 5850 overlocks well I don't see the need to spend $120 more either. Interesting...

Thanks for benching the differences Rob. You always seem to have the money saved up to do these hardware tests.
See less See more
Reply
Save
Like
Robilar
Hee hee, I forgot to enable HT on the 5850 xfire tests (it was on for the 5870 xfire test though).

Turning HT on pushed the 5850's over 27k and a couple of hundred points ahead of the 5870's...

(I updated the slides on the first page).
Reply
Save
Like
Etra1n
considering the scores of a single 5870 it will only be a matter of time before the xfire 5870's will scale better.

AMD most likely rushed the 5870 to production for the "Single fastest GPU crown" which resulted in a lack of xfire efficiency.

The best deal is still a single 5870.
Reply
Save
Like
N
different drivers? Really concerned about the sm2 score on the 5870s run now. =o\\. Too close to a single.
Reply
Save
Like
dir_d
Quote:


Originally Posted by Robilar
View Post

Hee hee, I forgot to enable HT on the 5850 xfire tests (it was on for the 5870 xfire test though).

Turning HT on pushed the 5850's over 27k and a couple of hundred points ahead of the 5870's...

(I updated the slides on the first page).

Are your 5870s Oced as well?
See less See more
Reply
Save
Like
Robilar
All of the cards are at stock.

All I did was enable hyper threading on the 5850's bench to match the 5870's test.
Reply
Save
Like
E
A friend of mine has crossfire 5870's and I have 5850's. Similar systems.

3DMark Vantage (performance) 5850's GPU score 21086 (stock)
5870's GPU score 21730 (stock)

I have mine now at 775/1125 and my 5850's are 22096
Reply
Save
Like
Gill..
Also please remember guys, there's more to life than 3d Vantage.

I'd be interested in seeing the performance of Crossfired systems running Milkyway/Collatz....there I think the extra threads would shine on the 5870.

2 wu's per system... but if the 5870 is finishing them 5 seconds less than the 5850 (say 40 seconds to 45 on the 5850)... that's 1/9 better than the 5850 (hypothetical)... those 5 seconds spread over an entire day would yield many more wu's for the 5870.... thereby making it scale and crunch much better - making your watts/ppd/price better on 1 or the other...

Point being - FPS might not translate well into pure computing power....CUDA is not the only kid on the block with a shiny bike anymore.
Reply
Save
Like
1 - 20 of 23 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Overclock.net
A forum community dedicated to overclocking enthusiasts and testing the limits of computing. Come join the discussion about computing, builds, collections, displays, models, styles, scales, specifications, reviews, accessories, classifieds, and more!
Full Forum Listing
Explore Our Forums
Hardware News Intel CPUs PC Gaming NVIDIA Water Cooling
Top