• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

bit-tech review 3870x2

I am disappointed that more reviews don't try 8xAA and don't bench 2560x. The card does well in those scenarios by comparison with the competition.

If you play 1680x with medium filtering you'd do very well with a 8800GT IMO. The 8800GTS is faster and also more cash. I believe the GTs are down to 239 now. That card should give you perfectly playable FPS under the conditions I've described. Heck I've seen some of them tackle 1920x-

http://www.driverheaven.net/reviews/8800GTs/index.php
 
lol417 said:
bit-tech review is up since yesterday. According to their review it seems like the 3870x2 is a bit slower or close to the 8800gtx. About the same conclusion as [H]

http://www.bit-tech.net/hardware/2008/01/29/amd_ati_radeon_hd_3870_x2/14
Click to expand...

Hmm how did they came to that conclusion from those graphs ?
It seems like 3870 lost only in WIC min framerates (which seems to be driver problem), and Enemy territory. Other are either very close oras with Crysis AMD win.

Anyway with that kind of cash to spend I'd get 8800 gt and then another one in SLI :D
 
kllrnohj said:
Except that different maps can cater to the strengths of the card. One map might be more shader intensive or be more spacious than another, etc... Especially in UT3, where despite being on the same engine, the maps can vary greatly in graphical load and look. My card tore it up in heatray, but got bogged down in shangri la... I was just throwing it out there as a possibility more than anything.
.
Click to expand...

Not only that, I got the card now and have tried to test it with same method as [H] in COD 4 . Something is wrong. I used Fraps 2.9.3 buld 6914 and tried to see what could change. Stood against a wall and moved quickly back and forth. Here's the result from a quick capture:

443
314
163
137
147
155
153
151
150
154
151
146
153
148
152
144
138
144
138
120
121
128
130
136
129
132
115
106
102
101
115
122
119
103
104

Frames Time (ms) Min Max Avg
5214 35428 100 443 147.172


Fraps seemed to cumulate the FPS. The faster I moved, the more FPS I got. Pretty useless for real time benchmarking in current state. Any idea what I did wrong?
 
kllrnohj said:
Yeah, you stared at a wall :rolleyes:
Click to expand...

Lol! :D

Seriously, if I can get an average of 3 times the amount [H] got by moving against a wall, something is either wrong with my settings or with the method itself. I got the same cumulative FPS by running fast around as well within the same area. Seems like the more speed I play in, the higher FPS I get.
 
Tamlin_WSGF said:
Seriously, if I can get an average of 3 times the amount [H] got by moving against a wall, something is either wrong with my settings or with the method itself. I got the same cumulative FPS by running fast around as well within the same area. Seems like the more speed I play in, the higher FPS I get.
Click to expand...

Um... you do realize games have crazy things called optimizations so shit you can't see doesn't get rendered, right? Staring at a wall will easily boost your FPS to obscene numbers, but who cares? CoD4 doesn't boast of "OUTSTANDING WATCHING-PAINT-DRY GAMEPLAY!!!!" on the box, does it?

Try this, have FRAPS record your play through on the same map as [H] did, and then report your results :rolleyes:
 
kllrnohj said:
Um... you do realize games have crazy things called optimizations so shit you can't see doesn't get rendered, right? Staring at a wall will easily boost your FPS to obscene numbers, but who cares? CoD4 doesn't boast of "OUTSTANDING WATCHING-PAINT-DRY GAMEPLAY!!!!" on the box, does it?

Try this, have FRAPS record your play through on the same map as [H] did, and then report your results :rolleyes:
Click to expand...

Of course I know this. No need for you to come with sarcastic remarks about this or some snotty rolleyes icon.
That doesn't explain the cumulative reaction that Fraps have when I increase the speed of my character (while running through the landscape not looking at a wall). With this reaction, the results can't be quantified satisfactory.
 
Hey tam, He wasnt trying to be sarcastic I dont think. Just saying that looking at the wall isnt stressing the system. Try to to join a multiplayer match with a lot of people and see if the results change.
 
FlyinBrian said:
Hey tam, He wasnt trying to be sarcastic I dont think. Just saying that looking at the wall isnt stressing the system. Try to to join a multiplayer match with a lot of people and see if the results change.
Click to expand...

It looked like it, but if he didn't then I apologize for assuming it.

The thing is that I get the same boost in FPS when I run through the same maps as [H] in high speed. My movement boost the FPS with 50-100 upon gameplay. The higher FPS gets to 250-300 fps and I'm not trying to boost the FPS, just playing fast.

The reason why I asked, is that I react upon these results and first I wish to figure out if something is wrong with my setup.

I was thinking also if ATI's powerplay might have something to do with it if the rest of the setup is ok. I hear the fan is more loud when moving fast then if going slowly and tactically through the map.
 
FRAPS is just recording the FPS, nothing more. It doesn't care how fast you are moving in game. I have no idea why moving fast would increase your FPS, but I do think that you are jumping to that conclusion a bit prematurely.

Try enabling the in game FPS display: "\cg_drawFPS 1 (choices are 0,1,2,or 3 where 0 = off)
 
jebo_4jc said:
Vista 64 bit is the future, so that's understandable.

There does seem to be something "else" impacting these reviews though. Sites that I trust show significantly differing results, both standing by their methods.

I guess all this mess really speaks to the necessity to check and re-check your facts about hardware before purchasing. I feel sorry for the poor schmuck that stands in the video card aisle at Best Buy or Fry's (or worse...wal-mart) and has to make a decision on the spot. It's hard enough for people like us, who are darn near "experts" on PC hardware, to make an informed decision.
Click to expand...

Actually they have it best - they buy bring it home and unless they find chosen card not working good in game x they are happy as they don't know they could have gotten better results.
 
Silus said:
It did win in UT3, but as usual, it's one of those games that are not tested with AA.
Click to expand...
Yes.. problem with that UT3 test was that HD3870 X2 were getting pretty freakish scores:
HD3870 X2 got 31.1% higher max fps
HD3870 X2 got 9.7% lower min fps
HD3870 X2 got 1% higher avg fps

We don't know how much did those peak fps's rise that average fps.
 
kllrnohj said:
FRAPS is just recording the FPS, nothing more. It doesn't care how fast you are moving in game. I have no idea why moving fast would increase your FPS, but I do think that you are jumping to that conclusion a bit prematurely.

Try enabling the in game FPS display: "\cg_drawFPS 1 (choices are 0,1,2,or 3 where 0 = off)
Click to expand...

Thanks! :)

I think I figured out why this is happening. In all my tests, whether the scene was upon first rendering or second, it seems that the 3870x2 pulls the textures already in memory when rendering a new scene. It perhaps stores some of the texture in the memory dedicated to GPU 2, while GPU memory 1 processes new frames. This gives an increadible boost on FPS 4-5 times first rendering and it stays that way until scene is changed.

I hope new drivers will have this more perfected. :D
 
With FRAPS you're better off recording frametimes, since those are reported on a per-frame basis, unlike FPS which are only polled every second.

Then just compute the FPS (1000ms divided by x ms reported by fraps), and expand that formula in excel, then graph it.
 
I'm getting superb results with my 3870x2

Crysis - is slightly temperamental though - often running better in windowed mode than full screen (which then initialises cf) - but even still getting much better frame rates compared to my previous 8800GTS (G92) - on assault level I get circa 20-26 fps solid frame rate everything set to high - 1900x1200

"High" benchmark, DX9 mode gives me min 25fps, 52 fps average frame rate

The Witcher the card excels with - get average of 40fps - a lot of the time 50-60 - no stutter or anything, 1900x1200 4xFSAA 16xAF

FSX runs like a dream with the card - its much better with FSAA than my GTS was

yet to try any other games yet - but impressed so far
 
This might be interesting for us waiting for newer drivers:
http://veritasindia.com/2008/01/30/update-radeon-845-v13-driver-to-improve-3870-x2-performance/

Maybe this is the beta?
http://files.filefront.com/Sample+R680+xp+vista+8+45aexe/;9552005;/fileinfo.html
Ati's digital signature on this file is 24. januar 2008 17:54:00, so it might be new.

Sapphire's drivers look newer as well:

http://www.sapphiretech.com/us/support/drivers.php

Beta from HIS (3870x2 only driver):
http://www.hisdigital.com/html/driver.php

French review with 8.1 drivers + fix (against Ultra and crossfire):
http://www.hardware.fr/articles/701-8/dossier-amd-radeon-hd-3870-x2.html
 
Theyre several days late if those new drivers were meant to hit on the 28th or the 30th jan =/
 
Chris_B said:
Theyre several days late if those new drivers were meant to hit on the 28th or the 30th jan =/
Click to expand...

I'm not suprised. ATI and Nvidia likes to delay stuff. All the new goodies from Nvidia that was supposed to be announced this month is delayed to march. Same goes with the crossfirex from ATI that was supposed to be released last month. :mad:
 
Sr7 said:
With FRAPS you're better off recording frametimes, since those are reported on a per-frame basis, unlike FPS which are only polled every second.

Then just compute the FPS (1000ms divided by x ms reported by fraps), and expand that formula in excel, then graph it.
Click to expand...

Thanks, but the results weren't so much different and a graph would be huge. It only gives me the average as well and its the same as in the avg. fps. When I summed it up, I got: 50,950016688197034861903007852964, while the AVG. FPS was shown as 50.944.

Did a test with new drivers:
8.451-071220a1-058596C-ATI link
Downloaded from HIS. Its the 28th of Jan beta drivers. It looks like they fixed the cumulative problem I had with Fraps, so I could finally test COD4 properly.

Test machine:
Not beefy
PSU: Corsair 620W
CPU: E6600@2997mhz 1333FSB
Memory: Corsair Value 4 x 1024 (Only 2048mb used) Timings@ 5,5,5,15 2T.
Harddrive: Samsung SP2504C 250GB SATA
OS: Vista home basic 32-bit

COD 4 - Modern warfare "The Bog" mission


1920x1200 all settings at max, 4xAA 16xAF (same settings as [H], but with 4XAA/Real AA, not transparency multisampled):

Frames Time (ms) Min Max Avg
29462 578318 8 163 50.944

Graph:
cod43870x2mo7.png


Haven't tested it at 2XAA yet, only 4XAA. :D

Looks like that driver was gold. Can't wait for the 8.2 :D
 
Tamlin_WSGF said:
Thanks for correcting this. :) It should read Transparency Multisampling (TRMS). Post corrected.
Click to expand...

FYI, the [H] test used 2X MSAA for the 3870X2 - the 8800GTX had the 4X TRMS :)

Must have been some driver update if it boosted FPS while allowing you to run with higher FPS though...
 
kllrnohj said:
FYI, the [H] test used 2X MSAA for the 3870X2 - the 8800GTX had the 4X TRMS :)

Must have been some driver update if it boosted FPS while allowing you to run with higher FPS though...
Click to expand...

I know. Thats why I used 4XAA. Performance hit is usually greater using conventional AA from within the game vs. TRMS.

I used different drivers, different machine and different OS. Still beta drivers though. :)
 
You must log in or register to reply here.
Back
Top