This 3870 IS F*&@#$& AWESOME [Seconded by S-N!]

[vixro]

I can't say it any better than that. Every single one of my expectations was met and more, this card is BAD ASS!

I am appalled at all the negative feedback the card was getting about it being less than the 8800GT. What nobody seems to mention is that it doesn't matter. The card is so fast, so slick and so good looking that it can run any game you throw at it, save Crysis, with high resolutions and high image quality settings.


I'm looking at a 864mhz/1225mhz overclock on the thing that it barely even breaks a sweat with the DuOrb cooler. 53c load temperatures and over 11,760 3dmark06 marks with only a dual core CPU.

I can't wait to push some voltage into this card and see what it can do.


Current games I play that have absolutely NO HICCUPS and run smooooooth:

See below for benches and screenshots starting with post #20.



I must say that overclocking this card is a must, I noticed over 1300 point gain in 3dmark06 when I went from stock to my max OC on the card. Once I'm able to modify the volts I don't think I'll have any trouble going over 900mhz on the core or even higher. :santa:


Since people like nerd porn, here's some pictures of the card and the cooler (the cooler has a dirty mark I couldn't remove but hasn't really done anything to hurt the overall temperatures from what I can tell):

http://images31.fotki.com/v1054/photos/8/873576/5678784/IMG_0960-vi.jpg

http://images31.fotki.com/v1042/photos/8/873576/5678784/IMG_0962-vi.jpg

http://images7.fotki.com/v110/photos/8/873576/5678784/IMG_0966-vi.jpg

http://images31.fotki.com/v1054/photos/8/873576/5678784/IMG_0968-vi.jpg

http://images112.fotki.com/v579/photos/8/873576/5678784/IMG_0970-vi.jpg

http://images32.fotki.com/v1060/photos/8/873576/5678784/IMG_0975-vi.jpg

http://images31.fotki.com/v1057/photos/8/873576/5678784/IMG_0977-vi.jpg

http://images32.fotki.com/v1059/photos/8/873576/5678784/IMG_0978-vi.jpg




The expectations the card met for me:

1) Runs smooth even with AA.
2) The stock cooler is quiet and actually works (especially if you turn up the fan with riva tuner).
3) It overclocks very well.
4) Fast enough to play all the current games.
5) Fences, trees, edges, ventilation shafts, and LOD distance are MUCH better than they were on my 8800 Ultra. I notice hardly any shimmering and the areas I did notice it went away when I turned on AAA with performance and I saw hardly any performance hit (a couple frames).
6) Tabbing out from games is faster.
7) Good 3dmark06 scores and benchmarks.
8) Anything else I can't think of right now.




Basically the best card for the money that you can buy right now. I have NOT TRIED THE 8800GT so I have no opinion on the card, I just know that this one does run great so if it is indeed slower then that doesnt really seem to matter. I can't wait to crossfire the card once I have more money and can actually afford to purchase a new motherboard and everything.

The only disappointment in this card that I have is my Memory overclock isn't as high as other people's. I started to artifact at anything over 1233mhz, but it runs great at 1225. I'm also sad to have lost my 8800 Ultra (even if it doesn't seem that way), but this is a great replacement for cheap.




If I had to say one thing about this card it's that they could have sold it for $300 and that would have STILL been a good deal. Close to $200 is absolutely amazing!

[Rattle]

glad your happy with it buddy.

today is a new day where for $300 and under you can get performance close to what you used to pay $600 for. I hope the trend sticks.

If nvidia/intel dont figure out whats up with this 45nm fiasco I may just go X38 and 3870 x-fire when i grab a new cpu next month.

[squashfx]

what about when the 780i hits? this may change the world

[Rattle]

well that may be an option but since i have to buy a mobo anyway, i may not go nvidia again. I love this setup but I would love it better on a intel mobo lol and thats what this si all about.

[squashfx]

so i guess im in the same situation as you

[Douken]

nice pictures, looks pretty good & nice price

[PhysX]

haha the orb is back in :) i was always a bluorb / corb user :D

[Rav]

How were your temps affected by going to the DuOrb, and how is noise level compared to the stock cooler at higher fan speeds?

[Jacobman]

Very nice, will the duorb work with the 3850's?

[HousERaT]

it should work...... looks like the same pcb.

[Cheffster]

I couldn't agree more, the 3870 is absolutely amazing, especially for the low price.

It OC's beatifully, and with the 7.11 catalyst drivers it outperforms the 8800gt on most tests.

Here's a recent 3dmark06 I got from this thing:

I just got a new OCZ GameXStream 600W PSU to eventually run dual 3870's w/crossfire, can't wait :)

http://i34.photobucket.com/albums/d101/Shamshaw/untitled54.jpg

http://i34.photobucket.com/albums/d101/Shamshaw/currentsetup.jpg

[oakstave]

Thanks Cheffster!

I can't wait to get mine.

[vixro]

How were your temps affected by going to the DuOrb, and how is noise level compared to the stock cooler at higher fan speeds?

I used the stock cooler at 56% fixed fan speed which matches the noise level of the DuOrb. The stock cooler was anywhere from 71-76c on load at 56% and the DuOrb is at 53c on load with the same noise level.


At 100% I had 50c temps on the stock cooler but it was pretty noisy. Honestly the DuOrb will give you lower temperatures, but it's not necessary at this time. 76c is definitely not that hot for this card and would save you some cash.


Now on the other hand, once we get ATI tool with some software voltage or do a hard mod for more volts then the DuOrb will start to show it's value.

[Cheffster]

some other things to note about this card vs the 8800GT:

3870 runs on 55nm process vs 65nm process on the 8800gt
3870 uses DDR4 ram vs DDR3 ram on the 8800gt
3870 supports dx10.1 (not sure if 8800gt does or not but I don't think so and its not that relevant atm anyways)

I think the overall technology on ati's latest card is better then Nvidia's 8800GT, I expect even more perfermance gains with new drivers in the future.

edit: 3870 also uses less power then the 8800gt:
http://i34.photobucket.com/albums/d101/Shamshaw/HD3870_60.jpg

[Super Nade]

Enough of this crap, all of us have HD3870's, so WTF are we waiting for?? It's time to roll out the first OCF collaborative review. The net is full of useless BS reviews and it looks like even the big sites like Anand, xbit have decided to give this card a miss.

Well, let us step up and take charge. This is a good opportunity to show everybody the power of a no BS community generated review. Who's down for it?

I'll do all the spade-work, you guys just keep sending me links to your tests and observations. I'll gather what I can till Sunday and roll something out on Monday for Ed to put up on the front page.

HouseRaT and crew are doing an excellent job keeping this stuff organized. Now I just need some input to get the ball rolling.

Keep posting in this thread...

[vixro]

Ok, I will do some benchmarks with fraps. Any specific "OFFICIAL" benches you want me to run, or is a 30 second fraps of playing the games and multiplayer fine?

[Super Nade]

Yep, that is cool. Basically any game, benchmark you guys think is worthwhile. :thup:

I'm installing Cat 7.11 now and I'll be running COD4 tests.

[vixro]

In my opinion, don't even bother with stock SN. Just go straight for the OC it's pretty much guaranteed 860mhz core out of the box.

[Super Nade]

I'll do that for sure. I just want to record the stock BS to set a bottom line and also for the sake of completeness. No way in hell will I be running stock or with the stock cooler...:D

[vixro]

BENCHMARKS

http://service.futuremark.com/compare?3dm05=3760558


Here is my 3dmark05 complete page, with all advanced tests run with OC'd speeds.


http://service.futuremark.com/compare?3dm06=4074678

3dmark06 complete page with all advanced tests.


---


Lightsmark results:

1280x1024 - 137.7fps
http://images32.fotki.com/v1061/photos/8/873576/5680874/lightsmark1-vi.jpg

1680x1050 - 126.6
http://images32.fotki.com/v1047/photos/8/873576/5680874/lightsmark2-vi.jpg


---


Catalyst settings for upcoming gaming tests:

http://images31.fotki.com/v1042/photos/8/873576/5680874/catsettings1-vi.jpg
http://images31.fotki.com/v1042/photos/8/873576/5680874/catsettings2-vi.jpg


--


RESULTS

-
All test results done in full firefights for 30 seconds fraps demo
-

Team Fortress 2 - full 28 player multiplayer server while fighting - 30 seconds:
All maxed settings
1680x1050
4xMSAA - 16xAF
Very High textures
http://images31.fotki.com/v1042/photos/8/873576/5680874/tf2settings-vi.jpg

Min Max Avg
46 150 74.87
51 128 86.5
54 129 76.37

http://images30.fotki.com/v43/photos/8/873576/5680874/hl22007120718102680-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/hl22007120718102680.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

-

Counter-Strike Source - 32/32 multiplayer server cs_office
All maxed settings except water is reflect world (no water during the area I capped)
1680x1050
4xMSAA -16xAF
HDR off (looks horrible in CS:S)
http://images32.fotki.com/v1044/photos/8/873576/5680874/cssettings-vi.jpg

Min Max Avg
81 198 124.9
99 227 133.57
112 213 161.73

http://images32.fotki.com/v1059/photos/8/873576/5680874/hl22007120719014595-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/hl22007120719014595.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

[vixro]

Half Life 2: Episode 2 - Last strider battle zone
All maxed settings
1680x1050
4xMSAA - 16xAF
http://images30.fotki.com/v43/photos/8/873576/5680874/hl2ep2settings-vi.jpg

Min Max Avg
48 82 61.83
51 80 63.63
50 89 65.67

http://images30.fotki.com/v43/photos/8/873576/5680874/hl22007120718200383-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/hl22007120718200383.html)
http://images30.fotki.com/v43/photos/8/873576/5680874/hl22007120718212435-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/hl22007120718212435.html)
http://images32.fotki.com/v1046/photos/8/873576/5680874/hl22007120718231644-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/hl22007120718231644.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

-

Call of Duty 4: Single player - Random single player missions in firefights only.
1680x1050
2xAA - 16xAF
Mostly High settings, see screenshot for full details.
http://images31.fotki.com/v1055/photos/8/873576/5680874/iw3sp2007120722105094-vi.jpg
http://images32.fotki.com/v1059/photos/8/873576/5680874/iw3sp2007120722110512-vi.jpg

Min Max Avg
33 77 56.97
22 76 55.27
35 89 58.37
47 78 65.53
73 136 112.47
36 73 51.33
47 73 58.53
42 91 58.33

It fluctuates a lot depending on the situation, but as you can see the average is in the upper 50's and very smooth for the type of detail and fighting going on.
The middle test with the highest scores was when there wasn't any fighting but just running through the trees and town.

http://images31.fotki.com/v1053/photos/8/873576/5680874/iw3sp2007120721494788-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/iw3sp2007120721494788.html)
http://images12.fotki.com/v217/photos/8/873576/5680874/iw3sp2007120722072362-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/iw3sp2007120722072362.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

[vixro]

Bioshock - random single player levels
DirectX 10
1680x1050
No AA
High where applicable, see screenshot for full details.
http://images32.fotki.com/v1044/photos/8/873576/5680874/Bioshock2007120722222796-vi.jpg

Min Max Avg
36 98 50.43
36 79 56
46 105 80.13
41 97 60.3
42 96 62.6
38 86 54.2
43 77 60.77

http://images31.fotki.com/v1055/photos/8/873576/5680874/Bioshock2007120722331779-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/bioshock2007120722331779.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

-

Unreal Tournament 3 - Multiplayer 32/32 Vehicle CTF
1680x1050
No AA
High/Max details.
Vivid Post Processing (Intense is max, 1 below that)
http://images32.fotki.com/v1048/photos/8/873576/5680874/UT32007120722534037-vi.jpg
http://images31.fotki.com/v1053/photos/8/873576/5680874/UT32007120722540587-vi.jpg

Min Max Avg
43 63 54.23
41 63 54.13
35 61 49.07
40 63 55.63
42 63 54.3
(I did not remove the 62fps cap).

http://images31.fotki.com/v1054/photos/8/873576/5680874/UT32007120722564278-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/ut32007120722564278.html)
http://images14.fotki.com/v256/photos/8/873576/5680874/UT32007120723003794-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/ut32007120723003794.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

-

Guild Wars - First 3 are in a very full town. Second set is in a zone during battle.
1680x1050
4xAA
Max settings.
http://images32.fotki.com/v1044/photos/8/873576/5680874/gwsettings-vi.jpg

Full Town:
Min Max Avg
59 173 94.2
46 139 69.47
66 130 98.9


Explorable PVP Zone:
Min Max Avg
73 188 92.17
59 97 80.63
80 153 104.97
55 131 80.37

http://images30.fotki.com/v43/photos/8/873576/5680874/Gw2007120723123103-vi.jpg (http://public.fotki.com/vixro/gaming-screenshots/3870-benchmarks/gw2007120723123103.html)
Click picture then click "Get original uploaded photo" below it to see the full 1680x1050 version.

[Super Nade]

IDK WTF happened but now it looks like my Windoze is trashed..:-/

It gets to the Windows screen and then freezes....I'm off to bed now, I'll mess with this tomorrow.

PS# GDDR4 runs a LOT cooler than GDDR3!

[vixro]

Well, screw all dial up users, for those that still exist and those on Aussie internet. I just loaded up all of my games I play and benched them with Fraps.


Each one contains 30 seconds of footage done multiple times in different areas, fights, and zones to make sure that the average showed the game framerate in real time instead of a benchmark that doesn't tell you how it actually plays.


I hope this helps! Scroll up for all the details!


by the way, the Golden Compass movie rocked!@ As good as the book.

[Super Nade]

Kick ass bro. :thup:

Here are a few of my scores:

Stock HSF, stock everything, no BIOS mods, no OC no nothing (baseline)

If 3DMark Hangs follow the procedure given below:

http://www.ocforums.com/showthread.php?t=538525
Validation:http://www.techpowerup.com/gpuz/7a8kb/
3DM 2006 (9285)http://service.futuremark.com/compare?3dm06=4083751

http://img518.imageshack.us/img518/7333/stockfirstrun06ji1.th.jpg (http://img518.imageshack.us/my.php?image=stockfirstrun06ji1.jpg)



3DM 2005 (15139)http://service.futuremark.com/compare?3dm05=3763133

http://img521.imageshack.us/img521/5731/stockfirstrun05md6.th.jpg (http://img521.imageshack.us/my.php?image=stockfirstrun05md6.jpg)
Fraps Log for COD4 (Act II Heat)2007-12-08 11:19:40 - iw3sp
Frames: 969 - Time: 60512ms - Avg: 16.013 - Min: 1 - Max: 20

2007-12-08 11:20:47 - iw3sp
Frames: 738 - Time: 45860ms - Avg: 16.092 - Min: 11 - Max: 30Settings:
http://img259.imageshack.us/img259/107/cod41ef2.th.jpg (http://img259.imageshack.us/my.php?image=cod41ef2.jpg)
http://img145.imageshack.us/img145/5479/cod42hn2.th.jpg (http://img145.imageshack.us/my.php?image=cod42hn2.jpg)
Joint Operations International Conflict ModJoint Ops:

2007-12-08 11:55:39 - Jointops
Frames: 3756 - Time: 22449ms - Avg: 167.313 - Min: 30 - Max: 1410 (<--ignore it is 2D)

2007-12-08 11:56:53 - Jointops
Frames: 3514 - Time: 60000ms - Avg: 58.567 - Min: 55 - Max: 65 (<--VSync cap by the game)
Default Game settings:
http://img249.imageshack.us/img249/7053/jointops200712081154421ym0.th.jpg (http://img249.imageshack.us/my.php?image=jointops200712081154421ym0.jpg)
HL2 Benchmarking:Using this canned version:
http://downloads.guru3d.com/downloadget.php?id=927&file=4&evp=736758669731c72054e9e92e51901b30

Resolution: 1024×768
Demo: HardwareOC - Anticitizen
Detail: Maximum quality
Hardware DirectX level: DirectX 9
High Dynamic Range: None
Heapsize: 64 MB - default
Antialising mode: None
Filtering mode: Bilinear
Fast Z: Disabled
Score = 169.5 FPSSystem
Operating System: Windows XP Professional (5.1, Build 2600) Service Pack 2 (2600.xpsp_sp2_rtm.040803-2158)
Processor: Intel(R) Core(TM)2 Duo CPU E6550 @ 2.33GHz (2 CPUs)
Memory: 3328MB RAM
Card name: ATI Radeon HD 3870
Display Memory: 512.0 MB
Driver Version: 6.14.0010.6734 (English)

The benchmark started at 12/8/2007 12:14:02 PM

Resolution: 1280×768 (HD WideScreen)
Demo: HardwareOC - Anticitizen
Detail: Maximum quality
Hardware DirectX level: DirectX 9
High Dynamic Range: Full (if available)
Heapsize: 64 MB - default
Antialising mode: 8×
Filtering mode: Anisotropic 4×
Fast Z: Enabled
Score = 168.3 FPSIn game FRAPS Log:


http://img221.imageshack.us/img221/1010/hl22007120812172209zc5.th.png (http://img221.imageshack.us/my.php?image=hl22007120812172209zc5.png)
http://img155.imageshack.us/img155/3973/hl22007120812180560zz1.th.png (http://img155.imageshack.us/my.php?image=hl22007120812180560zz1.png)



2007-12-08 12:11:32 - hl2
Frames: 680 - Time: 3344ms - Avg: 203.349 - Min: 146 - Max: 311

2007-12-08 12:23:44 - hl2
Frames: 846 - Time: 14124ms - Avg: 59.898 - Min: 59 - Max: 61

2007-12-08 12:24:30 - hl2
Frames: 3594 - Time: 60000ms - Avg: 59.900 - Min: 59 - Max: 61

[Super Nade]

O.k guys, I would gladly second the title of the thread. I am impressed with the beast of an OC on Stock cooling, no mods.

Overclocked, Stock Cooling, PLL Mod BIOS: [No other Tweaks]
3D Mark 2006 (12937)http://service.futuremark.com/compare?3dm06=4087220

http://img135.imageshack.us/img135/9956/85213012mv9.th.jpg (http://img135.imageshack.us/my.php?image=85213012mv9.jpg)
http://img145.imageshack.us/img145/6067/8521301tc3.th.jpg (http://img145.imageshack.us/my.php?image=8521301tc3.jpg)

862/1316 :drool:

http://img513.imageshack.us/img513/8934/8621316lb8.th.jpg (http://img513.imageshack.us/my.php?image=8621316lb8.jpg)
http://img155.imageshack.us/img155/3508/86213162gp2.th.jpg (http://img155.imageshack.us/my.php?image=86213162gp2.jpg)


885/1386 :drool:
The freaking stock fan died...lol. I junked the stock HSF and put the TR-HRO3. Note that to max out BIOS clocks like below, you have to have fan at atleast 75% duty cycle. I just decided to have it at 100% and it died..lol.

http://img150.imageshack.us/img150/8060/maxochr03id0.th.jpg (http://img150.imageshack.us/my.php?image=maxochr03id0.jpg)

3D..scores at this freq coming up. But wow, the mem and core are clocking like mad! 777-->885 and up on the core (dunno how to go higher on CCC), 1126-->1386 and higher on the mem. :drool:

[vixro]

Use Riva tuner for oc and you can go higher than that maximum if it's stable. Notice how high your scores went after the OC? It's amazing how much more overhead you get with the free OC from the card! That's why I hated every single benchmark I read, because NONE of them clocked the card, it was all stock (which in my opinion is kind of weak).

[Super Nade]

I'll give it a shot for sure! This card rocks the socks of anything I've had so far, especially the way the core is clocking.

13162 3DM06

http://img225.imageshack.us/img225/9249/maxochr033dmui1.th.jpg (http://img225.imageshack.us/my.php?image=maxochr033dmui1.jpg)

[vixro]

Also, you have a lower res on 3dmark, it was confusing me why it was more than mine :p

should be 1280x1024 eh? Your scaling is hella nice though, much better than mine.

[Audioaficionado]

Subscribed :thup:

If you guys get some time in between games and benchmarking, could you test out the video playback quality?

I've been looking forward to the next gen die shrinks. This should eventually end up getting GPUs that use much less power but scale well as you add more GPU cores.

[Super Nade]

Also, you have a lower res on 3dmark, it was confusing me why it was more than mine :p

should be 1280x1024 eh? Your scaling is hella nice though, much better than mine.

Res is 1440 x 900. I dunno why 3DM is doing this. I had the same problem with my X1900XTX, res was always reported as something weird, no matter where I set it. This is why I didn't post all these scores in the 3DM benching thread as hwbot scores cannot be with any such weird resolution. :(

Subscribed :thup:

If you guys get some time in between games and benchmarking, could you test out the video playback quality?

I've been looking forward to the next gen die shrinks. This should eventually end up getting GPUs that use much less power but scale well as you add more GPU cores.

How do I quantitatively check Video playback quality? I could rely on my eyes, but then it would be a really dodgy benchmark.

[vixro]

Haha, all I can say is that with my eyes I notice slight lines on color shade changes if there is fast motion and it's a low res-stretched video. Other than that, it looks pretty much the same as the 8800 ultra, how can we test this for you?

[Super Nade]

Just checked the 8800GT scores. Looks like we are giving the NVidia boys a big run for their money. :beer:

[vixro]

Don't tell VJ about this thread, he will make up some excuse for these cards "sucking".

[Super Nade]

The GT's are good cards, but anybody who says these cards suck have probably looked at stock scores. I'm really happy because it runs cool and I snagged one for cheap. Did you notice how every store worth its salt is jacking up the prices sky high for christmas? Even ncix jacked up prices. :mad:

In any case I'm very tempted to go CF. :D

[vixro]

If I had even close to a stable income I already would have but right now its going towards real life priorities. :p

[Maverick0984]

If I had even close to a stable income I already would have but right now its going towards real life priorities. :p

Should fix your priorities.

Computers > breathing > food > sleep > ... > work

[Audioaficionado]

How do I quantitatively check Video playback quality? I could rely on my eyes, but then it would be a really dodgy benchmark.These guys test video quality on LCD monitors rather well. Perhaps reading some reviews might give you ideas.

http://www.behardware.com/articles/693-1/report-amd-s-radeon-hd-3800.html

[satandole666]

This seems like a fairly appropriate place to ask.

Are there any volt mod/WCing results for these bad boys? I'm thinking of grabbing one of these when the price gouging ends. Maybe two if the price is right The deal will be sweeter if some extra volts and better cooling yield impressive results. It seems like these things are volt limited (not temp) right now. I'm curious to see how it clocks scale with voltage.

Thanks in advance,

[vixro]

I have not seen a WORKING hardware volt mod yet and ATI tool is expected to release some minimal software volting in the future.

[brakezone]

If you are talking about 3870 hardware vmod, they exist though some soldering is required, but one has not been accomplished through an edited vgabios. So far there is confirmed results of 1ghz core on aircooling. Even at 1ghz the voltmodded 8800GT's still seem better; however, it is pretty powerful at 1ghz. On dry ice somebody had one at 1200mhz over at extremesystems I think.

[vixro]

I will be doing all the video tests again (that I can) with XP for some more open results. Turns out my xp 32bit system is a good 10% faster than the Vista 64bit scores.


Already, my 3dmark06 score was 400 points higher, at http://service.futuremark.com/compare?3dm06=4189391

12,136 compared to my 11,761 score before. Everything went up, the cpu, sm2 and sm3 scores.



Over 600 point increase in 05 scores too.

[gvblake22]

How much did you guys pay for your 3870's? I want to make the upgrade, but they all seem way overpriced. I thought they were supposed to be priced at about $219, not $250! Is this just price gouging on Newegg's part or did I get misinformation about the suggested price of these things?

[satandole666]

How much did you guys pay for your 3870's? I want to make the upgrade, but they all seem way overpriced. I thought they were supposed to be priced at about $219, not $250! Is this just price gouging on Newegg's part or did I get misinformation about the suggested price of these things?

Price gouging.

[Coreyhm1]

How much did you guys pay for your 3870's? I want to make the upgrade, but they all seem way overpriced. I thought they were supposed to be priced at about $219, not $250! Is this just price gouging on Newegg's part or did I get misinformation about the suggested price of these things?

i paid $255 after shipping :-/

[gvblake22]

Hmmm, looks like I'll be waiting a month or two for prices to get back down closer to the $200 mark. Stupid supply and demand :bang head

[TTP]

Hmmm, looks like I'll be waiting a month or two for prices to get back down closer to the $200 mark. Stupid supply and demand :bang head

Dell had them for $208.

[gvblake22]

Dell had them for $208.
Hmmm, well it's jumped $80 (http://accessories.us.dell.com/sna/products/Games/productdetail.aspx?c=us&l=en&cs=19&sku=A1350903) then... :bang head

[OnDborder]

In any case I'm very tempted to go CF. :D

That's what I did and ended up ordering an x38 board... x38 have the 2 x16pci-e slots.. The p35's only 1 x16 and the other x4. As I was told, to make the benefit of having cf(benching), it's better to have the 2 x16's..

[Coreyhm1]

It is F*&@#$& awesome!

Im temped to get a X38 / crossfire these too OnDborder! maybe late jan / early feb i can make it happen :attn:

[ajrettke]

heh can't really call supply and demand with video cards price gouging, but they have gone up. I grabbed 2 3850's that had 512 of RAM...gonna OK the **** outa them and see how they fair in Xfire. Hoping for good things, I run 1920x1200 so I need 512 of ram for each card.

Once I get them setup I'll post a nice review. As for benchmarks in games...kind of hard to compare because each person benching say, CS, isn't going to be on the same spot on the map or under the same fire...etc. But I'll see what I can do.

[vixro]

Well, 5 straight hours of load and my card is still running at 47c. Looks like the DuOrb actually works well after settling in.

[lharp]

Just Got A 3870 the other day from GeCube ....
Was really lucky last one on shelf .....
So Here are my 3d06 benchies ...
Was wondering if i could break the 10,000 with my set up ...
Thanks for any info . Plus Would two 3870´s in cf bottleneck my x2 5200@3ghz???

[vixro]

Processor speed is most definitely a bottleneck, but still good score!!!

Are you enjoying it?

[lharp]

Love the card was a super switch from x1950 pro !!!!!
Crappy with bottleneck kinda wanted to go crossfire ......

Thanks 4 the info ...

[RollingThunder]

I'll give it a shot for sure! This card rocks the socks of anything I've had so far, especially the way the core is clocking.

13162 3DM06

http://img225.imageshack.us/img225/9249/maxochr033dmui1.th.jpg (http://img225.imageshack.us/my.php?image=maxochr033dmui1.jpg)

S/N,

You guys need to decide on common test programs and get 'er on. And I agree, these guys here are our best testers with "hands on" everyday use while not getting into an nVidia - ATI flame war.. :thup:

[shadin]

This is a great thread. I'm so sick of ATI vs NVIDIA and AMD vs Intel flaming going back and forth. I want threads with real pictures and real-world game performance with screenshots. Perfect.

Someone should make a collective thread for the 3850 256/512 as well.

[Audioaficionado]

Any decent reviews of the new 8800GT/S vs the 3870 with both cards pushed to the max?

That would be the info I needed to pick one.

@ stock the 8800GT would win but who among us does that LOL? :p

[oakstave]

So my Radeon 3870 XXX Edition from ExtremePC FINALLY showed up. (Two months later.)

Here's my 3DM06 scores, compared to my old setup, an ATI X1950XTX crossfire.

Honestly, I'm disappointed...

With my max overclock on the E6700, (3.5ghz, RAM @ 875)

X1950XTX Crossfire (695/1088): 11239

3870 XXX Edition: (875/1350): 11819

http://service.futuremark.com/orb/resultanalyzer.jsp?projectType=14&XLID=0&UID=13091416

I don't know... I expected a lot more, even though this is 2 cards against 1.

I will say that the general appearance of games is smoother and more crisp, and it is MUCH QUIETER. My 1950's sound like a jet engine taking off when I run benchmarks. The 3870 fan noise is hardly noticable by comparison.

I suppose I'm glad I got a DX10 card, or rather I will be someday... What can I say... In final analysis... Meh... It definitley doesn't suck. I think it made me appreciate my 1950's a lot more.

:-/

[vixro]

You doubled your performance and you're disappoiinted?

[oakstave]

You doubled your performance and you're disappoiinted?

:beer: Happy New Year Vixro!

Yeah, just a little. I really thought this thing would be hittin' more like 13k, not 11k. Why would I buy a new card for the same performance?

[BossBorot]

just bought one for 213.79 after tax, shipping, and the use of a gift card. Hopefully I dont regret it in the morning. Anyways Im going to put the s1 on it with two 120mm fans and also will use the free diamond thermal paste that I got off of these forums like two weeks ago. It should be fun to bench at the least :)

[monohouse]

anyone with 1000/1100 mhz on the core on water with voltmod can comment on the performance of it from a non-3dmark point of view ? what happened ? why no more posts ?

[vixro]

Because this thread is a year and a half old! Why did you bump it?! lol.

My 3870 has sat in my drawer unused since I installed my 4870. :)

[xtkxhom3r]

lmao i didnt think anyone still used 3870's... i still have my old 3870x2 somewhere around there under all my other old computer parts

[monohouse]

"old computer parts" ?
year and a half is the time since I bought my system, for at least the first half year there were no BIOS editors clocking utilities or voltmods, it takes time to take full advantage of your hardware, im surprised it's a year and half old hardware and I still didn't see that 100% of the people here that use(d) it have reached it's full speed.

personally I don't consider buying "new" unless I got the old full speed, but that may be difficult because I still didn't get a chance to put proper ramsinks on the 3870, or a proper water block.

after some looking around, the minimum freamerates (not in 3dmark, which is not a game) in games haven't increased that much from a 3870 to a 4890 (maybe 8 fps at most) and the 800 shaders are (apparently) too far from scaling, that combined with the bandwidth of DDR5 and an improved internal design - I'd say the 48 series doesn't even deserve it's number.

sure - it's so much faster in the maximum fps measures, but why do you care about wasted frames beyond 60 fps when your monitor has a refresh of 60 hz ?

I have a monitor with higher refresh rates, but even that does not give me a good enough reason because even with the higher refresh rates, the minimum fps stays almost the same - and the less said about crossfire scaling the better (aside from the stutter there are some games that actually have a LOWER minimum fps for both nvidia and amd)

[moocow]

"old computer parts" ?
year and a half is the time since I bought my system, for at least the first half year there were no BIOS editors clocking utilities or voltmods, it takes time to take full advantage of your hardware, im surprised it's a year and half old hardware and I still didn't see that 100% of the people here that use(d) it have reached it's full speed.

personally I don't consider buying "new" unless I got the old full speed, but that may be difficult because I still didn't get a chance to put proper ramsinks on the 3870, or a proper water block.

after some looking around, the minimum freamerates (not in 3dmark, which is not a game) in games haven't increased that much from a 3870 to a 4890 (maybe 8 fps at most) and the 800 shaders are (apparently) too far from scaling, that combined with the bandwidth of DDR5 and an improved internal design - I'd say the 48 series doesn't even deserve it's number.

Well, apparently you haven't read results properly if you think thats the case. I've owned a 3870, 4850, and 4890 and the 48xx series is far better than the 3870. Sorry to say it but theres a reason everyone upgraded to the 48xx cards over the 38xx.

[JLK03F150]

I'm still running a pair of HD3870s & they're great from a gaming standpoint.

From a benching standpoint, on the other hand...

[monohouse]

are you sayin this forum cares about nothing but benchmark points ? and people here only buy video cards and computers only for their benchmark point numbers ? o well, in that case forget everything I said and never mind me in the future, I guess I am in the wrong place

and moocow I don't read and never will read 3dmark benchmark scores not here and not anywhere and if you want to know why:

all video card makers are not stupid, and they most certainly know the most famous program that people use to evaluate their products, therefor here is the simple answer:

int main(char prog, int xe, int ye, int ze){
char progname[255];
int i=0, x=0, y=0, z=0;
progname=prog;
if (strcmp("3dmark.exe", progname) == 0)
{
for (int i=0; i<100; i++)
{
x=xe;
y=ye;
z=ze;
specially_optimized_drawing_function_for_benchmark s(x,y,z);
}
} else
for (i=0; i<100;i++)
{
do {
x=xe;
draw_function_normal(x, NULL, NULL);
y=ye;
draw_function_normal(y, NULL, NULL);
z=ze;
draw_function_normal(z, NULL, NULL);
if (z == 56) draw_supplemental_function(x,y)
.
.
.
} while(wait_for_input != 0)
}
return 0;

the point being, in a real game, there is never any way to predict exactly what the video card needs to do, and in a demo or a test program, there is NOTHING to prevent the video card makers from knowing what is the next thing they need to do, and they (in case of, for a example a demo/benchmark program) know just as you do exactly what to do at any given time - when you know your target code there is no limit on how well you can write a specialized code for it.

and I can tell you for sure right now - benchmarking programs that affect the sales of the company's products are MORE IMPORTANT to the company because they directly affect their sales than their performance in games which only affect them indirectly.

[JLK03F150]

You're not in the wrong place at all, monohouse. :) If you look above my avatar you'll see I'm a member of the benching team. Like I said, from a gaming standpoint the HD3870s are still very good cards for me. I don't have a need to upgrade for gaming, they are still making good framerates on all the games I currently play.

I am moving to a HD4890 in hopes of picking up some benching points though.

We all run different hardware for various reasons. If you are a gamer (& not a bencher) and the HD3870 is still giving you great framerates, then great! I know the HD3870 - either single or crossfire - meet my gaming needs. ;)

[vixro]

The 3870 no longer meets my expectations in terms of high quality and high framerates. It was a massive increase from 2x00 series as is the 4x00 series over the 3x00 series. For sub $100 you can double the performance almost, why wouldn't you? That's not that much money in terms of entertainment value. That's the cost of bringing a girl out to dinner and a movie pretty much, only a 4x00 series card would last you 1 year of enjoyment at least (I already have 13 months out of mine).

The same will more than likely be true of the 5x00 series which is only 1-3 months away.

Why struggle with a lower end card when you can be cranking all the juice out of a higher end card? I always thought I wasn't the only one that couldn't ever have enough. While I can understand getting every last ounce out of y our 3870, we've all been there and done that in 2008. We've all moved on to new hardware to do it all over again.

[monohouse]

For sub $100 - under these conditions, yes.
however the price of the 4870 currently is approximately equals to the price of the 3870 when it first came out (350$), I bought my card when it first came out for 1350 ILS 1.5 years ago, right now the 4870 costs 230$ (900 ILS) so apparently a little cheeper since the last time I checked it's price was a month or two ago (which was equal almost to the last $).

but to pay another 230$ for something of 10 fps faster in the best of cases in the minimum framerate in practice in games compared to the 350$ (rates of 1/1/2008) already spent on currently present hardware - well, what can I say ? only for the rich ? buying cards every year easily means the same amount spent as a year ago, and for what ? up to (in some cases it's the same as before) 10 fps higher minimum which is already low enough to make the game run slow for that scene anyway, it won't make the slow scenes fast, it will only make them slightly less slow).

I wouldn't mind spending 350$ again (current rates) if you told me right now that in every currently present game, all the slow scenes cease to exist and all settings can easily be turned on the highest possible.

I guess what I am saying is the practical value of what you are buying - that is my real point, not the "twice the performance gain" you are talking about which is only in synthetic tests where hardware is utilized to maximum as a function of the fact I mentioned above AND as a function of the fact that benchmarks (not company drivers) in their own right are a single program that constantly developed to support specific cards and adds to that list all the time - like very much a driver constantly makes changes to the code to support the various games (if not 90% of the work of the driver programming team, which is 65% or more bigger than it's hardware engineering team).

so yes one point in your favor - if you need the card for the benchmarks - sure you got double the performance, in fact you should probably get more than double - double is just the theoretical bandwidth difference between the cards, and what about the shader differences ? there is 800 shaders in the 4870 and in the 3870 "only" 320, and the architecture change made to it also increase performance - but all this fancy paperwork - gets nothing when it comes to actual usefullness in the game that you play - slow scenes are still slow, fast scenes are MUCH faster (but those frames are already above 60 fps anyway) so who really cares about anything beyond 60 fps if their monitor is only 60hz ? - but yhea sure why not - if you need to draw useless frames and that is your goal - why not.

oh and almost forgot - the features, like the antialias improvements mentioned before, there is *some* value to this, and perhaps some might argue there is value in the combination of abit added performance and some antialias improvements and the architecture more efficient - but if you weight that against 350$ every year on cards for a negligible benefit across the board in actual games - you are quite, quite wrong - and you know I am right about the actual performance in low framerates and you probably also know I can easily pop out 10 or 20 in-game performance comparisons between the two showing the minimum framerates ([H] comes to mind).

so yhe I would agree that if the price of a video card every year was < 100$ for the 10 minimum frames per second improvement, i'd say I am stupid to even suggest to do otherwise - especially for hardcore gamers where the minimum is the most important above all else.

edited: the 3870 still has some advantages also, for example power requirements, it is a single-connector card, which is a feature that probably no longer exists in current video cards, along with a MUCH smaller die - which in turn means less heat, and less power per performance - which in turn means higher overclockability for less voltage which in turn means you can increase voltage higher and because of the less demand on voltage from the core and the higher cooling efficiency because of less load on the water block. all those do amount very easily in terms of overclocking to compete with the frequencies even of the latest and gratest 4890, there are 3870's with core speeds of 1100 on water with voltmod at 1.6v, and as of yet I have not seen either the 4870 or the 4890 reach those frequencies at any configuration

[ekoh]

necrothreading