My Spider

[merkk]

Hi Guys,
I've been hanging out in the forum for a few weeks so i thought i'd make a post and say hello. I've just finished off my all AMD SPIDER rig. It's a basic 3 video card rig. I posted some
links to some pics for you. So far my best 3Dmark-06 score has been 15129 using
8.3 drivers. I am very happy with that score. I'm using a 9850BE cpu and 3850 video
cards. Take care all.

[Mussels]

not a bad system and it looks fairly tidy. try turning the sata HDD's around and have the cables behind them, it'll help with that.

that 3dmark score seems low somehow for a quad core, tri-fire rig - i get over 14,000 with a single card.

[HTC]

not a bad system and it looks fairly tidy. try turning the sata HDD's around and have the cables behind them, it'll help with that.

that 3dmark score seems low somehow for a quad core, tri-fire rig - i get over 14,000 with a single card.

Trifire with 3850 cards.

[Mussels]

yeah i thought 3850's were faster, i saw threads with people modding them to 3870's and so on.

I was simply surprised that three of them, barely matched my single card.

merkk: try the new cat 8.6 drivers, they may help.

[Bytor]

Nice setup... Love the 3 cards ....

Adding my 3rd soon.

[HTC]

yeah i thought 3850's were faster, i saw threads with people modding them to 3870's and so on.

I was simply surprised that three of them, barely matched my single card.

merkk: try the new cat 8.6 drivers, they may help.

Dunno about that but, looking @ freaksavior's topic:

http://img.techpowerup.org/080618/2008-06-19_040703.png

You can see how well this trifire is scoring!

[Mussels]

I guess his system is just at stock. Time to OC :D

[OzzmanFloyd120]

Your spider is missing a leg, but still it's a very nice build.

[merkk]

Hi guys
my videos cards are at all stock clock ATM just the cpu was over clock for the 15129 score .

I wish i could use 8.5 drivers but ATI add that fix for power play stating at 8.3 hot fixs and
all the newer driver make my rig black screen crash running any 3d app just one to two mins
into it .

Yes i need to do some about all the dam wire off the power supply i see if i can run the be hind
the hard drives

Thank You All !!for the replies

[Mussels]

8.6 was just released. It may work differently to 8.5

[merkk]

Mussels
Hi dude. Here's something to think about, i'm not trying to be rude but my crossfire X rig plays games in crossfire so you need to look at my 24/7 clock of just 2.7 mhz on the cpu i can knock down 91 FPS in canyon flight test in 3dmark-06 and that what my rig uses to play games with. Your 14890 is a great single card score but your only in the 50s fps range in canyon flight. My
canyon flight score on the 15129 was 94.9 fps.

[batmang]

Update to 8.6 merkk, has a very nice improvement performance wise.. plus, BETTER tri-fire support. Your scores should definitely rise.

[Mussels]

Mussels
Hi dude. Here's something to think about, i'm not trying to be rude but my crossfire X rig plays games in crossfire so you need to look at my 24/7 clock of just 2.7 mhz on the cpu i can knock down 91 FPS in canyon flight test in 3dmark-06 and that what my rig uses to play games with. Your 14890 is a great single card score but your only in the 50s fps range in canyon flight. My
canyon flight score on the 15129 was 94.9 fps.

you dont mention the 2.7GHz clock in your specs, so i wasnt aware of that.
Your point about crossfireX doesnt make much sense... games playing in crossfire doesnt seem to mean anything? I'm saying your score feels low for a tri-fire solution, and you're telling me... what?

You do realise the 3dmark score is based on FPS, so theres no way your canyon score would be nearly double mine, unles your SCORE was also double mine. also have no idea how you can even guess what my FPS would be.

heres a photo i just took of a 1280x1024 (its stretched as its not a native res for my TV) doing canyon flight in 06
http://img.techpowerup.org/080619/DSCF3777.jpg

[merkk]

Batmang
Hi nice SRT-4 you got there. It's lowered the right way getting the ass end of the car down. I got a 2003 EVO 8, i almost put money down on one of those twice but every time i stood back and looked at the car i saw how high in the air the butt sat and i couldn't bring my self to putting my money on the line.

[merkk]

Mussels
All you got to do is go to ORB
look up Phenom cpu
3850 video card
3dmark-06
you will see Merkk1 3x3850 15129 i'am number 1 in my class by 3dmark and you see for your self
that my canyon flight score was 94.9 fps it is what it is

[Mussels]

i'm not arguing the score, or anything about canyon flight.

Congratulations on being #1 with your hardware config, btw.

All i was saying, is that your overall score seems low. I beleive it should have been higher, and that possibly something is holding it back - possibly drivers as has been mentioned a few times.

[batmang]

Batmang
Hi nice SRT-4 you got there. It's lowered the right way getting the ass end of the car down. I got a 2003 EVO 8, i almost put money down on one of those twice but every time i stood back and looked at the car i saw how high in the air the butt sat and i couldn't bring my self to putting my money on the line.

Thanks! Coilovers make the SRT-4 a whole lot better. :rockout:

[merkk]

Mussels
Yupper you are right i'am the one holding my rig back with low cpu
clock speed and not overclocking my video cards . It get good fps rates in
crossfire with a low cpu clocks speeds so i didn't see the point of realy trying to push the thing. I have to water cool the cpu to get it up over the 3.0 range and put alot volts into cpu .That a lot harder on your gear i'am a old fart i'am way past the mine is biger than your thing.That why i when with AMD good FPS speed with low cpu clocks. I do think it could make the 16000 point range if i were to put its balls to the wall. Sorry
if i came across like a dick to you.

[Mussels]

you apologised, so it doesnt worry me. i just didnt want my point to get lost, that i felt something was off. have you tried the 8.6 drivers yet? people are saying crossfireX and trifire have got good boosts from them.

[merkk]

Mussels
I'am going to try the 8.6 driver when i get home form work to night
see what happens .

[Mussels]

awesome.

What would be a good test, is to try one, two and then three cards - maybe the 3rd one isnt actually kicking in?
I remember people saying you had to install the cards one by one to get them working on some PC's (as in, putting all three cards in, then the drivers was a bad idea. you had to reboot between cards)

[batmang]

That plane avatar is funny hehe. Merkk, I'm interested in seeing the results with 8.6. I notice a nice increase in CoD4, 3DMark Vantage and 06. Definitely worth the updat: http://forums.techpowerup.com/showpost.php?p=844752&postcount=49

[merkk]

Hi
At all stock clocks speed cpu (2.5) and video cards using 8.3 drivers the third video card is a 15 fps pick up in canyon flight test score .It was 655
point pick on my over all 3dmark-06 score . The point pick up gose up higher with the clock speed of the cpu. The PCIe slot set-up on DFI
mother board run all 3 slots at 2.0 PCIe 16x16x4 .

[HTC]

Mussels
I'am going to try the 8.6 driver when i get home form work to night
see what happens .

Apparently, 3D06 won't get a boost but games and vantage will. Not 100% sure, though!

[merkk]

Hi guys
Well that was a waste of 5 hours try to install 8.6 drivers in my pc. It went into crashing reboot cycle. It would not reboot after they were installed. Did a clean install of vista-64 same thing all over again. So i did a clean install of vista again used the 8.3 drivers and the're going in trouble free. This is why when poeple ask me if they should do a ATI pc i tell them to stay with a nvidia, they are just a way more trouble free set-up right now. You build an all AMD rig it's like a trip to LAS VEGAS you are taking a roll of the dice at the crap tables. You hit all 7 and the rig runs dam good, you get snake eyes and you got a pile of junk.

[erocker]

Hi guys
Well that was a waste of 5 hours try to install 8.6 drivers in my pc. It went into crashing reboot cycle. It would not reboot after they were installed. Did a clean install of vista-64 same thing all over again. So i did a clean install of vista again used the 8.3 drivers and the're going in trouble free. This is why when poeple ask me if they should do a ATI pc i tell them to stay with a nvidia, they are just a way more trouble free set-up right now. You build an all AMD rig it's like a trip to LAS VEGAS you are taking a roll of the dice at the crap tables. You hit all 7 and the rig runs dam good, you get snake eyes and you got a pile of junk.

That is a strange problem! You should deffinitely submit a ticket to Catalyst Feedback. I'm using a 790fx board and crossfired 3850's with Vista and had no problems..

[Mussels]

that really does sound like a pain in the ass, i have no idea how to fix that other than email ATI repeatedly. Trifire wouldnt be very common, so its possible they just dont know about the problem.

[jammy86]

Is this your dads computer? :p

JAmes.

[merkk]

Jammy86
Ha ha ha ha dude my dad 87 years old he don't even know how to trun one on ehehehe .

[Gam'ster]

Hey merkk thats a cool rig dont give up on it :toast: AMD/ATI :rockout:.

lol @ mussels avatar nom nom nom :roll:

[merkk]

Hi Guys
Here some info. for thoughts that want it.

AMD 9850BE running at stock clock speed (2.5) video cards at stock clock speeds (3x3850)

single video card 3Dmark-06 9,835 ,sm2 3867, sm3 4213 ,cpu 3425
canyon flight score 38.27 fps

two video cards crossfire 3Dmark-06 12,626 ,sm2 4877, sm3 6172, cpu 3398
canyon flight score 71.57 fps

three video cards crossfireX 3Dmark-06 13,280 ,sm2 4897, sm3 6912, cpu 3386
canyon flight score 86.52 fps

CPU over clock to 2.895 three video cards crossfireX 3Dmark-06 15,129 ,sm2 5653,sm3 7751
cpu 3903 canyon flight score 94.9 fps

[p_o_s_pc]

I guess his system is just at stock. Time to OC :D

Could be that or he could be running the cards with 1/2 the SP's of the 3870

are the cards 256 or 512mb?

[merkk]

Hi
The 3 HIS video cards are 256mb the base 3850 at stocks clock speed

[p_o_s_pc]

I know why your getting lower scores then. The cards that 3850 cards that are getting higher then you have 512mb of ram and 320 Stream Processing Units and the 3850s that you have i believe have 120 Stream Processing Units.

[Mussels]

I know why your getting lower scores then. The cards that 3850 cards that are getting higher then you have 512mb of ram and 320 Stream Processing Units and the 3850s that you have i believe have 120 Stream Processing Units.

really? that would explain a lot...

[From_Nowhere]

Nice Spider.

Oh and:

I do believe all Radeon 3800 Series have 320 Stream processing units. The ones with 256MB of RAM don't perform as well because they have 256MB of RAM.

Unless you're counting the 3830, I have no idea what the specs are on that card.

[merkk]

Hi
The 256mb do have the same stream processing units as the 512mb cards. What i find funny
is that pic. of canyon flight showing 71 fps . At that same point on my rig i'am getting over
150+ fps with my cpu clock at 2.7 with my cpu clock at 2.895 it's over 160+ fps so if poeple
think 150+ fps is slow at the same point in the test so be it . But the last time i look
150>71 but why try to piss poeple off.

[Mussels]

Hi
The 256mb do have the same stream processing units as the 512mb cards. What i find funny
is that pic. of canyon flight showing 71 fps . At that same point on my rig i'am getting over
150+ fps with my cpu clock at 2.7 with my cpu clock at 2.895 it's over 160+ fps so if poeple
think 150+ fps is slow at the same point in the test so be it . But the last time i look
150>71 but why try to piss poeple off.

no its merely that you made comments about what FPS i would get, when the FPS is a very irrevant way to look at it. comparing SM 2.0, 3.0 and CPU scores is an easier and more reliable method for that.

[p_o_s_pc]

merkk could you please give us a SS of GPU-Z or better so we don't spam the thread PM it to me please. Thanks

[merkk]

Mussels
This about how SLI and Crossfire work when playing a game and canyon flight test is a test of
sli and crossfire if your rig will do it .Your rig is running as a single card in the test your fps rates
are going to be slow end of story ,with the card you have now .The same thing will happen in
games that support SLI/CROSSFIRE my rig more than double it fps rate in games that support
crossfire as to my single card score in the same game .Your rig is stuck at the same fps rate because it a single card. If you were to go to a SLI set-up you see in games that support SLI your fps double to your sinlge card rate . It's why poeple run SLI and Crossfire in the first place. Now if the game dosn't have SLI/CROSSFIRE support and there a lot of older games that don't your single 8800gt card will kick my ass becase it a better single card than a 3850 is. But in any games that support SLI or CROSSFIRE the SLI / CROSSFIRE will be able to run faster fps rate than the single card can. CPU type, clock speeds and types of video cards all play into how
well your SLI / CROSSFIRE rig will run in that app. And that what the canyon flight test ,but
you have to have a rig that can run SLI/CROSSFIRE in the frist place to see it.

[p_o_s_pc]

Merkk don't getting too cocky you don't even have your facts straight. Mussels rig isn't going to get "slow" compared to yours on the canyon flight test. You won't get 2x the FPS by adding another card (1 8800GT to 2 8800GT) SLI/Crossfire won't give you 2x the power because it doesn't scale that high.Yes 2 8800GT would run faster then 1 8800GT if it is supported by the app BUT it doesn't that your rig would be better then his in everything just because of the video cards. some apps/games do better on Nvidia. So in somethings Mussels would kill your Spider not only because the game/app runs better on Nvidia but his CPU would kick your CPUs ass any day. BTW if any of you think i am being a fanboy look in my specs before you say anything!

[merkk]

P 0 S PC

Hi dude i don't think your a fan boy and i'am not madd at Mussels were just takling about what
we feel works and what don't. You have some good points and 8800gt cards don't sacle
as good in SLI as my 3850 do in crossfire . It take big cpu power to make that sec. 8800gt
card kick in but Mussels cpu is clock at 3.6 so i think he could do a dam good job running
sec. card . And again your right some games Nvidia works better and other ATI but for the
most part a game that support SLI and Crossfire the SLI / Crossfire rig will run at a highter fps rate in that game than a single card will . A big part of Mussels 3Dmark-06 score is his cpu
score and not how many fps he getting in the video tests. On my rig a big party of my score comes from my canyon flight score but that what counts when you are running a game in SLI /Crossfire that why i went with 2 video cards than add a thrid video card to my rig . I'am using what worksbest for my rig to get better game play with just enought cpu power to drive all three video cards . Mussels is using cpu power to get the most he can get out of single card .

[Gam'ster]

pospc is right a phenom @ 2.8 isnt really matched with a q6600 @ 3.6, At stock speeds its close q6600 just in front but at 3.6 q has the advantage, also a single 3850 is no match for an 8800gt and i think it can hold its own against a CF 3850 too, But tri-fire Ive no clue how well 3 x 3850 would scale. And a lot more games are optimized for nvidia cards and that helps quite a bit.

[p_o_s_pc]

Sorry if i came across kinda rude.... Just trying to point out a few things. and i see that Gam'ster agrees with me on some of it too. Merkk could you please PM me a screen shot of GPU-Z?

[Mussels]

merkk is trying to say that any crossfire/SLI is faster than any single card.

Its not true. he's confusing the fact that he has three lower end cards in crossfire - THREE, to match my single. not to mention he only has 256MB of ram (crossfire and SLI arent additive for system ram) so in fact, for actual gaming my FPS would be a lot higher.

[merkk]

Mussels
Hi dude go post your devil may cry benchmark in the devil may cry benchmark thread mine
are there .You find in the graphics card thread . It all in fun lest see what you got big boy.

[Mussels]

its downloading now. lets try a few resolutions more than just 1280x1024 however... i can go upto 1080i on here.

edit: and as said in the thread, you didnt even use the same test as everyone else. you need to use the settins specified in the first post, and currently the latest post in the thread by eastcoast. default settings are auto detected, so they're irrelevant between systems.

[merkk]

Hi
Sorry my 19" crt only gose up to a 1280x1024 just keep it simple run defaults
setting here mines. Can post yours in this thread to please. I'am trying to get some run with
4aa trun on but get black screen on the screen shot . O by the way 4AA>3850 it is what is.
my rig drops from a big S on my chest to a A but what the hell it all in fun .

[Mussels]

Hi
Sorry my 19" crt only gose up to a 1280x1024 just keep it simple run defaults
setting here mines. Can post yours in this thread to please. I'am trying to get some run with
4aa trun on but get black screen on the screen shot . O by the way 4AA>3850 it is what is.
my rig drops from a big S on my chest to a A but what the hell it all in fun .

defaults arent going to happen. i'm using max settings as requested in the thread. we cant compare defaults, as your defaults are going to be lower than mine.

[p_o_s_pc]

here is mine LMAO i need to get a video card badly
http://img295.imageshack.us/img295/4186/devilmaycry4benchmarkcq4.th.jpg (http://img295.imageshack.us/my.php?image=devilmaycry4benchmarkcq4.jpg)

[merkk]

Ok so as to keep apples to apples as best we can run these setting please.

1280x1024
msaa off
full screen on
vsync off
frame rate variable
shadow quality super high
quality super hight
dx9

this is my run at these setting post in this thread please.

[Mussels]

Ok so as to keep apples to apples as best we can run these setting please.

1280x1024
msaa off
full screen on
vsync off
frame rate variable
shadow quality super high
quality super hight
dx9

this is my run at these setting post in this thread please.

i'll do so after work. the bench is refusing to run above 1360x768 for me, so i'll have to swap screens (which takes a bit of effort, and i start work in 30 min)

[merkk]

Hi here a benchmark at the rez. you said you can run. Setting are.

1280x720
msaa off
vsync off
frame rate variable
texture resolution super high
shadow quality super high
quality super hight

[Mussels]

i ran with 4xaa as listed in the original thread. otherwise settings are teh same as yours.

http://img.techpowerup.org/080622/DMC4%20test%20720p.jpg

[merkk]

Mussel
Hi dude here my directX 9.oc and directX 10 runs same setting as you with 4aa 1280x720
sorry no screeen shot as post in other thread some of us are getting black screen on the screen
shot with the aa trun on .No data show up. All the runs were made with my cpu at stock clock
speed (2.5) and video cards stock speed to.

directX 9.0c 1280x720 4aa
1. 76.07
2. 53.41
3. 98.34
4. 60.05

directX 10 1280x720 4aa
1. 187.44
2. 136.67
3. 218.22
4. 102.59

Grats dude you beat me when the aa are trun on in directX 9.0c AA>3850
would like to see how DirectX 10 on Nvidia card are working now days can
you do a run and post in the benchmark thread. As you can see it night to
day diff. in aa run in DX9.0c and DX10 for me.

[Mussels]

that is a surprise that your DX10 results are FASTER, and not slower.

I'll go do a run in DX10 now.

http://img.techpowerup.org/080623/DX10 test.jpg

same settings as before. i wonder if this is what people were talking about in the thread on this, that AA wasnt working properly on all systems in DX10? Its either that, or your system is more optimal in DX10 mode.