Arcmanov's rig evolves...

[Arcmanov]

Well, the merchant (Computer Brain) is taking their cool time to ship it out, so next week for that.

Anyway...on to the benchies:

3DMark06                              3DMark Vantage - Performance             3DMark Vantage - High
                       


Based on my research, the results are kind of what I expected for these apps.  It seems that 3DMark06 does not like
this configuration in Vista so much.  When I OC'ed the cards to 785MHz core/1100MHz memory, the score actually
went down.  What is significant however is that the result is still better than one OC'ed HD4870 (which is to be expected).

The Vantage scores tell a completely different tale though.  Performance seems to scale very well with Crossfire compared
to these scores.  However the results are still significantly constrained by the 16x/4x configuration.  In addition to the 8x/8x
setup on the Asus mobo, the other advantage is that the PCI-E slots are gen 2.0, which are equal to 16x gen 1.1.

I'm predicting a significant improvement with that Asus P5Q-E.

Gaming benchmarks to follow.

[Spazosaurus]

It has just occurred to me...I know where you work so i'll follow you home heh heh heh.

I'm particularly interested to see what kind of gains you see in Crysis. But personally, if i'm to go with a multi gpu setup, it would be in the form of the 4870x2.

[MessiaaH]

Nice upgrade arc, those 4870s kick ass, im supprised how well my 8800GTX SLI holds up in comparison.

The 4870X2 is the shiz doh. a single X2 performacne on par/better than CF 4870. So im patiently waiting on the final release.
But with beta drivers and a preview card, the X2 kicks major cahoonas. so the final release can only be better.

so i think CF 4870X2 mite be the next upgrade for me, but then ill have to get a 1KW PSU errrr. PC Power 1.2KW in mind.
but den id have to change my board as well, grumble.

[Arcmanov]

Benchmark at max CPU OC:



Just to let certain other n1gga$ know nah...

Whats that?  Believe in the what.....?

The king is dead, long live the king.

Indeed.

[Spazosaurus]

Impressive score there Arc. Imagine what will happen when you remove the bandwidth bottleneck!

[Arcmanov]

Ent, and what about the OC?  Remember, those biatches are still at stock there.
I'm still figuring out how to OC em both effectively.

[MessiaaH]

Interesting that my 2 year old GTXs is only being beat by 300 points. This asures me even more thant upgrading to CF 4870s make no sense for me. CF X2 just 1 more month.

Hey arcman, is that your idle temparature in your systray? if so, the next bar-b-q is at your house, , if i OC my cpu to those temps, we'll have a new king but nobody going to be playing games with that kinda heat coming out they rig. ill think about if i wana go home and OC my cpu some more, but im leaning towards X2 ownage in a month or 2

Nice upgrade from what u previously had though, we'll see some more compition pretty soon, cuase alot of guys are upgrading, and the price of these cheap ass performaning cards, the compition mite boil down to who can achieve better CPU/GPU overclocks, which wil basicaly boil down to, who can volt mod / cool there sh&t properlly

[Arcmanov]

Unfortunately, Everest was playing up when I took that screenshot.  I noticed that idle temp and was like 'WTF!!!'
One reboot and 1 hour of Prime 95 later, I realised that the idle temp at 3.3 GHz is approx 43C and the load
temp is 54C.

...and yes, you are quite correct when you say 'CF4870 make no sense for me', as the performance difference
between 2 x 8800 GTX in SLI and 2 x HD4870 in CF is bound to be a small one, but consider the fact that the above score
was achieved with a 16x/4x 'limitation' and also, consider that the cards' frequencies remained at stock.
Also consider that the CPU is chugging along quite nicely at 3.3 GHz.  After you consider the above facts,
compare my rig in its current form to its former position and reconsider your statements.

I think 2 x HD4870 is quite a worthy upgrade from a single 8800 GT, especially because they are 'cheap-ass'.
I certainly cannot afford 2 x HD4870X2 AND the 1 or 1.2KW PSU needed to power those beasts, so I actually
welcome the challenge of seeing what mammoth performance increases can be gained by employing such a setup.

'Bring it on' I say.

[MessiaaH]

With proper cooling of the CF 4870s, proper OCing, id recond your score will increase about 500 points, a little over 1000 with liquid and volt mods. Wont expect too much doh. and i dont think that 16/4 limitation will affect the benches much either. 100 points give or take.

and keep in mind, as i had previously posted, i planned on going 3way sli, so a 1kw psu was always in the works, and the corsair 1kw is pretty afordable. so i was going to get a 1kw + 3rd 8800gtx anyway. but i decided to hold off for 3 gtx280s instead, but no i looks like its going to be 2 X2s.

Gotta sell my current 2 8800gtx, then foot the rest for QuadOwnage

and yea ill bring it on, dont worry. im patiently awaitin the release, price drop, and driver improvements of the X2 before i jump ship.

i realy dont want to go ddr3 since its expensive as hell, and i already runing 8gigs, so ill see what kind of board upgrade ill do when the X2 is finaly out.

[Arcmanov]

Now, if I had a lil extra cash, the Asus P5E Deluxe would be my mobo of choice, but I got the P5Q for about
185 US, so that will have to suffice.


EDIT!!!  Mih P5Q-E reach.  GG Ezone 

Super Upgrade Weekend starts tonite.

[W1nTry]

I have one word FCUK!

[Arcmanov]

Right, so here we have the Asus P5Q-E.  This mobo is wonderful.  Its even easier to OC
than the Abit IP35-Pro.  The BIOS is rather extensive, some stuff there I shall leave
well enough alone until I find out what all the settings do.

Board layout is excellent.  I was still able to use my two PCI-USB cards in addition to the CrossFire setup.
The SATA port layout is very well done, and not blocked by any of the graphics cards.  All the relevant connectors
are along the board edges, so I was able to route all cables around.

Enough about that, you wouldn't be reading this unless you wanted to know about how it performs, right?

On to the results:

3DMark06



3DMark Vantage - Performance



3DMark Vantage - High




What is interesting to note from these results is that it seems that 3DMark06 was unaffected by the 16x/4x configuration, because
the resulting score here is less than 100 points higher than previous, at the same CPU speed.  It is safe to assume that 3DMark06
does not provide sufficient 'throughput' to be bottlenecked by a PCI-E 1.1 setup.

Take a look at the Vantage numbers however.  I predicted a significant jump with the 8x/8x PCI-E 2.0 config, and the numbers there
speak for themselves.  The GPU score has jumped more than 2000 points alone.  WOW!!!  So it seems that a significant bottleneck
has been removed here, and the cards can now fling that PCI-E data with gay abandon.   Also, I should point out that the horribly
skewed graphics I noticed during the Jane Nash benchmark with the Abit mobo are gone now with this Asus mobo.  Another side effect
of the bottlenecking?  Perhaps.  Its certainly in line with what I've seen on another forum.

I haven't run any actual game benchmarks yet, but of course I had to take a run through Crysis.  The horrible stuttering during gameplay
I got with the Abit mobo has now disappeared.  Silky smooth gameplay abounds , although I get the feeling that Nvidia (the bastards)
intentionally did not want Crytek to optimise the game for ATI cards and multi-GPU setups.  After all thats 'the way its meant to be played' right?

I have all day, so I shall do some proper gaming benchies today...

[phoenix31tt]

good to know that Asus still has it... overclocking for me was a breeze with my p5b...

looks like my next upgrade is goin to be another asus yay!!...

keep the reviews coming... i want to know if there are any little bugs with the board (more interested with the board than anything else)

[Arcmanov]

[c] The Joker: ...'and here...we...go'.


Call Of Duty 4: Modern Warfare

Test was done using FRAPS v2.8.2.  Resloution was set to 1680 x 1050.
Anti-aliasing was set to 4x, and Anistrophic Filtering set to max (16x I think)
All other settings set to max.



Frames: 153629
Time: 1291328ms
Min: 19
Max: 390
Avg: 118.970

Needless to say, this game positively screams on CrossFire.  It is safe to say that this game
will be CPU-limited way before it becomes GPU-limited here.



Crysis

For some strange reason, I cannot get FRAPS to run with Crysis in DX10 mode, so I relied on the
built in GPU benchmark (GPU.bat) found in the game installation's 'Bin32' folder.  Resolution set to 1680 x 1050.
Anti-aliasing was set to 'Off' and all settings set to 'High'.


Hmmm...a wonderful 35 fps average.  Just goes to show how much work still needs to be done on this engine,
or probably drivers.  Maybe we will see some performance improvement with future driver releases.  Won't hold
my breath waiting though.








Racedriver GRID

My oh my, this game is a beauty.  Codemasters have outdone themselves on this one.
Test was done with FRAPS.  Results were recorded with a 1 lap race.
Resolution was set to 1680 x 1050.  AA set to 8x.  AF set to 16x.
All other settings set to max values, 'High' or 'Ultra' where applicable.

         

Frames: 14683
Time: 141996ms
Min: 48
Max: 133
Avg: 103

Another game that Crossfire seemingly makes mincemeat out of.  Its unfortunate that I didn't do these kinds
of tests on my former rig (with the 8800 GT) but I think this shows adequately what CrossFire can do when a
game engine is optimised correctly.  No GPU limitation here.




Half-Life 2: Episode 2

The Source engine at its best.  Nuff said.
Test was done using FRAPS.  Results were recorded during a run-through of the 1st level.
Resolution set at 1680 x 1050. AA set to 8x.  AF set to 16x.
All other settings set to their maximum value.
Don't know what went wrong here, only managed to garner a single screenshot.


Frames: 88916
Time: 10269027ms
Min: 0 (due to a save point)
Max: 173
Avg: 87

i'll have to re-run this one and record the results without that save point, as the average has been
skewed by the non-rendering of any frames during the save itself.  I can safely say though, there
was absolutely no drop in performance throughout this test.






This is only a small sampling of games, but apart from Crysis, this is a very good indication of the kind
of performance possible with a proper CrossFire setup.  These tests were not really meant to compare
single-card vs dual card performance, but as a raw max settings test at a 22-inch monitor's native resolution.

The ATI Radeon HD4870 makes a compelling case for multi-GPU setups, and I recommend them for CrossFire
on a 'budget' as a proper CrossFire certified mobo and PSU (800W or more) are also required to run it properly.
The Asus P5Q-E has made installing and running these cards an absolute dream.



Arcman approved!!!





"Sir, we are now in orbit above EPIC WIN"   

[Prowl]

hummm, I have 4 16x slots, wanna lend me those cards for a weekend :XD

[Philosophical45]

great reviews arc.. those 4870's are the hottest things out right now.. can't wait to see ah final 4870x2 in action.. who knows.. i might get to lick one soon..lolol

[Arcmanov]

Pics of the rig in its present form.  I think SPK will appreciate what I've done here... 

 


Cold Cathode Flourescent Lighting = hawtness (if you mount them right).

[Spazosaurus]

Wow. Pure brilliance. Well done man. If only I had that kind of patience.

[Arcmanov]

A lil update to the score dey.



CPU at 3.3 GHz
GPUs at 790 MHz core/1100 MHz memory.

Yep, thats 18K+.  if I can only figure out how to OC these bastards and keep the heat down,
I might well pass 20K.

[Spazosaurus]

Very nice, Arc. But you should be running Vista x64.