My 4870 CF results

[AriciU]

These were ran at my daily 24/7 settings. No tweaks. ~1 year old XP / Vista.

E8500 @ ~4500MHz
Team Xtreem PC6400 2Gb @ 473Mhz @ 4-4-4-4 PL7
2 x Sapphire 4870 @ 790/1100 with 8.7 beta driver
DFI LP LT X48 T2R

You can enlarge the pictures.


3DMark01: 86562




3DMark06: 22187




3DMark Vantage Performance: 13416 - I need PhysX driver




Crysis GPU 1680x1050 / High / NoAA / XP 32: 61FPS




Crysis GPU 1680x1050 / Very High / NoAA / Vista64: 29FPS This was run on old cpu @ 4000Mhz... 500Mhz would mean jack in Very high anyway.

[AriciU]

Now for the confusing part. Check this out.

3DMark03 @ 4500Mhz with 8.7 drivers @ 830/1150: 86530




vs


3DMark03 @ 4000Mhz with 8.6 drivers @ 750/900: 90037




I know you will say that the drivers clearly had to do with that mess up but i've just tryed it again with my new CPU using the 8.6 drivers and got only 84000 With these clocks and "working" 8.6 drivers i should've got around 96-97k or so (by checking what people are getting on HWBot).

No idea what's up with 03.

[Everett]

Very nice scores.

[fubarswe]

nice goin there.., cards are very nice and handles well.., keep it up and we will see more from you! :-)

[ahmad]

Thanks for sharing. 03 could be bugged... or another possibility is when drivers are first installed (no changes made) some ATI default setting is just ideal for 03 OR 8.7 introduces a system change (registry or otherwise) that 8.6 cannot override???

Only a fresh install w 8.6 would tell, because you are not the first to notice this.

[AriciU]

Thanks for sharing. 03 could be bugged... or another possibility is when drivers are first installed (no changes made) some ATI default setting is just ideal for 03 OR 8.7 introduces a system change (registry or otherwise) that 8.6 cannot override???

Only a fresh install w 8.6 would tell, because you are not the first to notice this.

Yeah. I'm pretty sure a clean windows install + 8.6's + 03 would bring the old scores back. If someone else with this "problem" has some free time they could try it out

[AriciU]

Re-tested with the 8.8 beta drivers. They improved my scores quite a bit. 3dMark03 is back where it should be and Crysis very high gave me a big surprise.

Click on pictures to enlarge them if you want.


3DMark01 - 87180





3DMark03 - 95925





3DMark05 - 30314





3DMark06 - 22197





3DMark Vantage Performance - 13745 - Nice GPU score





Crysis - DX10 - Vista64 - 1680x1050 - VeryHigh - NoAA - 40.3FPS

[STaRGaZeR]

maybe Crossfire is finally working right with DX10 now they have to make the X2 card look right in the reviews

[bonzo_catalin]

Nice work mate . Maybe we can do something to improved those results

[overcrash]

Dude can you run please using 8.8 beta catalyst on xp 32bit, 4500 Mhz Cpu, 830/1150 ( using Xoc bios ) a 3dmark06 run.

[AriciU]

Nice work mate . Maybe we can do something to improved those results

Hehe. Too hot for serious stuff + i'm at the pool hall ~10hrs a day

Dude can you run please using 8.8 beta catalyst on xp 32bit, 4500 Mhz Cpu, 830/1150 ( using Xoc bios ) a 3dmark06 run.

The runs were made with 8.8 beta catalyst on XP 32... the cpu was 3mhz shy of 500. The cards don't clock that high. 815/950 is the max i can get stable.

[ahmad]

WOW nice crysis numbers. I think i better get those betas...

[HousERaT]

AriciU: when you got your "normal" 03 scores back did you reinstall the OS or just upgrade to the 8.8 betas? I'm getting the same weirdness with 4850s.

[AriciU]

AriciU: when you got your "normal" 03 scores back did you reinstall the OS or just upgrade to the 8.8 betas? I'm getting the same weirdness with 4850s.

Installed 8.8 beta's, rebooted, ran 03. If i run 01 for example and then try to run 03 i get the same 85k ty score for some reason Dunno what's up with that but there's something wierd with the drivers. In 03, i'm talking about the Troll's Lair test. That's where the FPS drop massively for some reason... normal for 95k is around 580-590FPS when the two troll's are sitting at the table, 15-20seconds into the test. When i get 85k i only get 390fps in the same part for some reason.

[HousERaT]

Installed 8.8 beta's, rebooted, ran 03. If i run 01 for example and then try to run 03 i get the same 85k ty score for some reason Dunno what's up with that but there's something wierd with the drivers. In 03, i'm talking about the Troll's Lair test. That's where the FPS drop massively for some reason... normal for 95k is around 580-590FPS when the two troll's are sitting at the table, 15-20seconds into the test. When i get 85k i only get 390fps in the same part for some reason.

Yes, I've experienced the same thing. Didn't connect it with running other tests though. I'll have to see if thats the cause. My scores suffer in Proxycon and Trolls lair to the tune of about 100FPS on each. Everything was cool on 8.6. Since upgrading to 8.7, that's when I noticed it. I'll report back if I find out any more.

[stargate125645]

Nice! I thought the 4870 was better with AA compared to the 3000 series, so did you just not do it for Crysis because it is so Nvidia optimized that ATI scores low regardless? Also, I assume Crossfire for Crysis is still defunct? That was a nice score increase with drivers for Crysis, but I'd like to think that if Crossfire were enabled it would be higher.

[Turrican]

great results bro.
i also got my 2 hd4870 today

[AriciU]

Eh? AA and Crysis doesn't work on very high, regardless of what card is used. Too low fps.

Of course CF works for Crysis. I get 26fps at 1680 very high, single card vs 40fps with crossfire enabled. Almost 50% increase with 2 cards so i say that Crossfire is definately working like it should.

Crossfire, 1680x1050, very high




Single card, 1680x1050, very high

great results bro.
i also got my 2 hd4870 today

Thanks and good luck with it!

[Mumid]

These were ran at my daily 24/7 settings. No tweaks. ~1 year old XP / Vista.

E8500 @ ~4500MHz
Team Xtreem PC6400 2Gb @ 473Mhz @ 4-4-4-4 PL7
2 x Sapphire 4870 @ 790/1100 with 8.7 beta driver
DFI LP LT X48 T2R

You can enlarge the pictures.

So much for X48 having much better crossfire performance than P45! Here are some similar benchmarks i ran with P45. Doesnt look like there are any bottlenecks here (allthough the extra bandwidth offered by X48 might come in handy for 4870x2 Xfire!). Here are the tests - setup details in screenie and in my sig:

[makaka]

Nice score men , i am waiting for my HD 4870 X2 i hope to see similar result .

[stargate125645]

Of course CF works for Crysis. I get 26fps at 1680 very high, single card vs 40fps with crossfire enabled. Almost 50% increase with 2 cards so i say that Crossfire is definately working like it should.




Thanks and good luck with it!

I asked because countless reviews suggest Crossfire does not work well with Crysis. Here is just the most recent:
http://www.extremetech.com/article2/...2327868,00.asp

So you are saying that dual GPU cards have issues with Crysis, but dual GPUs (two cards) do not? I imagine that the GX2s handle Crysis just fine, and it's another case of TWIMTBP at work against ATI regarding CrossfireX.

[Delex]

Hi! How much benefits with ASUS Bios instead Sapphire?

Tnx

Alex

[stargate125645]

Hi! How much benefits with ASUS Bios instead Sapphire?

Tnx

Alex

Negligible, if any.

[Mumid]

Hi! How much benefits with ASUS Bios instead Sapphire?

Tnx

Alex

None performance wise mate, you just get a better fan profile, better stock speeds (if your card can handle them) and the ability to overclock higher in CCC. I like the new bios tbh.

[Venom-Crusher]

Yes, I've experienced the same thing. Didn't connect it with running other tests though. I'll have to see if thats the cause. My scores suffer in Proxycon and Trolls lair to the tune of about 100FPS on each. Everything was cool on 8.6. Since upgrading to 8.7, that's when I noticed it. I'll report back if I find out any more.

I meet a problem like you in 3DMark03 only, it happens with all driver every at I try, sometimes the points of GT2&GT3 go up, sometimes the points down in same speed & config, I puzzling is a lot of,