Anand Review: ATI Radeon HD 3870 X2

ATI Radeon HD 3870 X2

Quote:

---

snip

The Radeon HD 3870 X2 manages to outperform every other single card solution in our tests, being bested only by the 8800 GT SLI which is a good $150 more expensive than the ATI solution. It's worth noting that it looks like AMD had finally solved its DirectX 10 performance issues under Bioshock; it's not just that the 3870 X2 performs well, but AMD's driver team seems to have done some good work over the last few driver revisions.

snip

---

http://anandtech.com/video/showdoc.aspx?i=3209&p=3

Countdown until Nvidia decides to releases whatever monstrous ace they have up their sleeves?

Quote:

---

Originally Posted by Faktor

Countdown until Nvidia decides to release whatever monstrous ace they have up their sleeves?

---

Which has been delayed at least a month...

Quote:

---

The Radeon HD 3870 X2 manages to outperform every other single card solution in our tests, being bested only by the 8800 GT SLI

---

Does 8800GT SLI out-perform 8800GTS 512 SLI?

Quote:

---

Originally Posted by MuffinFlavored

Does 8800GT SLI out-perform 8800GTS 512 SLI?

---

Of course not. 8800GTS 512MB SLI wasn't included in the tests.

Quote:

---

Originally Posted by MuffinFlavored

Does 8800GT SLI out-perform 8800GTS 512 SLI?

---

The Radeon HD 3870 X2 manages to outperform every other single card solution in our tests, being bested only by the 8800 GT SLI

I've highlighted some important terms for you.

What of the word on New Catalyst Drivers by the 28th????????

The CrossFireX drivers won't be available until CeBit apparently...

yay for a BS laggy ass degraded 8x pci-e2 slot for the ati card on a 780i MB, that such a fair bench look at the expreview its got like 20-60% better performance on theirs and they were at 600mhz higher clock

ARE THEY SERIOUS????!!!!!

BOTH tests are done on a 780i motherboard....thats like saying "hey lets see how this Phenom 9900 performs on the the asus maximus!!!"

this is so much f-ing proof of has biased anandtech really is.

Quote:

---

Originally Posted by MuffinFlavored

Does 8800GT SLI out-perform 8800GTS 512 SLI?

---

Does a 8800GT outperform a 8800GTS 512?!

Quote:

---

Originally Posted by MuffinFlavored

Does 8800GT SLI out-perform 8800GTS 512 SLI?

---

Quote:

---

Originally Posted by zanzabar

yay for a BS laggy ass degraded 8x pci-e2 slot for the ati card on a 780i MB, that such a fair bench look at the expreview its got like 20-60% better performance on theirs and they were at 600mhz higher clock

---

Just curious, where DO you get this info from?

Quote:

---

Originally Posted by zanzabar

yay for a BS laggy ass degraded 8x pci-e2 slot for the ati card on a 780i MB, that such a fair bench look at the expreview its got like 20-60% better performance on theirs and they were at 600mhz higher clock

---

Quote:

---

Originally Posted by SNiiPE_DoGG

ARE THEY SERIOUS????!!!!!

BOTH tests are done on a 780i motherboard....thats like saying "hey lets see how this Phenom 9900 performs on the the asus maximus!!!"

this is so much f-ing proof of has biased anandtech really is.

---

Chipsets most often don't effect performance.

However, there is something fishy with the 8800GT SLI results on Anandtech...

its how the sli based pci-e2 controller works on the 780i (should be the same on the skull trail) were it has a split 32x single lane (dual 32x on the skull trail and some others that are coming) and only uses an 8x unless u have an NV card that has a signed with an sli approved card to change it to 16x, unless there is a bios "fix" applied and thats not going to be on the 780i unless u go trough HP/voodoo


the 780i isnt a chipset its a 680i + pci-e2 controller and reference 8 phase digital pwm, its not a chipset platform like say x38 v p35

wow....and how much is nvidia paying anand...??? i mean.. GEEEEZZ!:down: bs

Quote:

---

Originally Posted by Atal

wow....and how much is nvidia paying anand...??? i mean.. GEEEEZZ!:down: bs

---

Agreed to that:shakes:

Quote:

---

Originally Posted by Zytek_Fan

The CrossFireX drivers won't be available until CeBit apparently...

---

I seriously doubt that will be the case cuz there will be some pissed of Homies if they don't get some drivers out!

hahaha clever anad tring to feed XS members with BS.

+1 for new review please using neutral motherboard

Mind you... a test with the ATi on a X38 board & the GTs on a 680i/780i will give pretty much the same numbers...

What is lame for both companies ( ATi & nVIDIA ) is the whole delay game ( which is "right" in their way ) and ATi's inability to compete in the high end sector...and in a timely manner.

so they tested ati card with nvidia mobo how about x38 ? i think it is better than nvidia solution

Quote:

---

Originally Posted by BenchZowner

Mind you... a test with the ATi on a X38 board & the GTs on a 680i/780i will give pretty much the same numbers...

What is lame for both companies ( ATi & nVIDIA ) is the whole delay game ( which is "right" in their way ) and ATi's inability to compete in the high end sector...and in a timely manner.

---

doesnt change the facts here....

nvidia card tested with nvidia board
ati card tested with nvidia board

in the interest of a fair review they should have tested with a neutral board for both cards such as the X38

most people I imagine wont even bother reading further once they see the test bed used in this review lol

this shows just how Anad is lame reviewer:

"A surprisingly successful FPS on the PC, Call of Duty 4 also lacks any sort of in-game benchmark so we benchmark the cut scene at the beginning of the first mission. We start the frame rate counter as soon as we're in the helicopter and stop once the man in the chair gets shot."

1. success of CoD4 is all but not surprising
2. this way of testing is completely flawed in terms of real gameplay performance!

Quote:

---

Originally Posted by hecktic

doesnt change the facts here....

nvidia card tested with nvidia board
ati card tested with nvidia board

in the interest of a fair review they should have tested with a neutral board for both cards such as the X38

most people I imagine wont even bother reading further once they see the test bed used in this review lol

---

What matters the most is real life.

For example...user 1 has a nice system up & running, and he's looking for a new graphics card.
He goes out and grabs a new card... is he going or really need to change the board ? Nope.

Quote:

---

Originally Posted by Zytek_Fan

Chipsets most often don't effect performance.

However, there is something fishy with the 8800GT SLI results on Anandtech...

---

had the same thought when i saw that 8800gt sli performance, saw other reviews on the net of 3870cf against 8800gt sli where cf was the most interesting, however looking at anand review.....

Most likely, peeps who want to run a HD 3870X2 card will use Intel chipset for Intel Platform and AMD chipset for AMD platform. So go ahead, test 8800 GT SLI with nVidia SLI board, but also using that particular board to test ATi card is just a biased review demonstrated right in front of us.

Techreport (use a X38 chipset) and Guru3d's reviews also show 8800GT SLI pwning 3870x2 left and right, stop seeing only what you want to see (ex: some benchmarks from some chinese review site), it's clear that 8800GT SLI is faster. The HD 3870X2 is a nice product, provide power for the masses but hardly an enthusiast product with a measly overclocking headroom.

One thing that is good is that the 8800GT in SLI outperforms the 3870X2 which means that hopefully the 8800GX2 coming out should be abit more powerful than the SLI'd cards as they will have had time to optimize card design and drivers.

Quote:

---

Originally Posted by Faktor

Countdown until Nvidia decides to releases whatever monstrous ace they have up their sleeves?

---

why every thread about ATi card ends talkiing about Nvidia and Vice-versa????



btw : nice review , now we need a better driver (same old story)

my HD3870X2 arrive in 2 days , i will post some tests later (in review section)


regards

To summarize Anand's testing methodology:

Bioshock: "Our Bioshock test involves a quick run through a section of the Medical Pavillion. Since the game has no built-in benchmark, our runthrough is manual and we rely on FRAPS to measure the average frame rate. All of the enemies are already dead in the area so we don't run the risk of running into one of them during our test."

CoD4: "lacks any sort of in-game benchmark so we benchmark the cut scene at the beginning of the first mission"

Crysis: "Crysis ships with a built-in GPU benchmark"

Enemy Territory: "We used our island benchmark for ET: Quake Wars as always"

UT3: "We used the built-in vCTF flyby in Unreal Tournament 3"

Witcher: "ran FRAPS during the game's first major cutsceneat the start of play"

Take it how you will

P.S. I don't know what the island benchmark is in Quake Wars, and I don't care

argh no CF results :(

Quote:

---

Originally Posted by Morais

argh no CF results :(

---

errmm you mean QuadCF results? Those can only be done once drivers are there that support it. Wait a little while more m8, they are comming.

Quote:

---

Originally Posted by Morais

argh no CF results :(

---

because theres NO catalyst Driver yet with suport for HD3870X2 crossfire !

Vrozne says the new catalyst Oficial will be released 30th Jan. >> http://www.vr-zone.com/articles/Rade...ance/5530.html


However no mention if the new driver have CrossfireX support ....


regards

Quote:

---

Originally Posted by Faktor

Countdown until Nvidia decides to releases whatever monstrous ace they have up their sleeves?

---

thank u for that useless contribution of NVIDIA fanboyism...

They say 8800GT SLI is $150 more than an HD3870x2. I see prices around $480 for the HD3870x2, while I can get an 8800GT for $260.

Try more like, $40 more expensive.

[H] gets in on the review action as well.
http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==

Quote:

---

Originally Posted by Hockster

[H] gets in on the review action as well.
http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==

---

theirs reflects the http://en.expreview.com/?p=219 review and if u compair the expreview 8800gt sli were the ultra was 1% faster so the 3870 x2 should be about 5% faster

Quake Wars is a joke with this card

i gotta say, its their first nice card in a while. but its depressing that technology has made so little progress. the performance is basically that of 2900xt crossfire 9 months ago, albeit it is much cheaper now.

Quote:

---

Originally Posted by safan80

Quake Wars is a joke with this card

---

That review used the old drivers must be. Here is Techpowerup's summary of the HD3870X2's performance in QW.

http://www.techpowerup.com/reviews/H...3870_X2/8.html

Aside from [H], who else actually sits their a$$ down for some time to PLAY the game and use Fraps to show min's and avg's?

Even the ATI release notes Anand listed talk about improving "Island GPU test up to 1920x1200." :( IMO, pick 2-4 cards that compete in same $ range and play some games. Then tell me which card did a better job. Throw some noise/power/driver/$/avail info in there and call it a day.

Quote:

---

Originally Posted by Jodiuh

Aside from [H], who else actually sits their a$$ down for some time to PLAY the game and use Fraps to show min's and avg's?

Even the ATI release notes Anand listed talk about improving "Island GPU test up to 1920x1200." :( IMO, pick 2-4 cards that compete in same $ range and play some games. Then tell me which card did a better job. Throw some noise/power/driver/$/avail info in there and call it a day.

---

Bit-tech does.

nice...

if this review trustworthy ???

Im looking @ selling my gtx if these results are actually true...

Quote:

---

Originally Posted by munim

To summarize Anand's testing methodology:

Bioshock: "Our Bioshock test involves a quick run through a section of the Medical Pavillion. Since the game has no built-in benchmark, our runthrough is manual and we rely on FRAPS to measure the average frame rate. All of the enemies are already dead in the area so we don't run the risk of running into one of them during our test."

CoD4: "lacks any sort of in-game benchmark so we benchmark the cut scene at the beginning of the first mission"

Crysis: "Crysis ships with a built-in GPU benchmark"

Enemy Territory: "We used our island benchmark for ET: Quake Wars as always"

UT3: "We used the built-in vCTF flyby in Unreal Tournament 3"

Witcher: "ran FRAPS during the game's first major cutsceneat the start of play"

Take it how you will

P.S. I don't know what the island benchmark is in Quake Wars, and I don't care

---

As long as the cards are tested under exactly, or as close to exactly, the ame circumstances and through the same sections of a game, I have no problem whatsoever with them doing it like that...

Quote:

---

Originally Posted by safan80

Quake Wars is a joke with this card

---

Why? Notice the 2560x1600 resolution and highest settings. Your post is a joke. 54 fps is enough. Anything above is nice but not needed. A card "X" that would have at least 50 fps in every single game at that resolution would be a winner over any other card that has 150+ fps in one game but less than 30 in another.

so you still cant play crysis on high settings smoothly. fail

Quote:

---

Originally Posted by aintz

so you still cant play crysis on high settings smoothly. fail

---

yea, it performed worse than the 8800gtx according to hardocp. at least that card had the excuse of coming out before crysis was released.

Dude, When did we start thinking that ATI was going to stomp the yard with this card?
As far as I knew, this was the latest greatest technology, because of the POWER IT SAVED while still producing a comparable performance ratio to a card already released using TWICE the power. Now we get crossfire on one card for people who dont have dual slots, and we still complain?

Come on guys.

It's a mixed bag really. I've read a few reviews and there is no doubt it is a serious bit of kit and does perform well in most games, however it has 3 major flaws.

1) Power Draw.... it is one hungry power sucking card. (Perhaps an external power source ala Voodoo5 6000 might be a good idea?)

2) Crossfire, yes I know not all will agree with me, but whether we like it or not Crossfire does require a bit of praying to the Driver Gods (as does SLi) for a game to work correctly

3) it's not that much faster than the single GPU solutions from nVidia which have been out for well.....over a year now?

Aside from those 3 blemishes I would say the HD3870X2 is a good card, but I am not going to buy one... I am happy with my 8800GTS 512MB and will hang onto it until the real Next Gen come out (so probably keep it until October or something).

I think we may see more power efficient, clever and serious polygon pushing GPU's coming out from both ATi and nVidia later this year....which would be nice.

Until then it is sort of like the hack job 9800GX2 Vs the pretty clever yet brute force HD3870X2.

John

Quote:

---

Originally Posted by JohnZS

...I would say the HD3870X2 is a good card, but I am not going to buy one... I am happy with my 8800GTS 512MB and will hang onto it until the real Next Gen come out (so probably keep it until October or something).

---

Me too. COD4 plays fine. I've zero desire to kick up power usage and heat for more performance in the only 3 games that already play fine for me...also IMO canned benchmarks fail when compared to actually playing the game and giving a subjective opinion like we do here.