Sapphire HD 4870 X2 2048 MB

[Megasty]

wow!!!!

wow is right I think someone just needs some attention We need to give him some

 

[jbunch07]

dude wizz is one of best hardware reviewers around, go get a life kid

seriously! show some respect to the Wizz!

 

[Culex]

dude wizz is one of best hardware reviewers around, go get a life kid

Um, yeah, that's the problem. I DO have a life, obviously with a little more experience than the one you live, kid. I find the results hard to stomach because every review I've seen shows that 1. The 4850 beats the 9800GTX+ (The opposite was shown on the review).
2. The 4870 beats the GTX260 (The opposite was shown in the review), especially in HL2 and other source games, for it out performs the gtx280 in these as well (again, the opposite was shown in the review).
3. How can Opposing fronts score 199 fps when it's frame capped at 60?!?
4. Regardless of whether the 4870x2 has 1 or 2GB of GDDR5, it should cream not only a GTX280, but definitely a 4870 (It makes sense, 2x4870 GPUs on one diode would beat 1x4870 GPU NO MATTER WHAT!).
5. Do some research before you speak, otherwise you might understate your intellect in

 

[btarunr]

On a precautionary note, please leave W1zzard to respond to these comments by Culex (if he feels he should). Let's not spawn dozens of posts that basically mean nothing more than "how dare you?".

 

[alexp999]

Um, yeah, that's the problem. I DO have a life, obviously with a little more experience than the one you live, kid. I find the results hard to stomach because every review I've seen shows that 1. The 4850 beats the 9800GTX+ (The opposite was shown on the review).
2. The 4870 beats the GTX260 (The opposite was shown in the review), especially in HL2 and other source games, for it out performs the gtx280 in these as well (again, the opposite was shown in the review).
3. How can Opposing fronts score 199 fps when it's frame capped at 60?!?
4. Regardless of whether the 4870x2 has 1 or 2GB of GDDR5, it should cream not only a GTX280, but definitely a 4870 (It makes sense, 2x4870 GPUs on one diode would beat 1x4870 GPU NO MATTER WHAT!).
5. Do some research before you speak, otherwise you might understate your intellect in failing to do so

You have to remember some games have terrible crossfire scaling, same as Sli. Same as dual's and quad CPU's. I can guarantee across all games, the 4870X2 will NEVER be twice that of a single 4870, due to the scaling. And as for opposing fronts, I have never heard or seen a single game, which cannot have an FPS limited disabled, in some shape or form. I you sure you are not just running with vsync on?

 

[Culex]

On a precautionary note, please leave W1zzard to respond to these comments by Culex (if he feels he should). Don't spawn dozens of posts that basically mean nothing more than "how dare you?".

Well, the first comment that hasn't been a mindless shouting down of someone with a different opinion (one well based, I might add). Thanks for the refreshing knowledge that there are people with sufficient EQs, btarunr. I direct you all to this review, one of many which support my comments:

http://en.hardspell.com/doc/showcont.asp?news_id=3807&pageid=3116

 

[Culex]

You have to remember some games have terrible crossfire scaling, same as Sli. Same as dual's and quad CPU's. I can guarantee across all games, the 4870X2 will NEVER be twice that of a single 4870, due to the scaling. And as for opposing fronts, I have never heard or seen a single game, which cannot have an FPS limited disabled, in some shape or form. I you sure you are not just running with vsync on?

True, SLI and Crossfire most definitely does not scale to 100%. What I was stating is that a 4870X2 would not under any circumstances be beaten by a single 4870GPU, unless it were faulty. Just like a 9800GTX would not outperform a 4850 (nowhere near) especially in HL2. The same goes for a 4870 vs a GTX 260. Also, I'm not aware that CoH:OP has vsync as an option.

 

[W1zzard]

Dear culex,

1) look at the summary page
2) look at the summary page again
3) when the fps cap is removed and it's not opposing fronts
4) look up "single" in a dictionary, the test setup page might also be of assistance
5) right back to you

 

[Culex]

Dear culex,

1) look at the summary page
2) look at the summary page again
3) when the fps cap is removed and it's not opposing fronts
4) look up "single" in a dictionary, the test setup page might also be of assistance
5) right back to you

Dear W1zzard

1) My mistake, I should have viewed the test setup.
2) I simply assumed Opposing Fronts was being used, as it is the standard for all good reviewers. My mistake.
3) Regardless of whether or not you disabled one of the cores, it should have still beaten the GTX260, which it did not (especially in TF2). Quite bluntly, if the 4870 does not CREAM a GTX260 in source games, let alone a 280, the review instantly lacks credibility and should not be used as a reference. Check out this review:http://en.hardspell.com/doc/showcont.asp?news_id=3807&pageid=3116
4) Right back to you

 

[Mussels]

what makes you think a 4870x2 is in fact, exactly the same as a 4870 if one card is disabled? The cards share PCI-E bandwidth (less total, for one GPU) and its common for dual GPU cards to use lower clocks/higher latency ram to get their heat/cost down.

 

[btarunr]

Yup, even with CFX disabled, the disabled GPU is still 'active' though not participatory in graphics processing. In essence the one that is doing the graphics processing ends up with only 2.5 GB/s bandwidth, not that with one GPU not participating in Crossfire (doesn't mean disabled) means that all the PCI-E x16 bandwidth is at the disposal of the single GPU. It's really not the same as a single HD 4870 sitting cozy on a x16 slot, system bandwidth matters (to performance) more than the amount of memory that's available to a GPU so don't let that "1 GB / GPU" factor mislead you.

 

[Exavier]

I'd be enjoying my X2 if it worked with this DFI x48 mobo..
still, RMAing the mobo, now waiting to use it though £320 at release was an excellent price

 

[Tatty_Two]

there you go sir: http://www.techpowerup.com/reviews/Powercolor/HD_4850_PCS/26.html 30 pages for a vga review ..

we also changed the algorithm which builds the summary percentages slightly to be more intuitive. discuss the new per-resolution-graphs in the pcs 4850 comments thread please.

Thats interesting....thank you, so contrary to some popular belief, the higher res you go, the the smaller the gap between the 280 and x2 becomes, that surprises me a little!

 

[Mussels]

Thats interesting....thank you, so contrary to some popular belief, the higher res you go, the the smaller the gap between the 280 and x2 becomes, that surprises me a little!

its (something) of common knowledge that higher resolutions are more CPU bound than GPU limited. The gaps always close, and of course an x2 is going to need a monster of a CPU to feed it.

 

[Chewy]

Im curious as to how the 4870x2 is doing with driver updates? anyone know? I may just get this card someday but not if its still lagging behind the 280 overall

 

[ShadowFold]

Well my 4850 was already getting max fps in all my games with 8.7 and they did the same with 8.8 so go with a 4870X2

 

[Chewy]

I plan on getting this card down the road sometime maybe, I was planning on getting it on release, but than it was released for $600+taxes where I live and I spent my money on something else anyway.

time will tell, this is the best card out though atm from what someone told me on msn.

 

[ShadowFold]

Wait for the non refs man, it will be worth it

 

[Ourasi]

Thats interesting....thank you, so contrary to some popular belief, the higher res you go, the the smaller the gap between the 280 and x2 becomes, that surprises me a little!

The HD4870x2 widens the gap to GTX280 the higher the resolution get's. How you can read those graphs so wrong is odd. Look them over one more time, and you'll se that "the popular belief" are infact the truth in this case...

The bad results the HD4870x2 recieve in this review, seem to be down to the P35 board he uses. There are popping up other reviews with low scores, and the P35 seems to be the common factor. Why this is the case, I can only speculate, but PCIe 1.1 might be the culprit. Newer drivers have surfaced since this review, but they do not seem to eliminate this.. There are a few reviews that shows that a Crossfire setup of 2xHD4870 scores lower on a 8x-8x PCIe 2.0 P45 motherboard compared to a 16x-16x PCIe 2.0 x38/x48 M/B. 8x PCIe 2.0 have the same bandwidth as a 16x PCIe 1.1, but the 8x PCIe 2.0 are improved on some other areas compared to 16x PCIe 1.1, so if 8x PCIe 2.0 are a limiting factor in some cases it wouldn't be surprising that the 16x PCIe 1.1 is that aswell, especially when a beast like HD4870x2 is used...

In Wizzard's review there are some awefull scores in a few games, and it will benefit a potensial buyer of HD4870x2 to check out other reviews on full 16x PCIe 2.0 motherboard's like x38/x48/P45 before drawing a conclution...

 

[Ourasi]

its (something) of common knowledge that higher resolutions are more CPU bound than GPU limited. The gaps always close, and of course an x2 is going to need a monster of a CPU to feed it.

He is wrong, the X2 widens the gap the higher the resolution get's. And you are wrong aswell, in modern games, CPU will be less of a limiting factor the higher the resolution and AA/AF you use. If you game in high enough resolution in most of todays games, you will not see any difference between a 200$ CPU and a 1000$ CPU....

 

[Tatty_Two]

He is wrong, the X2 widens the gap the higher the resolution get's. And you are wrong aswell, in modern games, CPU will be less of a limiting factor the higher the resolution and AA/AF you use. If you game in high enough resolution in most of todays games, you will not see any difference between a 200$ CPU and a 1000$ CPU....

You are right if you say the x2 widens the gap in most....but certainly not all, there is no FPS growth difference between the x2 and 280 between 1024 and 1920 in both COD4 and Team Fortress 2 and the gap actually narrows between the 2 in Call of juarez and Supreme Commander, add to that in this review the 280 also wins in 2 of the game benches, there is 6 already so yes, the x2 does "widen the gap in more" but not as you have intimated in all, although my comment in an earlier post was also misleading, I only looked at Call of juarez lol!

Ohh by the way, just found a really sweeeetttt deal for a 4870x2 in the UK and my finger is hovering over the "buy it now" button

 

[Kwod]

I can guarantee across all games, the 4870X2 will NEVER be twice that of a single 4870, due to the scaling

In future{and in some games right now}, the 4870x2 2 gig will be 500% faster than 4870 512 CF because of the 512 ram......obviously, 2 x 4870 1 gig will solve that problem.

 

[Kwod]

There are a few reviews that shows that a Crossfire setup of 2xHD4870 scores lower on a 8x-8x PCIe 2.0 P45 motherboard compared to a 16x-16x PCIe 2.0 x38/x48 M/B. 8x PCIe 2.0 have the same bandwidth as a 16x PCIe 1.1, but the 8x PCIe 2.0 are improved on some other areas compared to 16x PCIe 1.1, so if 8x PCIe 2.0 are a limiting factor in some cases it wouldn't be surprising that the 16x PCIe 1.1 is that aswell, especially when a beast like HD4870x2 is used...
...

P35 and P45 are ineffective for maxing CF{but P45 should be okay for any x2 single card}.......the action starts with 38/48 with full 16x2.

 

[Ourasi]

P35 and P45 are ineffective for maxing CF{but P45 should be okay for any x2 single card}.......the action starts with 38/48 with full 16x2.

There is no difference in P45 and X48 when single PCIe is used, both has 16x PCIe 2.0. P35 uses PCIe 16x 1.1, witch is the same or slightly slower then 8x PCIe 2.0, and might be the reason for the bad scores in this review.
P35 suffers badly in crossfire with it's 16x-4x PCIe 1.1 solution, and I would not recommend crossfire on this M/B, same goes for the HD4870x2 IMHO...

 

[Kwod]

There is no difference in P45 and X48 when single PCIe is used, both has 16x PCIe 2.0. P35 uses PCIe 16x 1.1, witch is the same or slightly slower then 8x PCIe 2.0, and might be the reason for the bad scores in this review.
P35 suffers badly in crossfire with it's 16x-4x PCIe 1.1 solution, and I would not recommend crossfire on this M/B, same goes for the HD4870x2 IMHO...

I've seen articles on P45 performance with CF and it sux........but as you say, there's no problems with P45 and x2 cards.

P35=dinosaur.

X2= one card with 2 GPU's.
CF= 2 seperate cards.