Fermi reviews


I love being able to read russian at times like this.

edit: "Именно поэтому многим читателям кажется, что задержка с выпуском была слишком большая, что-то около полугода. На самом деле перенос был всего лишь один — трёхмесячный. Изначально планировался выход новых решений в последнем квартале 2009 года, который затем был передвинут на квартал вперёд. "

"This is exactly why many readers believe that the delay to release was too long, something around a half year. In reality the carry over [time from scheduled release] was just one - 3 months. Initially the planned release was in the 4th quarter of 2009, which was thereupon moved one quarter."

Haha ok, wait, so they're no longer competing with AMD? Because if they are, it's in their interest to release no later than AMD. This is complely wrong. They've had 2 delays. The first was the release they should've made, and the second is the revised date they had to transform into a Tesla release.
 
Last edited by a moderator:
techPowerUp benched with Quake 4 and no anti-aliasing? Makes sense.
 
Most of these review sites are using older catalyst versions. One site even had 9.12?! What?
 

They vary settings inside of the same tests :eek:? Jeez.

The PCper, Firingsquad, and Anandtech reviews seem best to me so far... some of these other sites are using old versions of drivers which for the 5870 makes a small difference. Seems in apples-to-apples tests the 480 oftentimes owns the 5870, while other times just a slim 10-15% advantage (especially note the minimum FPS on these #'s).
 
I thought the 470 would put up a better fight, but I guess not. Major disappoint in that area... I guess that 5850 twin frozr is it for me.
 
They vary settings inside of the same tests :eek:? Jeez.

The PCper, Firingsquad, and Anandtech reviews seem best to me so far... some of these other sites are using old versions of drivers which for the 5870 makes a small difference. Seems in apples-to-apples tests the 480 oftentimes owns the 5870, while other times just a slim 10-15% advantage (especially note the minimum FPS on these #'s).

So still buying the card or did you already cancel your order? ;)
 
So still buying the card or did you already cancel your order? ;)

PCperspective

Friingsquad (with OC results!)

Anandtech's


Hardware Canucks
According to our conversations with NVIDIA, their current beta driver (and the one which will be launched with the card) exhibits an issue in this game (BFBC2 and DiRT 2) where framerates take a massive plunge at 2560 x 1600 resolution. They expect a fix to be forthcoming sometime in the weeks following launch.

Seems quite solid a value to me, as I said it TROUNCES the 5870 in some tests (35-40%+) and in most of the others at least hits 10-15% better... that's not out of whack with the 25% price difference for a top-end part... and apparrently there are some driver issues in a couple of the main games that are holding it back and being fixed ASAP... so, my order is in-place, but I switched to Amazon for $4 prime overnight shipping ;).

I thought the 470 would put up a better fight, but I guess not. Major disappoint in that area... I guess that 5850 twin frozr is it for me.

Yeah I was surprised at the 470's results... but the 480... (homer simpson drool).
 
Cherry pick 3 reviews? What about the other 20+ out there?

I mean, even HardwareCanucks and Anandtech pegged the card at only 10-15% faster, not 25% as some claim

And come on GoldenTiger, you've been on [H] long enough you should know their methodology by now

So still buying the card or did you already cancel your order? ;)

Of course he's still buying the card, he gave it a 9/10 in the poll thread

May his card last longer than those GTX 280's did
 
They vary settings inside of the same tests :eek:? Jeez.

The PCper, Firingsquad, and Anandtech reviews seem best to me so far... some of these other sites are using old versions of drivers which for the 5870 makes a small difference. Seems in apples-to-apples tests the 480 oftentimes owns the 5870, while other times just a slim 10-15% advantage (especially note the minimum FPS on these #'s).

Yeah, I honestly think [H] has the worst reviews of any major site. Graphs are nice but it's just flat out stupid to bench card X with one set of settings and then change the settings when benching card Y. I want to know which card is more powerful and by what percentage, and the vast majority of [H] review space is dedicated to useless pursuits.
 
Looking at all the reviews, it looks like Fermi is fast with light load but when resolution gets higher, the gap to HD 5800 gets smaller. I think that 2x HD 5870 would do better that 2x GTX 480 in multi monitor gaming.
 
Have any of you even tried to read some of these reviews? Like the guru3d one? Hard to read. Like some guy with ADD wrote them during his "crack break." I read his review of the 470.


Anyway, think I'm going to stick to Anandtech's review just curious what they wrote. (Ugh why did I read the guru3D one, I want my 10 minutes back.) HardOCP's review was really stellar so far. It almost felt like I had one in front of me.
 
Yeah, I honestly think [H] has the worst reviews of any major site. Graphs are nice but it's just flat out stupid to bench card X with one set of settings and then change the settings when benching card Y. I want to know which card is more powerful and by what percentage, and the vast majority of [H] review space is dedicated to useless pursuits.

Useless persuits? Explain which part of the review is useless.

Worst of any major site? Do you consider Guru3D a major site? If you do, that is all i needed to know...
 
Looking at all the reviews, it looks like Fermi is fast with light load but when resolution gets higher, the gap to HD 5800 gets smaller. I think that 2x HD 5870 would do better that 2x GTX 480 in multi monitor gaming.

It's like the opposite of what you would expect. Fermi does seem to scale pretty well though, and one review even shows sli 480's beating sli 5970.

Not sure what I'm gonna do. Too many variables. The reviews all have the 5970 at stock when everyone knows that card is worthless without a massive OC. Nvidia's first run drivers are being compared to refined ATI drivers. I didn't see any reviews comparing sli 480 to my setup, and definitely none comparing OC sli 480 to OC tri-fire.

If I could just walk into a store and swap my setup for a pair of 480's I'd do it, but selling my cards will be a pain in the ass and ebay fees on expensive items just went WAY up.
 
Useless persuits? Explain which part of the review is useless.

Worst of any major site? Do you consider Guru3D a major site? If you do, that is all i needed to know...

The four I linked are old, major review sites.

P.S.:

ZerazaX said:
Cherry pick 3 reviews? What about the other 20+ out there?
I mean, even HardwareCanucks and Anandtech pegged the card at only 10-15% faster, not 25% as some claim

LOL, I linked those as part of my post, who's cherry-picking now? I think you're misreading the graphs and articles, because it's a pretty decisive win for the 480 over the 5870 in those reviews...

Plus, you have to remember:

HardwareCanucks review on the Battlefield Bad Company 2 page: said:
According to our conversations with NVIDIA, their current beta driver (and the one which will be launched with the card) exhibits an issue in this game (and DiRT 2) where framerates take a massive plunge at 2560 x 1600 resolution. They expect a fix to be forthcoming sometime in the weeks following launch.

So you have to take those benches and assume they're lower by a good chunk than they will be once the card actually launches.
 
Any reviews at 5760x1080 or 5760x1200? I'm really curious how well their version of Eyefinity works since it's a much different implementation.
 
Any part of the review that tries to compare framerates between cards using different levels of AA. Which is to say, almost the entire review.


You've proven no point. The charts compare gaming experience. One shows your beloved 480GTX allows you to have a better experience then a 5870.

How are they useless? What? Do they not allow you to troll forum post any harder?
 
You've proven no point. The charts compare gaming experience. One shows your beloved 480GTX allows you to have a better experience then a 5870.

How are they useless? What? Do they don't allow you to troll forum post any harder?

No, they don't compare gaming experience. They compare disparate software settings on a tiny number of games. Sometimes it gets especially egregious, like in past reviews where they showed comparisons of one card running Crysis on high vs another card running it medium.

Most people don't buy a brand new graphics card to play 5 already released games. They buy new cards with the expectation of playing whatever comes out in the next year or two. Therefore, the most important thing a review can do is provide a good estimate of the relative power of the cards so someone can find the best choice for the next couple of years. That requires a high number of apples to apples benchmarks over a wide range of games. [H] reviews fail miserably in that regard.
 
Back
Top