Hardware AMD ATI Radeon HD 5970 Review

Wow why is there so much BT hate lately? My poor bit-tech always getting slammed on every review lately >.<

 

... I love bit-tech but those are some seriously low fps numbers

 

I have to agree with two things that have been stated in this thread.

1. The CPU needs much more juice

2. It's a little odd that BT is getting 10-30% lower scores for fps as compared to almost every other site out there.

 

I love bit-tech, but sometimes constructive criticism can make a good website even better or warn against worrying trends. It's hard to know if you're doing the right thing if no one complains.

My personal beef, I will reiterate it one last time, is BRING BACK THE CRYSIS VERY HIGH RESULTS. This is the fastest graphics card in the world. Ever. Crysis Very High is still the most challenging game yet released. A match made in heaven right? The maddening thing is bit--tech provided Crysis Very High results back with the 295, 275, 4890, etc. But now, just as soon as we start getting cards that can realistically handle Very High, the benchmark disappears, never to be seen again. I said this in the 5870 review, how hard could it have been to toggle the setting and run the test one more time? Like I said, people aren't going to splurge on a $600 graphics hard to just play Crysis on "High".

Also even with Stalker: Clear Sky, they say "we tested on High not Maximum due to intensity of engine...", when you're getting 125 FPS on High it might be time to test the card on Maximum. Give this card a challenge guys, let's see what this thing can really do!

 

Crysis very high would be a given surely. Enabling 4xaa is around the same fps drain as going all very high from regular high. Theres quite some difference in favour of all very high imho.

 

Although not that it really matters when the card itself is so horribly broken. This card is a disaster. Either the drivers are horrible, or we have the new FX 5800 on our hands. Honestly you'd need a total re-review when (if??) ATI fixes this monster to get any idea of what potential it really has.

 

Bit tech use a section of game that most stresses cards. What's the pout in saying a game is playable when the next scene crawls? None! Bit-techs are exceptionally accurate. I've replicated tests even to system spec before and found maybe a 5% difference at most. These other sites are clearly not using the most strenuous part of the game and as such cards appear quicker.


Andy

 

We do a graphics with new drivers refresh whenever is necessary. Partner cards with a bit of an overclock? Not really necessary as you only need to guage how much faster it is than a stock card. New GPUs (like the HD 5870, 5850, 5770?) We do a refresh. New uber GPU (like the HD 5970) we do a refresh (and switch to Windows 7 too). It takes an enormous amount of time to do a full refresh - 3 cards a day if you're quick, which is a lot of time when there's other stuff to cover. The fact that no new architectures were released since January sort of time is why we weren't frantically retesting GPUs through the spring and summer, but when it started to necessitate it, we updated, rebuilt the test rigs and got it done.

For those saying we need a faster CPU - what, you all have 3GHz Core i7 processors at home? The reason we don't OC our system is that, following a LOT of experience with test systems, OCing the core components is asking for stability trouble further down the line. I think our test system is pretty representative of a high end system.

 

In which game?

All the results you see here are 100% repeatable. If you were to come into the lab, I could have cards repeating these numbers in 15 mins. We've got nothing to gain posting incorrect numbers and I take my GFX testing very seriously - those results are correct.

Could it be those "other" sites don't test with 16xAF forced in the driver as we do?

 

We've obviously posted very high numbers previously (last year iirc), but the fact is they were some of the least viewed numbers on the site, and take over an hour a card to run. Now cards are getting to the level that they might be able to deliver playable FPS in Crysis Very High I'll consider adding them for when we finally get our hands on Fermi (hehehe).

t-t-t-triple post!

 

To be fair, i think that bit-tech do a good job of highlighting the inconsistencies that still occur with crossfire/sli. I used crossfire in the 2900/3800/4800 series of cards and although it was mostly successful, there was the occasional problem that used to frustrate me. Having invested in a single 5870, i think it's more than enough for me (at 1680 x 1050 certainly) and the image quality is excellent. I can't see myself going back to crossfire, in the short term at least.

It will be interesting to see one of bit-tech's graphics card roundups that they do regularly to see how the card performs after a couple of driver revisions, and how it stands in terms of value for money in a month or so.

 

Huh. I think we all make the mistake of assuming our own personal preferences apply to the larger public. Crysis Very High was the very first page I'd look at, even before reading the intro. But I totally believe you when you say most people ignored it.

Also, I had assumed adding the Very High would be perfunctory given you already do Crysis at High. Shows you what I know about benchmarking.

 

Yeah, that's exactly what got me interested in bit-tech. Other sites just give you an FPS average of an arbitrary segment. At Bit-tech you get minimum and average in realistic benchmarks. Also I really like how bit-tech treats 16x Anisotropic Filtering as standard these days, because really it should be. 0x and 4x anti-aliasing numbers are informative too, some sites use stupid things like 8x or 16x AA which IMO is pretty much just a waste of graphics resources.

 

Great review

Good Review

Fastest card on earth gets this rating from BT?!

Does it have to be the fastest in the Federation, or the entire Alpha quadrant to get a 9?

Also, the price on your little island should not affect your rating of the card. Believe it or not people from around the world enjoy your page (myself included). I think it truly sucks that the card costs more on the Foggy Island, but you cant penalize the card for that.

As mentioned you need a beefier cpu for this card and you need more dx10/11 games to test (see link to the HardOCP pick of games). DX9 is dead since incredible games like Crysis and Shattered Horizon are coming out more regularly. No one is going to spend 600$ on the best graphics card ever made to play dowdy old dx9 games.
And as to Joe's question, the same people who can afford a 600$ gpu ARE the EXACT people who will use eyefinity.

In any case it is a time to celebrate, ATI/AMD has again devilverd the best of the best, with the added wonder of being able to run 3 screens at once. What fps would competing cards get on 3 monitors? 0, they simply can not do it.

Yours in soon to be the proud owner of a ATI Radeon HD 5970,
Star¤Dagger

P.S. I have started to add 3 points on to your rating to come up with the true score, cheers!

 

It is difficult to compare the reviews as the TPU review done by Wizzard doesn't state whether the built in benchmark or custom timedemo was used. If it is the former then that would explain the higher results (time demos are usually more demanding than the built in benchmark).

This doesn't explain how the HD5870 scores 39fps in the Bit-tech review vs 31.7fps in the TPU review (same settings as above). The GTX295 results are wildly different as well (higher on Bit than TPU).

If anything its the TPU review that doesn't make sense, the H5970 scales more than 100% on the very same test as above when compared to the HD5870. Whilst I know this isn't impossible I have never seen this issue before with Crysis (a game not known for its great CF scaling).

On a seperate note, Bit - I see the Nvidia driver problem plauging the GTX295 in Fallout 3 has now been fixed (no more 0 FPS minimums).

I forgot to mention in my first reply as well, thanks for the review guys, keep up the good work!

 

Glad to see you think Modern Warfare is worth a 12/10!

Seriously though, all your posts seem to involve you crying about Bit-tech. If the staff and their reviews bother you so much, why continue to read/visit the site?

As for your 'Foggy Island' comment, this is a UK based site, with UK staff, owned by a UK company with most of its visitors from the UK (I think). So yes, UK retail prices factor very much into their reviews.

Seeing as you're so obsessed with arbitrary numbers, maybe you'd be better off finding a German based site that doesn't actually review anything. Instead they just take the top three scores from other websites and write down the average? Would save many knickers getting twisted.

 

Are there mature drivers for this? Because what I'm seeing is more driver-side than anything else, no way the card would perform that badly, especially since one of the charts say that the HD5870 outperform it.

 

Im english and im not offended by foggy island. But yeah this is an english site so ...get used to it buddy! German review sites and pc mags are universally s.h.1.t. except pcgh.de which does interesting tit-bits. Maybe nu-tech might have some points going have u thought?

move to Warhead and dump stalker clear sky. Ive got both shadow of chernobyl and clear sky and clear sky is pretty damn lame imho. And thats without counting S.o.C with the oblivion lost mod. Its stable but its just pants.

On a seperate note, have bit-tech considered integrating some 'highest playable' settings into the roundups? Even something fairly simple say the highest practical res. and detail settings used? not pertaining to this review but just in general.

 

Take out call of duty and the enormously nvidia loving crysis (warhead doesnt have this at all imho) and replace with 2 other titles and you'd see less comments on mature drivers. Just a general point, not a particular dig at bit-tech, after all root around some other reviews and get a general picture of things. No ati card will look good against its nearest and fairest (i.e. 5970 vs. Fermi) nvidia competitor in the original Crysis and no amount of driver tweaking will close that gap. The story is much different in Warhead. Also, testing either title at very high settings leads to a much tighter race, not just because the fps is lower, neither.

 

No i'm just commenting on the old drivers that BT tends to use.