Reviews - NVIDIA GeForce GTX 580 graphics card review

[Terbinator]

Most of us figured that a balls-to-the-wall 460gtx Press Edition would beat the 6800's, but that didn't stop them from being included in reviews.

There is a very disturbing lack of consistency around right now.

I don't think it was guaranteed that a GTX 460 O/C would be faster (fast as) than a 6870 - but your right, some sites can't seem to settle on a test bed at the minute.

Anandtech did a 6850 O/C card round-up but the didn't include a O/C 460 - the mind boggles

[CAT-THE-FIFTH]

Anandtech did a 6850 O/C card round-up but the didn't include a O/C 460 - the mind boggles

They did not include the overclocked GTX460 as many people were annoyed they had included an overclocked GTX460 in the original HD6850 review.

[Terbinator]

They did not include the overclocked GTX460 as many people were annoyed they had included an overclocked GTX460 in the original HD6850 review.

I know, but if you were to include an O/C 460 in any review surely it would be against other O/C'd cards.

[Jimbo75]

I know, but if you were to include an O/C 460 in any review surely it would be against other O/C'd cards.

They already included the O/C 460 vs non-overclocked cards, that's what the fuss is all about.

I don't see any O/C 5970's vs the 580 either.

[Terbinator]

They already included the O/C 460 vs non-overclocked cards, that's what the fuss is all about.

I don't see any O/C 5970's vs the 580 either.

I know what the fuss was about, but in this case its perfectly logical - even the moaning posters from the stock 68xx review can't deny that, IMO.

[Jimbo75]

I know what the fuss was about, but in this case its perfectly logical - even the moaning posters from the stock 68xx review can't deny that, IMO.

This poster on AT got it right.

Comparing stock-clocked cards from a LAUNCH article of a new series to OC variants is questionable (again the key word being launch), but I believe that few of us were asking to not have any factory OC cards included (especially when this review focuses solely ON factory OC cards).

Just keep the comparisons consistent; that's all most of us readers without an agenda are advocating.

[CAT-THE-FIFTH]

The funny thing is that an HD6850 1GB overclocked from around 775MHZ to around 940MHZ to 960MHZ for the core and a RAM overclock still consumes less power than a GTX460 1GB at stock clockspeeds:

http://www.anandtech.com/show/4002/a...asus-xfx-msi/8

[Terbinator]

Conversely it would be consistent to maintain the use of Brand X O/C'd 460 in the review.

Without trying to deviate away from the thread, apparently the Asus 580 is capable of hitting 1000+ on the core

[CAT-THE-FIFTH]

Without trying to deviate away from the thread, apparently the Asus 580 is capable of hitting 1000+ on the core

The HD4890,HD5850 and HD5870 have reached 1GHZ on the core with air cooling. Even the HD6870 has reached over 1GHZ with air cooling too.

[Terbinator]

The HD4890,HD5850 and HD5870 have reached 1GHZ on the core with air cooling. Even the HD6870 has reached over 1GHZ with air cooling too.

This is a ~200MHz jump though from stock and quite a feat given the original GF100 O/C problems.

IMO of course.

[Jimbo75]

I don't think anybody would argue that nVidia has done a decent rescue job here, and at current prices you'd take one of these before a 5970 any day. Should be interesting to see how the 6970 competes.

[Domestic_Ginger]

Great review and really pleased to see you taking cards apart. A couple more pics of the cooler block would be nice close up. I like that kind of thing!

[Pob255]

Ummmm I think there's something wrong with this graph on page 5

GTX 580 is significantly faster than GTX 480 and HD 5870 and is only a little slower than the twin-GPU HD 5970.

Esp as you list the 580 in a table as -3% compaired to the 5970

http://www.hexus.net/content/item.php?item=27307&page=5

EDIT: and it's nice to see nvidia finally get the real fermi card out not the respun bodge job that was the gtx480, ok this probably is still based mainly on the respun silicon and finally addressed some of the real issues, unless it's just a matter of some very careful silicon binning.

[Tarinder]

Pob255,

Thanks for that. The percentage difference is based on the GPU score and not on feature test 4. Quoting from the article:

'Going by the GPU score, here's how it breaks down:'