Hardware Sapphire Radeon HD 4870 2GB Vapor-X

http://www.bit-tech.net/hardware/graphics/2009/04/14/sapphire-radeon-hd-4870-2gb-vapor-x-review/1

Sapphire has not only added an excellent Vapor-X cooler to this Radeon HD 4870, but it has upped the memory from 1GB to 2GB. Does the extra RAM make a difference, and how effective is that cooler? Read on to find out.

 

Well depending on if you want quiet or speed id just spend £188 and get a HD4890 including delivery

 

At the speeds the 4870, 4890 and GTX 275 and GTX 295 operate and render in commonly used environments using the "I'd go for speed over silence" argument is the most ridiculous, fanboyish rubbish argument you could possible have used. Honestly, who is gonna play games on those high resolutions you have been using for testing? Those 3 people (besides you) certainly have afforded an SLI or Crossfire + X2 setup anyway.

 

Seems like RV770 has ran out of firepower.

Judging by the pictures on the first page it does have rear exhaust, doesn't it?

 

From the STALKER benchmarks:

"at the lower resolutions, the GTX 275 is a good bit faster. For example, while the Vapor-X can run the game at a minimum of 25fps at 1,920 x 1,200 to the 23fps minimum of the GTX 275, the GTX 275 manages a 27fps minimum and 35.5fps average at 1,680 x 1,050 to the 23fps minimum and 31.6fps average of the Vapor-X. "

Surely thats not worth mentioning? NO card in this review is a 1600x1200 item, and the Vapor-X is playable at 1900x1200 while the GTX 275 is not / otherwise really pushing it. So why make a point of 1600x1200 perfrmance?

Please explain.

 

I second that post A_P. I have been dissapointed in some of BT's reviews as of late (I can explain further if anyone is interested - but kinda tired right now so I am also probably a little grumpy).

However, apart from that there were a couple of typos, etc. From page two, captioning the pictures of stock a vapor-x mosfets:

"The more basic reference design VRMs on the right versus the more complex Sapphire VRMs on the right"

And this one is a little more interesting - it changes the results slighty depending on where the mistake was made, either is quite possible. Page 5, last graph (CRYSIS, DX10, HQ, 2560x1600, 0xAA, 16xAF). You are obviously talking about the blue bars - however the blue bars are labeled 4890, and the red bars above are labeled Sapphire Vapor-x 2Gb 4870? Not sure which is which. At this higher resolution, is more memory used, and as such enough to make such a big difference? I know that if AA was turned on, then it certainly would use more memory.

 

I'm not quite sure what you mean by 1,600 x 1,200 because we haven't tested anything anywhere at 1,600 x 1,200 in this review. I think it's very relevant to talk about 1,680 x 1,050 and 1,920 x 1,200 performance considering the price range of these cards and the monitors they're likely to be used with.

The 275 is quite a bit faster at 1,680 x 1,050 and we think that is worth mentioning for the large majority of our readers who use monitors with that resolution (there's lots of data we've got saying 1,680 x 1,050 is much more widely used by bit-tech's readers than 1,920 x 1,200).

 

1920? That's 24" monitor size and readily run to plenty of people. When we ran our readers survey last year many of you gamed at 20-24" with surprisingly many indicating that they were interested in 30" screens and testing at 30" resolutions. 30" 2560x1600 is a natural extension of what we review anyway and it's just going to show the complete picture.

When you game do you not wear headphones or turn up your speakers? Otherwise we might as all buy passive 4670s and not worry about performance, as long as you're sitting there in silence

The 1920x1200 numbers for the Vapor-X are marginally different at 2FPS and BOTH are barely playable - 25 versus 23FPS.

However at 1680 the GTX 275 stretches a few more FPS out and has a much higher 6FPS minimum: almost the difference between smooth and stutter. I beg to differ with your statement that these cards are not 16x12 items - it's a perfect resolution for them for 20-22" monitors, or 24" at lower resolutions with the AA/AF slightly higher.

 

LOL, so my 285 still rocks for the res I bought it for (1920). Has a better lower frame rate than any other single card (which is the real measure) and has a better top end FPS in enough games.

 

It also cost a lot more =P

 

Well yes sorry i meant 1680x1050 its just that in my mind they are the same res given that Im stuck with a 1400x900 max res. monitor for quite some time to come, anything higher than that is in my 'realm-of-fantasy' resolution list lol.

Erm point taken regards pairing monitors with graphics cards, although perhaps the gtx 275 and 4890, although not the subjects of the review, would perhaps be overkill for anything less than 1900x1200 / 1920x1200. Dont get me wrong overkill is very nice in my book, thats why i splashed 200 quid for a 4870 512mb in July last year even though im on a 1400x900 monitor. Thats not quite a max-anything-out situation for me however because the rest of my rig is a bit wimpy (cannot even max Far cry 2 out under dx9 lol, its all stuttery and horrible!).

 

From BindiBagi:

"The 1920x1200 numbers for the Vapor-X are marginally different at 2FPS and BOTH are barely playable - 25 versus 23FPS."

Its a very narrow band but you will notice that extra 2fps I would think. It is kind of splitting at hairs but a minimum of 25fps would suggest playability even if in practice it would be a half stuttery half fairly smooth experience. Performance at the points where competing cards 'crap-out' is surely the #main# point to be making in benchmark summaries? Anyway, two small points:

1) Your sister mag Custom PC would declare the game playable on the Vapour-X and unplayable on the GTX 275, but then again thats me pointing out their snotty attitude towards components on occasion =P more than making any serious points/judgements.

2) It might have been a good point to make that some titles that cannot be maxed out and playable with todays cards (STALKER, GTA4) are not soo much falling short in the shader/texture/bandwidth/driver department but in sheer frame buffer size, and that, taking Crysis as an exceptional title hardware requirements wise where frame buffers over 1gb have little or no impact, the future direction of Games could well be to require quantities of video ram beyond what a 32bit operating system can realistically deal with (1gb at the very most if at least 2gb of system ram are to be installed, at least availabl / 'visible' to the OS). Vista might be the last OS where 32bit versions are the most prevalent.

 

I would say that the #tone# of the graphics reviews seem to have changed. Cant quite put a finger on it but i can feel a change (in the bit-tech force).

 

Not a bad review by any means, but... The issue of comparing different cards from the same company, but with differing drivers, removes the possibility of objectively measuring their performance, relative to their brethren.

The article mentions Forceware 182.50, yet is not used in any of the benchmarks?
Would it not be preferable, to measure each card with the same drivers, even it means using a slightly older release.
Such testing may not fully establish the absolute performance of an individual card, however it would allow you to gauge, to what extent the actual hardware affects performance.

 

When you get a chance... test this pretty little thing here:

http://translate.google.com/transla...e/nyhet,15542.html&sl=sv&tl=en&hl=en&ie=UTF-8

 

I would have to agree. Since you guys got merged/ taken over by custompc it seems like they have totally taken over your site both in articles and writing style. They clearly write like IT magazine journalists ( can be taken as a compliment or not) whilst the original BT team wrote good, interesting and funny articles backed up with solid knowledge and testing procedures. I can even tell that regulars like Joe, Harry, Richard and Tim are having their articles heavily edited by someone at Dennis Publishing/ Custom PC.

My two cents. FWIW I still respect the original BT crew.

 

Thank you for the the benches, it makes me just want to have a GTX275 even more. It got the lowest idle and thats very important when i am not playing a game but do other stuff like browsing the web. It does get hot but since i am using WC (just need a GTX 275 wc block ).

 

Vista won't be the last 32-bit prevalent OS because most business' still use 32-bit and many, many people don't bother with graphics cards or 64-bit apps. Win7 will still sell predominantly 32-bit at launch until 4GB or more of memory becomes "normal".

I have to disagree about the 2FPS difference and I'd doubt anyone looking at a game to notice between 23 and 25FPS for a local minimum. The average has as much weight because a higher average means it spends more time at increased FPS, whereas the minimum is a single dip.

 

Bindibagi,

the people using 24" or bigger monitors (particularly for gaming) still are a small minority. I think Valve Software's hardware are a pretty good reference for that.

 

That's true, but the data in Valve's Hardware Survey is also much lower-end than the data we acquired during our last few reader surveys. On average, our readers tend to be more high-end than the typical Counter-Strike player and spend a reasonable amount on upgrades every year.