Hardware Sapphire ATI Radeon HD 4850 TOXIC

http://www.bit-tech.net/hardware/2008/08/19/sapphire-ati-radeon-hd-4850-toxic/1

Sapphire, ATI's largest board partner, bring us the Radeon HD 4850 TOXIC. Featuring a significantly improved cooling solution over that of the stock 4850, and a healthy out of the box overclock, we take a look and see if the 4850 TOXIC is radio-awesome or just a TOXIC-waste

 

It's amazing that a simple cooler upgrade can push videocards to new heights. Makes you wonder why A(MD)TI / Nvidia use such ridiculous cooler solutions and letting cards run worrying high temps in the first place...

 

Exactly - an extra meaty cooler means much higher clock - you'd think with millions of dollar of R&D they'd be wise to this by now. I'd love to see some custom air coolers for the GTX 2xx series too - the 48xx's can, for the most part, just use 38xx custom coolers.

 

Nice to see a 4800 series card with a manufacturers fitted Zalman cooler! Shame about the price though!

Manufacturers fitted Zalman coolers seem to be rarer these days which is a crime if you ask me, quite a few manufacturers used to do them. Hopefully they will bring out a 4870 with one fitted, saves voiding the warranty to get good temps with a quiet fan!

Ati & Nvidia should get someone like Zalman to help them design a decent reference cooler that is both quiet and offers decent temperatures, most reference coolers at the moment are a joke!

 

Note that fitting an after-market cooler these days doesn't neccessarily void a warranty, since some of the manufacturers will still honour it so long as you didn't damage the card in the process. (evga for one) Though it may well be that as well as generally shorter warranties, the amd partners are less generous in areas like this.

 

http://www.specialtech.co.uk/spshop/customer/product.php?productid=6406&cat=0&page=1 tis £125 at specialtech tho not in stock right now.

Quite tempted by this card.

 

You sure? I guess if they don't know then that's fair enough but some manufacturers really don't like it. I remember one manufacturers site (I forget which one) has a gallery of images showing voided graphics cards.

 

That's a very good price, but they may well put it up once they have stock, it does seem quite low compaired to other retailers.

 

Am I the only person who's not really had any cooling problems with a 4850? My stock 4850 idles at 55-60 degrees; the fan is never on at idle either.. I guess I must be lucky to have really good airflow in my case or something..

 

...waiting for a 4870 with a decent cooling solution...

 

I'm hopefully watercooling my 4870 when I get it so no worries here.

Still, it looks like a great card. It's not often it's worth getting a non-reference card.

 

dammit!

Finally a 4850 without any dumb pictures of Ruby on it, and they make it blue! ATI should stick to red alltogether IMHO.

Anyway, it's still a good looking bit of kit, and seems well worth the extra money.

Richard, is there any way you could ask Sapphire A question for me? Have they aquired some sort of "golden sample" of better overclockers (close to centre of the wafer, pre-tested, etc etc) to build these cards, or are they using standard chips?

 

Whats the point in testing a mid range card on a Core 2 Extreme?.

I want to know what the performance will be on a sensibly priced C2....

Anyone who can afford a C2 Extreme, isnt (shouldnt) be putting a 4850 with it.

 

To try and eliminate any system bottleneck as much as possible, so we can have like-for-like comparisons between all the cards tested at BiT, without the effects of the rest of the system coming into the mix.

In the end, you'll never replicate these results at home anyway.

 

We're testing the card, not the system. For us to be able to compare every card fairly, we have to try and remove any other limiting hardware factor, be it CPU, memory or otherwise - hence the high end setup and why we try to use GPU intensive tests (like our core demo in Crysis) and not CPU intensive ones (E.G Company of Heroes)

 

As mentioned above, bring on the 4870's with the decent coolers and thats where my money goes. Maybe we should wait for the forthcoming Sapphire online shop, some time in September if I recall correctly, that offers various extras and add ons.

 

But these results will mean nothing to the 99% of people who will run a 4850 with a mid range C2 CPU....

You should do a secondary set of results with such a mid range system that a 4850 is likely to be inside.

 

well, as said by krikkit, you'll never duplicate these results at home.

DONT use these numbers as a way to decide which setup you need to buy to get a certain FPS count. The numbers are only a way to compare cards to eachother, they mean nothing else.

Even if you buy identical setups, your PC at home might run a little hotter, lose efficiency because of antivirus software, etc etc etc.
The tests conducted are fine and really helpful if you know how to interpret the results

 

That and running 2 tests takes twice as long.

 

No, because most people who buy Q6600s (of which there are MANY!!) will run them at 3.0GHz - that's a simple overclock from 1066MHz FSB to 1333MHz FSB. It's entirely applicable, and anyway - it's apples to apples explaining the differences on a particular system

Correct! Never ever take our reviews to give you a certain FPS count - we do real world testing so the chances are that unless you play the exact same level with the exact same settings the exact same way Tim and Harry does, it won't directly reflect. Real World testing gives you a precise idea of actual game numbers, and the way we do them is consistent (after plenty of retests along the way) but you can't use them to scale actual FPS, just "this is actually 5/10/50% faster than this in this game".