Sapphire 4850 Toxic graphics card
Solving the regular Radeon's heat problems
Review Sapphire’s HD 4850 Toxic graphics card tackles a perceived problem of the reference AMD ATI Radeon HD 485 by changing the cooling package to lower the temperature.
The standard HD 4850 runs at a toasty 80-90°C as a result of having 800 unified shaders humming away inside the GPU. AMD uses a variable-speed fan that spins up to speed as the graphics load rises, a strategy which successfully maintains a constant temperature.
Sapphire's 4850 Toxic: dispersing the regulation 4850's heat issues?
However, you can distinctly hear the fan working away during a gaming session. It’s not loud enough to cause concern but the high GPU temperature seems wrong, despite this being a deliberate design decision on the part of AMD. The HD 4850 chip has a high power draw and despite its high efficiency there's plenty of heat to be dispersed. The £125 HD 4850 uses a single-slot design that radiates waste heat inside the PC case, or you can choose a £175 HD 4870 which is a big, dual-slot card that exhausts through a slotted bracket to the outside world.
Under load, the HD 4850 operates in the high 80s Celsius while the HD 4870 is in the low 80s. While neither card will suffer stability problems, the thought of all that heat may cause you some concern and it also reduces the cards' potential for overclocking.
The Sapphire Toxic takes a surprisingly direct approach to this issue, as it’s off with the standard cooler and on with a Zalman VF900 unit that includes passive coolers for the board's eight memory chips.
If you fancy a bit of DIY, you can buy the Zalman for £24, which bumps up the cost of a stock HD 4850 from £125 to £149. You’d be invalidating your warranty, so it probably makes sense to pay Sapphire £153 for the Toxic.
I don't understand the Crysis figures and it's bloomin' annoying.
I started using a Crysis run-through with FRAPS out of idle curiosity some time ago when it became apparent that the built in benchmarks are hopelessly inadequate.
My methodology is this: I choose the game settings, load a Crysis saved game and play through a short section part of the game.
At first the test results were fairly predictable as more powerful graphics cards gave better results in terms of frame rates and Nvidia had an advantage over ATi, doubtless thanks to the Way It's Meant To Be Played.
The fly in the ointment is that dual graphics chip cards and SLI/CrossFire perform far worse than a decent single graphics card which goes against logic. It's not just a low FRAPS score - the game plays like you're running through treacle.
Here's the thing; a number of readers have picked up on this apparent contradiction in the test results of some of my reviews - high 3D Mark scores and low Crysis scores - but no-one has said that they run HD 4850 CrossFire or a GeForce 9800 X2 and have a great time in Crysis.
re: crysis and crossfire
crossfire still causes dodgy results with some games. mostly due to ati's crap drivers (linux user here, so I especially feel the pain...)
Erm, do I really want to buy a product that describes itself as toxic? No.
Leo; any idea what's up with the Crysis dual 4580 results?
Assuming the labels are correct, it appears that running a dual 4580 degrades performance in Crysis to far below that of a single card running the same driver version.
Is that correct? If so, any idea why?
thats X1550 sorry
im not sure if i misstyped it as its not gone up yet,what i meant was "thats X1550/512" sorry