Sapphire’s Radeon HD 5850 Toxic Edition: Our First Fully-Custom 5850
by Ryan Smith on February 18, 2010 10:00 PM EST- Posted in
- GPUs
Power, Temperature, & Noise
On the one hand, by overclocking this card Sapphire has ultimately increased the power draw of the card and the amount of heat it will be generating. On the other hand with the Vapor-X cooler, they’re better equipped to siphon out that heat, and to do so without making much in the way of additional noise. So let’s see how things pan out.
Right off the bat, the Toxic is at a disadvantage for power usage as we expected. At idle it operates at the same voltages and clocks as a reference 5850, so our 9W difference ultimately comes down to board differences; remember, the Toxic has a number of additional components compared to the reference card, particularly capacitors. Under load the difference is 17W, due to a combination of components and clock speeds. This is ultimately in-line with where you’d expect it to be based on the higher clock speeds.
Moving to temperature, we can see the difference the Vapor-X cooler makes. The 5850 was always a particularly cool card, and surprisingly the Toxic does worse here. We’re measuring the temperature of the GPU, so any extra heat produced by the Toxic’s component selection shouldn’t be factoring in. The difference likely comes down to the coolers – a shrouded blower that fully exhausts hot air looks to be a more efficient option under these circumstances.
It’s under load where the Vapor-X cooler on the Toxic shines. Even though the card is overclocked and drawing an additional 17W, it still bests the reference cooler by 3C. It’s not a massive amount, but then again it’s only 3C warmer than even the 5670.
So the reference 5850 cooler may be a bit better at keeping temperatures down at idle, but it’s not the quietest option. Here the Toxic can do 2dB quieter. It’s a similar story under load, where it’s 2.4dB quieter than the reference cooler, and once again take in to consideration the fact that the Toxic is dispersing 17W of additional heat in the process.
As for our fully overclocked Toxic, the 895MHz/1175MHz clocks push power consumption up by another 27W to 339W under load. Even with the now 44W difference between it and the reference 5850, the noise generated by the Toxic and the GPU temperatures are in a dead-heat with the much slower reference 5850. There’s no question that the Toxic’s Vapor-X cooler is a superior cooler, and this leaves us wondering just how much more it can take if we could overvolt the GPU.
71 Comments
View All Comments
Ryan Smith - Friday, February 19, 2010 - link
For temperature testing, we use a fully assembled Thermaltake Spedo. So yes, it's a well ventilated case.Being an open-ended cooler, a well-ventilated case is something that's of greater importance than on a reference card. But at the same time, you need a well-ventilated case anyhow just to avoid having negative air pressure in the case, because a reference 5800 card is going to be blowing out quite a bit of air.
Mr Alpha - Friday, February 19, 2010 - link
Are we going to get to see new idle power numbers with Catalyst 10.2 at some point?What about the availability of the 5970? In this part of the world I haven't seen a single one yet, and prices have been creeping up and are getting to the €700 mark. Is it turning into vaporware?
Ryan Smith - Friday, February 19, 2010 - link
ULPS only impacts the idle power performance of CF setups. It doesn't impact single cards.Calin - Friday, February 19, 2010 - link
As it appears in the pictures, it's wrong (Celsius is the correct text, not Celcius as written)Patrick Wolf - Friday, February 19, 2010 - link
Well, I'm sold. Where do I sign?NO! Must...wait...for Fermi. Ugh.
phaxmohdem - Friday, February 19, 2010 - link
Sign now! By the time Fermi actually gets released, it WILL be in super limited quantities, and I highly doubt we'll see any lower end derivatives than the 470/480 models from the original Fermi series. It is already 6 months or so late, so it is encroaching on Fermi 2's timeline.My way of thinking, HD5K series now, Fermi 2, when nVidia gets their act together and works out the bugs. It's not like we'll see any useful apps that can harness the purported GPGPU power of Fermi before the HD6K/Fusion series and Fermi2 lines are out anyway.
spidey81 - Friday, February 19, 2010 - link
I'll have to agree with him on waiting for Fermi. Not that I want a Fermi based card but that I imagine we will see a $20 to $50 price cut off the top and probably $10 to $20 MIRs on the 5850's and 5870's. At the MSRP I would buy a 5870, but not at the market inflated prices that they are at currently.whatthehey - Friday, February 19, 2010 - link
Okay, a few things I have to get off my chest:Page 2: "If any significant number of them could go higher, then AMD would have released them as a higher-end bin."
I call bunk! What's far more likely is that AMD doesn't want any overclocked GPUs coming out that will end up being as fast as the inevitable 5890 or whatever the speed bumped version is called.
Second, why are there no power and noise results for the 5870? These really need to be included.
Finally, a grammar rant. You get these enough I'm sure, so feel free to ignore me. Anyway, "in to" versus "into". I don't think you're getting it quite right a lot of the time. I've noticed Ryan never uses "into". There are places to use "in to" (i.e. "I turned in to the driveway." because "turned in" is a phrasal verb), but there are also times where you need to use "into".
Taking something into consideration is a tough call; I prefer "into" and you use "in to", and I'm not sure which is correct. However, I'd say you definitely throw a wrench INTO gears/recommendations, not throw a wrench IN TO recommendations. Into is typically directional in use. But I could be wrong.
At any rate, you should probably read this:
http://grammartips.homestead.com/into.html">http://grammartips.homestead.com/into.html
Here's one source that agrees with "take in to consideration" as correct:
http://dictionary.reference.com/browse/take+in+to+...">http://dictionary.reference.com/browse/take+in+to+...
But they also say "take into consideration" is correct (and in both cases they list "take into consideration" lower on the page):
http://dictionary.reference.com/browse/take+into+c...">http://dictionary.reference.com/browse/take+into+c...
I still don't like "take in to", but maybe it's because I subscribe to this sort of writing philosophy: Use "into" unless it doesn't make sense that way, like turning a paper into your teacher.
http://en.allexperts.com/q/General-Writing-Grammar...">http://en.allexperts.com/q/General-Writing-Grammar...
jajig - Sunday, February 21, 2010 - link
Thanks for the grammar tip.7Enigma - Friday, February 19, 2010 - link
Agreed. Ryan, call it like it is. We know with no competetion they do not want to be eroding the price premium of the 5870. You can see with your OC you already almost have parity between the 2 in some games. It's all about money, and with no competition it's their right, no matter how much it stinks.