AMD New Radeon 5870, 5850

[nova]

One thing I'm wondering is with the high VRM temps, the card might attempt to throttle the GPU. More info here:

http://www.anandtech.com/video/showdoc.aspx?i=3643&p=11

EDIT: More info - looks like 120C is the point where the VRMs start throttling on the 5970, so it's probably the same on the 5870 and 5850:

http://www.anandtech.com/weblog/showpost.aspx?i=657

Do a load test and look at the temps of the VRMs and the clock speed of the card, and you should be able to verify it.

I tried OCCT while running GPU-Z in the background logging everything and the card never throttles down. The VRMs reaches a scorching 125 degrees C. OCCT and furmark really is the worst case scenario though and if the card manages that you can rest easy it will handle any game you throw at it.

Since I bought this card for gaming I won't be running these "power viruses" as AMD calls them after I initially proof the setup. I'll actually start using the card as intended and be happy with it. You'll never see these kind of temperatures in real life gaming. So next I'll go enjoy myself and log temperatures while gaming instead.

[flinx]

Hmm.. that's interesting. I guess I'll find out what the threshold is for the 5870. I'll use OCCT, but only to "preheat" the VRMs and GPU to a decent temperature before switching to a game to see how it performs.

[flinx]

By the way it seems some people are experiencing some problems with their ATI cards:

http://forums.amd.com/game/messageview. ... did=124747

Anyone getting a card might want to do some proper stability testing here before attempting any mods that void the warranty (such as replacing the cooler).

[mkygod]

By the way it seems some people are experiencing some problems with their ATI cards:

http://forums.amd.com/game/messageview. ... did=124747

Anyone getting a card might want to do some proper stability testing here before attempting any mods that void the warranty (such as replacing the cooler).

Good advice. Thats what I am planning to do when i get my card on Wednesday.

I made sure to get an XFX card so that replacing the heatsink won't void the warranty.

I have the Accelero s1 and vrm-4 on order at the same time and will install it after a few days of load testing with the stock heatsink.

The "gray stripes" problem in that link is the first time ive heard of it. It's got me a bit worried. Have you experienced anything like that Nova?

[flinx]

How would XFX be any different warranty wise than say, PowerColor? Replacing the cooler would void your warranty either way I'd think. I suppose you could fit the regular cooler back on and hope they don't notice that you swapped them if doing an RMA, but you have to be really careful that you don't leave any marks. I think a saw a mention in here somewhere of someone using a dremel to remove a screw, and that's not what I'd call a good idea.

[nova]

The "gray stripes" problem in that link is the first time ive heard of it. It's got me a bit worried. Have you experienced anything like that Nova?

Nope, not a single problem for me. Actually I've been trying a few things while I've been away.

First I tried undervolting the card to 1.038V (1.088 stock) Furmark maxed out at 58 degrees C and VRMs at 77 degrees.

I also tried ducting the card and moving the fan to the front of the Solo. Being cautious I tried 3DMark Vantage and GPU topped out at 77 degrees and VRMs where in the 100's. The position of the fan in proximity to the front of the case made it too noisy anyway.

Next I mounted a nexus fan to the sink and used the duct and that got me 63 degrees GPU and 83 on the VRMs. All running Furmark.

I'll try slowing down the fan and see where I end up. I also did a short try running at 1V and everything seemed stable but Furmark dropped out after a while. Note that I haven't downclocked the card to stock 725MHz while doing this.

I'll keep experimenting and will post results and a few pictures of my duct.

[mkygod]

These are my 5850 results with stock heatsink/fan running Furmark extreme for about 15 mins:

STOCK HSF - IDLE vs LOAD


I just recently installed Accelero S1 and VRM-4 on the 5850. The VRM-4 cannot fit with the baseplate installed, so I took off the baseplate and used the heatsinks that came with the Accelero.

I'm using an Antec P150 case with 12cm @ 1200rpm exhaust, 12 @ 1000rpm cpu, and 92mm @ 1300rpm intake fan. PSU is Seasonic x750:



PASSIVE Furmark results (15 minutes):

PASSIVE - IDLE vs LOAD


I didn't have time to do 1 hour tests, so these temps may be a little lower than it should be. But one thing I notice is that the card isn't running at full speed according to GPU-Z. (speed should be 725/1000mhz, but GPU-Z shows 400/900mhz at load). I don't know why this is, since the VRM temperatures seem okay. Could the VRM be throttling the GPU?

UPDATE : I managed to re run furmark and this time the video card was going at full core speed, and the temperatures went up to over 90 C, and that was when i stopped the test. Guess the temps i got before were too good to be true. I'm disapointted, even though people said it couldnt be done. Guess its time to slap a fan on this thing..

[Rushdie]

Well, as you've found out, its the volterra vregs that are the hardest ones to cool. Not just because they get hot, but also due to their small size and possible interference from caps/resistors close to them. The resistors* are about 0.5-1mm higher than the vregs and if you mount a flat heatsink onto the vregs the contact will be poor when using regular thermal grease and not pads. That's why the stock cooler is shipped with that thick gunk on the vregs and mems. Take a look at the bottom of any vreg heatsink, they're all flat, including that Thermalright. This is a common mistake, and a possible source for errors as well (the heatsink might short circuit the resistor).

Using the stock plate to cool the vregs and mems would work, but one must keep in mind the fact that this uses thermal 'gunk' rather than paste, so if you were to change it to lets say arctic silver, you might worsen the temperatures.

Since there was no reliable coolers for the 5850 when I bought it, I ended up building a custom one. I had an old HR-03 laying around (designed for the G80-based Geforce 8 ) which I thought would be nice to re-use in order to save some bucks. The final HSA looks like this:

(It's a clickable thumbnail)
It's built around a couple of old heatsinks I had laying around. The vregs are cooled by an old PentiumII-cooler with some ram coolers from an old Geforce3. The rest is: heatsink from an Geforce4, nforce2-chipset cooler, thermaltake mem heat sinks, copper plate, stainless steel plate, epoxy and arctic silver

The GPU heat sink is cooled by the same fans (Nexus) that cool the CPU. The rest of the 5850 is cooled by an intake fan (Nexus). The most important temperature, vregs, idle at about 29°C (84°F). Load about 50-55°C (130°F) depending on overclock.


* I actually think they are capacitors... not sure though.

[Tzupy]

@Rushdie: no offense meant, but your voltage regulation terminology may confuse some people. AFAIK:
- the Volterra chips are playing the same role as discrete Mosfets, so I presume they integrate 2-3 Mosfets, which makes them even hotter than discrete ones.
- if you happen to know that they work in a different manner, please do tell, I'd like to know what's new.
- what you call 'resistors* (or capacitors)' are probably the inductors (coils), one per phase and ~4x larger than the Volterra chips.

[Big Pimp Daddy]

...I just recently installed Accelero S1 and VRM-4 on the 5850.... temperatures went up to over 90 C.....

Poop, wish I'd read that before I ordered my VRM R4, the damn thing cost more than my Accelero S1. Was hoping to get my system back down to two fans again. I suppose a nexus blowing on it will still be quieter than the stock heatsink. Let us know what you try and what results you get. Cheers.

[Rushdie]

@Rushdie: no offense meant, but your voltage regulation terminology may confuse some people. AFAIK:
- the Volterra chips are playing the same role as discrete Mosfets, so I presume they integrate 2-3 Mosfets, which makes them even hotter than discrete ones.

None taken.

- if you happen to know that they work in a different manner, please do tell, I'd like to know what's new.

The power to the GPU is controlled by three VT1157SF PWMs(four on the 5870, and both the 5850 and the 5870 have a spare slot not occupied by a circuit), connected to a VT1165MF controller on the back side of the pcb (on the 5850, the 5870 have its controller located on the same side). Above them, there is two VT242WF that is used for the memory (vDD+vDDQ). The lone component located close to the crossfire connectors are, to my knowledge, used for the IMC (the 5870 have two of these AFAIK).
If my terminology fails its probably due to the fact that I call them what most people do, regulators / vregs / mosfets. That might be wrong (to be honest I'm not 100% sure about the said terminology), but sounds better than 'those hot black thingies'. And just call them 'phases' doesn't quite do it, now does it?

- what you call 'resistors* (or capacitors)' are probably the inductors (coils), one per phase and ~4x larger than the Volterra chips.

Well, here I know what I'm talking about. Since they are marked Cxxx (in contrast to resistors which are usually marked Rxxx), I would say they are capacitors. I've marked the ones that are troublesome on this picture to clarify:

The ones marked with red are actually 0.5-1mm higher than the components you intend to cool. That's why [whoever assembles the cards] uses the thick thermal pads to ensure thermal conductivity.

And if you take a look at this picture, the components are marked with Cxxx: http://www.pcgameshardware.com/screensh ... -PCB07.jpg

[Tzupy]

The inductors are the 3 components under the 'CPL2-3' marking, probably the same type as the other ones marked R23.
The components you market with red are IMO too small to be power capacitors, so they shouldn't require extra cooling.
Does the stock cooler provide cooling for these? If so, then they probably need extra cooling.
PS. the 'terminology issue' is mainly caused by the new Volterra chips, that you called 'vregs', and I'm not sure how to call. Typically each phase of a buck converter has one inductor and 2-3 Mosfets, my 4850 and 5770 use this discrete approach. But the 4870 and the new 5850 and 5870 use newer chips, with presumably the same functionality as multiple Mosfets, and maybe more efficient. But their smaller (total) size makes them even more difficult to cool properly.

[Rushdie]

The inductors are the 3 components under the 'CPL2-3' marking, probably the same type as the other ones marked R23.
The components you market with red are IMO too small to be power capacitors, so they shouldn't require extra cooling.
Does the stock cooler provide cooling for these? If so, then they probably need extra cooling.
PS. the 'terminology issue' is mainly caused by the new Volterra chips, that you called 'vregs', and I'm not sure how to call. Typically each phase of a buck converter has one inductor and 2-3 Mosfets, my 4850 and 5770 use this discrete approach. But the 4870 and the new 5850 and 5870 use newer chips, with presumably the same functionality as multiple Mosfets, and maybe more efficient. But their smaller (total) size makes them even more difficult to cool properly.

Ok now you need to start reading and stop assuming things. I've never said that they need cooling. I said that they interfere with a normal flat heatsink, making the thermal conductivity bad. Thats it

The reason why 4870 / 58xx / 5970 uses the Volterra setup is due to their ability to be software controlled, and offer a very high switching frequency (so I've heard).

[mkygod]

...I just recently installed Accelero S1 and VRM-4 on the 5850.... temperatures went up to over 90 C.....

Poop, wish I'd read that before I ordered my VRM R4, the damn thing cost more than my Accelero S1. Was hoping to get my system back down to two fans again. I suppose a nexus blowing on it will still be quieter than the stock heatsink. Let us know what you try and what results you get. Cheers.

The VRM R4 is great for what it does, which is too cool the VRMs. The VRM runs cooler than the stock setup, if you have the heatsink propped up within the vicinity of your tower CPU fan.

It is the Accelero S1 that is not efficient enough to cool WITHOUT a running fan cooling it. I have a 12cm 1000rpm fan running on it now and it is down to about 84 C running Furmark; although when running Crysis loops, i couldn't even get it to 70C. Furmark really is overkill, IMO.

What I'm going to do now is try to connect the 12 cm fan straight to the video card and have it so on idle, the fan runs slow or not at all, and during load, the fan will spin up to normal speed. I just have to purchase the right pwm fan and the appropriate fan adapter cable to connect the fan to the vidcard.

[Big Pimp Daddy]

...The VRM R4 is great for what it does, which is too cool the VRMs. The VRM runs cooler than the stock setup, if you have the heatsink propped up within the vicinity of your tower CPU fan.

It is the Accelero S1 that is not efficient enough to cool WITHOUT a running fan cooling it...

Ah right, now I see. I assumed you were talking about the VRM temps being too high, as that is what I understand most people have problems cooling, my mistake. Unfortunately I don't use a CPU cooler fan, just a case fan, also the VRM R4 will, according to my estimates, end up right behind the backside cooler from my IFX14, blocking the airflow somewhat. I will just have to wait until Royal Mail decide to deliver the VRM R4 and see what the state of play is. I guess hoping I could cool my CPU and graphics with just the one fan, like when I had a 3870, was a bit optimistic.
Will post results here in case anyone else is interested.

[mkygod]

Ok now you need to start reading and stop assuming things. I've never said that they need cooling. I said that they interfere with a normal flat heatsink, making the thermal conductivity bad. Thats it

The reason why 4870 / 58xx / 5970 uses the Volterra setup is due to their ability to be software controlled, and offer a very high switching frequency (so I've heard).

I think that for the 5850, you are incorrect in saying that they interfere with the regular heatsink. The regular heatsink is not actually flat. It has raised points (approx 1mm high) which are covered by the thermal pads. These are marked in this photo in orange outline:



As you can see in the pic, the raised points and thermal pads never actually touch the capacitors that you marked in red.

I've looked at my stock heatsink and can clearly outlines of where the VRMs and RAM have contact with the thermal padding. The RAM has thick thermal padding, but 3 VRMs have very thin padding where you can see clean imprints made by these chips..\ There is also another unit (maybe another VRM) on the far top left of the card next to the first RAM chip. This chip was in contact with the heatsink and had some thermal gunk on it. This is a part that I assum needs cooling as well and so I stuck the VRM sink that came with the Accelero on that point.

[Rushdie]

Ok now you need to start reading and stop assuming things. I've never said that they need cooling. I said that they interfere with a normal flat heatsink, making the thermal conductivity bad. Thats it

The reason why 4870 / 58xx / 5970 uses the Volterra setup is due to their ability to be software controlled, and offer a very high switching frequency (so I've heard).

I think that for the 5850, you are incorrect in saying that they interfere with the regular heatsink. The regular heatsink is not actually flat. It has raised points (approx 1mm high) which are covered by the thermal pads. These are marked in this photo in orange outline:



As you can see in the pic, the raised points and thermal pads never actually touch the capacitors that you marked in red.

I've looked at my stock heatsink and can clearly outlines of where the VRMs and RAM have contact with the thermal padding. The RAM has thick thermal padding, but 3 VRMs have very thin padding where you can see clean imprints made by these chips..\ There is also another unit (maybe another VRM) on the far top left of the card next to the first RAM chip. This chip was in contact with the heatsink and had some thermal gunk on it. This is a part that I assum needs cooling as well and so I stuck the VRM sink that came with the Accelero on that point.

(Will this discussion never end?)
It might be the fact that I'm unable to explain the 'problem' properly. But since you have disassembled the stock HSA, you know how it's designed. Now take a close look at the thermal pad for the vregs / VRM:s. You will find imprints in the actual thermal pad made from the capacitors that I pointed out. That said, if you remove the pad, and replace it with something else, the heatsink will make contact with this capacitor(s). This also applies to any 3rd party heatsinks you might be using. This is not uncommon, and it not only a 'problem' for the 5xxx-cards. A lot of computer hardware uses this setup. Laptops might have thermal grease for the cpu, but a thermal pad for the gpu (while still using the same HSA). HP uses this on some of its designs, so do IBM. If you replace the pad with grease, it will not make a proper contact. This also applies to the reference design cooler mounted on Geforce 8800. Grease for the gpu, pads for the memory modules. If you replace the pads with grease, they will not make a proper contact, if any at all.

What I'm trying to accomplish here is to inform people that you cant just mount a flat surface vreg / VRM-cooler directly over the VRM:s, without taking the other components into consideration. Pads are the easy way (that's why it's used by the manufacturer), but if you want to use a thin layer of thermal grease then you must make sure that the heatsink doesn't interfere with something else.

(the lone capacitor / resistor I marked out on the picture that's located between the memory modules are just for information that might be useful for anyone building waterblocks or custom air coolers).

[flinx]

It might be the fact that I'm unable to explain the 'problem' properly. But since you have disassembled the stock HSA, you know how it's designed. Now take a close look at the thermal pad for the vregs / VRM:s. You will find imprints in the actual thermal pad made from the capacitors that I pointed out.

Hmm.. let's see *checks his 5870 which has the same layout* nope. Not only is there no imprint from them, but the area with the thermal pads is raised up from the head spreader plate so there's more room.

Here, look for yourself:

http://picasaweb.google.com/lh/photo/pu ... directlink



I actually used the flat VRM heatsinks that came with my Twin Turbo Pro, and had no issues with them coming close to being in contact with the components you mention.

[mkygod]

What I'm trying to accomplish here is to inform people that you cant just mount a flat surface vreg / VRM-cooler directly over the VRM:s, without taking the other components into consideration. Pads are the easy way (that's why it's used by the manufacturer), but if you want to use a thin layer of thermal grease then you must make sure that the heatsink doesn't interfere with something else.

The VRM-4 I am using now has one bar that vertically goes down the VRMs, like the stock heatsink, and overhang part of the heatsink is raised so as to not touch the caps or other components. Installation is foolproof with no possibility of it touching the capacitors.

I can agree that the capacitor thing is something you should look out for, but this is clearly not an issue with the stock heatsink, the Accelero S1, or Thermalright VRM-4.

As far as I know, most coolers for these cards are only for the GPU, plus stickon ram/vrm sinks to complete it. There is no single flatbased heatsink solution where this could become a problem. I think it can only become a problem if someone uses a heatsink in the way it was not meant to be used, such as using ramsinks on the VRMs, which are too wide for that application. VRM heatsinks tend to be pretty skinny, so they don't hit those capacitors if installed properly. I just think that you would have to be very incompetant or incredibily sloppy to mount these type of sinks in a way that it would hit the capacitors.

[Big Pimp Daddy]

In case anyone was wondering, I upgraded the stock cooler on my 5850 to an Accelero S1 + VRM R4, full details in gallery: viewtopic.php?p=497701#497701

Long story short; with no fans at idle it's cooler than the stock cooler, at load with no fans - fiery death. Slap a low speed 120mm fan on and it's fine at load, as well as obviously being quieter than the stock cooler.

[Scott J]

In case anyone was wondering, I upgraded the stock cooler on my 5850 to an Accelero S1 + VRM R4, full details in gallery: viewtopic.php?p=497701#497701

Long story short; with no fans at idle it's cooler than the stock cooler, at load with no fans - fiery death. Slap a low speed 120mm fan on and it's fine at load, as well as obviously being quieter than the stock cooler.

Thanks very much for posting your results, looks like it worked out really well. I just received Thermalright's VRM-R4 and VRM-R3 today. I needed both in order to figure out how to make them work together in a 'crossfire' set-up with two 5850s, although I just have the single 5850 at the moment.

I read somewhere (forgot to mark the link) that Thermalright said the R-3 & R-4 won't work together in 'crossfire', but I think that depends on:

A) the PCI-e slot layout/position of the motherboard in question, i.e., the distance between the slots used for the 1st and 2nd video card, and

B) one's willingness to improvise. I think the R-3 is going to run into the bottom of the heatpipes of the R-4 in a 'crossfire' set-up, but not by much, at least not with the EVGA 760 Classified MB. I haven't had a chance to measure carefully yet, but it looks like I'll only need to cut a few fins out of the top of the R-3 in order to create "slots" just wide enough for the heatpipes of the R-4 to fit in between, allowing the R-3 to lay "flat" in the natural position it takes when attached to a video card. Either that, or heat it up and bend the heatpipes of the R-3 upward slightly, so that the radiator goes over the top of the R-4's heatpipes.

If the second video card (in PCI-e slot 3 on the 760 Classified) was a 5870 instead of a 5850, the longer 5870 may position the VRMs lower, which in turn would position the VRM-R3 lower, which in turn may remove the need for any modifications to the R-3 in order to make it work in a dual-card set-up.

Lots to figure out...

[mkygod]

In case anyone was wondering, I upgraded the stock cooler on my 5850 to an Accelero S1 + VRM R4, full details in gallery: viewtopic.php?p=497701#497701

Long story short; with no fans at idle it's cooler than the stock cooler, at load with no fans - fiery death. Slap a low speed 120mm fan on and it's fine at load, as well as obviously being quieter than the stock cooler.

I can confirm those findings. I had the exact same results with my S1 and VRM-4

[nova]

It seems we're all getting about the same results. I knew from the start that running this card passive would be a stretch. However even passive cooling needs airflow and a really slow fan can do wonders.

I promised I would post pictures and results of my ducting. One thing I've noticed is while the VRMs are the hardest to cool, moving the fan closer to the VRMs in fact gave worse results for me and the GPU got a lot hotter.

Since last time I've reduced my fan speeds even further and now my CPU fan runs at 680 RPM, case fan 645 RPM and the Nexus on the S1 is running at 605 RPM. Temperature when idle is 33 C, I'm not sure about furmark load in this configuration but I can say that while gaming the GPU has never reached more than 67 C.

I'm a happy camper and my computer is actually quieter than when I was running the 8800GT passive.

I've been asked if I tried to run a PWM fan connected to the card controlled by MSI afterburner, however I'm with MikeC on prefering fixed fan speeds at all times.

Now without further ado, pics of my current setup. The duct for the CPU is still in place after I tried removing the fan on the Thermalright, I guess I'm just too lazy

[grandpatzer]

One thing I'm wondering is with the high VRM temps, the card might attempt to throttle the GPU. More info here:

http://www.anandtech.com/video/showdoc.aspx?i=3643&p=11

EDIT: More info - looks like 120C is the point where the VRMs start throttling on the 5970, so it's probably the same on the 5870 and 5850:

http://www.anandtech.com/weblog/showpost.aspx?i=657

Do a load test and look at the temps of the VRMs and the clock speed of the card, and you should be able to verify it.

I tried OCCT while running GPU-Z in the background logging everything and the card never throttles down. The VRMs reaches a scorching 125 degrees C. OCCT and furmark really is the worst case scenario though and if the card manages that you can rest easy it will handle any game you throw at it.

Since I bought this card for gaming I won't be running these "power viruses" as AMD calls them after I initially proof the setup. I'll actually start using the card as intended and be happy with it. You'll never see these kind of temperatures in real life gaming. So next I'll go enjoy myself and log temperatures while gaming instead.

I was thinking about cooling ATI 5850 with a 550rpm 120mm slipstream but a 125c temperature is unacceptabel

Can anyone confirm if the 125c is with or without a fan, anyone else use S1 R2 and ATI original VRM cooler?

[zyrobs]

But one thing I notice is that the card isn't running at full speed according to GPU-Z. (speed should be 725/1000mhz, but GPU-Z shows 400/900mhz at load). I don't know why this is, since the VRM temperatures seem okay. Could the VRM be throttling the GPU?

I had that bug too, pretty annoying.
http://www.gamingheaven.net/showthread. ... ost1346382

I've yet to try the DXVA fix mentioned there. But, using the AMD GPU Tool with the -restore parameter will reset the card clocks, which will get your card out of the 400/900 phase. It also deletes overclock settings, though.

The funny thing is, even at 400/900, the 5850 can run any game I try maxed out in 1680x1050. As long as I turn off AA. Crazy powerful card.

[mkk]

The funny thing is, even at 400/900, the 5850 can run any game I try maxed out in 1680x1050. As long as I turn off AA. Crazy powerful card.

My initial reaction was the same. Was playing modern warfare 2 and only occasionally encountered a dip in framerate at 1920. Wasn't until playing Dirt 2 and Mass Effect 2 that it became a problem. Now with DXVA off for that Microsoft codec I'm problem free while waiting for it to be fixed. Having noticed that Media Center is a little bit snappier in non-accelerated mode for TV I'll keep that setting in the future as well.

[robbull]

The vapor-x toxic version (slight factory overclock) will run hotter than a normal vapor-x card, right?
Would that make the fan spin faster and cause more noise?

If so,
Could a vapor-x toxic be clocked down to match the temps and noise of a normal vapor-x card?

I'm asking 'cause I can't seem to find a vapor-x 5850 available in sweden, only the toxic version. And the reason I'm looking at the vapor-x cards is because of their improved noise level.

[notslow]

Does anyone here have the HD5850 Vapor-X? I'm curious if it is really as quiet as it is supposed to be. I'd like to upgrade to a 5850 card, but don't want to have to mod it after spending 300+ on just the card.

[JohnMK]

Does anyone here have the HD5850 Vapor-X? I'm curious if it is really as quiet as it is supposed to be. I'd like to upgrade to a 5850 card, but don't want to have to mod it after spending 300+ on just the card.

I have the Sapphire Radeon HD 5850 Toxic, which is pretty similar to the Vapor-X. At idle, it satisfies me completely. I can't honestly pick it out from the ambient home/computer noise I'm normally exposed to. Under load, it does ramp up a bit but nothing I have any problem with.

I love this card. I've modded the BIOS so that the standard clocks on load are 833/1150, instead of the default Sapphire 765/1125. Remember, the 5850 is normally 725/1000, so my card was factory overclocked. This gives me performance that's slightly above mid-way between the 5850 and 5870. I consider my card a 5860. Games are buttery, silky smooth. I know how much power is contained within this video card, and it impresses me to see this kind of performance at such a non-offensive level of noise.

Nvidia have really fumbled this product cycle. ATI, however, -- my hat off to them.

-John

[notslow]

Hi John,
It looks like the Toxic is an overclocked Vapor-X. I could be wrong, but they look exactly the same. I wonder if the Vapor would be quieter, maybe it runs the fan slower since it isn't overclocked. I found a review that showed the newer Sapphire 5850s to be quite a bit lower dbl than the reference 5850 cards. I wonder how true this is? I'm considering the Vapor X 5850 or the Sapphire 5850 Cypress Pro, trying to decide if the vapor is worth the 40.00 premium.