ATI Silencer 5 Rev.2 BEWARE NOISE COMES BACK!

[nutball]

But, my card takes around 150W... and let's say that 75-100W goes to heat.

a) What makes you think that your card alone is consuming 150W?
b) Where do you think the other Watts are going? Unless your card is maybe creating matter out of thin air, then it's pretty simple: electric power in = heat out.

[warriorpoet]

"just warriorpoet" here

I've had the cooler now for about 3.5 months (NV Silencer 5 Rev.3) and it has yet to produce any sort of odd noise. However, given the ovewhelmingly negative response to AC Silencer fan noise in the Rev. 1 and 2 models caution is appropriate.

How many others here have the Rev. 3? Are we just projecting experiences with Revs. 1 and 2 to 3, or has anyone, besides "just me" tried it for an extended length of time?

[JVM]

"just warriorpoet" here

I've had the cooler now for about 3.5 months (NV Silencer 5 Rev.3) and it has yet to produce any sort of odd noise. However, given the ovewhelmingly negative response to AC Silencer fan noise in the Rev. 1 and 2 models caution is appropriate.

How many others here have the Rev. 3? Are we just projecting experiences with Revs. 1 and 2 to 3, or has anyone, besides "just me" tried it for an extended length of time?

Are you running Rev 3 at stock speed of 2000 rpm?

[thetoad30]

a) What makes you think that your card alone is consuming 150W?
b) Where do you think the other Watts are going? Unless your card is maybe creating matter out of thin air, then it's pretty simple: electric power in = heat out.

Nutball,

I crudely compared my UPS power settings at idle load for my GPU and then at full load using ATI TOOL. ATI TOOL is known as one of the most extensive loader for the GPU, so I know the card is at full tilt.

Using my UPS load meter I saw that the card at full load took around 150-175W of power. There was no CPU load that I could see from task manager in XP, and my CPU temps did not rise, so I crudely assumed negligable power loss to the CPU.

In terms of power in vs. power out:

I was stating for the GPU heat output. As you are well aware, some of that power gets lost in the PCI bus, some gets lost in all the wiring to and from the card, some of it goes to the VGA/DVI terminal, and some goes to the mem chips.

The VGA card is not a direct resistance. There is some reactance and such in the card. Because of this, not all the electrical energy is turned into heat, if I recall correctly. I might not be stating this correctly, so if I'm not please do correct me for my knowledge and for the knowledge of the other forum users.

Anyway...

I assumed that the GPU chip put out around 100W of heat from random readings I did on the internet.

Here

You can see that they have a total of a little more than 100W for the whole card at full tilt. I have my card overclocked, and bios flashed, so I know that extra 50W I'm seeing is probably real and not due to error. To be safe, though, I just take the 100W for the GPU and call it even. I'm sure it's nowhere near perfect, and I don't claim it to be, but to me it seems "close enough" in terms of relative power and cooling potential.

[warriorpoet]

Are you running Rev 3 at stock speed of 2000 rpm?

No. Read my review for more specifics, I don't have more time than that ATM, sorry.

[thetoad30]

Do you think I should consider the ACS 5 Rev 3 or is that noisy as well?

I would honestly say that the AC5R3 would be a choice to consider.

It's nowhere near perfect, but I find that having the fan plugged into the card instead of having a set 5 or 12V that the Zalman comes equipped with allows the fan to automatically be adjustable to graphics temps without any extra equipment or purchases.

Be forewarned, however, that as of right now I am questioning the warranty validity on the AC coolers. I have not been successful in reaching the company despite 2 e-mails from their website specifically asking for warranty information as NewEgg won't accept the RMA since it's past 30 days. Don't get me started on NewEgg, either....

[Tephras]

Power draw measurements of VGAs done by comparing total system consumption at idle and at load can be used in relation to other VGAs power draw if they are measured under the same conditions to find out how different video cards stack up to each other, but it's not a very useful way to determine the actual power draw of a VGA, out of context the results of the power consumption tests at gamepc.com should not be relied upon.

[thetoad30]

Power draw measurements of VGAs done by comparing total system consumption at idle and at load can be used in relation to other VGAs power draw if they are measured under the same conditions to find out how different video cards stack up to each other, but it's not a very useful way to determine the actual power draw of a VGA, out of context the results of the power consumption tests at gamepc.com should not be relied upon.

I was merely using this site as a quick google.com example.

Many other sites have shown 100+W as a good measurement to how much the X850XTPE uses. If there are a lot of places showing the same measurements, you can be pretty sure that it isn't because of coincidence.

And yes, using total system power is still a perfectly fine way of seeing how much wattage a card uses from the AC line. In terms of DC power draw, you would need a much better way to do it. So far, I haven't seen a meter that can plug into both the PCI slot and the PCIE power slot to measure total card draw.

That being said, if you have an active PFC PSU that is close to .9 or .99 PFC, then you can be reasonably sure that the power draw at the AC line is close to the power draw inside the DC computer.

These are rough estimates here. They are pretty close in terms of looking at a HSF for potential cooling power, and are more than accurate enough to do simple comparisons.

[Tephras]

Maybe if more review sites would put some effort in putting together their own measurement devices we could have more VGA power consumption measurements as those provided by X-bit labs.

[thetoad30]

Maybe if more review sites would put some effort in putting together their own measurement devices we could have more VGA power consumption measurements as those provided by X-bit labs.

I'm sorry, I just don't buy words. In no way am I saying X-bit Labs is misleading or not to be trusted, but I just don't think that their method of testing is complete. I'm not saying mine is either.

There are MANY connectors on the PCI-E slot and in order to test all of these leads would tend to be cumbersome and not easy at all.

Remember that the PCI-E slot, the last time I checked, allowed up to 75W of power. Based off of that chart, the PSU would only need to provide up to an additional 40W at best for a high-end card, mainly the new x1xxx series. Rubbish.

My card is required to have the connector, even in non overclocked, standard 12 pipeline mode. It's equivilent to an X800XL in terms of power usage in that mode.

My PSU is supposed to allow a whole 75W per card (150W in SLI) per card, and one company was asking to up that total power to 150W per PSU connector.

For X-Bit's power usage to be correct, I don't even see the need for a power connector until now, much less a year ago when it first came out. Real world, however, is much different.

As for the external connector, that's easy enough. Just measure the wire connections.

I would need to see a picture of their test bed to actually support that they are doing it the best way possible. Again, not saying they couldn't do it, but just want to make sure they are actually measuring only that PCI slot, and all available power going into that slot.

[GHz]

Oscar3d: No offense, but you really shouldn't be drawing such far and wide conclusions about a product you've only had one sample of. I've been through a dozen Silencers and VF700CUs, so as you can see I don't pass judgement based on 1 sample. I know it's not practical for the majority of people to test multiple samples, but you shouldn't go out of your way to trash a product with only 2 samples (1 of each).

In my experience, the NV5 Rev. 2 (don't have the Rev. 3 so I can't say if this is fixed) gets noisy over time (a few months). I'm guessing that AC's "ceramic" bearing is just a fancy sleeve, and probably just wears out fast due to the intense heat of a high-end video card. In comparison, the dual-ball bearing fan on the VF700CU starts out noisier than the AC Silencer's fan, but the noise stays consistently low. I've had no trouble at all keeping 6800GTs well-cooled with only 5v. At least with the Zalman, you can replace the fan if it bothers you... no such luck with the AC.

[Tephras]

I would need to see a picture of their test bed to actually support that they are doing it the best way possible. Again, not saying they couldn't do it, but just want to make sure they are actually measuring only that PCI slot, and all available power going into that slot.

The VGA power tests done by X-bit labs is nothing new to this site, they have often been used as a reference for over a year, I don't see why these measurements all of a sudden should be regarded as something fishy. The test method for PCI Express cards are the same as the method used for AGP cards.

[depravedone]

I'm very near replacing the stock fan on my VF700Cu with a Panasonic FBA08T12L. It's a 15mm thick 80mm fan that appears to be a good candidate for the swap.

I'll post pics once it's installed!


Panasonic Panaflo FBA08T12L
Volts: 12
Operating Voltage: 7 - 13.8
Rated Current: 79ma
Rated Input: 0.95 Watts
Nominal Speed: 2000rpm
Maximum Airflow: 21.9cfm
Noise: 23dba

[thetoad30]

The VGA power tests done by X-bit labs is nothing new to this site, they have often been used as a reference for over a year, I don't see why these measurements all of a sudden should be regarded as something fishy. The test method for PCI Express cards are the same as the method used for AGP cards.

All I'm saying is that I personally don't think it's taking into account all the current and power. There's a reason that a so called 71W card needs the second connector: It's gotta be using MORE than 75W. My calculations, however crude they may be, are showing double.

All measurements are relative, and if you aren't careful what you are doing you can actually influence the measurement. Hell, taking a measurement influences the measurement. It's a law of physics.

Just because some website that I like uses some other website's measurements doesn't mean I have to accept them verbatim. I can think for myself and will not take things that are just spoon-fed to me. If something doesn't seem right, then I will question it and try to figure it out. Some things I can admit I'm way over my head, and this is probably one of them, but I can tell from experience and my measurements that there is no way my card is only pumping 71 W of power. Regardless of what their measurements say.

EDIT: I just re-read their AGP method, and I personally think that's not the way they should do it. They are introducing more resistance into the circuit, and taking mathematical measurements. I don't think that's the way to do it. You need to put the ammeter inbetween the lines for the best result. Period. I know you can use Ohm's law and it "should" be the same, but you are introducing errors with the resistors and such. I just think it is not the best way to do it.

[gbohn]

J
Even so, you will still need to run the VF700CU at high for a high-end graphics card. I have an X800GTO^2 flashed to 16 pipes and OC'ed to XTPE speeds, and without the Zalman on high the card overheats.
...

I think this also varies with how good your case airflow is, and what your ambient temps are as well.

I have a Sapphire X800 GTO2 that I modded to 16 pipes, upped the frequency to 490 core/560 memory, and replaced the heatsink with a Zalman VF-700Cu.

I ran this in two different cases (with the same ambient temps).

In my 'cooler' case, I can run at full 3D load with the Zalman set to 54% speed (in ATI Tool with the Zalman using the video cards fan header) and get a core temp of only 65 degrees Celsius (even after prolonged testing).

The case with better airflow (120 MM Nexus case Fans, punched out grills) runs the video core at least 10 Celsius degrees cooler (at full 3D load) than when in the 'warmer' case (for the same fan speed).

[thetoad30]

I think this also varies with how good your case airflow is, and what your ambient temps are as well.

I have a Sapphire X800 GTO2 that I modded to 16 pipes, upped the frequency to 490 core/560 memory, and replaced the heatsink with a Zalman VF-700Cu.

I have tested this, both with my case side panel on, and off. Both times the card did overheat, even when I was only on regular core/memory clocks of 400/490. Not sure what the difference is between you and I, but an open case should have prevented this if you can get 54%.

As for overclocking, I am using almost 90 MHz more on the core than you, and 40 MHz more on the memory. This more than requires the fan to be set on high.

I have the Antec P180 case FWIW.

The only difference I saw between the case panel on and off was that with the case panel on, the heat of the video card stayed trapped underneath it. The Zalman fan just blew the air out the side of the card, and the heat could not rise above it. I'm considering the V1 from ThermalRight, but until I know whether or not it will fit, much less the fact that if I turn the fan to blow the correct direction - up - it might blow the heat right into my Scythe Ninja.

Ahh the joys of airflow management.

EDIT: When I test temperatures, I use ATI TOOL and the 3D Window/Scan for Artifacts as this is, to the best of my knowledge, the most intense 3D application.

[Tephras]

The VGA power tests done by X-bit labs is nothing new to this site, they have often been used as a reference for over a year, I don't see why these measurements all of a sudden should be regarded as something fishy. The test method for PCI Express cards are the same as the method used for AGP cards.

All I'm saying is that I personally don't think it's taking into account all the current and power. There's a reason that a so called 71W card needs the second connector: It's gotta be using MORE than 75W. My calculations, however crude they may be, are showing double.

All measurements are relative, and if you aren't careful what you are doing you can actually influence the measurement. Hell, taking a measurement influences the measurement. It's a law of physics.

Just because some website that I like uses some other website's measurements doesn't mean I have to accept them verbatim. I can think for myself and will not take things that are just spoon-fed to me. If something doesn't seem right, then I will question it and try to figure it out. Some things I can admit I'm way over my head, and this is probably one of them, but I can tell from experience and my measurements that there is no way my card is only pumping 71 W of power. Regardless of what their measurements say.

EDIT: I just re-read their AGP method, and I personally think that's not the way they should do it. They are introducing more resistance into the circuit, and taking mathematical measurements. I don't think that's the way to do it. You need to put the ammeter inbetween the lines for the best result. Period. I know you can use Ohm's law and it "should" be the same, but you are introducing errors with the resistors and such. I just think it is not the best way to do it.

I suppose that if the power draw of a video card is on the borderline of what the slot can provide an external power connector will be mandatory in the specifiction from the GPU manufacturer because they feel that there is a need for a safety margin so they can disclaim responsibility for any failure. Other causes for supplementary power sources are mentioned in X-bit labs AGP article; "Even if the peak power consumption of a device doesn’t exceed [available power through the slot], but approaches this point, additional power is required – long exploitation under a strain never made any computer component live longer.", "The two power connectors on the GeForce 6800 Ultra is not a consequence of a crazy appetite, but a desire of the manufacturer to ensure the stability of power supply by dividing the currents into two connectors." - the latter could also be said about those VGAs supplied with power both through the slot and through a singel external power connector.

Maybe there's a variation in power draw between video cards even though they use the same GPU? There is no mention in the X-bit labs articles on the number of tested cards so I assume they only tested one sample of each VGA, they should test more cards to see if the results are consistent and to give their measurements a higher tenability. So far, however, I haven't seen any review site going to such great length in measuring VGA power consumtion as the X-bit labs site does, review sites that use the total-system-power-draw method are put to shame, while those sites are doing guesstimates X-bit labs are trying to do direct measurements of the actual VGA power consumption.

The reason as to why I mentioned that X-bit labs measurements has been referenced in this forum for more than a year is because if there is something incorrect in their methods I would have expected someone to raise their voice during that period since there are som many SPCR forum members with high technical/electrical knowledge (and I am not one of those).

[oscar3d]

No offense, but could you please move the conversation about VGA Power Consumption to some place else?

It's good to know some details, but from a Silencing VGA thread, this had become another issue that I suggest you detail in another thread, to avoid misleading the whole point about the Arctic Silencers.

[Tephras]

You're right, sorry for sliding away from the thread topic, I wont persist.

[thetoad30]

No offense, but could you please move the conversation about VGA Power Consumption to some place else?

It's good to know some details, but from a Silencing VGA thread, this had become another issue that I suggest you detail in another thread, to avoid misleading the whole point about the Arctic Silencers.

While I understand and respect your request, I do have to say that power consumption for ANY component in a computer is the whole reason that SPCR is here!

More heat output = harder to silence a machine. The more heat you have, the more effiecient your heat removal has to be in order to be effective at low/no airflow.

I would actually like to see tests done on future HSF to see what the theoretical max's are for heat input and removal, much like SPCR does with their test beds and such.

I currently own the AC 5 for the ATI... and I have had clicking since the beginning. It's not a bad click, but enough to hear in low noise levels.

Hopefully it won't persist or get worse!

[gbohn]

I think this also varies with how good your case airflow is, and what your ambient temps are as well.

I have a Sapphire X800 GTO2 that I modded to 16 pipes, upped the frequency to 490 core/560 memory, and replaced the heatsink with a Zalman VF-700Cu.

I have tested this, both with my case side panel on, and off. Both times the card did overheat, even when I was only on regular core/memory clocks of 400/490. Not sure what the difference is between you and I, but an open case should have prevented this if you can get 54%.

...

EDIT: When I test temperatures, I use ATI TOOL and the 3D Window/Scan for Artifacts as this is, to the best of my knowledge, the most intense 3D application.

My point was that your setup/environment can have a lot to do with how cool it runs, and therefor if it's adequate. In my case, in a 66 Degree F environment I got either 65 C or 75 C core based on which case (of the two cases I've tried) I used (at 54% fan setting).

In my cool (tower) case, I have a constant low-impedence flow of air from in at the front bottom (through a 120 MM Nexus with the case grill removed), up past the video card, and out at the back through a 120 MM Nexus (removed case grill) and the 120 MM fan in the Seasonic S12-500.

The same card in the less breezy case setup ran 10 degrees C hotter in the same ambient temp. (I presume because the hot air wasn't removed as effectively causing more of the same hot air to be recycled through the heatsink.)

It's also possible that an open case can run hotter than a closed one if the closed one causes an additional air-flow past the heatsink that's not there with the side off. (I'm not saying it has to run cooler, just that it's possible given the right setup).

I also use ATITool and the 3D artifact tester since, just as you have seen, this heats my cards core and memory more than any other test I have tried so far (rtidribl, 3D games, etc.)

[thetoad30]

My point was that your setup/environment can have a lot to do with how cool it runs, and therefor if it's adequate.
...
The same card in the less breezy case setup ran 10 degrees C hotter in the same ambient temp. (I presume because the hot air wasn't removed as effectively causing more of the same hot air to be recycled through the heatsink.)

I completely agree with that.