Are Reference Cards OK?

[timmay]

I dare say you've never used the Titan or 780 shroud.

I used it for about 2 hours while verifying that the card was working, before replacing it with a watercooler. I always buy reference to make sure the waterblocks fit.

It uses expensive and high cost parts for the BOM, and the results show - it is, by far, BY FAR, the best and quietest reference shroud ever designed.

That may well be.

And it is quiet at stock and 2000rpm.

We must have a different definition of "quiet". For example: I think any GOOD 120mm case fan (Scythe, Noiseblocker, Noctua) above 800rpm is too loud.

In fact, I remember PCPerspective did a comparison of the 290x at 2000rpm and the 780 reference shroud at 3500rpm. The 3500rpm 780 shroud was quieter

Do you have a link for that?

 

[xorbe]

A single dB reading is less than useless without a recording, as to hear the frequency shape of the noise.

I would prefer to hear some moderate air whooshing sound, rather than a quiet rrrrrrrrrrrrrrrrrrrrrrrrr. You know what I'm talking about!

 

[Unknown-One]

We must have a different definition of "quiet". For example: I think any GOOD 120mm case fan (Scythe, Noiseblocker, Noctua) above 800rpm is too loud.

Finally, someone who gets it...

I've been doing pretty well with Arctic F12 fans. Swapped out all of the 120mm fans in my case for F12's and have totally satisfied with them + temp-based speed control keeping them under control.

 

[biggles]

One takeaway of this thread is the reference cooler on the gtx 780 is one of the best, or perhaps the very best, of the reference coolers out there. Which makes me happy because I just bought a system with the EVGA superclocked gtx 780, which uses the reference cooler.

 

[limitedaccess]

How does the following however affect the perception of coolers used on video cards? Not just in terms of reference vs. custom coolers but are even custom vs custom cooler comparisons.

1) The vast majority of reviews are tested on an open bench. While allowing more control of variables it also doesn't replicate the typical usage scenario of the end user.

Can we actually assume that all coolers (even among nonblower custom designs) will have the same delta change in temperature once moved into a case? I'm not so sure about that . I do recall a fairly recent set of data from Tom's (yes I realize how people feel about the source, which is why we need more sources of test data) that showed how several different coolers had varying deltas in performance in case vs open bench.

2) As related to the above sound measurements are then typically taken without accounting for the other system noise (case fans in this scenario, but maybe also the CPU or other fans) as the video card would have an effect (greater effect) on its surroundings in an encased environment (case) vs open environment.

3) Temperature measurements are mostly limited to the GPU die only. While some reviewers have moved to including more data on other component temperatures the sources for this is still relatively limited due to the cost of equipment required and added complexity of measurements.

4) Numbers are not necessarily given in context and/or understood by the reader, although this isn't really completely an issue with reviews but also the reader as well. This refers to understanding of what dBA measurements actually mean and what they don't encompass (for example tonal differences). However even temperature data might not be fully understood.

 

[xoleras]

1) The vast majority of reviews are tested on an open bench. While allowing more control of variables it also doesn't replicate the typical usage scenario of the end user.

So much this. SO MUCH THIS.

An open test bench is not even remotely close to the real world. I've seen countless reviewers shrug blowers off , but I have to question their intelligence because frankly, blowers have their place. I'm not saying custom cards aren't great, they ARE, but they must be used in specific situations. The one website that comes to mind is guru3d. All of their tests are open test bench. I like guru3d, but I don't like their testing on an open test bench.

As an example, I used 680 msi lightnings in SLI. I cannot even begin to describe the difficulty in getting it work work fully stable for hours at a time, and this is in a FULL TOWER EATX CASE with 9 case fans. I know what the fuck i'm doing with air flow. And it was just a pain.

Now you can make the above work if you have x79/x99 which allows 2 slot spacing between cards, but I had 1 slot spacing between cards and what would happen is this: the upper card would slowly cook over time and the GPU temps eventually, after 2 hours of gaming, would get so hot that it would cause TDRs. In the end, I got it to work. Could not overclock (whats the point of the lightning there???) This was a CM Cosmos II case which, if you google image it, is fucking HUGE. And I had proper airflow. So don't give me the bullshit that custom cards are fine for SLI. Unless you have 2-3 spacing between cards, it's going to be a PITA. Unless you have weak GPUs which don't kick out heat, you can probably get cheesy shit like GTX 660s or R9-270s to work. But not 250W TDP cards. For situations like this, blowers are SLI it and forget it. Or CF and forget it. You don't have to deal with stupid shit or hassle with heat.

I don't care what anyone says, when you have 2 cards at 100% utilization kicking out 500-600W TDP (more if OC'ing) of heat inside your case, good luck getting any number of case fans to get that out. Case fans help for sure, but the upper card will still roast over time. 1 hour, 2 hours later, crashes from high temps.

Guess what? That shit doesn't happen on a test bench. I guess reviewers somehow think that a test bench is even remotely applicable to real world, but the fact of the matter is, it isnt. For SLI or CF custom cards you probably want x79 since it has more PCIE lanes which allows WAY MORE spacing between custom SLI/CF cards. 1 space is typically all that is allowed on z97/z87 motherboards unless you buy a 450$ PLX quad SLI board. And not many people do that> If you're buying a 450$ board, why not just get fucking X79/.X99? Doesn't make sense.

Anyway, I wish reviewers would test these things in an actual closed typical real world case to replicate real world situations. Test bench is bullshit. Now, I like custom cards. But reference cards ARE absolutely needed in some situations: SLI (for most people), CF, SFF, I could go on. And the fact of the matter is, the 780/titan shroud is outstanding which helps.

 

[zerodamage]

the 880 will not be a pure reference cooler, its going to be a customized new cooler.

Reference of course means blower style cooler 99% of time AND stock clocked.. Blower style coolers work well for multi-card setups cause each card vents its own hot air instead of exhausting into the case.

Problem with reference blower style they don't tend to cool very well, get dirty easier, tend to be quite loud though at the same time the reference coolers also tend to cool the entire card gpucore/memory/vreg well enough sometimes even better then many "non" reference coolers.

So better for temps in case yes if it vents properly, but room heats up so it sucks in hot air that much sooner cause its not cooling as well as fast its a double edged sword.

One "mod" I have seen folks do with decent success is to snip the back vent where the hot air blows out to make it less constricted.

The exception to the rule are the 290 and 290x cards. The reference coolers are just terrible IMO and I waited a long time for 3rd party solutions before I purchased my 290 and even then I waited for a deal before spending the money on one.

 

[RCantw3ll]

My GTX580 and 680 reference coolers were/are absolutely fantastic. Just had to set a manual profile for the fan, and temps max in the 70s with a mild OC and fan noise that isnt any more audible than my 120mm 1500rpm case fans. They arent silent, but they cannot be picked out of a crowd.

That being said, this has only been the case since the 500 series released. My 470 was incredibly loud, but manageable with a fan profile to a degree. My 5870 was also reference, and im not certain, but i do not recall it being terrible loud at the time.

 

[ManofGod]

The exception to the rule are the 290 and 290x cards. The reference coolers are just terrible IMO and I waited a long time for 3rd party solutions before I purchased my 290 and even then I waited for a deal before spending the money on one.

That is your opinion and that is fine. However, I have not found that to be the case at all in my direct experience. I own an XFX R9 290 that I ran for 7 months before I flashed it to an R9 290x. In both cases, the fan noise was minimal at idle and load. Of course, I am not running my card on an open air test bench.

Also, I listening to some of the posted videos of the "fan noise". It was clear that there were defective fans that needed to be replaced because I could clearly hear the tick or the fan hitting against something.

 

[rennyf77]

That is your opinion and that is fine. However, I have not found that to be the case at all in my direct experience. I own an XFX R9 290 that I ran for 7 months before I flashed it to an R9 290x. In both cases, the fan noise was minimal at idle and load. Of course, I am not running my card on an open air test bench.

Also, I listening to some of the posted videos of the "fan noise". It was clear that there were defective fans that needed to be replaced because I could clearly hear the tick or the fan hitting against something.

you've posted this many times already, and as an owner of a reference xfx 290x and reference powercolor 290x purchased several months apart, who is very pleased with his purchase especially since moving to 4k, you are certainly in the minority regarding your 'quiet' 290. I'm not wishing to attack you here, but yours is an experience contrary to the experiences of many other reference 290/x owners, and the entire reviewing press at large, and it is not ok that you use your singular example to obfuscate this issue.

 

[biggles]

What does TDR mean?

 

[rennyf77]

the darkknight returns?

 

[Whach]

I just wish there were 3rd party, ultra quiet, blower-type cooling solutions - especially for AMD cards. Don't get me wrong, the current nvidia solutions are great IMO, but to have an alternative would be nice (not the msi/asus massive fan solution which are still great)

 

[xoleras]

I think HIS has in the past created blower type coolers with factory OC'ed cards, as has EVGA, but neither company has done that with the current gen.

At least, I think it was HIS? Some company did it. I can't recall , it's been a while.

For certain situations, blower coolers definitely are better. On the NV side I don't think it can get much better than the Titan shroud, on the AMD side i'm not sure if HIS has plans to do blowers in the future.

 

[rennyf77]

the iceq cards were equipped with quiet blower style coolers up until the hd 79xx series. for the longest time they used a design reminiscent of the old arctic cooling style aftermarket gpu coolers. their iceq products use to also be blue uv reactive.

 

[Unknown-One]

Guess TaintedSquirrel has bowed out of this discussion. Lots of people in here giving kudos to the Nvidia reference blower cooler now

I finally got an AIO water cooler to work correctly, and it's pretty much the only thing I've found to be superior in all aspects to the 780 / Titan blower:

• Cooler idle and load temperatures
• Same noise level at idle
• Lower noise level at load
• Still exhausts directly out the back of the case
• GPU load leaves case and component temperatures unaffected.

In fact, putting the GPU under load now drops case temps (because the GPU fan doubles as the rear-exhaust fan for the box), so that's nice. Only black-mark is that I went through three of these darn AIO water coolers and ALL of them had horrible quality-control issues. If you're going to attempt a setup like this while seeking silence, make sure you buy the cooler locally so you can easily return it and try another one.


For reference, I ended up with an H75 mounted to my GTX 780 with a NZXT G10 bracket. The following modifications were done:

• Front-plate from an EVGA ACX cooler modified to fit an AIO cooler properly (4 metal tabs had to be removed).
• Additional heatsinks added to the ACX front-plate over memory and VRM areas.
• H75 was opened and additional coolant was added (unit was received under-filled)
• H75 Pump has been under-volted to a fixed 5v.
• Included Corsair fans not used. Radiator cooled by a single Arctic F12 fan.
• 120mm radiator fan and 96mm NZXT bracket fan (VRM cooling) spliced together and speed controlled based on temperature.

Total cost to, hands-down and no-questions-asked, beat the reference blower? $176 (or $136 if you already had an ACX cooler to steal parts from).

 

[phorkz]

i think if they are indeed using the "titan style cooler", i will go ahead and grab 2x 880's on release.

 

[MongGrel]

I still have several AC fans in here myself, but the vast majority are Noctua's I guess.

I just like them personally,

Haven't even tried the newer generation ones out a lot.