R290X Crossfire V NVDA's SLI 780s/Titans from [H]

[tviceman]

No. A case with good airflow is just as good if not better. Any enthusiast would know this.

OH KNOES I am not an enthusiast! LOL ok LOLOLOLOLOLOLOL I completely get the pc master race slants with comments like yours.




Knock off the threadcrapping. If you don't have a proper question/answer to the post,
don't post ridiculous statements like "OH KNOES I am not an enthusiast! LOL ok LOLOLOLOLOLOLOL "

Serves no purpose in this forum


esquared
Anandtech Forum Director

 

[Rvenger]

OH KNOES I am not an enthusiast! LOL ok LOLOLOLOLOLOLOL I completely get the pc master race slants with comments like yours.

An open test bench isn't as good since you really don't have any active airflow blowing towards the graphics card.

Brent, Do you remember me from Speedguide.net over a decade ago?

 

[AnandThenMan]

OH KNOES I am not an enthusiast! LOL ok LOLOLOLOLOLOLOL I completely get the pc master race slants with comments like yours.

Rvenger is right of course, if you've ever build a machine using a decent case you know the extra airflow can actually lower component/peripheral temps versus open air. I thought this was common knowledge myself...no?

 

[tviceman]

Rvenger is right of course, if you've ever build a machine using a decent case you know the extra airflow can actually lower component/peripheral temps versus open air. I thought this was common knowledge myself...no?

I'm sure components that are passively cooled (like ram) benefit with case fans, but with active cooling I can't see how cas airflow can actually be "better" than an open environment since the a fan spinning (on a graphics card) at X speed can only move so much air and on an open test bench fans should not at all be starved for air flow. Perhaps open-air coolers might benefit from surrounding active cooling, but a closed blower???

 

[OVerLoRDI]

Wow, that is a seriously glowing review for Crossfire.

I don't think my lady would be that impressed if I threw down for 2 of these bad boys with water blocks, but this review makes me definitely consider it.

 

[FalconHorse]

I'm sure components that are passively cooled (like ram) benefit with case fans, but with active cooling I can't see how cas airflow can actually be "better" than an open environment since the a fan spinning (on a graphics card) at X speed can only move so much air and on an open test bench fans should not at all be starved for air flow. Perhaps open-air coolers might benefit from surrounding active cooling, but a closed blower???

I would imagine that a temperature increase from the rest of the components on the test bench is possible, and would cause a higher ambient temperature in the area surrounding the video cards. In a closed case with good airflow, ideally (theoretically?) you are bringing in fresh cool air constantly, so the air the card is drawing into the shroud (regardless of cooler style) is going to be cooler in well-ventilated case, after a few minutes of warmup at least. Just a guess, though.

 

[BrentJ]

In my open air test setup I have no active air moving across or over the video cards, like there would be in a case. A case would have front intakes blowing air across them, and potentially a side panel fan blowing air onto them. With moving air inside a case with proper intake and exhaust, the ambient temperature could be lower around the cards.

 

[blackened23]

OH KNOES I am not an enthusiast! LOL ok LOLOLOLOLOLOLOL I completely get the pc master race slants with comments like yours.

Serious reply: With a blower cooler it does not make a difference. With an aftermarket card, it makes a very dramatic difference - basically, the hot air is going to flow out of the rear exhausts of the card whether the card(s) are in a test bench or an enclosed case. Keep in mind, the hot air doesn't circulate around the cards with a closed blower, this is true whether an enclosure is used or not. Conversely, open air cards tend to do way way better on an open air test bench. I think that was his point. Although, he was facetious in the wording of it with the "any enthusiast would know this" comment.

Please note that i'm not saying that the blower cooler on the 290X is good at all. I think it sucks and drags the otherwise great aspects of the 290X (performance) down. But the point stands that blower coolers don't matter on an open air test bench, while it does change results for aftermarket coolers to a degree. Then again, how do you test for this with an aftermarket cooler? There are thousands of cases so if a reviewer is testing an aftermarket card, he can't possibly replicate 100% of the conditions that a user may have. You might have an aftermarket open air card in a mITX case - so if a reviewer tests on an EATX monstrosity, his results will differ from yours. So I guess it's really a no win situation for reviewers even with an aftermarket card, you know? But the point remains that with a blower, it doesn't affect results, really.

edit: As others have mentioned, blowers do better in enclosures than they do on test bench environments because of fans and air flow as well.

 

[Rvenger]

290x cooler is terrible. No doubt about that. I can say that 70% fan speed is extremely loud on the 290x. But an open test bench is pretty common sense and has been preached numerous times in this forum that it's inferior compared to well ventilated cases. So my response was not ridiculous... it was from years experience.


BTW, if you were feel the air discharging out of the back on the 290x, you don't feel very much air like it is getting trapped in the shroud somehow.

 

[blackened23]

Yeah. I acknowledge that and agree with it (about test benches). You know, I still cannot believe that AMD put that stupid cheap cooler on the 290X. I mean, under the hood from what i've heard it is no different than the 7970 cooler - and that cooler was the same as the 6970 and 5870 coolers as far as I can tell. I just can't believe it. AMD could have prevented all of this and they had to cheap out in the worst way possible with the 290X, I mean it would be different if aftermarket cards were available on day one. But they're not, while they are for the 780OC cards.

Unbelievable and unfortunate that the performance is outstanding, but the "total package" with all metrics considered gets dragged down. Oh well. Did your card have coil whine btw?

 

[The Alias]

Yeah. I acknowledge that and agree with it (about test benches). You know, I still cannot believe that AMD put that stupid cheap cooler on the 290X. I mean, under the hood from what i've heard it is no different than the 7970 cooler - and that cooler was the same as the 6970 and 5870 coolers as far as I can tell. I just can't believe it. AMD could have prevented all of this and they had to cheap out in the worst way possible with the 290X, I mean it would be different if aftermarket cards were available on day one. But they're not, while they are for the 780OC cards.

Unbelievable and unfortunate that the performance is outstanding, but the "total package" with all metrics considered gets dragged down. Oh well. Did your card have coil whine btw?

it's not like aftermarket coolers will never exist .

 

[blackened23]

it's not like aftermarket coolers will never exist .

I'm sure AMD's goal is to sell cards, and when buyers see, hmm, OC'ed 780s on the market now for possibly slightly less than the 290X versus the 290X with the blower, i'm sure some will opt for the OC'ed 780s. Know what I mean? This is why I say I cannot believe that AMD did not improve it. It would be different had aftermarket designs been available from the start, but they're obviously not.

When aftermarket cards come out for the 290X that will change things, but that still doesn't excuse AMD from not improving the reference shroud. I'll also add that reference shrouds are a pre-requisite for some buyers using small form factors, aftermarket cards don't get along with small enclosures. Anyway, AMD should have known better since NV raised the bar so high in that respect - like I said before, if they wanted to go all out they should have...gone all out with the 290X. But they just went cheap on the cooler.

 

[Rvenger]

Yeah. I acknowledge that and agree with it (about test benches). You know, I still cannot believe that AMD put that stupid cheap cooler on the 290X. I mean, under the hood from what i've heard it is no different than the 7970 cooler - and that cooler was the same as the 6970 and 5870 coolers as far as I can tell. I just can't believe it. AMD could have prevented all of this and they had to cheap out in the worst way possible with the 290X, I mean it would be different if aftermarket cards were available on day one. But they're not, while they are for the 780OC cards.

Unbelievable and unfortunate that the performance is outstanding, but the "total package" with all metrics considered gets dragged down. Oh well. Did your card have coil whine btw?

My card had 0 coil whine which was surprising since 75% of my AMD cards screamed!

 

[Rubycon]

A real liquid cooling solution should be able to keep the GPU under 50C with ease and keep the VRM parts cool enough to allow harder driving (as long as SOA margins are preserved!).

However it's disappointing out of the box with those temps and the noise. That's going to add a LOT of heat to the case with those aftermarket style coolers that blow all the heat into the case!

 

[5150Joker]

I'm going to revise my recommendations (ignoring G-SYNC). If you're in the market for a new card and don't plan to purchase a G-SYNC display, the 780 Ti variant with 6GB of memory would be the #1 choice. Second choice would be a R9 290X and third choice a used Titan. I wouldn't touch a 3GB 780 because I'm finding BF4 already hitting 3GB of VRAM utilization and it's only going to get worse when newer games made for PS4 get ported to the PC.

Most games on the market right now and especially on the PC were designed with older gen consoles in mind, these newer consoles (esp. PS4) will change all that.

 

[Teizo]

it's not like aftermarket coolers will never exist .

And that justifies AMD cheaping out on their cooler how?

If AMD wants to drop the 'budget' moniker that sticks to them like ugly on a redneck then they need to pony up and start putting better coolers and fans on their top tier reference cards.

Open slot AIB designs will be worse for CF, I imagine. All that heat is going to be dissipated into your case. So, a good quality reference design on a high end gpu is not something a company with good management should overlook.

I'm going to revise my recommendations (ignoring G-SYNC). If you're in the market for a new card and don't plan to purchase a G-SYNC display, the 780 Ti variant with 6GB of memory would be the #1 choice. Second choice would be a R9 290X and third choice a used Titan. I wouldn't touch a 3GB 780 because I'm finding BF4 already hitting 3GB of VRAM utilization and it's only going to get worse when newer games made for PS4 get ported to the PC.

Most games on the market right now and especially on the PC were designed with older gen consoles in mind, these newer consoles (esp. PS4) will change all that.

You saying this for 1080p or 1440p and up? I'm still not sure if 2GB will be a significant limiting factor at 1080p in BF4, but I most certainly could be wrong. I do imagine at resolutions above 1080p that 3GB is the bare minimum though.

 

[blackened23]

I'm going to revise my recommendations (ignoring G-SYNC). If you're in the market for a new card and don't plan to purchase a G-SYNC display, the 780 Ti variant with 6GB of memory would be the #1 choice. Second choice would be a R9 290X and third choice a used Titan. I wouldn't touch a 3GB 780 because I'm finding BF4 already hitting 3GB of VRAM utilization and it's only going to get worse when newer games made for PS4 get ported to the PC.

Most games on the market right now and especially on the PC were designed with older gen consoles in mind, these newer consoles (esp. PS4) will change all that.

I hope you're right, but keep in mind that PS4 games are going to be 1080p or 900p without MSAA - MSAA is generally the #1 contributor to high VRAM use. So I have difficulty in seeing 3GB or 6GB fully utilized unless you're going for 4k to surround. Maybe 3GB for the high resolution crowd.

That said, when I think back to 2010 - 1GB of VRAM was considered more than enough. 1.5GB on the GTX 580 was considered enough. People who bought 3GB cards at that point and time were considered nuts. So you may be onto something with that thought! The way I figure, more VRAM will eventually happen but will probably take a year or two - and by that time, the 290X and Kepler will probably be a distant memory. Just IMHO. I do hope you're correct, though! I'm all for higher resolution textures, if that happens it would be fantastic. So I guess, I hope my prediction is wrong and that yours is correct. I hope more VRAM use happens sooner rather than later.

 

[Teizo]

I hope you're right, but keep in mind that PS4 games are going to be 1080p or 900p without MSAA - MSAA is generally the #1 contributor to high VRAM use. So I have difficulty in seeing 3GB or 6GB fully utilized unless you're going for 4k to surround. Maybe 3GB for the high resolution crowd.

That said, when I think back to 2010 - 1GB of VRAM was considered more than enough. 1.5GB on the GTX 580 was considered enough. People who bought 3GB cards at that point and time were considered nuts. So you may be onto something with that thought! The way I figure, more VRAM will eventually happen but will probably take a year or two - and by that time, the 290X and Kepler will probably be a distant memory. Just IMHO. I do hope you're correct, though! I'm all for higher resolution textures, if that happens it would be fantastic. So I guess, I hope my prediction is wrong and that yours is correct. I hope more VRAM use happens sooner rather than later.

I'm interested in seeing with COD Ghosts has to offer in that regard. 50GB? Good gosh that is a large download.

 

[toyota]

lol just 6 to 7 years ago we were arguing that 256mb of vram was enough.

 

[5150Joker]

And that justifies AMD cheaping out on their cooler how?

If AMD wants to drop the 'budget' moniker that sticks to them like ugly on a redneck then they need to pony up and start putting better coolers and fans on their top tier reference cards.

Open slot AIB designs will be worse for CF, I imagine. All that heat is going to be dissipated into your case. So, a good quality reference design on a high end gpu is not something a company with good management should overlook.



You saying this for 1080p or 1440p and up? I'm still not sure if 2GB will be a significant limiting factor at 1080p in BF4, but I most certainly could be wrong. I do imagine at resolutions above 1080p that 3GB is the bare minimum though.

Yes at 1440p and up. Although brightcandle found 2GB utilization at 1080P + 4xMSAA and other settings he used. So if I had to guess, 3 GB will hit a wall sooner than later now that next gen consoles are here.

 

[blackened23]

lol just 6 to 7 years ago we were arguing that 256mb of vram was enough.

True, even though the PC has been getting yearly doses of console ports from 2007-2012, VRAM use steadily rose at a consistent pace over those years. In 2007, 768MB was enough. And we all know how poorly those 768MB cards are doing now.

So i'm sure VRAM use will steadily increase, but i'm of the opinion that by the time 6GB is fully utilized, that we'll be well into the 20nm GPU cycle. Again, I hope i'm wrong on this - if VRAM use rises dramatically early next year i'm all about it. I hope i'm wrong. But I just can't see it happening that fast, I think 3GB will be a "sweet spot" for the next year or so at a minimum. Of course this would not apply to surround users, they will need every bit of VRAM they can get both now and in the future.

 

[5150Joker]

True, even though the PC has been getting yearly doses of console ports from 2007-2012, VRAM use steadily rose at a consistent pace over those years. In 2007, 768MB was enough. And we all know how poorly those 768MB cards are doing now.

So i'm sure VRAM use will steadily increase, but i'm of the opinion that by the time 6GB is fully utilized, that we'll be well into the 20nm GPU cycle. Again, I hope i'm wrong on this - if VRAM use rises dramatically early next year i'm all about it. I hope i'm wrong. But I just can't see it happening that fast, I think 3GB will be a "sweet spot" for the next year or so at a minimum. Of course this would not apply to surround users, they will need every bit of VRAM they can get both now and in the future.

Blackened, check it out: http://forums.anandtech.com/showpost.php?p=35682986&postcount=41

That's from BF4 so that's why I think these new games are gonna start crossing 3GB on a single high resolution display like mine. Surround will probably go well beyond that figure.

 

[BrightCandle]

Yes at 1440p and up. Although brightcandle found 2GB utilization at 1080P + 4xMSAA and other settings he used. So if I had to guess, 3 GB will hit a wall sooner than later now that next gen consoles are here.

I found that the game used about 1.1GB (no MSAA) or 1.5GB (MSAA 4x) and that the game would then climb up its usage to the amount of VRAM the card had. The fact is that the game caches old assets and keeps them around so the VRAM readings from GPU-Z/MSI Afterburner are useless except for right at the beginning after you have just loaded the level.

This is important because a lot of people are showing traces after playing the game for a while and that is a flawed approach for this game as it misleads others into thinking that BF4 uses more VRAM than it does. BF4 runs fine on 2GB of VRAM, infact I don't have enough GPU stream processor grunt to push all the effects and MSAA + PPAA, but without PPAA I can run MSAA just fine. Its not VRAM that is the issue.

The PS4 and xbox1 might have a unified memory access system with the potential for VRAM usage to be in the 5-6GB range but in practice they don't have the GPU grunt power in their stream processors to really utilise it all. The cards they were based on typically ship with 1-2GB of VRAM and I suspect we'll continue to see console ports targeting that for much of their lifetime.

 

[blackened23]

I see! I didn't realize BF4 would use that much. Anyway, like I said - if future games raise the bar on VRAM usage i'm all about it. I hope that does happen with more future games.

 

[5150Joker]

I found that the game used about 1.1GB (no MSAA) or 1.5GB (MSAA 4x) and that the game would then climb up its usage to the amount of VRAM the card had. The fact is that the game caches old assets and keeps them around so the VRAM readings from GPU-Z/MSI Afterburner are useless except for right at the beginning after you have just loaded the level.

My VRAM usage jumps to 2.9GB+ upon loading the level almost immediately. So there is no caching of old assets in that case.