[WCCFtech] AMD and NVIDIA DX12 big picture mode

[RussianSensation]

Well that seems like good news.

I hope this "issue" gets resolved soon. My GTX 580 is on its way out, requiring more and more voltage to stay stable at stock speeds, so I need a replacement soonish.

Really wish Fury didn't have only 4gb of ram, seems short sighted.

Be great if I can get another 5 years out of a card.

I would personally recommend that you change your GPU upgrading strategy altogether. What you have done should have been a great learning lesson when it comes to GPU ownership -- i.e., buy a new $500 GPU and keep it for 5 years but find it out that in 1.5 years from launch a $250 HD7870 is faster and in another 1 year a $300 R9 280X is 65-70% faster and then in by Sept 2014, a $300-330 970/290 is 2.25X faster. Imo, you would have been better off buying a $299 HD6950/GTX560Ti and then a $300 R9 290 /$330 GTX970 in late 2014. Instead, the GTX580 was barely faster than the HD6950 2GB (unlocked)/GTX560Ti OC, because back then all of these cards were fast enough and provided great performance. Yet, when next gen games got demanding, GTX560Ti/6950 and 580 became equally too slow. In fact, just 8% separates an unlocked 6950 and a 580 today.

Using the above, today you could grab a $280-300 970/R9 390 8GB and then in 2.5-3 years another $300 GPU (in fact you could probably grab a $350-400 GPU since you'd have some $$ from the resale value of the 970/390 card that can be put towards your 5-year mid-life cycle GPU upgrade). Over the course of 5 years you'd end up with a more satisfactory overall performance/experience. More optimistically, if you score a B-stock GTX980 for $370, and get a 2nd GPU in the summer of 2018 for $400, reselling the GTX980 for just $120, that would be very similar to buying a $650-700 flagship card now and running it into the ground by 2020-2021 so to speak.

The best way to see what happens to flagship cards vs. 2nd/3rd tier cards long-term is to simulate intense GPU workloads (or look at many old reviews because they all paint the same picture). The situation frankly becomes the same regardless as (all of them are too slow once overwhelmed). So really the ONLY to truly future-proof for 5 years is to buy a $300-400 GPU now, sell it in 2.5 years and buy another $300-400 GPU. Buying a late 2015 $650-700 GPU to try to future-proof for DX12 / 5 years (i.e., to late 2020) is not going to work well because this strategy has never worked. Not even 2 years has passed since we can have GTX780Ti's $699 level of performance in a $230 R9 290 or a $280 GTX970.

 

[EarthwormJim]

I would personally recommend that you change your GPU upgrading strategy altogether. What you have done should have been a great learning lesson when it comes to GPU ownership -- i.e., buy a new $500 GPU and keep it for 5 years but find it out that in 1.5 years from launch a $250 HD7870 is faster and in another 1 year a $300 R9 280X is 65-70% faster and then in by Sept 2014, a $300-330 970/290 is 2.25X faster. Imo, you would have been better off buying a $299 HD6950/GTX560Ti and then a $300 R9 290 /$330 GTX970 in late 2014. Instead, the GTX580 was barely faster than the HD6950 2GB (unlocked)/GTX560Ti OC, because back then all of these cards were fast enough and provided great performance. Yet, when next gen games got demanding, GTX560Ti/6950 and 580 became equally too slow. In fact, just 8% separates an unlocked 6950 and a 580 today.

Using the above, today you could grab a $280-300 970/R9 390 8GB and then in 2.5-3 years another $300 GPU (in fact you could probably grab a $350-400 GPU since you'd have some $$ from the resale value of the 970/390 card that can be put towards your 5-year mid-life cycle GPU upgrade). Over the course of 5 years you'd end up with a more satisfactory overall performance/experience. More optimistically, if you score a B-stock GTX980 for $370, and get a 2nd GPU in the summer of 2018 for $400, reselling the GTX980 for just $120, that would be very similar to buying a $650-700 flagship card now and running it into the ground by 2020-2021 so to speak.

The best way to see what happens to flagship cards vs. 2nd/3rd tier cards long-term is to simulate intense GPU workloads (or look at many old reviews because they all paint the same picture). The situation frankly becomes the same regardless as (all of them are too slow once overwhelmed). So really the ONLY to truly future-proof for 5 years is to buy a $300-400 GPU now, sell it in 2.5 years and buy another $300-400 GPU. Buying a late 2015 $650-700 GPU to try to future-proof for DX12 / 5 years (i.e., to late 2020) is not going to work well because this strategy has never worked. Not even 2 years has passed since we can have GTX780Ti's $699 level of performance in a $230 R9 290 or a $280 GTX970.

Well it wasn't $500 when I bought it, they went on sale pretty quickly. Was closer to $400. So your examples don't really pan out too well, you're talking nearly double what I paid over 5 years. Plus, the cheaper cards back then had less RAM, which is a pretty big deal. Your benchmarks that you linked show this at 1080p.

Plus I have it water cooled, I can't stand any of the air coolers that come on graphics cards, they all make too much noise. So I'd have to buy a new block for a card each time? No thanks. I also got a pretty damn good overclock (almost 900 mhz core) due to water cooling, that lasted up until this month.

 

[tential]

Yup, I'm happily playing all my games now on my r9 290 and it's silent on my sapphire trix card. Did I have to wait a little to get it? Yes, but for $200, I still have great performance for the next year and can sell it back for over $150 and get whatever new card out.

As RS is describing it's simple. Find the best bang for your buck deals on cards that aren't going to lose that much value, then resell those cards and upgrade to a far faster card. I got enough games now to play to occupy me til Volta/whatever AMD has coming up. Hopefully by then, Gsync will have died as a closed standard like that will never have enough variety of monitors to be in products I need while Freesync has quickly hit a wide variety of products because of the ease of implementation.

 

[EarthwormJim]

Yup, I'm happily playing all my games now on my r9 290 and it's silent on my sapphire trix card. Did I have to wait a little to get it? Yes, but for $200, I still have great performance for the next year and can sell it back for over $150 and get whatever new card out.

As RS is describing it's simple. Find the best bang for your buck deals on cards that aren't going to lose that much value, then resell those cards and upgrade to a far faster card. I got enough games now to play to occupy me til Volta/whatever AMD has coming up. Hopefully by then, Gsync will have died as a closed standard like that will never have enough variety of monitors to be in products I need while Freesync has quickly hit a wide variety of products because of the ease of implementation.

I think we have different definitions of silent.

 

[tential]

I think we have different definitions of silent.

Do you have any proof that the Sapphire Tri-X R9 290 is a loud card?

Or that the Sapphire Tri-X cooler is a poor cooler? I'm pretty sure Sapphire's coolers have gotten better from the HD7000 series to the Sapphire Fury Tri-X we see today.

RS, you have the same Dual-X cooler I did on my HD7950 (but then again you recommended the card and I'm sure I mentioned noise as a factor).

The Tri-X cooler on my R9 290 at the same fan settings is cooler, quieter, and obviously faster of course. I just need a 4K monitor now! And maybe another R9 290. I just saw another go for 180 on ebay of course lol. In fact, when I went to pick it up actually, I didn't realize his system was running at loud with the card in lol.

 

[EarthwormJim]

Do you have any proof that the Sapphire Tri-X R9 290 is a loud card?

Or that the Sapphire Tri-X cooler is a poor cooler? I'm pretty sure Sapphire's coolers have gotten better from the HD7000 series to the Sapphire Fury Tri-X we see today.

RS, you have the same Dual-X cooler I did on my HD7950 (but then again you recommended the card and I'm sure I mentioned noise as a factor).

The Tri-X cooler on my R9 290 at the same fan settings is cooler, quieter, and obviously faster of course. I just need a 4K monitor now! And maybe another R9 290. I just saw another go for 180 on ebay of course lol. In fact, when I went to pick it up actually, I didn't realize his system was running at loud with the card in lol.

I never said it was a bad cooler, it's one of the best air coolers for a graphics card, but it is still louder than I like.

Nothing in my case is audible. An air cooler can't compete with the cooling surface area I have on my radiators. Took me a ton of Gentle Typhoons to find ones that didn't groan at low voltages. Everything that moves is decoupled from the case.

 

[tential]

I never said it was a bad cooler, it's one of the best air coolers for a graphics card, but it is still louder than I like.

Nothing in my case is audible. An air cooler can't compete with the cooling surface area I have on my radiators. Took me a ton of Gentle Typhoons to find ones that didn't groan at low voltages. Everything that moves is decoupled from the case.

I see. That is a different definition of quiet.

I think what RS said still applies though. You can still get cards with new blocks for cooling, or pick them up used. In fact, since I've been looking around on ebay for cards, the ones with blocks usually go cheaper since there is less of a market. So you can at least pick up a used card and swap out the 580 but again, WC isn't my expertise so I wouldn't know whether it's compatible only that there is a large market out there for cards.

What RS is trying to say though is that you can:
A) Buy a $500 card now and add a block to it and WC it.
Keep for 5 years. After 2 years though, a $300 card will be far faster.
Then still hold onto it another 2 years. Where now it's even slower than the $200 cards out. Then hold onto it even another year.

Or

B) Buy a $300 card, WC it. Then, sell that card on Ebay/craigslist/whatever 2 years from now for $150. Then, buy another $300 card. and have far better performance. And even then, you can still sell that new $300 card 2-3 years down the line, and upgrade to a new card again.

The buy and hold til the asset depreciates to zero value is not a good strategy.

 

[EarthwormJim]

I see. That is a different definition of quiet.

I think what RS said still applies though. You can still get cards with new blocks for cooling, or pick them up used. In fact, since I've been looking around on ebay for cards, the ones with blocks usually go cheaper since there is less of a market. So you can at least pick up a used card and swap out the 580 but again, WC isn't my expertise so I wouldn't know whether it's compatible only that there is a large market out there for cards.

What RS is trying to say though is that you can:
A) Buy a $500 card now and add a block to it and WC it.
Keep for 5 years. After 2 years though, a $300 card will be far faster.
Then still hold onto it another 2 years. Where now it's even slower than the $200 cards out. Then hold onto it even another year.

Or

B) Buy a $300 card, WC it. Then, sell that card on Ebay/craigslist/whatever 2 years from now for $150. Then, buy another $300 card. and have far better performance. And even then, you can still sell that new $300 card 2-3 years down the line, and upgrade to a new card again.

The buy and hold til the asset depreciates to zero value is not a good strategy.

You're still looking at a flat $80-100 added to each card you buy if you get a new full coverage block. Blocks depreciate even harder than the video cards themselves.

GTX 580's aren't worthless right now, they still surprisingly go for $100 on ebay (no clue why someone would pay that much, maybe for Adobe Premier?), sadly mine is dying.

 

[xpea]

Robert Hallock posted several times using the phrase "AMD's gamble paying off", and I do think it was a nice move given their limited R&D $$ and financial situation. They need the consoles to survive as a business. Lisa Su in the recent analyst talk focus heavily on AMD GCN powering over 200 million gaming consoles. With Nintendo NX, it will sustain them, and now with PC having closer to the metal DX12, it's a huge pay-off too. It means they just design the best architecture for consoles and also reap the reward on PC.

Reality check: AMD's marketshare in PC plumbed like hell in the last year and they are now below 20%... so much paying off...

I just think people assuming because AMD hardware is in the new consoles this will auto give AMD an advantage. AMD hardware was in the generation 7 leading console - didn't do jack for AMD pc gpu performance.

+1000

AMD is a sinking boat and their 14/16nm gen dGPU may be the last one if they don't do something dramatic... or if NV screws up big time...

 

[Silverforce11]

AMD is a sinking boat and their 14/16nm gen dGPU may be the last one if they don't do something dramatic... or if NV screws up big time...

Their console wins will keep them going on and on. There's talk of Sony working with AMD on the PS5 already.

PC dGPU is just going to use console GPU tech, with PC APIs being console-like, it won't hurt AMD one bit.

 

[xpea]

Their console wins will keep them going on and on. There's talk of Sony working with AMD on the PS5 already.

Totally wrong !
it's been a while since AMD graphic division doesn't make any money. And on console business, the profit is so low that they can't pay the r&d and survive with it alone...

 

[Silverforce11]

Totally wrong !
it's been a while since AMD graphic division doesn't make any money. And on console business, the profit is so low that they can't pay the r&d and survive with it alone...

Yes it's *so wrong* that they will be around for many years to come.

And who knows, if Pascal is crippled like Maxwell with software schedulers & driver async compute, it could be a very disturbing next few years! Buckle up!

 

[ShintaiDK]

Yes it's *so wrong* that they will be around for many years to come.

We can hope so. But their financials and free fall market share doesnt inspire to such hopes.

Then you can make whatever PR claims. But f the market share and money isnt there...

 

[tential]

You guys are talking about the past, silverforce is talking about the future. So yes, you can say "how's that working out for amd?" over and over again....

But when you draw a long term plan up you take losses. Someone like myself sits there and calculates the losses to future earnings, discounts that and sees whether it's profitable. It's obvious what silverforce is trying to explain....

Amd has had a strategy the execution of it has just been poor.....

 

[ShintaiDK]

Oh yes, the "just wait and see".

AMDs outlook for Q3 says it all. Another 100M$ loss and cash burn.

60% of the company revenue is consoles.

 

[Silverforce11]

You guys are talking about the past, silverforce is talking about the future. So yes, you can say "how's that working out for amd?" over and over again....

But when you draw a long term plan up you take losses. Someone like myself sits there and calculates the losses to future earnings, discounts that and sees whether it's profitable. It's obvious what silverforce is trying to explain....

Amd has had a strategy the execution of it has just been poor.....

Their biggest screw-up was the reference cooler on the R290/X. That completely ruined an EXCELLENT piece of GPU hardware for the entire generation due to the bad stigma.

And yes, AMD had no choice but go long-term strategy, limited R&D money means they have to be very wise in how they spend it. Spending it to make a great console GPU architecture guarantees their survival despite lower efficiency in DX11. But for the future, that's all changing with DX12 being console-like.

 

[Silverforce11]

Some interesting talks on VR:

https://www.youtube.com/watch?v=kvZAVrMuklE

 

[3DVagabond]

Their biggest screw-up was the reference cooler on the R290/X. That completely ruined an EXCELLENT piece of GPU hardware for the entire generation due to the bad stigma.

And yes, AMD had no choice but go long-term strategy, limited R&D money means they have to be very wise in how they spend it. Spending it to make a great console GPU architecture guarantees their survival despite lower efficiency in DX11. But for the future, that's all changing with DX12 being console-like.

Agreed. Except their biggest screw up was not fixing the original cooler. They could have released a "390" anytime they wanted to. Say, right after the mining craze?

 

[monstercameron]

Agreed. Except their biggest screw up was not fixing the original cooler. They could have released a "390" anytime they wanted to. Say, right after the mining craze?

The stock reference was just a forum/review site thing. How long was the reference cooler even on the market before custom coolers were available? A few months, so saying all they had to do was release/rebrand the 290/x to 390/x wouldn't have meant much. Using it as a stop gap was the better strategy IMHO.

 

[EarthwormJim]

Oh yes, the "just wait and see".

AMDs outlook for Q3 says it all. Another 100M$ loss and cash burn.

60% of the company revenue is consoles.

It's a big company being weighed down in other areas. 60% of their revenue being from consoles is a testament to their GPU division. That revenue is due mainly to their GPUs. They would not have scored those contracts from both Sony and Microsoft without good hardware designs.

We'll see how this pans out in the PC market once DX12 sees wider use, they really need it to work out in their favor though.

 

[RussianSensation]

You're still looking at a flat $80-100 added to each card you buy if you get a new full coverage block. Blocks depreciate even harder than the video cards themselves.

GTX 580's aren't worthless right now, they still surprisingly go for $100 on ebay (no clue why someone would pay that much, maybe for Adobe Premier?), sadly mine is dying.

If you have to buy right now, you have no choice but to go with the fastest card since that's what benchmarks show => 980TI OC > Fury X OC and it has 50% more VRAM as a bonus. If you must go WC, then that's probably the best option if you don't want to upgrade for 5 years. I don't see how it'll be way quieter though since GTX970 cards like MSI Gaming turn off their fans up to 60C and are uber quiet at load. I still think you aren't giving enough credit for some of the best after-market air cooled designs.

I mean wouldn't it be more practical to get a Fractal R5 case and insulate it with sound dampening material vs. spending hundreds of dollars on WC components on a $650 card that drops close to $100 in resale value over a period of 5+ years? Again, it's of course your choice but based on what you described I'd say 980Ti OC on paper appears like the safer bet. IMO both the 980Ti and Fury X will become obsolete at roughly the same time. I expect in 5 years, a flagship card will be ~3.5X faster than the 980Ti. Therefore, no matter what you buy today, it's not going to be more future-proof. GTX580 was slightly more future-proof than HD6950 (BTW if you got your GTX580 for $400, chances are HD6950 was $250 or less and I also recall when GTX580 was $400 it was closing in around November 2011, just mere months before HD7970 came out and crushed it), but ultimately both of those cards got outperformed by so many cheaper cards not long after.

 

[Shmee]

I would actually think 6950 is a bit better for longevity, at least the 2 GB version, and many of them would unlock shaders.

 

[3DVagabond]

The stock reference was just a forum/review site thing. How long was the reference cooler even on the market before custom coolers were available? A few months, so saying all they had to do was release/rebrand the 290/x to 390/x wouldn't have meant much. Using it as a stop gap was the better strategy IMHO.

Even today, when you look at reviews that include the 290's they still report 94°C and ~60db for temp and noise. They could have released a higher clocked card with a better cooler anytime. I don't understand why you think them waiting until they had 18% quarterly market share was the opportune time. The current 390 series would have beaten Kepler, just like it does today.

 

[Headfoot]

imo the best, easiest, cheapest way of getting a "silent" pc is getting decently quiet parts and really long cords and put it far away from you. The inverse square law quickly makes any PC inaudible. http://hyperphysics.phy-astr.gsu.edu/hbase/acoustic/isprob2.html

 

[monstercameron]

Even today, when you look at reviews that include the 290's they still report 94°C and ~60db for temp and noise. They could have released a higher clocked card with a better cooler anytime. I don't understand why you think them waiting until they had 18% quarterly market share was the opportune time. The current 390 series would have beaten Kepler, just like it does today.

There is a difference in maybe circulating aftermarket cooled 290/xs to reviewers against a complete rebrand.