What a relief and I stand corrected, Pascal does improve with DX12 Async Compute feature (at least with GTX 1070 confirmed, eagerly waits for GTX 1080 results to trickle in). May not be as high as the percentage increase gained by AMD GPUs, but for me looks significant enough, and will get better as Pascal is still relatively new. As for Maxwell, pretty much what I thought will happen when I decided to sell my GTX 970.



Excerpt from PCWorld 3DMark Spy Test Results

"First impressions: The power of the dedicated async shader hardware inside Radeon graphics cards shines in this test. Excluding the ridonkulously powerful GTX 1070, Radeon hardware dominates Nvidia’s GeForce cards in this DirectX 12-focused test, lending yet more credence to the idea that AMD’s cards have an advantage in next-generation graphics technologies. Even the $200 RX 480 thoroughly trounces the GTX 980. (Note that Nvidia tends to hold an advantage in DirectX 11 performance, however, which still accounts for the vast majority of game releases.)"

This post has been edited by adilz: Jul 15 2016, 02:25 AM

wow, chugs like hell during entire benchmark

CPU - stock
GPU - 1003/1802MHz



EDIT: retested with latest Nvidia drivers (the previous one was 358.87) and man, the score dropped to 1.5k

not to mention, the drivers died halfway again while running

im really suspecting something is wrong with Kepler and 36X.XX series drivers, any fellow Kepler users here can give me more input on whats happening?

This post has been edited by marfccy: Jul 15 2016, 03:45 AM

hahahaha notty notty

http://www.pcper.com/reviews/Graphics-Card...ute-Performance

http://www.guru3d.com/articles-pages/futur...k-review,1.html

With GTX1080

FuryX performance is...lacking...it should be able to zip through GTX1070 with its shader perf..

This post has been edited by Demonic Wrath: Jul 15 2016, 08:14 AM

I don't know what rumour or hype you heard, but this is the first time I heard someone say that Fury X was expected to zip through GTX 1070. Its expecting too much from a card built on 28 nm tech, a generation behind GTX 1070 16 nm architecture. Fury X should be compared to its generation, likeGTX 980 Ti or GTX Titan X, Trust me, tech reviewers will be talking about what performance gain the cards got with DX12 rather than where it sits in the Time Spy benchmark table.

If you say Fury X (a card that retails for USD 589 around Xmas 2015) performance is lacking in this DX12 Time Spy results, than what word can you use to describe the performance of GTX 980 Ti (USD 599 around the same time) or the beast GTX Titan X (USD 999).

Again, point of my earlier post, what are you willing to spend for the performance you want. Those benchmark table are nice if you can get all the cards for free. Put current price (used or new) of those card in that table, and means a lot more to a person a building a new PC with RM 3K budget, than a person who have no qualms plonking RM 3K+ just for brand new shiny GTX 1080 GPU alone. Head on to LYN market, there's a used R9 290X asking for RM 500, and compare its position in those Time Spy bench, to GTX 970 GPU with asking price between RM 900 to RM 1200 used. So does that mean that all GTX 970 owner should sell their GPUs, and get a used R9 290X and still have extra money for a 256GB SSD? Definately no. I say go GTX 1070.

Give credit where credit is due. Either by luck or by purpose, AMD's decision to keep faith with their GCN years ago paid off. So now, they might still be able to optimize and extract a bit more from their older gen cards to add on their already impressive gains with the new APIs, but there's only so much more they can get out of that 28 nm tech. Focus will be more on their Polaris now and the upcoming Vega.

As for Nvidia, I think its pretty much at the end of what they can get out of their Maxwell GPUs. Those are still good cards with DX11 games, which will still be released alongside DX12 version of the game for at least the foreseeable future. Nvidia should pour all their effort into Pascal now. It has proved me wrong because it does gain from DX12 Async Compute. There's potential for Pascal to make the same amount of performance gain like AMD if they really try.

So maybe its time to move on from this Async Compute debate. I'm so looking forward to DX12 EMA with SFR implementation, because that will be the best thing gamer can expect from the DX12 shift.

This post has been edited by adilz: Jul 15 2016, 10:09 AM

Hype or rumors (is there even a hype or rumor on this lol?).. doesn't matter.. it's a fact that Furies have higher shader performance.. just saying..

I was also comparing it against its own 28nm brethen R9 390X. It is weird that Fury having much higher shader perf (1.27x higher) doesn't have much higher perf compared to R9 390X.. mainstream cards are performing as expected though.. if I'm not wrong, it should be tessellation bottleneck..

There's also this.. well, AMD cannot push their shader intrinsic stuff into this benchmark as it is AMD specific codes.


This post has been edited by Demonic Wrath: Jul 15 2016, 10:23 AM

Its also a fact that my i7-4930K Ivy-Bridge CPU has extra 2 cores/ 4 threads compared to i7-6700K Skylake. But I will be really disappointed if I expect my CPU to beat the Skylake in Firestrike or Time Spy bench just because its higher core count. Don't expect too much from old tech compared to the newer ones.

I'm not sure what you're getting at here . Higher core counts and higher clocks will benefit in heavily threaded app if the difference in arch is not much.

In heavily threaded app, your i7-4930K will beat i7-6700K Skylake. Skylake is faster because it has higher IPC and clock speed in games (prefer faster core compared to more cores). In Firestrike Phsyics score, 4930K will score higher than 6700K. I'm pretty sure video encoding will be faster on your 4930K too. The keyword here is heavily threaded app. Please don't be disappointed

First, going from Ivy Bridge to Skylake, Intel has increased the clock speed and IPC performance. R9 390X and Fury has the same IPC performance. R9 390X and Fury also has similar clock speed. So the difference is core count and raw shader performance (TFlops). Now it is only 5% higher on the Time Spy benchmark. Clearly something is being bottlenecked for Fury series. You wouldn't expect GTX1060 to beat GTX980Ti just because it is new architecture right?

Secondly, graphics are heavily threaded app. So higher cores and higher clocks will beat lower cores and lower clocks with same architecture. R9 390X and Fury has the same GCN architecture. Heck, AMD has increased Fury efficiency in tessellation too. So Fury should have a bigger difference in performance compared to 390X.

Cheers.

This post has been edited by Demonic Wrath: Jul 15 2016, 11:24 AM

Sorry, my bad. Misunderstood your statement about the shaders.

As for my i7-4930K, am not disappointed at all. Even at matching OC clock speed to i7-6700K, and the higher Physics score, overall score will still be lower than Skylake. Skylake just make better use of the GPUs. Underlines how much improvement the better IPC made, at least to gaming. As I posted earlier, i7-6700K still matches even beastly i7-5960X in games. For multi-threaded apps like video editing, the extra cores helps. Cinebench score higher than Skylake.

As for the R9 390X vs Fury, agree with you. Something's bottlenecking and I suspect its the 4GB HBM.

I don't expect GTX 1060 to outperform GTX 980 Ti even with the new arch. It doesn't make business sense for any of the GPU manufacturer to produce cheaper card that outperforms a vastly more expensive previous gen card (I suspect technically they can). They want to maintain a certain price/ performance tier-ing so that they can milk our money as much as the possibly can.

Haha, its ok.

My next upgrade will need to be at least 4x GTX970 perf. So, just wait. Now GTX1080 is only 2x GTX970 perf. Even GTX1080Ti with 3584 cores is only 1.4x GTX1080, so only 2.8x GTX970.

Now need to upgrade monitor first.

Almost everywhere is out of stock for MSI cards.

I'll just wait for Christmas and hope they got good Christmas deals for the cards and get it as a Christmas present for myself.

Already no mood to get the card now since birthday already past 1 week ago. Badly wanted to get it as a birthday present.

Just now wanted to buy a tissot watch as a birthday present but the guy told me no stock.

If there is one thing to hate about retailers, it's the words "NO STOCK".

Then how about my scenario? from GTX670 jump to GTX1070?

Things are often out of stock when you have the $$$$ for it.

Yeah, MSI cards is hard to find. I place an order for GTX1080 Gaming X almost three weeks ago. Still no news. Suppose to come in this week, looks like have to wait further more. Sienzzzzz...

Maybe I should have get Palit instead. Hiya, patience patience.

This post has been edited by CPURanger: Jul 15 2016, 03:34 PM

Haha rugi lorh

Should buy bigger chunk gpu first then only jump 4k

Haha happens all the time

That one surely times 10 boost

Haha i feel your pain brah

When 970 came out it was no where to be seen, no stock everywhere. I managed to grab one last unit in digital mall before they raised the price by 100myr.

got to wait a while until the demands get higher. they normally don't overstock.

This post has been edited by r47z: Jul 15 2016, 03:42 PM

Awww sowwee 2 hear that