ya so in the end only FPS matters, just cannot expect the lower end Maxwell to perform as well as the higher end 980ti in terms of newer API. strange though you would think 1070 would perform way better than 980ti in Vulkan.

https://en.wikipedia.org/wiki/Kepler_(micro...ction_Scheduler

btw if you see the Kepler whitepaper, they mentioned replacing the complex one with simple software scheduler..on the die itself.

from what i understand, the Gigathread Engine is a serial engine compared to the 8 hardware ACE units found on AMD. 8 engines which gives 8 threads/queues = 64 capability instead of a single engine with 1x1x32 threads.

so this part is true for the 970 and perhaps 960/950?: "Forcing it on in Maxwell causes inefficiencies (performance loss) since the software scheduler is already dynamically assigning tasks and now you're overriding what it does. Pascal is better, but mostly just because it can be forced to reassign tasks in a more granular manner (by SM, not by entire GPC)."

So yes, Nvidia can "do" async compute, no they don't have async shaders, no there is no some performance gain on some Maxwell cards (sometimes a detriment), yes Pascal can see minor gains.

This post has been edited by svfn: Jul 14 2016, 07:05 PM

I found some feedback on they FB. Seem like trusted seller. Shop name is GTech Solution.

https://www.facebook.com/gtech.solutions/

makes me feel sad again

edit: are u sure? because they shared their fb on thread https://www.facebook.com/VenomX.My/

This post has been edited by napster2142: Jul 14 2016, 07:21 PM

What is meant by simpler scheduler is this:
In a hardware/dynamic scheduler, there will be a hardware to re-order instructions at runtime for performance reason. In other words, there's a hardware in the scheduler to optimize the algorithm.

In a compiler/static scheduler, this re-ordering/optimization of instruction is done during code compilation. (As it is possible to predict the instruction anyway.. e.g. an add operation will always go through the same instruction) It is using compiler to create the best optimized order of instruction. Hence, why it is labelled as software scheduler.

You can read more on static vs dynamic scheduler for more info.

But the hardware to perform task allocation is still there. Don't worry Even a warp scheduler in each SM is also considered as a scheduler. It still need some hardware to decode the instruction and send it to SM to crunch.

The Gigathread is a well kept secret on NVIDIA side. No one knows (except NVIDIA) how it is internally. They don't clearly label it as 32 graphic processors engine or so in any presentation. It is one of the reason why it can crunch DX11 games so efficiently too.

actually we are both wrong, that Gigathread is instruction level scheduler it seems, NOT the scheduler responsible for async compute.

you need a hardware scheduler to run Async & you need independent engines for each queue, which is what ACE engines is.

i think if you ask about it in other forums, you will get a better, more informed answer than what i can give.

This post has been edited by svfn: Jul 14 2016, 07:22 PM

I think you can stay with your CPU for much much longer than a year. Only if you plan going SLI, then your i7-3770 will fall by the sideway. Or if newer games can utilize more threads, which I think will take sometime to happen as game developers learn their way around the new APIs.

VenomX.my is distributor Gaming Chair to GTech

Confirm by GTech

Now I feel like kantoi because become an stalker LOL

Tried to get MSI GTX1070 Gaming X.

It's freaking out of stock everywhere.

sorry i'm not familiar with them either, but the price is too good to be true, i'm also beginning to wonder what the catch is, because these cards are selling like hotcakes. even asus own info next shipment is out soonest by next week estimated, and even then supply will be limited. so it doesn't quite make sense why they would drop the price

i even checked the news to see if maybe news of a 1080ti release date was confirmed, or if Asus supply issues for this card was ramped up to cope or something xd...

This post has been edited by Moogle Stiltzkin: Jul 14 2016, 09:23 PM

Those with Pascal cards and VR headsets.



Nvidia's VR Funhouse is out. #PlayMoreGames

so now all new game developer going vulkan? Look like AMD gonna revived.

What is dropping price?
Where where?

I don't need more fps. The G-sync is buggy too.

new driver. 368.81

http://us.download.nvidia.com/Windows/368....tional-whql.exe

Thanks for the heads-up bro. Going to download it tomorrow.

someone need to buy 1st then we know if they are trusted and genuine. BTW their FB est 2015, dunno if they actually have a lot successful trade but didn't have record on that

just for you, dont tell anyone
https://forum.lowyat.net/index.php?showtopic=3749533&hl=

3DMark Time Spy Benchmarking Utility is out. Update your 3DMark app on Steam.

Results from Time Spy 3DMark DX12 benchmarking on my GTX1080, running at 2012Mhz core 10800 memory and 4.8Ghz 4790K.

Demonic Wrath


svfn

On this topic of the new Vulkan API and Doom game, I think we can all agree that generally AMD GPUs so far, had more significant performance gain compared to Nvidia GPUs . On why it turns out that way, some of us have different and conflicting views. Driver lah, shaderslah, async compute lah, engine lah. All this technical "presentations" is good, and I say, no one is absolutely right, no one is absolutely wrong, its just individual understanding and perspective on the the technical aspects. So maybe for awhile, we just leave it as it is.

At the end, fps matters (personally to me, gap between min fps and avg fps is more important that just looking at avg fps only). But how much are you willing to spend to get those fps? And for me, the impact of these new APIs like DX12 or Vulkan on GPU performance (past and the new gen), and the future on gaming development, is more significant to those on a tight budget, than those with thick wallets to burn.

Fictional scenario. Say roughly around a year ago, Guy A and Guy B, both completely clueless about computers, decided to get a gaming computer each. They both had roughly the same budget (been saving for 3 hard years), and ended up buying almost identical PC. Same Intel i5 CPU, mobo, RAM, budget sub RM 500 FHD 60 Hz LCD monitor and what not. Except Guy A got a GTX 970 because he got advise from his Team Green die hard friend, and Guy B got R9 390 because his friend is Team Red troll. Difference in price maybe around RM 100 - 150 only.

So both went on their gaming journey, playing a mix bag of games, but over time turned into an avid First Person Shooter gamer. CS:GO, Call of Duty, Battlefield etc. And at the same time turned into their respective GPU brand diehard fans. All this while, Guy A and Guy B gaming experience is almost similar too (give and take 3-5 fps difference depending on the game). The came Doom 2016 and both guys rejoiced at getting to try a new FPS game to satisfy their shooting urge.

Up till a few days ago, GTX 970 was around 12 fps faster than R9 390 at 1080P (we use the Compterbase.de Vulkan Doom Analysis but not that it matters to these 2 guys because they only have 60 Hz FHD monitor anyway, just that maybe Guy A can brag about his faster card. Also it happens to be that both of them got duit raya bonus, RM 1,000 just before raya, and already contemplating what PC upgrade they can get, but both are really on tight budget for any additional fund required.

Then came de Vulkan update. The GTX 970 remained around 93 fps@1080P and around 61 fps@1440P. The R9 390 jumped from 81 fps to 119 fps@1080P, and from 56 fps to a comfortable 77 fps@1440P. Now from upgrade option scenario if they really need to spend that RM 1,000 pronto (maybe worry if keep than girlfriend will kikis).

1. They both can sell their FHD monitor say, for RM 250, and get brand new FHD 144 hz monitor for around RM 1,200 (Hz matters more than resolution to FPS gamers, from what I was told). But Guy B he will have a better Doom gaming experience because his fps is closer to the 144 Hz speed.

2. They both can sell their FHD monitor say, for RM 250, add roughly another RM 250 of their money, and get a brand new 1440P monitor for around RM 1,500 (if they want to experience sharper looking graphics than FHD). Again same thing. Guy A GTX 970 may have 61 avg fps, but the min fps could 50 fps and he'll have a not so smooth gaming experience. Whereas Guy B can pretty much enjoy his new monitor smoothly because min fps most probably above the monitor 60 Hz.

3. Guy A can sell his GTX 970, say for about RM 900 (though I've seen a post offering a year old Asus GTX 970 Turbo OC for RM 850, and the Vulkan update results won't help the pricing). And get GTX 1070. But what benefit can he get using his old FHD monitor.

Okay, so this is over simplistic scenario. But my point is, those having previous gen GPUs on a tight upgrade budget will be more impacted by what the new DX12 and Vulkan APIs brings to current and upcoming games. EA already confirmed that upcoming Battlefield 1 will use Async Compute (and I mean DX12 Async Compute which so far only AMD GCN can take advantage of) and have "DX12 fireworks". Guy B will probably benefit from this, but for Guy A, I think most probably not and will lag behind the R9 390. How much difference is yet to be seen, but I expect Battlefield 1, being developed with DX12 already in mind, will most probably have more performance gain than what we've seen with ad-hoc DX12 patch like RotTR.

As for future of Vulkan here's and excerpt from Eurogamer

"However, in terms of raw GPU performance improvement, our numbers show that Vulkan is a big deal for AMD. The turnaround with the R9 Fury X in particular is remarkable - while GTX 1080's sheer brute force in terms of GPU power keeps it comfortably at the top of the pile, the Fury X pulling ahead of both GTX 1070 and 980 Ti is a seriously impressive result for a software-only upgrade.

id Software itself is pretty clear about the advantages of Vulkan and async compute. We asked the team whether they see a time when async compute will be a major factor in all engines across platforms.

"The time is now, really. Doom is already a clear example where async compute, when used properly, can make drastic enhancements to the performance and look of a game," reckons Billy Khan. "Going forward, compute and async compute will be even more extensively used for idTech6. It is almost certain that more developers will take advantage of compute and async compute as they discover how to effectively use it in their games."

If you have RM3K to splash on GPU upgrades, all this won't matter much. Async, no async, DX12, Vulkan, just get that GTX 1080 and you'll be just fine. But there are many, who has that same 3K budget, but for whole buying a whole rig.

I hope moving forward, Nvidia will try its best to get its Pascal GPU to take advantage of DX12 and Vulkan APIs too, instead of just relying on its own version of Async Compute implementation. AMD is 1 year behind Nvidia in releasing its high end Vega GPU, but for me it also means it has 1 year extra to fine tune it. If Vega comes close to Pascal performance minus its Async Compute capability for GPUs within the same price bracket, I worry that Pascal will lag behind as more games take advantage of the new APIs Async Compute as what I've seen recently. If that happens, I hope by that time, more games has DX12 and Vulkan Explicit Multi-Adapter feature built-in.

This post has been edited by adilz: Jul 15 2016, 02:23 AM