Video Card Rumours

[LucShep]

8 hours ago, LucShep said:

Until then, perhaps this....

https://videocardz.com/newz/radeon-monster-profile-volt-freq-optimization-tool-for-rdna2-gpus-is-under-development-for-hydra-software

 

8 hours ago, schmiefel said:

Never heard about that ... is this maybe something similar to the already existing MPT from the German-based Igor's Lab?

https://www.igorslab.de/en/the-new-morepowertool-1-3-8-final-is-now-available-for-download/

No, not that one (I think).
That guy is well known for the 1usmus Custom Power Plan for AMD Ryzen (AMD Zen2 power tweak tool), released in late 2019, when that stuff was really welcome.

He's legit, and may again provide again a tool that, this time - so he says - can provide a pretty substanciable and reliable performance increase (300Mhz+ ?!) in RDNA2 GPUs, and maybe also upcoming RDNA3 as well.
I've heard inumerous times that RDNA2 unfortunately doesn't have much performance tweaking room (OC'ing or otherwise) being hardlocked or power limited, somewhere and somehow. So, that alleged peformance boost would be quite welcome if true, not just for current users but also for potencial buyers of RDNA2 GPUs, now that their prices are decreasing so considerably (both new and used).

Better keep an eye on this "RMP - Radeon Monster Profile" tool once and if it's launched (it's still WIP).

Edited October 13, 2022 by LucShep

[SkateZilla]

5 hours ago, schmiefel said:

With a RiftS there is no need for OpenXR as its supported directly in DCS. In the civil FS I use it already but it doesn't make much difference. Its a well known 'problem' that AMD seems to give a s**t on VR and doesn't care about improving its else good drivers for that, too.

OpenXR is most beneficial to anyone using SteamVR/WindowsMR, as the APIs for those horrible and crash prone.

Oculus users wouldnt need to move to OpenXR, oculus literally has the most options and the most stable api.

[LucShep]

........... LOL

https://www.nvidia.com/en-us/geforce/news/12gb-4080-unlaunch/
 

That's it.
RTX 4080 12GB is cancelled.

The huge (justified) backlash of the "4080 12GB is just a rebranded 4070 to justify its stupid inflated price"  all over the tech communities was just too big.

Now, imagine all the AIBs producing their version of this model, as Nvidia was not producing a FE version of this model. Which means this only hurts the AIB partners, not so much Nvidia themselves. Makes you wonder how much of the EVGA story could have anything to do with it......

Edited October 15, 2022 by LucShep

[BitMaster]

Starting to crawl back, a bit late for my taste.

 

Let this ripe, the last word on "what GPU can I actually afford with the future electricity bills in mind.." hasn't been said yet nor have those astronomically high kW/h rates reached all those who will eventually get them and open their eyes, paired with a new Gas bill that will scale likely. With -5k€/year on the energy side you will have to rethink, if you like it or not.

There is no room for such idiotism in the near future for most users. A Gaming Rig that will consume ~500w total while .listen...GAMING.. will become a different taste for most in central Europe where energy prices skyrocket. The US might be different in that regard but if it hits Europe hard it will hit & affect every market.

Edited October 16, 2022 by BitMaster

[Baldrick33]

3 hours ago, BitMaster said:

There is no room for such idiotism in the near future for most users. A Gaming Rig that will consume ~500w total while .listen...GAMING.. will become a different taste for most in central Europe where energy prices skyrocket. The US might be different in that regard but if it hits Europe hard it will hit & affect every market.

 

You are right in the sense that the energy consumption of pretty much everything we use will be under scrutiny. That said gaming is still a very low cost form of entertainment, you can have an awful lot of hours of 500W usage compared with a pint of beer or going to the cinema etc.

[LucShep]

AMD RDNA 3 “Navi 31” Rumors: Radeon RX 7000 Flagship already with AIBs, 2x faster Rasterization and over 2x Ray Tracing improvement:

https://wccftech.com/amd-rdna-3-radeon-rx-7000-gpu-rumors-2x-raster-over-2x-rt-performance-amazing-tbp-aib-testing/


------------------ \ \ ------------------
 

AMD RDNA 3 “Navi 31” Rumors: Radeon RX 7000 Flagship With AIBs, 2x Faster Raster & Over 2x Ray Tracing Improvement:

The latest rumors come from Greymon55, who suggests that AIBs already have AMD Radeon RX 7000 GPUs based on the RDNA 3 architecture being tested in their labs.
The leaker didn't quote partners but did give us some early performance estimates which sound really good.
We do know that AMD is working on its Navi 31 flagship GPU for launch next month on the 3rd of November, so it is likely that these are the flagship chips that are being tested at the moment to tackle NVIDIA's GeForce RTX 4090 & RTX 4080 graphics cards.


AMD RDNA 3 "Radeon RX 7000" GPU With 2x Raster & Over 2x RT Performance?
As per the leaker, the AMD RDNA 3 GPUs featured on the Radeon RX 7000 series graphics cards are delivering up to a 2x performance increase in pure rasterization workloads and over 2x gains within ray tracing workloads. It is not mentioned if the RDNA 2 GPU is the RX 6900 XT or RX 6950 XT but, even if we look at the 6900 XT, the RDNA 2 chip offered superior performance in raster vs the RTX 3090 and came pretty close to the RTX 3090 Ti, while the RX 6950 XT excelled over it.
A 2x improvement in this department would mean that AMD would easily compete and even surpass the performance of the RTX 4090 in a large section of games.

In ray tracing, a gain over 2x means that AMD might end up close to or slightly faster than the RTX 30 series "Ampere" graphics cards, depending on the title. The Ada Lovelace RTX 40 series cards do offer much faster ray tracing capabilities, offering close to 2x gains in ray tracing performance over the RTX 30 lineup. So ray tracing will see a major improvement but it may not be able to close the gap with RTX 40 series.
There's no word on new AI-assisted capabilities featured on RDNA 3 GPUs to help with upscaling technologies such as FSR.  


Reference TBP of AMD RDNA 3 GPUs Reportedly Looks "Amazing"
Lastly, the leaker states that the reference TBP looks great and we don't know if that's comparing it against the RTX 40 series or the RDNA 2 lineup.
AMD has already said that the RDNA 3 "Radeon RX 7000" GPU lineup will have much lower power figures than the competition.

AMD confirmed that its RDNA 3 GPUs will be coming later this year with a huge performance uplift. The company's Senior Vice President of Engineering, Radeon Technologies Group, David Wang, said that the next-gen GPUs for Radeon RX 7000 series will offer over 50% performance per watt uplift vs the existing RDNA 2 GPUs.
AMD's SVP & Technology Architect, Sam Naffziger, has highlighted that the next-generation RDNA 3 GPUs featured on the Radeon RX 7000 GPUs and next-gen iGPUs, will going to offer a range of new technologies including a refined adaptive power management tech to set workload-specific operation points, making sure that the GPU only utilizes the power required for the workload. The GPUs will also feature a next-gen AMD Infinity Cache which will offer higher-density, lower-power caches and reduced power needs for the graphics memory.

The AMD Radeon RX 7000 "RDNA 3" GPU lineup based on the Nav 3x GPUs is expected to launch later this year, with reports pitting the flagship Navi 31 launch first, followed by Navi 32 and Navi 33 GPUs. A recent rumor also highlighted that the graphics cards will hit retail shelves in December.

 

Edited October 22, 2022 by LucShep

[reece146]

NVIDIA peeps, update your drivers. Apparently the latest driver gives a big uplift on DX12 - but of course that doesn't help us on DCS (DX11).

How much tech debt is Eagle Dynamics dragging behind itself again...?

[SkateZilla]

1 hour ago, reece146 said:

NVIDIA peeps, update your drivers. Apparently the latest driver gives a big uplift on DX12 - but of course that doesn't help us on DCS (DX11).

How much tech debt is Eagle Dynamics dragging behind itself again...?

ED is Moving to Vulkan.

[reece146]

Yes.

Migration to DX12 should have happened 5 years ago.

[tomcat_driver]

4 hours ago, reece146 said:

Yes.

Migration to DX12 should have happened 5 years ago.

Migration to Vulkan should have happened 5 years ago. Not even Microsoft Flight Simulator 2020 had DX12 5 years ago...not even a year after first release and up to very recently, DX11 was still performing better in that game.

[schmiefel]

8 hours ago, LucShep said:

AMD RDNA 3 “Navi 31” Rumors: Radeon RX 7000 Flagship already with AIBs, 2x faster Rasterization and over 2x Ray Tracing improvement:

Hopefully AMD finds some 'magic' within its drivers. too, so that they get on par in VR with NV. The raw perfomance of their GPUs is really great but if they don't can catch up with their drivers that doesn't help much at least when it comes to VR perfomance

[LucShep]

More news/rumours on RDNA3: https://wccftech.com/amd-radeon-rx-7900-xt-rdna-3-graphics-card-to-pack-20-gb-gddr6-memory/
 

-------------- \\ --------------


AMD Radeon RX 7900 XT To Feature RDNA 3 "Navi 31" GPU Core & 20 GB GDDR6 VRAM



The AMD RDNA 3 GPU lineup will kick off first with high-end offerings based on the Navi 31 MCM chip. The chip will be featured in enthusiast-class Radeon RX 7000 graphics cards including the RX 7900 XT which we have received new information about from our sources. As per the details, the AMD Radeon RX 7900 XT graphics card won't be the top model but it will feature 20 GB of GDDR6 VRAM.

Our source reports that AMD's Radeon RX 7000 "RDNA 3" GPUs may exceed expectations & while there are conflicting rumors (here and here), it looks like the end result might be a far better product than anticipated. We have also come to know that the AMD Radeon RX 7900 XT originally featured 24 GB of memory capacity before being downgraded to 20 GB. The Full-Fat 24 GB & Top Navi 31 bin will be aimed at NVIDIA's full-Fat Ada die (the RTX 4090 Ti). AMD seems very confident that with 20 GB and a slightly cut-down MCM chip, they will sit in a comfortable position against the RTX 4090 and may even outperform it in pure rasterization performance while bringing a big jump in RT performance versus the existing RDNA 2 GPUs.

With the Radeon RX 7900 XT utilizing 20 GB of memory, it would indicate that the rumors regarding a higher-end SKU seem to be true. AMD has already started using the *950 XT convention in its last RDNA 2 generation of GPUs so an RX 7950 XT is surely happening. Angstronomics also reported on similar details a while back in their juicy article over here which you should definitely check out.
 

The Navi 31 GPU will also carry 6 MCD's which will feature 16 MB Infinity Cache per die and are also likely to carry the 64-bit (32-bit x 2) memory controllers that will provide the chip with a 384-bit bus interface. While this equals 96 MB of Infinity Cache which is lower than the 128 MB featured on the current Navi 21 GPUs, there's also a 3D-Stacked solution in the works which was pointed out recently and that would double the Infinity Cache with 32 MB (16 MB 0-hi + 16 MB 1-hi) capacities for a total of 192 MB of cache. This is a 50% increase versus the current Navi 21 design and it also makes Navi 31 the first GPU with both, chiplet and 3D stacked designs. These chiplets or MCD's will be fabricated on TSMC's 6nm process node and measure 37.5mm2 each.

Now, this is going to result in a higher power draw and AMD seems to have confirmed this much that their next-generation graphics card lineup will feature higher power consumption, but they will still be a more efficient option than what NVIDIA has to offer.
The AMD Radeon RX 6950 XT already has a TBP of 335W so for a >2x performance gain. The cards are expected to retain their dual 8-pin plug input for power and feature an updated triple-fan cooling design, which is slightly taller than the one currently in use.

AMD will be unveiling its RDNA 3 GPU architecture and the Radeon RX 7000 graphics cards on the 3rd of November.
 They have a full Livestream planned which you can read more details about here.

Edited October 22, 2022 by LucShep

[Thinder]

Looking at the specs of the new gen GPU, I realised that none of the top cards fitted in my box, this plus the power consumption pushed me to look for an alternative solution, all I want is to upgrade from my GTX 1080Ti and being able to do some 4K VR without having to break the piggy bank with a case and PSU upgradse on top.

I did find one which more or less fits the bill. Recommended PSU 750 W. Card Dimension (mm) 267 x 120 x 50mm.

[Doncho]

11 hours ago, Thinder said:

Looking at the specs of the new gen GPU, I realised that none of the top cards fitted in my box, this plus the power consumption pushed me to look for an alternative solution, all I want is to upgrade from my GTX 1080Ti and being able to do some 4K VR without having to break the piggy bank with a case and PSU upgradse on top.

I did find one which more or less fits the bill. Recommended PSU 750 W. Card Dimension (mm) 267 x 120 x 50mm.

Its not the hardware that needs to fit in the case, but the case that must fit to the hardware. Don't be limited by the case, its the cheapest of all components.

[Thinder]

8 hours ago, Doncho said:

Its not the hardware that needs to fit in the case, but the case that must fit to the hardware. Don't be limited by the case, its the cheapest of all components.

The case isn't the only one of the upgrades requiered, plus I clearly recall NVIDA saying less power for increased performance, from the Radeon RX 6800 XT 16G to the Radeon RX 6900 XT 16G, it's clearly not the case, just an example.

850 W  Recommended PSU is above MY requiered specs, another alternative fitting them would be the NVIDIA GeForce RTX 3080 Ti.

I don't shop around trying to fit my budget to manufacturer specs, I shop around trying to find a fit to my budget AND requiered specs, I'll spend the extra 250£ in a VR headset.

The new generation GPU are designed like monsters, I'm not going to spend an extra 250£ just because manufacturers can't get the advertized performancs increase on a card of the same size and with the same power consumption, that's not what I call progress, those are trade-off which will cost you more for the GPU only.

As an example, an NVIDIA GeForce RTX™ 3080 SUPRIM 10G is 336 x 140 x 61 mm and requiers 850W. = Double fail.

My actual case, motherboard, PCU, PSU and RAM will stay until I can find a set of viable solution for a playable 4K VR, and in view of the actual trend, it might take time, and then some.

NVIDIA GeForce RTX 3080 Ti

Edited October 22, 2022 by Thinder

[BitMaster]

To be fair it must be said that if you undervolt aka powerlimit the 4090 it consumes less than most high end 3000 series cards while still having more fps.

the elephant in the room is not the card itself. It’s the price.
If tuned right I have to admit it’s a step towards more efficiency. Just out of the box it’s tuned to max performance and not common sense. 
 

 

[VR Flight Guy in PJ Pants]

Even if you can afford, it is the availability. I guess the UK always hits with the short straw in terms of inventory.

[Thinder]

6 hours ago, BitMaster said:

To be fair it must be said that if you undervolt aka powerlimit the 4090 it consumes less than most high end 3000 series cards while still having more fps.

the elephant in the room is not the card itself. It’s the price.
If tuned right I have to admit it’s a step towards more efficiency. Just out of the box it’s tuned to max performance and not common sense. 
 

 

And there is more to it, although I wouldn't risk running it with my 750W PCU. The NVIDIA card looks like it fits in my case.

GeForce RTX 4090

 

Good infos in this video...

 

Edited October 24, 2022 by Thinder

[SkateZilla]

AMD RX7K will not use the ATX 3.0 12 Pin Power Connector

[schmiefel]

6 hours ago, SkateZilla said:

AMD RX7K will not use the ATX 3.0 12 Pin Power Connector

I think that's a wise decision. The design of those 12VHPWR connectors seems to not suit the power demands of the new graphics cards very well and there are plenty of possibilities to make a faulty connection that could lead to a failing and even burned card. If the manufactorers had choosen a different position on the PCBs that would at least prohibit bending the cables too much, but it seems that those tiny card boards with that big cooling solutions on top didn't allow a different design very easy.

[Thinder]

3 hours ago, schmiefel said:

I think that's a wise decision. The design of those 12VHPWR connectors seems to not suit the power demands of the new graphics cards very well and there are plenty of possibilities to make a faulty connection that could lead to a failing and even burned card. If the manufactorers had choosen a different position on the PCBs that would at least prohibit bending the cables too much, but it seems that those tiny card boards with that big cooling solutions on top didn't allow a different design very easy.

It seems to validate what I think of this industry, basically, they give us less technology and more Premiums to fund their R&D, it's been going this way since the days Silicon Graphics persuaded the military and top CGI industries to go COTS and started to sell their products to high street consumers, by memory, around 1995.

I read an article in Flight International of the time, saying precisely that: Computing, CGI in particular was too expensive and R&D too slow, they needed to cut the cost of both, so the reasoning behind going to the high street market was to sell premium in order to finance R&D, which means that the top rech had to be shared with the military (COTS).

3 years later, a 3D house (Saint Agnes Studios in Britain) would have to pay £35.000 per seat on a SG machine, one license per seat, my ex-boss, Alan Marques, decided to go PC-based 3D Studio max loaded networks, where a network of Pentium II with 128Mb of RAM could do the same rendering job with more seats available for modeling, animation, maping, special effects etc, and this under ONE rendring licence at a fraction of the price.

It's a time where manufacturersd started to develop stuff like SLI (Voodoo 2), faster graphic buses, larger capacity and higher frequency RAM and I don't recall the prices going through the roof or parts setting the PC on fire, we were able to bluid our machines ourselves accodring to our requiered specs for CGI.

It could have gone well until competition made them cutting corners and sell us (high street clients) more Premiums and less technologie, nobody is gonna make me believe that those successive generations of cards have been putting the emphasis on tech rather than profit and that technology couldn't have made sure they could provide more performances for less volume and weight today.

Compare NVIDIA very own NV1 of 1995 to this RTX 4090 and ask yourself why with all the technology developed since 1995, a top graphic card inflated like a ballon filled with Mercury... And can burst into fire.

If you want to know what was possible to do with 1997 PCs, watch Lost in Space (1998), it was made at St Agnes Studios and at some point I  had every single app and plug-ins for special effects on my home PC, we helped KINETIX develop 3D Studio Max for the cinema industry and ran Intel, then AMD self- built PCs to cut costs, from where I'm standing, the industry created this inflation, and we're paying for it.

Lost in Space 1998

 

Edited October 26, 2022 by Thinder

[Thinder]

Complement of information: If you wonder what the money we spend in Premium is used for here is a clue.

Simulation Massive en Reseau. Massive Network Simulation.

French Armee de l'Air uses RAZBAM Mirage 2000C module for training at squadron level (of course with corrected datas for performances and system), NATO uses squadron level COTS (Components of the shelf, originaly Component of the shelf in military language), this is not uncommun among NATO squadrons and always have been the idea behind COTS.

Those platforms are COTS PCs but are developed so that a French pilot can train vs an American pilot in the USA or any other pilot from a member of NATO in a netwrok, despite this level of sophistication, this represents the low cost level of Air Forces training, the high end being the full cockpit simulators.

I'm certain that VR will complement the actual 2D systems because the module and simulation already have the capability, it is only a matter of budget and when it will be aproved.

So to get to this point, R&D cost have been shared between military budgets (and I can tell you that it is nowhere near what the military were paying previously) and the high end cunstomers paying premiums on top end cards.

Edited October 26, 2022 by Thinder

[cordite]

Are the AMD drivers really that bad for VR? I've read that the 6xxx gen has shown improvements.

 

I've been using Nvidia cards for some time now, and I know that some of NV's driver releases have been stinkers, for 2d or VR.

Edited October 26, 2022 by cordite
6th gen amd, not 5th gen

[schmiefel]

8 hours ago, cordite said:

Are the AMD drivers really that bad for VR? I've read that the 6xxx gen has shown improvements.

'Really that bad' is not the case, but they don't get the same perfomance out of the cards into the VR device that they show in 2D, where a 6900XT easily is on par with a 3080ti or 3090 in most use cases up to 4k resolutions and the selected 6900XTXH or 6950XT even can reach the performance of a 3090ti.

[Thinder]

11 hours ago, cordite said:

Are the AMD drivers really that bad for VR? I've read that the 6xxx gen has shown improvements.

 

I've been using Nvidia cards for some time now, and I know that some of NV's driver releases have been stinkers, for 2d or VR.